BISar 2018 e

BIS
Annual Economic
Report
June 2018
© Bank for International Settlements 2018. All rights reserved.
Limited extracts may be reproduced or translated provided the source is stated.
www.bis.org
[email protected]
Follow us
BIS
Annual Economic
Report
June 2018
Promoting global monetary

and financial stability
This publication is available on the BIS website (www.bis.org/publ/arpdf/ar2018e.htm).
Also published in French, German, Italian and Spanish.
© Bank for International Settlements 2018. All rights reserved. Limited extracts
may be reproduced or translated provided the source is stated.
ISSN 2616-9428 (print)

ISSN 2616-9436 (online)
ISBN 978-92-9259-175-5 (print)
ISBN 978-92-9259-174-8 (online)
Contents
Annual Economic Report 2018: Editorial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
The global economy: developments, prospects and risks . . . . . . . . . . . . . . . . . . . . . . . . . . ix

Special chapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Macroprudential frameworks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Cryptocurrencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
I. A stronger expansion: how to make it last . . . . . . . . . . . . . . . . . . . . . . . . 1
The global expansion strengthens amid low inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Financial conditions remained very easy for most of the year . . . . . . . . . . . . . . . . . . . . . . 4
The near-term outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Box I.A: Evidence of unusual late-business-cycle momentum – a historical perspective 10
The risks ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
How have financial cycles played out so far? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Box I.B: The changing nature of the business cycle and its link to the financial cycle . . 14
Snapback risk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Reversal of risk appetite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Risks from the further build-up of financial imbalances and debt . . . . . . . . . . . . . . 20
II. Monetary policy: a narrow normalisation path . . . . . . . . . . . . . . . . . . 25
Monetary policy normalisation: where do we stand? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

Monetary and financial conditions: imperfect transmission? . . . . . . . . . . . . . . . . . . . . . . . 28
Policy challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Box II.A: The decline in natural real interest rates: what do we know? . . . . . . . . . . . . . . . 38
III. The financial sector: post-crisis adjustment and pressure

points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Basel III: key elements of the completed framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

Stage 1 reforms: raising bank capital and liquidity buffers . . . . . . . . . . . . . . . . . . . . 44
Stage 2 reforms: completing the framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Banks’ adjustment: the long path to sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Implementation and bank business models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Bank resilience: moving ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Box III.A: Banks’ window-dressing: the case of repo markets . . . . . . . . . . . . . . . . . . . . . . . 49
Box III.B: The widening of Libor-OIS spreads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Box III.C: Factors driving bank equity valuations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Bank/non-bank interactions: new pressure points? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Institutional asset managers: bigger footprint, changing dynamics . . . . . . . . . . . . . 54
Snapback and interlinkages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Box III.D: When yields “snap back” – funds’ exposures and amplification effects . . . . . . 56
Implications for prudential policy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
BIS Annual Economic Report 2018 iii

IV. Moving forward with macroprudential frameworks . . . . . . . . . . . . 63
Macroprudential frameworks: elements and challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

When and how to act? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Identifying risks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Selecting and calibrating instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Box IV.A: Macroprudential approaches to capital market activities . . . . . . . . . . . . . . . . . . 69
Communication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Governance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Impact: the experience so far . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Towards an integrated macro-financial stability framework . . . . . . . . . . . . . . . . . . . . . . . . 74
Box IV.B: FX interventions in a macroprudential context . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
V. Cryptocurrencies: looking beyond the hype . . . . . . . . . . . . . . . . . . . . . . 91
Putting the rise of cryptocurrencies into perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

A brief history of money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
The current monetary and payment system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Cryptocurrencies: the elusive promise of decentralised trust . . . . . . . . . . . . . . . . . . . . . . . 94
A new petal in the money flower? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Distributed ledger technology in cryptocurrencies . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Assessing the economic limitations of permissionless cryptocurrencies . . . . . . . . . 98
Box V.A: Forking and the instability of decentralised consensus in the blockchain . . . . 103
Beyond the bubble: making use of distributed ledger technology . . . . . . . . . . . . . 104
Policy implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Regulatory challenges posed by cryptocurrencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Box V.B: Distributed ledger technology in trade finance . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Should central banks issue digital currencies? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Box V.C: Wholesale central bank digital currencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
iv BIS Annual Economic Report 2018

Graphs
E.1 The global economy strengthens further as inflation converges . . . . . . . . . . . . . . . x

E.2 Global debt continues to rise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
E.3 Narrowing room for policy manoeuvre and waning structural reforms . . . . . . . . . xiv
I.1 GDP growth improves, investment strengthens and unemployment declines . . . 2

I.2 Inflation remains low and wage growth subdued . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
I.3 Global financial conditions remain very easy in 2017 and early 2018 . . . . . . . . . . 5
I.4 A stronger dollar puts vulnerable EMEs under pressure . . . . . . . . . . . . . . . . . . . . . . 6
I.5 Stock market valuations remain stretched . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
I.6 The near-term outlook for growth and inflation is positive for most countries . . 8
I.7 The financial cycle supports growth in advanced economies . . . . . . . . . . . . . . . . . . 16
I.8 Areas of vulnerability: commercial property, household debt and EME dollar
debt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
I.9 Vulnerability of debt service ratios to rising rates varies by country . . . . . . . . . . . 19
II.1 An asynchronous policy normalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

II.2 Low policy rates and large central bank balance sheets worldwide . . . . . . . . . . . . 27
II.3 A tightening paradox? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
II.4 Factors offsetting monetary tightening . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
II.5 Central bank asset purchases weigh on long-term interest rates . . . . . . . . . . . . . . 31
II.6 The current tightening has been highly gradual and predictable . . . . . . . . . . . . . . 32
II.7 The natural interest rate is measured with considerable uncertainty . . . . . . . . . . . 34
II.8 Higher debt raises vulnerabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
II.9 Global spillovers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
III.1 Unsustainable risk-taking required decisive post-crisis adjustments by banks . . . 44

III.2 Basel III floors – addressing unwarranted RWA variability . . . . . . . . . . . . . . . . . . . . . 46
III.3 Implementation of new requirements and banks’ adjustments are progressing . . 47
III.4 Regulatory vs market-based metrics – resilience is up, but by how much? . . . . . 50
III.5 Bank credit risk and return-on-equity – further room for improvement . . . . . . . . 52
III.6 Exposure to valuation losses and concentration in asset management are rising . 55
III.7 Hidden liquidity risks not covered by additional buffers? . . . . . . . . . . . . . . . . . . . . . 58
IV.1 Macroprudential orientation moves to the mainstream . . . . . . . . . . . . . . . . . . . . . . 64

IV.2 Tightening actions are used more frequently as credit booms build up . . . . . . . . 66
IV.3 Macroprudential institutions and powers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
IV.4 Direct effects of macroprudential measures on bank credit and house prices . . . 73
IV.5 Effects of targeted macroprudential tightening measures on other types of
credit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
V.1 The money flower: a taxonomy of money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

V.2 Centralised ledger and permissioned/permissionless decentralised ledgers . . . . . 96
V.3 Valid transactions in a centralised ledger/bank account and in a permissionless
cryptocurrency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
V.4 Energy consumption and scaling issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
V.5 Transaction fees over time and in relation to transaction throughput . . . . . . . . . . 100
V.6 Volatility of select cryptocurrencies and number of cryptocurrencies . . . . . . . . . . 101
V.7 Mining concentration and bitcoin value during a temporary fork . . . . . . . . . . . . . 102
V.8 Indicators of the volume and cost of remittances . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
V.9 The shutdown of an illegal marketplace and the legitimacy of ICOs . . . . . . . . . . . 107
BIS Annual Economic Report 2018 v

Tables
III.1 Basel III phase-in arrangements: key standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
IV.1 Use of macroprudential measures by targeted credit, instrument type and region 67
IV.A1 Use of macroprudential measures by targeted credit, instrument type and region 88
This Report went to press on 14–15 June 2018 using data available up to 25 May 2018.
Conventions used in the Annual Economic Report
$ US dollar unless specified otherwise

mn million
bn billion (thousand million)
trn trillion (thousand billion)
% pts percentage points
bp basis points
lhs, rhs left-hand scale, right-hand scale
sa seasonally adjusted
yoy year on year
qoq quarter on quarter
... not available
. not applicable
– nil or negligible
Components may not sum to totals because of rounding.
The term “country” as used in this publication also covers territorial entities that are
not states as understood by international law and practice but for which data are
separately and independently maintained.
vi BIS Annual Economic Report 2018

Country codes
AE United Arab Emirates FR France NL Netherlands
AO Angola GB United Kingdom NO Norway
AR Argentina GR Greece NZ New Zealand
AT Austria HK Hong Kong SAR PA Panama
AU Australia HR Croatia PE Peru
BA Bosnia and HU Hungary PH Philippines
Herzegovina ID Indonesia PK Pakistan
BE Belgium IE Ireland PL Poland
BG Bulgaria IL Israel PT Portugal
BR Brazil IN India QA Qatar
CA Canada IS Iceland RO Romania
CH Switzerland IT Italy RU Russia
CL Chile JP Japan RS Republic of Serbia
CN China KR Korea SA Saudi Arabia
CO Colombia KW Kuwait SE Sweden
CY Republic of Cyprus KZ Kazakhstan SG Singapore
CZ Czech Republic LT Lithuania SI Slovenia
DE Germany LU Luxembourg SK Slovakia
DK Denmark LV Latvia TH Thailand
DZ Algeria LY Libya TR Turkey
EA euro area MK Macedonia, FYR TW Chinese Taipei
EE Estonia MT Malta US United States
ES Spain MX Mexico VE Venezuela
EU European Union MY Malaysia VN Vietnam
FI Finland NG Nigeria ZA South Africa
Currency codes
AUD Australian dollar EUR euro JPY Japanese yen
CHF Swiss franc GBP pound sterling USD US dollar
Advanced economies (AEs): Australia, Canada, Denmark, the euro area, Japan, New
Zealand, Norway, Sweden, Switzerland, the United Kingdom and the United States.
Major AEs (G3): The euro area, Japan and the United States.
Other AEs: Australia, Canada, Denmark, New Zealand, Norway, Sweden, Switzerland
and the United Kingdom.
Emerging market economies (EMEs): Argentina, Brazil, Chile, China, Chinese Taipei,
Colombia, the Czech Republic, Hong Kong SAR, Hungary, India, Indonesia, Korea,
Malaysia, Mexico, Peru, the Philippines, Poland, Russia, Saudi Arabia, Singapore,
South Africa, Thailand and Turkey.
Global: All AEs and EMEs, as listed.
Commodity exporters (countries whose average share of commodities in export

revenues in 2005–14 exceeded 40%): Argentina, Australia, Brazil, Canada, Chile,
Colombia, Indonesia, New Zealand, Norway, Peru, Russia, Saudi Arabia and South
Africa.
Country aggregates used in graphs and tables may not cover all the countries
listed, depending on data availability.
BIS Annual Economic Report 2018 vii

Annual Economic Report 2018: Editorial
Starting this year, the traditional BIS Annual Report – comprising an analysis of the
global economy and a presentation of the BIS’s activities – will be split into two
separate publications. In addition, the economic part, renamed Annual Economic
Report, has been restructured. The first three chapters review global developments,
prospects and risks, while two special chapters focus on topical issues, with analyses
of macroprudential frameworks and cryptocurrencies. This editorial summarises the
content and key messages.
The global economy: developments, prospects and risks
It is now 10 years since the Great Financial Crisis (GFC) engulfed the world. At the
time, following an unparalleled build-up of leverage among households and
financial institutions, the world’s financial system was on the brink of collapse.
Thanks to central banks’ concerted efforts and their accommodative stance, a
repeat of the Great Depression was avoided. Since then, historically low, even
negative, interest rates and unprecedentedly large central bank balance sheets
have provided important support for the global economy and have contributed to
the gradual convergence of inflation towards objectives. Still, central banks were
largely left to bear the burden of the recovery, with other policies, not least supply
side structural ones, failing to take the baton. These actions by central banks helped
lay the groundwork for the resumption of growth that we now see. But, in the
process, they have been one factor behind the legacy of swollen private and public
sector balance sheets and higher debts that shapes the road ahead. As the global
economy reaches or even exceeds potential, it is time to take advantage of the
favourable conditions to put in place a more balanced policy mix to promote a
sustainable expansion. However, the path ahead is a narrow one.
The dividends of past policies were clearly in evidence in the year under review
– a vintage one for the global economy (Graph E.1 and Chapter I). The expansion
strengthened and broadened. Global growth rates were roughly on a par with pre-
crisis long-term averages, and the expansion was highly synchronised across
countries (Graph E.1, left-hand panel). Unemployment continued to decline,
reaching multi-decade lows in a number of economies, including some of the
largest (centre panel). Overall, headline inflation rates moved closer to central bank
objectives, although core inflation remained more subdued. In fact, the year capped
a steady improvement in the global economy that has been evident for some time.
As already noted two Annual Reports ago, the picture then was considerably better
than the gloomy rhetoric indicated. And in last year’s Report we highlighted how
talk of secular stagnation had given way to renewed optimism and a revival of
animal spirits.
For the next couple of years, consensus forecasts see the trend continuing,
marking one of the longest postwar expansions (Graph E.1 and Chapter I). Despite
the softer patch in the first quarter of 2018 and some jitters in emerging market
economies (EMEs) (see below), the forecasters’ central scenario is still for global
growth to exceed potential, reducing unemployment further, with economies
testing capacity limits. Investment is expected to strengthen, boosting productivity
over time. And fiscal expansion should provide additional near-term stimulus: quite
BIS Annual Economic Report 2018 ix

The global economy strengthens further as inflation converges Graph E.1
GDP growth1, 2 Unemployment rates 2, 3 Inflation1, 4

yoy changes, per cent; sa Per cent; sa Per cent
9 9 8
6 8 6
3 7 4
0 6 2
–3 5 0
–6 4 –2
83 88 93 98 03 08 13 18 83 88 93 98 03 08 13 18 09 12 15 18
Global AEs EMEs
In the left-hand panel, the dots indicate Consensus Economics forecasts for 2018; the dashed lines indicate 1982–2007 averages.
1
Weighted averages based on GDP and PPP exchange rates. 2 For the euro area, weighted average across individual countries before
1995. 3 Weighted averages based on labour force levels; definitions may vary across countries. 4 Consumer prices.
Sources: IMF, International Financial Statistics and World Economic Outlook; OECD, Main Economic Indicators; CEIC; Consensus Economics;
Datastream; Global Financial Data; national data; BIS calculations.
apart from the measures in the United States, the OECD foresees an easier fiscal
stance in around three quarters of its members this year and next. At the same
time, inflation is forecast to edge up.
The current scenario is somewhat unusual in the postwar period (Chapter I). It
is not common to anticipate such strong growth so late in the expansion, when
capacity constraints start biting, with only modest signs of an inflation threat. The
reasons for this picture are much debated. There may be more slack than meets the
eye: the crisis may have left a legacy of discouraged workers ready to re-enter the
labour force as conditions improve; the investment pickup may be erasing some of
the crisis scars, raising potential; and longer-term demographic factors and pension
reforms may also be at work, as indicated by the widespread increase in participation
rates among the older segments of the population, with the United States a notable
exception. Moreover, as emphasised in previous Annual Reports, the persistent
influence of globalisation and technological advances on inflation should not be
underestimated, not least through their impact on workers’ and firms’ pricing power.
Buoying the expansion, and partly as a result of the heavy reliance on monetary
policy to support the post-crisis recovery, financial conditions once again played
a key role in the year under review (Chapters I and II). At least until recently,
global financial conditions remained very easy. In fact, they loosened further even
as US monetary policy proceeded along its very gradual and well anticipated
normalisation path. True, long-term US Treasury yields moved up. But term premia
remained historically low and equity price valuations quite rich, except when
assessed in relation to the prevailing low interest rates. Importantly, credit spreads
have been unusually compressed, often at or even below pre-GFC levels, and the
corresponding markets appear to have become increasingly illiquid. Moreover, for
most of the year under review the US dollar depreciated, supporting buoyant
financial conditions especially in EMEs, which post-crisis have borrowed heavily in
that currency and during the past year saw strong portfolio inflows. These buoyant
conditions in EMEs, however, reversed more recently (see below).
BIS Annual Economic Report 1

x BIS Annual Economic Report 2018
In the Report, we also assess the risks ahead in some detail (Chapter I). The
conclusion is that medium-term risks are material, although there are cross-country
differences. In some respects, the risks mirror the unbalanced post-crisis recovery and
its excessive reliance on monetary policy. Where financial vulnerabilities exist, they
have been building up, in their usual gradual and persistent way. More generally,
financial markets are overstretched, as noted above, and we have seen a continuous
rise in the global stock of debt, private plus public, in relation to GDP (Graph E.2).
This has extended a trend that goes back to well before the crisis and that has
coincided with a long-term decline in interest rates (Chapter II).
In some countries largely spared by the GFC, for quite some time there have
been signs of a build-up of financial imbalances. This is because, in contrast to
countries at the heart of the turmoil, no private sector deleveraging has taken
place, so that the financial expansion has continued. The signs of imbalances have
taken the form of strong increases in private sector credit, often alongside similar
increases in property prices – the tell-tale sign of the expansion phase of domestic
financial cycles, qualitatively similar to those observed pre-crisis in the economies
that subsequently ran into trouble. Fortunately, much has been done to strengthen
the financial system’s resilience. The post-crisis financial reforms, not least Basel III
and the implementation of macroprudential frameworks, have substantially bolstered
the banking system (Chapters III and IV). And in China, the largest economy where
the signs of imbalances are evident, the authorities have taken steps to rebalance
the expansion and rein in some of the more serious financial excesses.
Against this backdrop, a number of developments could lead to the
materialisation of risks, threatening the economic expansion in the medium term
(Chapter I). In all of them, financial factors seem destined to play a prominent role,
either as a trigger or as an amplifying mechanism. Indeed, the role of financial
forces in business fluctuations has grown substantially since the early 1980s, when
financial liberalisation took hold. And post-crisis the weight of non-bank
intermediaries, such as asset managers and institutional investors, has risen
substantially, and is likely to influence the dynamics of any future episodes of
financial stress, in familiar but also some unexpected ways (Chapter III).
Global debt continues to rise1 Graph E.2
USD trn % of GDP
160 300
266 269
135 259 250
233
217
110 196 204 200
179 176
85 150 150
133
113
100
50
0
AEs EMEs All AEs EMEs All AEs EMEs All AEs EMEs All
end-2007 end-2010 end-2013 end-2017
Global total (lhs) Rhs: General government Non-financial corporates Households

1
Weighted averages based on GDP and PPP exchange rates. The sample includes all reporting economies except Ireland, owing to large
data fluctuations caused by the restructuring of large multinational corporations.
Sources: IMF, World Economic Outlook; BIS total credit statistics; BIS calculations.
BIS Annual Economic Report 2018 xi

One possible trigger of an economic slowdown or downturn could be an
escalation of protectionist measures. Its impact could be very significant, if such
escalation was seen as threatening the open multilateral trading system. Indeed,
there are signs that the rise in uncertainty associated with the first protectionist
steps and the ratcheting-up of rhetoric have already been inhibiting investment.
Moreover, were the recent reversal in the US dollar depreciation to continue, trade
negotiations would become more complicated.
A second possible trigger could be a sudden decompression of historically low
bond yields or snapback in core sovereign market yields, notably in the United
States. This could take place in response to an inflation surprise and the perception
that central banks will have to tighten more than anticipated. In the United States,
the prospective heavy issuance of government debt, combined with the gradual
unwinding of central bank purchases, could add to this risk. Importantly, the
surprise need not be a large one, as indicated by the financial markets’ wobble in
February in response to slightly stronger than anticipated US wage growth. And the
impact would spread globally, given the weight of the US economy and the
dominant role of the dollar in global financial markets.
A third trigger could be a more general reversal in risk appetite. Such a reversal
could reflect a range of factors, including disappointing profits, the drag of the
contraction phase of financial cycles where these have turned, a souring of
sentiment vis-à-vis EMEs, or untoward political events threatening stability in some
large economies. From this perspective, recent events in the euro area are a source
of concern, as reflected in the widening of spreads on Italian and Spanish bonds. In
contrast to the snapback scenario, this third trigger would usher in a further
compression of term premia in those core sovereign markets that benefited from a
flight to safety.
Indeed, in April signs of strain did emerge in the most vulnerable EMEs,
beginning in Argentina and Turkey, as the US dollar began to appreciate and
financial conditions in international markets started to tighten. At the time of
writing, it is too early to tell whether the strains will remain contained or will spread
further. Most EMEs are better placed to confront financial stress now than they
were in the mid-1990s. They have taken steps to strengthen their defences, by
building reserves, adopting more systematic macroprudential measures (Chapter IV),
improving their current account positions and adopting more flexible exchange
rate regimes. This should provide them with some more room for manoeuvre were
global financial conditions to tighten further.
Nevertheless, some pitfalls remain. The shift in the pattern of financial
intermediation towards greater borrowing through the bond market has reduced
rollover risk but introduced greater duration risk. Portfolio investors with limited
tolerance for losses may amplify price fluctuations should they attempt to reduce
exposures simultaneously. More generally, non-banks have been the largest
borrowers; if they found themselves under financial strain, they might curtail
operations and employment. A slowdown in the real economy may be the risk to
watch for if EMEs continue to experience tightening financial conditions.
Looking further ahead, if the global economy successfully navigates the
choppy waters just described, the expansion could continue. But then, almost
inevitably, supported by easy financial conditions, financial imbalances and, above
all, the aggregate debt-to-GDP ratio could rise further. Financial market complacency,
low volatility and excessive risk-taking would continue. Limited market discipline
would induce further poor resource allocation, including through the survival of
ultimately unprofitable firms and weaker incentives for sovereigns to ensure fiscal
space. All this would make the subsequent adjustment more painful. Such a further
rise in global debt would be especially worrying (Chapter I). Not only would it make
xii BIS Annual Economic Report 2018

it harder to raise interest rates to more normal levels without threatening the
expansion, given the associated rise in debt service burdens – a kind of “debt trap”.
It would also narrow the room for manoeuvre to address any downturn, which will
come sooner or later.
What can policy do to ensure the current expansion is more sustainable and
balanced? The question is especially pressing, given that compared with pre-crisis
the room for policy manoeuvre has indeed narrowed considerably (Graph E.3).
Public sector debt in relation to GDP has increased further, constraining fiscal policy
(left-hand panel). At the same time, interest rates are significantly lower, constraining
monetary policy – a phenomenon exacerbated in advanced economies by the
much larger central bank balance sheets (centre panel). Building room is a priority.
There are several possible lines of action that, if combined, would support each
other (Chapter I). Their common theme is to focus firmly on longer horizons, since
both monetary and fiscal expansions work to a considerable extent by borrowing
demand from the future. And when the future becomes today, there is inevitably a
price to be paid. This puts a premium on taking advantage of the current highly
favourable conditions to redress the balance. Such policy adjustments would be
consistent with the implementation of a broader macro-financial stability
framework, in which the various policies would work in tandem to ensure
macroeconomic and financial stability while raising long-term sustainable growth
(Chapter IV).
The first line of action is to redouble efforts to implement structural policies –
the only way to raise sustainable growth without generating inflationary pressures.
This is essential, as economies are already operating at or beyond standard
estimates of full employment and potential output even though post-crisis growth
has been, on balance, rather disappointing – an indication of supply constraints. In
particular, structural policies can alleviate the dilemmas monetary policy is currently
facing and that are narrowing its room for manoeuvre. The essence of the reforms
is to make product and labour markets more flexible, enabling them to allocate
resources more efficiently and to absorb technical innovations more easily. One
important element here is also to safeguard the open multilateral trading order
that has served the global economy so well during the past decades. Unfortunately,
the post-crisis record in structural reforms has fallen far short of what is desirable:
since 2011 the pace has actually slowed down (Graph E.3, right-hand panel).
Moreover, recent protectionist rhetoric and actions do not augur well.
The second line of action is to strengthen further the resilience of the financial
system (Chapters III and IV). This requires completing and consistently implementing
the post-crisis financial regulatory reforms. Ideally, where appropriate, this should
be supported by steps to remove structural impediments to banks’ efforts to attain
sustainable profitability, which is critical to absorb any losses smoothly and swiftly
should these materialise at some point. Examples of such steps include tackling the
obstacles to the necessary consolidation and cost cutting. Sustainable profitability
is especially important at the current juncture: banks have been facing the dual
challenge of persistently and unusually low interest rates eating away at their net
interest margins, and growing competition from new technology-savvy players –
big tech and fintech. Strengthening resilience also calls for the active deployment
of macroprudential measures in those economies where financial imbalances have
been building up and the improvement of macroprudential frameworks more
generally. In both cases, the non-bank sector, notably asset managers and
institutional investors, deserves closer attention, to complete unfinished business
there.
The third line of action is to ensure the sustainability of public sector finances
and to avoid procyclical fiscal expansions. The importance of this issue cannot be
BIS Annual Economic Report 2018 xiii

Narrowing room for policy manoeuvre and waning structural reforms Graph E.3
Government debt1 Interest rates and central bank Structural reforms6

balance sheets
% of GDP Per cent USD trn Per cent
100 12 12 36
75 6 6 24
50 0 0 12
25 –6 0
02 07 12 17 03 08 13 18 05 07 08 09 10 11 13 15 177
–06 –08 –09 –10 –11 –12 –14 –16
Global AEs EMEs AEs: EMEs:2
Lhs: Nominal interest rates
3, 4
AEs EMEs8
Real interest rates4, 5
Rhs: Total central bank assets
1
General (if not available, central) government core debt at market (if not available, nominal) value. Weighted averages based on GDP and
PPP exchange rates. Backdated using central government debt (% of GDP) or total government debt securities (% of GDP) based on data
availability. 2 A few outliers for which the nominal interest rate exceeded 60% are omitted from the sample. 3 Policy rate or closest
equivalent. 4 Simple monthly averages across economies. 5 Nominal interest rate less consumer price inflation. 6 OECD reform
responsiveness indicator (RRI), measured based on a scoring system in which policy recommendations set in the context of Going for Growth
take a value of one if significant action is taken and zero if not. Simple averages across economies. The sample sizes for country groups vary
across time. 7 Sum of the RRIs for the reforms that are fully implemented or being implemented. 8 Except HK, MY, PE, PH, SA, SG, TH and
TW.
Sources: IMF, International Financial Statistics and World Economic Outlook; OECD, Going for Growth; Datastream; Global Financial Data; Oxford
Economics; national data; BIS policy rate statistics and total credit statistics; BIS calculations.
emphasised enough. Public debt has risen to new peacetime highs in both advanced
and emerging market economies. And, as history indicates, fiscal space is likely to
be overestimated in countries where financial imbalances have been building up.
With due regard for country-specific circumstances, fiscal consolidation is a priority.
The final line of action concerns monetary policy. Monetary policy normalisation
is essential in rebuilding policy space. It can create room for countercyclical policy,
help reduce the risk of the emergence of financial vulnerabilities, and contribute to
restraining debt accumulation. That said, as discussed in detail in Chapter II, given
the unprecedented starting point, the uncertainties involved and persistently
low inflation in many jurisdictions, the path ahead is quite narrow, with pitfalls on
either side. It requires striking and maintaining a delicate balance between
competing considerations, notably achieving inflation objectives in the short run
and avoiding the risk of encouraging the further build-up of financial vulnerabilities
in the longer run.
While the right approach will naturally depend on country-specific conditions,
a couple of general observations are possible. One is that treading the path will call
for flexibility in the pursuit of inflation objectives. This applies in particular to
moderate inflation shortfalls, given the benign structural disinflationary pressures
still at work. The other is that policymakers will need to maintain a steady hand,
avoiding the risk of overreacting to transitory bouts of volatility. After all, given
initial conditions, the journey is bound to be bumpy. Financial market ructions will
no doubt occur. Higher volatility per se is not a problem as long as it remains
contained; it is actually healthy whenever it helps inhibit unbridled risk-taking.
BIS Annual Economic Report 3

xiv BIS Annual Economic Report 2018
Special chapters
Macroprudential frameworks
The implementation of macroprudential frameworks represents a major and

welcome element of the post-crisis financial reforms. It has stemmed from the
recognition that the pre-crisis approach to regulation and supervision was
insufficient, as it was largely focused on an assessment of the risks incurred by
individual institutions on a standalone basis. Such a (micro-oriented) approach can
miss and fail to manage vulnerabilities at the system-wide level – the main source
of systemic crises with serious macroeconomic costs. A prominent source of such
crises is the procyclicality of the financial system, ie its tendency to amplify financial
expansions and contractions, which can in turn amplify business fluctuations. The
financial cycle is a reflection of such forces.
The activation of macroprudential measures is especially important at the
current juncture. It can help contain the financial vulnerabilities bequeathed by the
unbalanced post-crisis recovery and mitigate the build-up of further risks. And, in
the process, it can support monetary policy along its normalisation path, increasing
the room for manoeuvre. Current favourable economic conditions provide a
window of opportunity for active deployment that should not be missed.
Against this backdrop, Chapter IV takes stock of the experience so far with
macroprudential frameworks and explores the way forward. It reaches a number of
conclusions. First, while implementation issues are challenging, the authorities have
made substantial progress. Issues include identifying the build-up of systemic risks
in good time to take remedial action, choosing appropriate instruments, political
economy constraints on their deployment, and establishing effective governance
arrangements. Second, so far the tools at the authorities’ disposal have largely
targeted banks; there is a need to extend them to other financial institutions, not
least to the asset management sector. Third, macroprudential measures have
succeeded in strengthening the financial system’s resilience, but as deployed so far
their restraining impact on financial booms has not always prevented the
emergence of the familiar signs of financial imbalances. Fourth, this suggests that
macroprudential measures are most effective as part of a more holistic macro-
financial stability framework, also involving structural, fiscal and monetary policies.
Finally, there is scope to further strengthen international cooperation in this area.
Cryptocurrencies
Cryptocurrencies promise to replace trust in long-standing institutions, such as

commercial and central banks, with trust in a new, fully decentralised system
founded on the blockchain and related distributed ledger technology (DLT). The
transformative nature of this promise makes this development of core concern to
central banks.
Chapter V evaluates whether cryptocurrencies can deliver on their promise as a
form of money. Looking beyond the hype, it finds that this is not the case. Much
has already been said about the impractical nature of cryptocurrencies as a means
of payment, the scope for fraud and the enormous environmental cost. As the BIS
General Manager put it recently,1 cryptocurrencies have become a “combination of
a bubble, a Ponzi scheme and an environmental disaster”.
The chapter highlights the further economic limitations of cryptocurrencies.
These relate to their limited ability to satisfy the signature property of money as a
coordination device and their questionable promise of trust. Cryptocurrencies
BIS Annual Economic Report 2018 xv

cannot scale with transaction demand, are prone to congestion, and fluctuate
greatly in value. And trust in them can evaporate at any time, owing to the fragility
of the decentralised consensus mechanisms used to record and validate transactions.
Not only does this call into question the finality of individual payments, it also
means that a cryptocurrency can simply stop functioning, resulting in a complete
loss of value.
The decentralised technology of cryptocurrencies, however sophisticated, is a
poor substitute for the solid institutional backing of money through independent
and accountable central banks. DLT itself, however, does have promise in
applications other than cryptocurrencies. Examples include, in particular, simplifying
administrative processes in the settlement of financial transactions. But this still
remains to be tested.
The emergence of cryptocurrencies calls for policy responses. A globally
coordinated approach is necessary to prevent abuses and strictly limit interconnections
with regulated financial institutions. In addition, delicate issues arise regarding the
possible issuance of digital currencies by central banks themselves.
Endnote
1
See A Carstens, “Money in the digital age: what role for central banks?“, lecture at the House of
Finance, Goethe University, Frankfurt, 6 February 2018.
xvi BIS Annual Economic Report 2018

I. A stronger expansion: how to make it last
In the year under review, the global economy outperformed expectations. Growth
strengthened and broadened; inflation remained subdued despite a further drop in
unemployment rates; and, for most of the period, global financial conditions eased
further even as monetary policy inched towards normalisation. Despite some loss
of momentum in early 2018 and a deterioration in market sentiment, especially vis-
à-vis emerging market economies (EMEs), most countries are expected, at the time
of writing, to grow at above-potential rates in 2018 and 2019, and inflation is
expected to pick up only moderately.
From a longer-term perspective, the global economy has been reaping the
dividend from the post-crisis measures taken by monetary and regulatory
authorities. Prolonged very easy monetary policies have underpinned the global
recovery. And banking systems are now better capitalised and more resilient, and
thus better positioned to support the economy (Chapter III).
The key challenge now is to sustain the higher growth beyond the near term.
So far, the recovery has been too dependent on central banks’ actions and
unconventional policies, leaving some problems in its wake. Financial vulnerabilities
have been rising. Financial markets appear overstretched. In some economies,
credit has expanded strongly, often alongside large property price increases and
sometimes heavy foreign currency borrowing. Globally, aggregate total non-
financial debt has risen further relative to income. The room for fiscal and monetary
policy manoeuvre is more limited than pre-crisis and, partly because policy has
failed to address structural impediments, long-term potential growth rates are
lower. And more recently, increasing protectionist pressures have challenged the
international trade system that has buttressed global growth post-WWII. All this
suggests that downside risks to growth are material, as has recently been confirmed
by financial strains in some EMEs.
With this backdrop, policy should take advantage of the cyclical upswing to
mitigate risks and to rebuild room to address any future downturn. Specifically,
fiscal policy should be oriented at regaining space while structural policies should
boost growth potential. The precious open multilateral trading system should be
fully preserved. Macroprudential measures should be used to help strengthen
further the financial system’s resilience and mitigate financial excesses (Chapter IV).
And with due regard for country-specific circumstances, it would be desirable to
continue the process of monetary policy normalisation. The path ahead is a narrow
one (Chapter II).
The chapter first describes how the macroeconomic and financial landscape has
changed over the past year. It then discusses the near-term outlook and the policies
needed to make growth more sustainable. Finally, it deals with the risks ahead.
The global expansion strengthens amid low inflation
Over the past year, global economic activity accelerated. From 3.2% in 2016, global
GDP growth is estimated to have risen to 3.8% in 2017, 0.4 percentage points above
forecasts made at the end of 2016 and close to its long-run average. Despite losing
some momentum in the last quarter of 2017 and the first quarter of 2018, especially
in the euro area, growth remained above potential in most countries.
BIS Annual Economic Report 2018 1

The growth pickup was more synchronised and more evenly spread across
regions and countries than in the past few years. Growth in EMEs returned to rates
closer to historical averages, recovering almost completely from the lows in 2014
and 2015. It also resumed in major commodity exporters, such as Brazil and Russia,
after unusually long and deep recessions. And it beat forecasts in the euro area,
Japan and China, while being approximately in line with them in the United States,
Asia (excluding China) and Latin America (Graph I.1, left-hand panel).
The recovery was more evenly balanced also in terms of spending components.
Along with inventories, private fixed investment contributed a larger share of global
GDP growth than in 2016 (Graph I.1, centre panel). Its strong recovery, especially in
the non-residential sector, accounted for most of the higher growth in advanced
economies. In EMEs, private consumption was relatively more important. Investment
overall also rebounded somewhat, but its evolution varied more across countries. In
particular, investment fell as a share of GDP in China, as the economy continued to
rebalance, while remaining subdued among commodity exporters.
As a result of these developments, manufacturing output and global trade
expanded strongly. After several years of trailing behind services, manufacturing
activity recorded the highest growth rates since 2014. And, with an annual increase
of almost 5% in 2017, global trade increased at its fastest rate in years, benefiting
manufacturing exporters, especially in East Asia. All of this contributed to a rebound
in commodity prices, which in turn supported the gradual recovery in commodity-
exporting countries.
As growth gathered pace, unemployment rates plunged to post-crisis lows
(Graph I.1, right-hand panel). Remarkably, in some major economies, such lows were
GDP growth improves, investment strengthens and unemployment declines Graph I.1
GDP growth in 2017 relative to Decomposition of global real GDP Unemployment rate3
expectations 1 growth2
Percentage points Percentage points Per cent
0.8 6 9
0.6 4 8
0.4 2 7
0.2 0 6
0.0 –2 5
–0.2 –4 4
07 09 11 13 15 17 08 10 12 14 16 18
Global
Euro
area
Japan
China
Asia excl
China
Latin
America
AEs
EMEs
US
Real GDP Contributions: AEs EMEs Global

growth Private consumption
Public consumption
Private investment
Public investment
Inventories
Residual
1
Actual growth in 2017 relative to consensus forecast for 2017 in December 2016; aggregates are weighted averages based on GDP and PPP
exchange rates. 2 Weighted averages of real GDP growth rates and contribution rates based on GDP and PPP exchange rates. 3 Weighted
averages based on labour force levels; definitions may vary across countries.
Sources: IMF, World Economic Outlook; OECD, Economic Outlook; World Bank; Eurostat; Consensus Economics; Datastream; national data;
BIS calculations.
2 BIS Annual Economic Report 2018

also close to postwar troughs. For example, in April 2018 unemployment reached
3.9% in the United States, the lowest except in 2000, 1966–69 and in the 1950s; 2.5%
in Japan, the lowest in 25 years; and 3.4% in Germany, the lowest in almost 40 years.
In these economies, labour shortages became evident, especially among specialised
and highly skilled workers. Unemployment rates also declined rapidly in several EMEs.
Standard estimates of the output gap (albeit highly uncertain) corroborate the view
that several economies might have moved closer to full capacity. That said, the
recovery did not advance as well everywhere. Despite Germany’s strong performance,
and no doubt hindered by structural factors, the unemployment rate in the euro area
remained at 8.5% in April, with a wide dispersion across member countries.
Against this backdrop, headline inflation moved up to almost 2% in advanced
economies and slightly above 3% in EMEs. In both cases, the rebound in energy
prices pushed average headline slightly above core inflation (Graph I.2, left-hand
panel). While the increase in headline inflation has been relatively broad-based
across major advanced economies, core inflation has varied significantly. With a few
exceptions, core CPI inflation generally remains subdued. Over the past 12 months,
it rose slowly in the United States and Japan, reaching 2.1% and 0.4% respectively
in April. In contrast, over the same period, it oscillated around rates near 1% in the
euro area.
Relatively low core inflation has generally reflected subdued wage growth.
Over the past year, real wage growth in advanced economies picked up but
remained below average (Graph I.2, centre panel). In some economies, including
the United States and the euro area, real compensation growth was substantially
lower than productivity gains, while in other advanced economies it was either just
above or in line with productivity gains, thereby putting little or no upward pressure
on unit labour costs (Graph I.2, right-hand panel). Real wage growth in EMEs was
below historical averages.
Inflation remains low and wage growth subdued Graph I.2
Inflation1 Real wage growth1, 2 Real compensation and labour

productivity growth
yoy, per cent yoy, per cent Per cent
8 8 1.5
6 6 1.0
4 4 0.5
2 2 0.0
0 0 –0.5
–2 –2 –1.0
08 10 12 14 16 18 09 11 13 15 17 US EA JP DE FR IT ES GB
3 4
Headline: Core: AEs EMEs Real compensation: Labour productivity:
AEs 2016
EMEs 2017
The dashed lines in the centre panel indicate averages over the period Q1 2000–latest for AEs and Q1 2001–latest for EMEs.
1
Weighted averages based on GDP and PPP exchange rates; definitions may vary across countries. 2 Deflated by GDP deflator. For AEs,
compensation growth per employee as defined by OECD Economic Outlook. For EMEs, wage growth; definitions may vary depending on data
availability. 3 Compensation per employee as defined by OECD Economic Outlook; deflated by GDP deflator. 4 Real output per employee.
Sources: OECD; Datastream; national data; BIS calculations.

Low wage growth in the face of fast declining unemployment is somewhat
puzzling. True, historically wage growth has typically lagged behind inflation and
productivity growth, a pattern that has coincided with a secular erosion of labour’s
income share. But evidence also indicates that the link between wage (or unit
labour cost) growth and measures of labour market slack, albeit much weaker than
in the past, is still present. Based on historical correlations, wage inflation should
have been higher.
A number of factors may help explain the low wage inflation. To begin with,
labour market slack may be larger than headline figures suggest, partly reflecting
hysteresis effects from the Great Financial Crisis (GFC) as well as structural changes
in labour markets. In some countries, broader unemployment measures, which
include discouraged and part-time workers, are significantly higher than headline
unemployment rates (eg close to 8% in the United States). Post-crisis, participation
rates have generally increased, except in the United States, where they are more
than 3 percentage points lower than in 2008. Part of these increases reflects a
higher participation of workers over 55 years old.1 As a result, firms might still be
facing a relatively elastic labour supply, allowing them to fill vacancies without
having to offer significantly higher wages.
Another reason is that globalisation, even if it may have slowed in the last few
years, continues to make labour markets highly contestable.2 Since the integration
of China and the ex-Soviet bloc in the early 1990s, the world economy has been
able to count on a much larger labour pool. In addition, a much greater share of
production now takes place through value chains spread across many countries. To
the extent that production can be outsourced to countries where labour costs are
lower, workers face competition not only in their local labour market, but also
externally. Hence, assessing labour market slack only by looking at domestic
unemployment measures may be misleading. There is indeed some evidence that
global value chains have made local production costs more sensitive to foreign
factors and, relatedly, that unit labour costs have become more synchronised across
countries.3
At the same time, other structural forces such as the adoption of new
technologies have been gathering strength. The room for automating jobs has
been increasing not only in manufacturing but also in services sectors.4 Although
hard evidence is still scant, recent technological diffusion can surely put downward
pressure on wage growth by further reducing labour’s bargaining power.
Meanwhile, by improving productivity, technological diffusion may help prevent
inflation from rising, thus reducing wage demands. Technological diffusion could
also lead to changes in product market structure. There is evidence that
concentration in some economic sectors is on the rise, as fewer and more efficient
firms gain ground at the expense of others.5 Productivity gains by a small number
of firms may, at least initially, partly feed into lower prices, both directly and
indirectly, by putting pressure on incumbents.6
Financial conditions remained very easy for most of the year
The global recovery was supported by very easy financial conditions which, in fact,
eased further for most of the period under review. Only well into the first quarter of
2018 did signs emerge that a significant change in those conditions could be in the
offing, especially for EMEs.
The monetary policy stance of major central banks remained very accommodative,
although it diverged somewhat across areas (Chapter II). The Federal Reserve
continued its very gradual tightening. The ECB extended the time frame of its asset

purchase programme, albeit reducing its pace slightly. The Bank of Japan left its
easing stance essentially unaltered. And the People’s Bank of China kept its
benchmark policy rates unchanged while expanding its use of central bank lending
facilities.
The government bond markets’ reaction to these policy adjustments and the
unexpectedly stronger growth outlook was, for most of the period in review, rather
limited. In the United States, 10-year yields increased modestly until the turn of the
year, when they began moving up steadily, reaching values close to 3% at the end
of April. Even so, long-term yields in the United States and other major economies
remained very low by historical standards (Graph I.3, first panel). In the United
States, the yield curve remained relatively flat, reflecting unusually low term premia
(Chapter II), whereas term spreads in the euro area stayed relatively high after the
sharp steepening of the yield curve last June (Graph I.3, second panel).
Compressed long-term yields and term premia in the United States were to
some degree surprising, given the monetary policy tightening and fiscal expansion
there (Chapter II). At least two forces appear to have been at work: the persistent
impact of relatively looser monetary policy in the euro area and Japan; and
investors’ expectations that inflationary pressures would remain under control even
as growth strengthened, so that monetary policies would not need to adjust.7
Along with low bond yields, broader financial market indicators underscored
the very easy financial conditions. Based on a composite index, financial
conditions in the United States actually continued to loosen until the end of 2017
Global financial conditions remain very easy in 2017 and early 2018 Graph I.3
Long-term view of Term spread2 US financial conditions3 Corporate spreads 4

government bond yields 1
Per cent Percentage points Index Basis points
12 1.25 2.25 1,000
9 1.00 1.50 800
6 0.75 0.75 600
3 0.50 0.00 400
0 0.25 –0.75 200
–3 0.00 –1.50 0
16 17 18 03 08 13 18 10 12 14 16 18
IT
DE
GB
JP
EME local
EME USD
FR
US
United States High-yield: Investment grade:

Germany US
Maximum Japan Europe
Average
75th percentile
Latest
50th percentile
25th percentile
Minimum
The dashed lines in the fourth panel indicate averages over the period 1 June 2005–30 June 2007.
1
Yield to maturity. For AEs, long-term historical values of 10-year government bonds in local currency, since January 1993; for EME local,
JPMorgan GBI-EM Index, seven- to 10-year maturity, since January 2002; for EME USD, JPMorgan EMBI Global, seven- to 10-year maturity,
since January 2001. 2 Difference between the 10-year and the two-year government bond yields. 3 Federal Reserve Bank of Chicago’s
National Financial Conditions Index; positive (negative) values indicate financial conditions that are tighter (looser) than average. 4 Option-
adjusted spreads.
Sources: Bloomberg; Datastream; ICE BofAML Indices; JPMorgan Chase; BIS calculations.

(Graph I.3, third panel) and, despite some subsequent tightening, they remained
easier than during most of the pre-crisis period. In both the United States and the
euro area, high-yield and investment grade credit spreads narrowed further till
end-2017 (fourth panel) along with rising stock prices and low market volatilities.
Such very easy financial conditions underwent some reversal in the first quarter of
2018, after a spike in stock market volatility, related to the liquidation of funds
with values linked to volatility, led to a sharp drop in equity prices. While this
episode highlighted the potential for disruptive market dynamics (Chapter III), it
was a healthy correction that remained largely confined to equity markets.8
Until early 2018, the easing of global financial conditions had also gone hand
in hand with a prolonged depreciation of the US dollar (Graph I.4, first panel). From
December 2016 to March 2018, the dollar lost about 8% of its value in trade-
weighted terms, of which more than half corresponded to a strengthening of the
euro, the renminbi and the yen. It is unclear what explained the depreciation. The
tightening of US monetary policy and the implied widening of short-term interest
rate differentials relative to other major economies would have suggested an
appreciation. However, it is not unusual for the dollar to depreciate when monetary
policy tightens: in half of the past tightening cycles, the dollar depreciated,
including in 2004–06, when the dollar lost about 7% in trade-weighted terms. This
suggests that other factors are relevant. In 2017, one such factor was the unexpected
strengthening of the global economy relative to the US economy. This may have
boosted investors’ risk appetite for non-US assets, including fixed income in EMEs,
while bringing forward the expected timing (and possibly raising the expected
speed) of a withdrawal of monetary policy stimulus in other major economies,
A stronger dollar puts vulnerable EMEs under pressure Graph I.4
USD exchange rates1 EME spreads EME currency depreciation Flows into EME portfolio
funds7
1 Dec 2016 = 100 Basis points USD bn
105 600 KR 2 30
CZ
100 500 0 15
Fiscal deficit/GDP (%)5
PH TH
MX
95 400 PL RU –2 0
TR ID HU
CL MY
90 300 CO PE –4 –15
CN
ZA
85 200 –6 –30
AR
IN
80 100 BR –8 –45
2016 2017 2018 10 12 14 16 18 –5 0 5 10 15 2015 2016 2017 2018
EUR EMEs2 GBI3 CEMBI4 Current account balance/GDP (%)5 Bond Equity
4
JPY EMBI
Depreciation against USD:6
>10% 5–10% ≤5%
The dashed lines in the first panel indicate the long-term average for JPY (January 1987–May 2018) and EUR (January 1999–May 2018).
1
An increase indicates an appreciation against the stated currency. 2 Simple average of AR, BR, CL, CN, CO, CZ, HK, HU, ID, IN, KR, MX, MY,
PE, PH, PL, RU, SA, SG, TH, TR and ZA. 3 For JPMorgan GBI-EM index (local currency-denominated), spread over seven-year US Treasury
securities. 4 For JPMorgan EMBI Global (USD-denominated) and CEMBI (USD-denominated) indices, stripped spread. 5 2017
data. 6 Depreciation of the stated country currency against the US dollar over the period 1 February–25 May 2018; CZ, HU and PL adjusted
for euro depreciation over the same period. 7 Monthly sums of weekly data across major EMEs up to 23 May 2018. Data cover net portfolio
flows (adjusted for exchange rate changes) to dedicated funds for individual EMEs and to EME funds with country/regional decomposition.
Sources: IMF, World Economic Outlook; Bloomberg; EPFR; JPMorgan Chase; national data; BIS calculations.

especially the euro area.9 Uncertainties related to the US Administration’s policies,
especially concerning trade, but also fiscal policies, may also have played a role.
While supported by improved growth, financial conditions in EMEs largely
mirrored the depreciation of the dollar. They eased significantly until the early
months of 2018, as indicated by the large drop in the spreads on local currency
bonds (over 130 basis points from January 2017 to February 2018) as well as dollar-
denominated bonds (Graph I.4, second panel). Then, in the first quarter of 2018, as
the dollar reversed course and started to appreciate (Graph I.4, first panel) and US
long-term yields rose, conditions tightened considerably, with EME currencies
coming under pressure, especially those of countries with weaker current account
and/or fiscal positions (Graph I.4, third panel). Tensions were acute in Argentina and
Turkey, with the former turning to the IMF for support in May. After staying positive
for an unprecedented 16-month spell, portfolio inflows came to a sudden halt and
reversed in May (Graph I.4, fourth panel). Dollar-denominated bond spreads
widened more, on average, than local currency ones (Graph I.4, second panel).
At the time of writing, it is hard to tell how the tightening of financial conditions
will unfold across regions. In major advanced economies, credit spreads have
increased only modestly, while equity markets have resumed their upward trajectory
(Graph I.5, left-hand panel). Implied volatilities have also quickly subsided,
remaining below recent historical averages (Graph I.5, centre panel). That said,
concerns about stretched valuations remain, especially in the United States, where
the cyclically adjusted price/earnings ratio has exceeded its post-1982 average and
is almost twice its long-term 1881–2017 average (Graph I.5, right-hand panel).10 In
the euro area, the spread of Italian sovereign debt relative to German bunds
widened considerably in May, following political events. And in EMEs, conditions
Stock market valuations remain stretched Graph I.5
Stock markets Implied volatilities2 Equity valuation ratios

1 Dec 2016 = 100 Percentage points Percentage points Ratio Ratio
6
CAPE (lhs) Price/dividend (rhs)
145 12 30 80 160
130 10 25 60 120
115 8 20 40 80
100 6 15 20 40
85 4 10 0 0
Q2 17 Q4 17 Q2 18 Q2 16 Q4 16 Q2 17 Q4 17 Q2 18 US Eur7 JP GB US DE JP GB
S&P 500 Nikkei 225 Lhs: Exchange rates

3 Maximum
4 Latest 75th percentile
STOXX Europe 600 MSCI EM1 Bond futures
Rhs: Equities 5 Average 50th percentile
1881–latest 25th percentile
Minimum
1
MSCI Emerging Markets Index, in US dollars. 2 The dashed lines represent simple averages over the period
January 2010–May 2018. 3 JPMorgan VXY Global index, a turnover-weighted index of the implied volatility of three-month at-the-money
options on USD currency pairs. 4 Implied volatility of at-the-money options on long-term bond futures of DE, GB, JP and US; weighted
average based on GDP and PPP exchange rates. 5 Implied volatility of the S&P 500, EURO STOXX 50, FTSE 100 and Nikkei 225 indices;
weighted average based on market capitalisation. 6 For the period December 1981–April 2018; for each country/region, the cyclically
adjusted P/E (CAPE) ratio is calculated as the inflation-adjusted MSCI equity price index (in local currency) divided by the 10-year moving
average of inflation-adjusted reported earnings. 7 European advanced economies included in the MSCI Europe index.
Sources: Shiller database, www.econ.yale.edu/~shiller/data/ie_data.xls; Barclays; Bloomberg; Datastream; BIS calculations.

have continued to deteriorate, and could do so further should the US dollar extend
its appreciation.
The near-term outlook
The global economy’s unexpectedly strong performance over the past 12 months
had led analysts to repeatedly revise upwards growth forecasts for 2018 and 2019
in most countries. This pattern prevailed until the first quarter of this year, when a
number of indicators signalled a possible loss of momentum. While growth
expectations have since been revised down in a number of countries, the prospects
for the global economy overall remain upbeat. Based on consensus forecasts,
global growth is currently forecast to rise to 3.9% in 2018, from an estimated 3.8%
in 2017, before returning to 3.8% in 2019 (Graph I.6, left-hand panel).
The expected increase in global growth masks some differences across
economies. In the United States, forecasts have been upgraded substantially since
the announcement last December of the tax reforms and the spending stimulus:
GDP is currently expected to expand by 2.8% in 2018 and 2.6% in 2019, from 2.3%
in 2017. By contrast, euro area GDP is expected to grow by 2.3% in 2018, the same
as in 2017, followed by a slowdown to 1.9% in 2019, with forecasts revised down in
early 2018. In Japan, growth is expected to slow from 1.7% in 2017 to 1.3% in 2018
and 1.1% in 2019. In other advanced economies, growth is expected to decline over
the next two years. And in EMEs, excluding China, growth is expected to rise to
4.2% in 2018 and 4.3% in 2019 (Graph I.6, left-hand panel).
These near-term forecasts are above long-run potential growth estimates in
most countries, which are lower than pre-crisis and unlikely to go back up fully, given
demographic headwinds and other structural impediments. For instance, based on
long-term consensus forecasts (six to 10 years ahead), long-run growth is currently
estimated to be 2.1% in the United States, compared with over 3% pre-crisis;
The near-term outlook for growth and inflation is positive for most countries Graph I.6
GDP growth1 Headline inflation1 Commodity prices

Per cent Per cent 2 Jan 2015 = 100
6.0 4 120
4.5 3 100
3.0 2 80
1.5 1 60
0.0 0 40
2015 2016 2017 2018
Global
EA
JP
GB
excl GB
CN
excl CN
Global
EA
JP
GB
excl GB
CN
excl CN
Other AEs
EMEs
Other AEs
EMEs
US
US
Brent crude oil Grains

Industrial metals Livestock
Actual: Consensus forecast:2 Actual:3 Consensus forecast:2, 3
2017 2018 2019 2017 2018 2019
1
Aggregates are weighted averages based on GDP and PPP exchange rates. 2 As of May 2018. 3 Based on average of monthly
year-on-year changes in CPI, except for AR, BR, CL, CO, MX, PE and RU (December-to-December changes).
Sources: IMF, World Economic Outlook; Bloomberg; Consensus Economics; national data; BIS calculations.

1.3% in the euro area, against 2% pre-crisis; and 0.7% in Japan, less than half the
pre-crisis estimate. In EMEs, depending on the country, long-run growth expectations
have dropped by between one fifth and one half from pre-crisis levels.
The outlook for inflation is generally consistent with projected stronger growth
(Graph I.6, centre panel). Headline inflation is expected to increase in most countries,
reflecting partly the reduction in output and labour market slack and partly the
recent increase in the price of oil and other commodities. In particular, consensus
forecasts see headline (CPI) inflation reaching 2.5% in 2018 in the United States,
before dropping to 2.2% in 2019; and close to 1% in Japan in 2018 and 2019. In the
euro area, by contrast, headline inflation is expected to remain unchanged at around
1.5% in both 2018 and 2019, albeit with significant variations across member
countries. In line with the pickup in short-run inflation expectations, market-based
measures of long-term ones have also moved up over the past 12 months in both
the United States and the euro area; at the same time, consensus forecasts for
inflation six to 10 years ahead have remained remarkably stable. The increase in oil
and industrial metal prices over the past year should support the projected increase
in headline inflation going forward (Graph I.6, right-hand panel).
In the near term, a number of mutually reinforcing factors should help support
the global expansion. First, globally, low unemployment rates, some pickup in wage
growth and limited expected increases in inflation should sustain household
spending. Second, above-average business confidence readings and continued
very accommodative financial conditions should support fixed capital investment.
The recent reform of the tax regime in the United States is likely to be an additional
driver of investment spending, especially in the short run. Third, thanks to the post-
GFC financial reforms, banking systems are generally better capitalised and more
resilient (Chapter III). Fourth, over the next two years, fiscal policy is set to turn
strongly procyclical in the United States and mildly expansionary in Germany, while
becoming broadly neutral in other economies.
At the same time, looking beyond the near term, such a strong projected
momentum at an advanced stage in the upswing, alongside mild inflationary pressures,
is quite unusual by historical standards (Box I.A). It is possible, as argued before, that
there is more slack than conventional measures indicate. Moreover, to the extent that
current investment raises productivity, albeit with a lag, there may be room for the
economy to continue expanding at above-average rates beyond the near term. That
said, there are questions about the sustainability of the current expansion. Strong
investment and fiscal expenditure could, at a certain point, push the economy up
against capacity constraints. This could lead to the emergence of stronger inflationary
pressures than seen so far. And even without those pressures, it may become
increasingly difficult for firms to find quality inputs and meet earnings expectations.
Over long horizons, the only way to ensure sustainable higher non-inflationary
growth is through structural reforms. As examined in more detail in previous Annual
Reports,11 while the necessary measures are country-specific, they share a number
of features. Their common denominator is fostering entrepreneurship and the rapid
take-up of innovation, limiting rent-seeking behaviour and promoting the flexible
reallocation of all factors of production. Unfortunately, although the pace of
structural reforms picked up in the immediate aftermath of the GFC, especially in
countries that were hardest hit, it has slowed significantly since then. It is now the
slowest since 2011.12 Naturally, the incentive to carry out politically difficult reforms
wanes in good times, when they may appear less pressing. But it is then that any
short-term adjustment costs can be borne more easily.
Making growth more sustainable also requires rebuilding the room for
countercyclical policy, which has narrowed considerably relative to the pre-crisis
period. This involves work on several fronts.

Box I.A
Evidence of unusual late-business-cycle momentum – a historical perspective
Global growth has picked up and broadened over the past year. This box compares recent developments with those
of the past and finds that the current momentum in the recovery is unusually strong so late in the cycle.
Since last year, there has been greater confidence that both output and unemployment rates will far exceed
conventional benchmarks for potential output and full employment. Graph I.A.1 highlights these developments in
the euro area, Japan, the United Kingdom and the United States. Relative to the averages of previous cycles (blue
lines), these economies are forecast to carry far more momentum and hence to exceed those benchmarks much
Boxes
further in the years ahead (red lines).
Recovery carries more momentum at this late stage in the cycle than in the past1
In percentage points Graph I.A.1
Output gap Unemployment gap Inflation rate (core)3
1 0.0 1.0
0 –0.4 0.5
–1 –0.8 0.0
–2 –1.2 –0.5
1st 2nd 3rd 4th 1st 2nd 3rd 4th 1st 2nd 3rd 4th
Years after reaching full employment2

Current Past
1
Past cycles cover the period 1960–2008 for EA, GB, JP and US, and the current cycle 2009–19 for GB, JP and US only (OECD projections for
2017–19). Weighted averages based on current GDP and PPP exchange rates. For EA before 1990, weighted averages of DE, FR and IT based
on GDP and PPP exchange rates. 2 Measures the deviation of each series relative to that in the first year. Horizontal axis defined as starting
at the first year when the unemployment rate in each economy fell below the NAIRU. For “Current”, the first year is 2015 for GB, 2014 for JP
and 2016 for US. The dashed line includes the projections. 3 For JP, data start in 1971 and are adjusted for the effect of the 2014
consumption tax increase.
Sources: Bank of Japan; IMF, World Economic Outlook; OECD, Economic Outlook 102 and Main Economic Indicators; Datastream; national data;
BIS calculations.
Investment and fiscal spending are two key drivers of this late-cycle momentum. Graph I.A.2 shows the main
components of domestic demand for these major economies. While consumption growth exceeds the average of
past cycles, both investment and fiscal spending are unusually strong. The late-cycle surge in investment reflects a
delayed recovery after rather anaemic activity during most of the post-crisis period. The depreciation of the capital
stock, the rise in capacity utilisation and the need to adopt new technologies are continuing to support this leg of
the upswing. Similarly, current fiscal deficits are much higher than in previous cycles, and projections indicate a
much more procyclical stance this time around.
Other supportive conditions are in place, not least buoyant consumer and business sentiment. The recent levels
compare favourably with past cyclical highs (Graph I.A.3). As in the past, this heralds further gains in employment
and incomes that, in turn, will tend to boost confidence further. This mutually reinforcing process, especially in
periods of relatively easy financial conditions, suggests that there is more underlying economic momentum in the
pipeline.
That said, questions remain about whether the stronger momentum can be maintained. Admittedly, it is always
difficult to draw precise parallels between current macro-financial conditions and those that, in the past, derailed
recoveries. Moreover, as discussed in Box I.B, there are reasons to believe that the nature of the business cycle, and
in particular the role of inflation and financial factors, has changed over time. And, last but not least, there is
considerable uncertainty about the measurement of full employment and potential output as events unfold, ie in

Main components of domestic demand contributing to late-cycle strength1 Graph I.A.2
Consumption growth2, 3 Investment growth2 Primary fiscal balance

Percentage points Percentage points Percentage of GDP
0.5 3 –3.0
0.0 0 –3.5
–0.5 –3 –4.0
–1.0 –6 –4.5
–1.5 –9 –5.0
–2.0 –12 –5.5

1st 2nd 3rd 4th 1st 2nd 3rd 4th 1st 2nd 3rd 4th
Years after reaching full employment4

Current Past
1
Past cycles cover the period 1960–2008 for EA, GB, JP and US, and the current cycle 2009–19 for GB, JP and US only (OECD projections for
2017–19). Weighted averages based on current GDP and PPP exchange rates. For EA before 1990, weighted averages of DE, FR and IT based
on GDP and PPP exchange rates. 2 Measures the deviation of the growth rate relative to that in the first year. 3 For JP, adjusted for the
effect of the 2014 consumption tax increase. 4 Horizontal axis defined as starting at the first year when the unemployment rate in each
economy fell below the NAIRU. For “Current”, the first year is 2015 for GB, 2014 for JP and 2016 for US. The dashed line includes the projections.
Sources: IMF, World Economic Outlook; OECD, Economic Outlook 102; Datastream; BIS calculations.
Consumer confidence and business conditions suggest more momentum in the

pipeline Graph I.A.3
Consumer confidence PMI1 Capacity utilisation4

Long-term average = 100
101.5 20 20
Investment growth (%)3
3
Investment growth (%)
100.0 10 10
98.5 0 0
97.0 –10 –10
95.5 –20 –20

83 88 93 98 03 08 13 18 –5 0 5 10 15 –20 –10 0 10 20 30
PMI (deviation from 50)2 Capacity utilisation (deviation from historical mean)
United States Japan
United Kingdom Euro area United States Japan
Average United Kingdom Euro area
The dashed lines in the left-hand panel indicate the start of US recessions defined by the NBER. In the centre and right-hand panels, the
square (triangle) markers refer to 2016 (2017) data.
1
The series starts in 1992 for GB, 1999 for EA and US, and 2002 for JP. For EA, weighted average of DE, FR and IT based on GDP and PPP
exchange rates. 2 Purchasing Managers’ Index (PMI) varies between 0 and 100, with levels of 50 signalling no change on the previous month
in the manufacturing sector. Readings above 50 signal an improvement or increase and those below 50 signal a deterioration or decrease on
the previous month. 3 Real private fixed non-residential investment. 4 The series starts in 1961 for US, 1968 for JP, 1971 for EA and 1985
for GB. For EA before 1990, weighted average of DE, FR and IT based on GDP and PPP exchange rates.
Sources: OECD, Economic Outlook 102 and Main Economic Indicators; Datastream; IHS Markit; national data; BIS calculations.

1
The evolution around peaks in the business cycle in advanced economies Graph I.B.1
“real time”. Even so, historical experience tends to suggest that output and employment exceeding those
benchmarks tend to increase the likelihood of a subsequent downturn. Put differently, under those conditions it
may prove more difficult for policy to ensure a smooth transition back to a balanced, sustainable growth path.
  Estimates of potential output and the natural rate of unemployment are subject to real-time uncertainty. For example, the structural
changes discussed in the main text suggest that currently there may be more slack in the economy than conventionally measured. There
are also reasons to consider that real-time benchmarks may be biased upwards because of the way trends are calculated (the “endpoint
problem”). All else equal, if a recession were to materialise, current gaps would tend to be revised down. What happened in the wake of the
Great Financial Crisis was no exception to this pattern. See D Staiger, J Stock and M Watson, “How precise are estimates of the natural rate of
unemployment?”, in C Romer and D Romer (eds), Reducing inflation: motivation and strategy, University of Chicago Press, 1997; M Watson,
“How accurate are real-time estimates of output trends and gaps?”, Federal Reserve Bank of Richmond Economic Quarterly, vol 93, no 2,
Spring 2007; F Grigoli, A Herman, A Swiston and G Bella, “Output gap uncertainty and real-time monetary policy”, IMF Working Papers,
WP/15/14, January 2015; and E Rusticelli, D Turner and M Cavalleri, “Incorporating anchored inflation expectations in the Phillips curve and
in the derivation of OECD measures of the unemployment gap”, OECD Journal: Economic Studies, vol 2015/1, 2015.      See M Jackson and
T Pietro, “A forest fire theory of the duration of a boom and the size of a subsequent bust”, June 2017.
First, in countries where financial vulnerabilities are building up (see below),

macroprudential measures can be very helpful in rebuilding policy buffers and can
help rein in financial excesses (Chapter IV). Indeed, in many countries major efforts
to implement macroprudential frameworks have been an essential complement to
the financial reforms aimed at strengthening individual institutions and key
infrastructures (Chapter III). The measures are particularly well suited to target
specific vulnerabilities, such as in the mortgage market, and can significantly
improve the trade-offs the authorities need to make when articulating a balanced
policy response to the macro-financial challenges they face. At the same time, they
do not always address the root cause of the problems and have so far proved more
effective in strengthening resilience than in succeeding, on their own, in fully
preventing the build-up of financial imbalances (Chapter IV).
Second, fiscal space needs to be preserved or rebuilt, naturally with due regard
for country-specific circumstances. Public debt has risen to new post-WWII highs in
both advanced and emerging market economies. Against the backdrop of falling
long-run potential growth rates, such higher levels of debt are likely to have
reduced fiscal space. The need to build fiscal space is especially important in those
economies where fiscal solvency has already been called into question, as during
the euro area debt crisis, and where financial expansions may be disguising the true
state of public finances, not least owing to temporarily buoyant tax revenues.
Besides the need for fiscal space for the macroeconomy more generally, sound
public finances are also an essential backstop for the financial system.13
Finally, monetary policy normalisation, too, is essential. It would create room
for countercyclical policy when needed in the future, help reduce the risk of the
emergence of financial vulnerabilities and contribute to restraining debt
accumulation. That said, as discussed in detail in Chapter II, given the unprecedented
starting point, including high debts and persistently low inflation in many
jurisdictions, the path ahead for monetary policy is quite narrow. It calls for striking
a delicate balance between competing considerations while taking heed of country-
specific conditions.
The risks ahead
Against the backdrop of positive near-term prospects, what might be the risks
ahead? In order to understand them better, it is essential to pay particular attention
to financial factors. Since at least the 1980s, their relevance for business cycle

fluctuations has grown (Box I.B). The GFC is just the most recent and prominent
example. In addition, with financial conditions having been so easy for so long, the
possibility of a reversal with macroeconomic consequences cannot be ruled out.
The recent wobbles in EMEs confirm this possibility.
To be sure, the trigger for a materialisation of risks need not be financial at all.
Of particular concern today would be an escalation in trade tensions, which would
negatively affect business confidence and investment. Were this seen as threatening
the current multilateral trading system, the impact could be very significant.
Another possible trigger could be inflation surprising on the upside. Or worries
about fiscal sustainability may return, not least in high-debt slow-growing
economies. Political events may put some countries under strain. And spending
may simply flag as business profitability disappoints. But, even if not acting as
triggers, financial factors are likely to be powerful amplifying forces.
In order to explore these issues further, after examining the background risks
that could arise from the state of financial cycles around the world, we focus on
two specific risk scenarios: a snapback in bond yields sparked by an inflation
surprise in major economies; and a sharp reversal of risk appetite unrelated to such
a surprise. We then turn to the question of the longer-term evolution of risks,
should the non-inflationary expansion continue and, given historically low interest
rates, support the build-up of financial imbalances and debt – private and public.
All of these issues are especially pressing due to the more limited room for policy
manoeuvre.
How have financial cycles played out so far?
The term “financial cycle” generally refers to the self-reinforcing interactions

between perceptions of value and risk, risk-taking and financing constraints that
can amplify business cycle fluctuations and are reflected in the joint behaviour of
credit and asset prices (Box I.B). Empirical work indicates that the financial cycles
that pose the greatest risk for economic activity are best captured more specifically
by combining information from medium-term fluctuations in credit and property
prices, although equity prices naturally also play a role. Graph I.7 illustrates the
state of the financial cycle aggregating across different countries in broadly similar
phases, using a simple measure that combines the relevant information (see Box I.B
for further details).
The graph indicates that in major advanced economies financial cycles are at a
relatively early stage of the expansion. This set of countries includes some of those
at the heart of the GFC and that have seen some private sector deleveraging since
then. In particular, at least at the aggregate level, no worrisome boom is evident in
the United States and the United Kingdom: this is because, while property prices
have recovered strongly from their post-crisis lows, credit-to-GDP ratios remain well
below their pre-crisis peaks (not shown). Similarly, in some large advanced
economies less severely affected by the GFC, such as Germany and Japan, the
financial cycle is also on an upswing. This suggests that, in the near term, the cycles
will not act as a headwind to economic activity and could support it further.
By contrast, in several advanced small open economies that avoided the crisis,
strong financial cycle expansions seem to be coming to an end. After a period of
rapid increase, growth in credit to corporates and households has decelerated since
2016, while growth in property prices has slowed or turned negative. However,
household credit as a ratio to GDP remains at historical highs in Australia, Canada
and some Nordic countries (Graph I.8, left-hand panel). To mitigate these
vulnerabilities, national authorities have been encouraging banks to tighten their
lending standards or have adopted macroprudential measures (Chapter IV).

Box I.B
The changing nature of the business cycle and its link to the financial cycle
Consumer confidence and business conditions suggest more momentum in the
pipeline Graph I.A.3
Since the early 1980s, a number of important changes have made financial factors more important in driving
Consumer
business cycleconfidence
fluctuations while reducing PMI1the relevance of inflation as an indicator Capacityofutilisation 4
unsustainable expansions.
Long-term
First, financial markets average
have been = 100
liberalised, starting around that time. Without sufficient safeguards, this change
created the potential for larger booms, followed by busts, in credit and asset prices – that is, larger financial cycles.
Second, starting roughly at the101.5 same time, inflation-focused monetary20regimes became the norm. Central banks’ 20

focus on inflation control led them to gradually downplay the role of monetary and credit aggregates and to rely
almost exclusively on microprudentially 100.0 oriented supervision and regulation 10 to affect financial behaviour during 10
financial booms and hence minimise any fallout as booms turned to busts (Chapter IV). This meant that they had
little reason to tighten policy if inflation 98.5 remained low, even as financial 0imbalances built up. Finally, from the 1990s 0
on, the entry of China and former Communist countries into the world economy, the international integration of
product markets and technological 97.0
advances have boosted global supply –10 and productivity. Along with greater –10
central bank credibility, this has made it more likely that inflationary pressures would remain mute even as
expansions gathered pace. It also means that financial booms can build up further and that a turn in the financial
95.5 –20 –20
cycle, rather than rising inflation, may bring about a downturn in economic activity.
83 88 93 98 03 08 13 18 –5 0 5 10 15 –20 –10 0 10 20 30
These factors were clearly present also in the PMIrun-up
(deviationto from
the Great
50)2 Financial Crisis.
Capacity Short-term
utilisation (deviationoutput volatility
from historical mean)
as wellUnited
as theStates Japan
level and volatility of inflation remained low (the so-called Great Moderation). At the same time,
United Kingdom Euro area United States Japan
leverage in
Average
the financial and non-financial system rose. When the Kingdom
United financial cycle Euro
turned, areafinancial stress emerged
and the economy experienced a serious recession.
The dashed lines in the left-hand panel indicate the start of US recessions defined by the NBER. In the centre and right-hand panels, the
Graph I.B.1 illustrates some of these changes for a group of advanced economies, focusing on the behaviour of
square (triangle) markers refer to 2016 (2017) data.
key variables around business cycle turning points. In the period 1960–84, inflation was higher and tended to
1
The series
increase bystarts in 1992
several for GB, 1999
percentage for EA peaking
points, and US, and 2002after
soon for JP.output;
For EA, weighted average nominal
the short-term of DE, FR and IT based
interest rateonalso
GDP tended
and PPP
exchange rates. 2 Purchasing Managers’ Index (PMI) varies between 0 and 100, with levels of 50 signalling no change on the previous month
to increase by several percentage points, closely tracking inflation; and there was no credit boom – in fact, the
in the manufacturing sector. Readings above 50 signal an improvement or increase and those below 50 signal a deterioration or decrease on
credit-to-GDP
the previous month.gap 3tended to decline
Real private slightly after
fixed non-residential the business
investment. 4
The cycle turned.
series starts By for
in 1961 contrast,
US, 1968since
for JP,1985 inflation
1971 for has
EA and 1985
been lower and remarkably stable around business cycle peaks; the short-term
for GB. For EA before 1990, weighted average of DE, FR and IT based on GDP and PPP exchange rates. interest rate has increased only
modestly; and credit has boomed in the upswing, as indicated by the positive and large credit-to-GDP gap.
Sources: OECD, Economic Outlook 102 and Main Economic Indicators; Datastream; IHS Markit; national data; BIS calculations.
Interestingly, these patterns resemble those observed between the two world wars, given that the 1920s saw a large
credit boom against the backdrop of low inflation, a high degree of global trade and financial integration, as well as
a monetary regime that de facto kept a lid on inflation. Similar patterns were also in evidence pre-WWI (the
classical gold standard) – the previous era of globalisation.
The evolution around peaks in the business cycle in advanced economies1 Graph I.B.1
Inflation Short-term interest rate Credit-to-GDP gap

Per cent Per cent Percentage points
9 8 8
6 6 4
3 4 0
0 2 –4
–3 0 –8
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
Years around peaks

1923–39 1960–84 Since 1985
1
The horizontal axis denotes years around peaks in the business cycles, with the peak date set at zero (vertical lines). Lines show the median
evolution across countries (AU, CA, DE, DK, FI, FR, GB, NO, SE and US) and events in the respective time period.
Sources: National data; BIS calculations.

Financial cycles1
In standard deviations Graph I.B.2
United States United Kingdom
2 2
1 1
0 0
–1 –1
–2 –2
72 77 82 87 92 97 02 07 12 17 72 77 82 87 92 97 02 07 12 17
The shaded areas represent recessions based on ECRI.

1
Financial cycles are measured by frequency-based (bandpass) filters capturing medium-term cycles in real credit, the credit-to-GDP ratio
and real house prices. Financial cycles are normalised by country-specific means and standard deviations.
Sources: Economic Cycle Research Institute (ECRI); national data; BIS; BIS calculations.
One noteworthy mechanism behind the interaction between the financial and business cycles operates through
the accumulation of debt and the subsequent increase in debt service burdens. That is, in the upswing of the
financial cycle, new borrowing and rising asset prices boost economic growth. Over time, however, the accumulation
of debt implies ever larger debt service commitments. These commitments have a strong and long-lasting negative
impact on expenditures of indebted households and corporations. Hence, once the financial cycle turns, the positive
effects of new credit on spending fade while the negative ones of the debt service burdens grow. It is therefore
unsurprising that measures of financial cycle expansions, not least those that include the evolution of debt service
burdens, can be useful leading indicators of subsequent economic downturns and that they also help explain the
length and depth of the Great Recession.
When focusing on the financial fluctuations that cause the greatest damage to economic activity (and the
financial system), empirical research suggests that a promising strategy is to represent the financial cycle through
medium-term fluctuations in credit and property prices. In turn, these fluctuations can be identified through a range
of methodologies. A simple one, used in Graph I.B.2 and Graph I.7 in the main text, relies on statistical filters to
extract cyclical fluctuations over periods from eight to 32 years in real credit, the credit-to-GDP ratio and real
property prices. It then combines these cyclical components into a single series. To facilitate comparison across
countries, it is also useful to normalise the cyclical components by country-specific means and standard deviations,
so that a value of one indicates that cycles are, on average, one standard deviation higher than normal.
As an illustration, Graph I.B.2 shows the evolution of the financial cycle in the United States and the United
Kingdom. It is evident that the downswings of the financial cycle – characterised by high debt service, deleveraging
and falling asset prices – are closely associated with the economic downturns that have occurred in these countries
since the mid-1980s, with some of these coinciding with serious financial strains. This also holds true for other
advanced economies not shown here.
  For a discussion of policy regime changes and their implications for monetary and financial stability, see eg C Borio and P Lowe,
“Securing sustainable price stability: should credit come back from the wilderness?”, BIS Working Papers, no 157, July 2004; C Borio and
W White, “Whither monetary and financial stability? The implications of evolving policy regimes”, BIS Working Papers, no 147, February 2004;
and C Borio, “Monetary and prudential policies at a crossroads? New challenges in the new century”, Moneda y Crédito: Revista de
Economía, vol 224, 2007.      See eg B Eichengreen and K Mitchener, “The Great Depression as a credit boom gone wrong”, in Research in
Economic History, Vol 22, Emerald Group Publishing Limited, 2003, pp 183–237.      See eg W Huffman and J Lothian, “The gold standard
and the transmission of business cycles, 1833–1932”, in D Bordo and A Schwartz (eds), A retrospective on the classical gold standard, 1821–
1931, NBER, 1984; and C Goodhart and P Delargy, “Financial crises: plus ca change, plus c’est la meme chose”, International Finance, vol 1,
1998.      See eg BIS, 86th Annual Report, June 2016, Box III.A; M Drehmann, C Borio and K Tsatsaronis, “Characterising the financial cycle:
don’t lose sight of the medium term!”, BIS Working Papers, no 380, June 2012; and S Claessens, A Kose and M Terrones, “How do business
and financial cycles interact?”, Journal of International Economics, vol 87, 2012.      For evidence of the negative effects of high debt

service burdens on household consumption and investment, see M Drehmann, M Juselius and A Korinek, “Accounting for debt service: the
painful legacy of credit booms”, BIS Working Papers, no 645, June 2017. M Juselius and M Drehmann, “Leverage dynamics and the real
burden of debt”, BIS Working Papers, no 501, May 2015, show that, even with real-time estimates, the predicted adjustment to leverage and
the debt service burden from 2005 onwards implies paths for credit and expenditure that closely match actual developments before and
during the Great Recession in the United States. More generally, for evidence of the negative effect of debt on growth, see also O Jorda,
M Schularick and A Taylor, “The great mortgaging: housing finance, crises and business cycles”, Economic Policy, vol 31, January 2016;
A Mian, A Sufi and E Verner, “Household debt and business cycles worldwide”, Quarterly Journal of Economics, vol 132, 2017; IMF, “Financial
conditions and growth at risk”, Global Financial Stability Report, October 2017, Chapter 3; and M Lombardi, M Mohanty and I Shim, “The
real effects of household debt in the short and long run”, BIS Working Papers, no 607, January 2017.      See references in footnote 4.
In a number of EMEs, the financial cycle appears to have already turned.14 In

2017, credit and property prices expanded at a much lower rate than their average
since 2010, and in several cases even contracted. China is a case in point, as after
a very rapid increase its overall credit-to-GDP ratio peaked at the beginning of
2017. In particular, credit to the corporate sector fell sharply as the authorities
intensified measures to encourage deleveraging and reduce financial stability
risks.
Aggregate financial cycle measures can help to identify whether slow-moving
financial factors are supporting or depressing growth and to spot risks ahead. They
are, however, only a first step in the analysis of financial vulnerabilities. In several
advanced economies, even in those where the financial cycle is still on an upswing,
pockets of financial fragility have surfaced.
One example is the continuous deterioration of non-financial corporate
balance sheets in the United States, the United Kingdom and, to a lesser extent,
France and other European countries. The deterioration is evident in the steady
increase in leverage, a significant drop in interest coverage ratios and a decline in
the share of firms rated A or higher. In the United States, in particular, corporate
leverage today is at its highest level since the beginning of the millennium and
similar to that prevailing after the leveraged buyout boom of the late 1980s. This is
so even after accounting for large corporate cash balances.15 And the large share of
The financial cycle supports growth in advanced economies 1

In standard deviations Graph I.7
–1
–2
92 97 02 07 12 17
2 3 4 5
GFC countries Small open economies Germany and Japan EMEs
1
Financial cycles are measured by frequency-based (bandpass) filters capturing medium-term cycles in real credit, the credit-to-GDP ratio
and real house prices. Financial cycles are normalised by country-specific means and standard deviations before simple averages are taken
for country groupings. 2 ES, FR, GB, IT and US. 3 AU, CA, CH, FI, NO and SE. 4 Germany and Japan are aggregated together as their
respective cycles have been asynchronous with other AEs. 5 BR, CL, CO, HK, ID, KR, MX, MY, PE, SG and TH.
Sources: National data; BIS; BIS calculations.

Areas of vulnerability: commercial property, household debt and EME dollar debt Graph I.8
Household debt and DSRs 1 Real commercial property prices3 USD-denominated credit to EME
non-bank borrowers 6
Percentage points Percentage of GDP Q1 2013 = 100 Amount outstanding, USD trn
2 90 145 3.5
1 80 130 2.8
0 70 115 2.1
–1 60 100 1.4
–2 50 85 0.7
–3 40 70 0.0
02 07 12 17 07 12 17
02 07 12 17
DSR (lhs)2 Debt (rhs) US Other AEs4 7
EA EMEs
5 Bank loans to non-bank borrowers
Bonds issued by non-bank borrowers
1
Simple averages of AU, CA, FI, NO and SE. 2 Average difference of the debt service ratio (DSR) from country-specific averages since
1999. 3 Definitions vary across countries; deflated by CPI; data extended using residential property prices if ending prior to
end-2017. Aggregates are weighted averages based on GDP and PPP exchange rates. 4 AU, CA, CH, GB and JP. 5 BR, HK, ID, KR, PH and
SG. 6 Non-banks comprise non-bank financial entities, non-financial corporations, governments, households and international
organisations. 7 Loans by LBS-reporting banks to non-bank borrowers, including non-bank financial entities, comprise cross-border plus
local loans.
Sources: Bloomberg; national data; BIS global liquidity indicators and locational banking statistics (LBS); BIS; BIS calculations.
firms rated just investment grade (BBB) is especially vulnerable to a deterioration in

their ratings to below investment grade.
Another potential area of rising vulnerabilities is commercial real estate. Real
commercial property prices have risen significantly in advanced economies. In
particular, in the United States they are close to pre-crisis peaks (Graph I.8, centre
panel). At the same time, nearly 50% of banks’ real estate exposures are to
commercial real estate, up from 40% five years ago. Values there seem particularly
vulnerable to rising long-term yields. Some research suggests that, in the United
States, a 200 basis point increase in long-term yields would lead to a more than
25% drop in commercial property prices.16
A third example concerns foreign currency borrowing in EMEs. There, the post-
crisis financial cycle expansions went hand in hand with rapid growth in non-banks’
US dollar borrowing, which continued throughout 2017. According to the BIS global
liquidity indicators, the outstanding stock of US dollar credit to non-bank EME
borrowers has roughly doubled since 2008 and currently stands at $3.6 trillion. As
the dollar weakened in 2017, the annual growth rate of dollar-denominated credit
to EME non-bank borrowers almost tripled from 3.1% at end-2016 to 8% at end-
December 2017. Growth was especially pronounced in international debt securities,
which expanded at an annual rate of 17% in December 2017 (Graph I.8, right-hand
panel). Moreover, estimates indicate that borrowing through FX swaps, not covered
by these statistics, was of a similar magnitude to that visible on balance sheets.17
These trends mean that EMEs have become more exposed to an appreciation
of the dollar and to reversals in international investors’ risk appetite, as recent
events have confirmed. Consistent with this, there is a growing body of evidence
indicating that post-crisis the value of the US dollar versus a broad basket of other
currencies has become an important driver of global banks’ leverage and cross-

border capital flows, more so than the VIX (Chapter II).18 Meanwhile, the greater
participation of foreign investors in local currency markets compared with pre-crisis
might not necessarily act as a stabilising factor, as it may expose EMEs to a greater
risk of capital flight.
Several developments in EMEs have reduced risks relative to previous episodes
of large-scale foreign exchange borrowings, but they have not eliminated them
altogether. Important mitigating developments include the large accumulation of
foreign exchange reserves by EME central banks and, compared with the turbulence
in the 1990s, more flexible exchange rate regimes. The active deployment of
macroprudential measures should help too (Chapter IV). At the same time, these
economies are not immune to a more general tightening of financial conditions,
should the dollar continue to appreciate (see below and Chapter II) and a major
shift in portfolio diversification be triggered among institutional investors.
Snapback risk
So far inflation has been rather unresponsive to the continuing tightening in

product and labour markets, but at some point pressures could mount and inflation
could surprise on the upside. For example, the longer the expansion continues, the
more likely it is that capacity constraints will bite. This is especially so if, as projected,
slack declines further at the global level: this would limit the safety valves available
to individual countries and possibly generate additional pressures on commodity
prices. In addition, if, as some evidence suggests, workers’ and firms’ expectations
have become more backward-looking since the GFC, the lag before inflation
emerges may simply be longer.19
To be sure, while inflation surprises cannot be ruled out, they are unlikely to be
large. The secular structural forces keeping a lid on inflation will not vanish any
time soon (see above). The increasing relevance of shale oil, given its greater
responsiveness to prices, coupled with advanced economies’ lower oil dependence
than in the past, should dampen inflation spikes linked to sharp oil price increases
and make them less persistent.20 And while an escalation of protectionist measures
could well reverse part of these trends, a persistent impact on inflation would take
time to emerge.
That said, even small changes in the inflation outlook (or monetary policy
response) could elicit an outsize market response. Very compressed (even negative)
term premia point to the potential for a quick and sharp reversal – a snapback – as
illustrated during the market ructions in early February this year. Market participants
clearly see low inflation stretching out into the future, and may also have taken
considerable risks owing to investment strategies, such as benchmark-hugging and
other forms of herding, that can amplify market moves. For similar reasons, credit
and liquidity risks may also be underpriced. The underpricing may be more severe
in sectors and countries where debt levels and credit flows have grown strongly
post-crisis, including EMEs.
A snapback could be rapidly transmitted to other major bond markets,
especially if it took place in the market of the dominant international currency – the
US dollar. For one thing, term premia tend to be quite correlated internationally,
even when the expected interest rate component of bond yields is not.21 This could
lead to an undesired steepening of yield curves even in countries where output is
still at or below potential and inflation remains well below objectives. Moreover, the
increase in yields could be compounded by an appreciation of the dollar and
capital outflows from countries with large dollar-denominated liabilities.
The effects of a snapback in bond yields on individual countries would depend
on several factors. One is the size of debt and financial imbalances. The most

exposed to such risks are naturally sectors and countries where debt in relation to
income is high or short-run refinancing needs are large. Among these are several
small open economies and, in particular, EMEs where the financial cycle has peaked
(Graph I.7), dollar debt is high, current account deficits are large and foreign
exchange reserve buffers small. This assessment is confirmed by a simple sensitivity
analysis (Graph I.9). In some small open advanced economies that have seen the
largest increases in debt post-crisis, higher interest rates would push debt service
burdens well above long-run averages, thus dampening consumption and
investment (Box I.B). A second factor is the extent to which lenders could absorb
any credit losses. In most countries, banks are generally better capitalised than pre-
crisis, especially in countries hard hit by the GFC (Chapter III). However, post-crisis, a
greater share of credit has been intermediated by non-banks, especially the shadow
banking system.22 A third factor is the participation of foreign investors in local
markets, as the asset management industry has grown rapidly in recent years and
become more global in its investment. This makes markets more vulnerable to a
reversal in flows and an evaporation of liquidity in times of stress (Chapter III). While
these risks are hard to assess, a combination of these factors would make a country
especially vulnerable.
Reversal of risk appetite
A generalised sharp tightening of financial conditions may occur even if there is no

inflation or monetary policy surprise in the large economies that are home to
international currencies. In some EMEs it could be induced, for instance, by the
domestic financial cycles contracting, given signs that they may have turned
(Graph I.7). In advanced economies, worries about fiscal sustainability may return,
especially in countries with high debt and slow growth and/or facing politically
challenging circumstances. More generally, even in the absence of inflationary
pressures, sentiment-driven swings in business or residential investment could
Vulnerability of debt service ratios to rising rates varies by country1

In percentage points Graph I.9
Non-financial corporate sector Household sector

Projection Projection
15 2
0 0
–15 –2
–30 –4
01 03 05 07 09 11 13 15 17 19 21 01 03 05 07 09 11 13 15 17 19 21
Australia Canada United Kingdom United States Euro area Sweden
1
Difference of debt service ratios from country-specific averages since 1999. Projections keep the credit-to-income ratio fixed and assume
that the average interest paid on the stock of debt increases in line with historical experience if short-term money market rates gradually rise
by 150 basis points over 1.5 years and then remain unchanged until Q2 2021. The pass-through from money market rates to average interest
rates is based on simple regressions using earliest available data for each country and up to 2017. Projections start in Q1 2018.
Sources: Datastream; Global Financial Data; national data; BIS; BIS calculations.

initiate a contraction, not least if profits were to disappoint and undermine stretched
equity valuations.23
A severe tightening of financial conditions could play out somewhat differently
from a snapback in bond yields in the major advanced economies. In particular, as
international investors retreated from the countries affected, stronger flows to safe
haven countries could well depress term premia there.
Despite the positive effects of safe haven flows, growth in receiving countries
could be more adversely affected than in the past even if the shock originated in
EMEs. The latter now account for 60% of global GDP and have contributed more
than two thirds of its growth since 2010. In various adverse scenarios that hit
growth in EMEs, model-based simulations indicate that growth in major economies
could be reduced by up to 1 percentage point, possibly a conservative estimate.24
Given the currently lower potential growth rate of several advanced economies, the
risk of downturn could be material if the original contraction was large enough.
Risks from the further build-up of financial imbalances and debt
Even if a soft landing scenario in the global economy materialises in the near to
medium term, downside risks could increase over the longer term. In particular, the
combination of a non-inflationary expansion and low interest rates would be likely
to encourage the further, gradual build-up of financial imbalances and debt
accumulation more generally, creating the conditions for a more costly contraction
further down the road.25 In addition to private sector debt accumulation, procyclical
fiscal policies, facilitated by current low borrowing costs, could lead to a further rise
in public debt, especially if, as evidence indicates, the financial expansion has
flattered the fiscal accounts.26 While supportive of growth in the short run,
expansionary fiscal policies could force retrenchment in the future and further limit
any room for policy manoeuvre. Indeed, a growing body of studies documents how
higher leverage, in both the private and public sectors, can boost growth in the
short run, but at the cost of lower growth on average, including deeper and
prolonged recessions, in the future.27
From a long-term perspective, the continuous accumulation of debt is worrying
for at least two reasons. First, the higher the debt, the more sensitive the economy
and financial valuations are to higher interest rates, reducing the level of interest
rates an economy can bear. This, in turn, makes it more difficult to raise them,
favouring further debt accumulation – a kind of “debt trap” (Chapter II). Second,
higher debt – private and public – narrows the room for policy manoeuvre to
address any downturn.
This broad analysis of risks, financial and real, points to a clear message. While
the global economy has made substantial progress post-crisis and near-term
prospects are positive, the path ahead is a narrow one. The risks highlight the
importance of taking advantage of the current upswing to implement the necessary
measures to put the expansion on a stronger footing and to rebuild policy buffers.
Such buffers are essential to regain the room for policy manoeuvre to tackle the
next downturn, which will surely come at some point.

Endnotes
1
The participation rate of workers aged 55 to 64 increased from 56% in 2008 to 62% in 2016 on
average across OECD countries. In Germany, it increased from about 60% in 2008 to over 70% in
2016. Given their shorter expected tenure, this group of workers tends to command lower wages
relative to younger groups. See B Mojon and X Ragot, “The labor supply of baby-boomers and
low-flation”, Sciences Po OFCE Working Paper, no 9, 2018-01.
2
For example, the diffusion of global production chains has slowed post-crisis. That said, costs have
not fully converged across countries, suggesting room for further integration. Even without the
greater contestability of labour markets due to globalisation, existing competitive pressures may
interact with other factors to keep inflation weak. One example concerns the formation of inflation
expectations embedded in wage negotiations. Due to existing competitive pressures, workers may
be (temporarily) more reluctant to ask for wage increases in line with inflation targets. Thus,
inflation expectations may have become more inertial (or appear as de-anchored), reflecting more
heavily past inflation outcomes than inflation targets.
3
See BIS, 87th Annual Report, June 2017, Box IV.B. Evidence on the relevance of measures of global
slack in Phillips curves is mixed. C Borio and A Filardo, “Globalisation and inflation: new cross-
country evidence on the global determinants of domestic inflation”, BIS Working Papers, no 227,
May 2007, and R Auer, C Borio and A Filardo, “The globalisation of inflation: the growing
importance of global value chains”, BIS Working Papers, no 602, January 2017, found that the
expansion of global value chains accentuated the importance of global factors relative to domestic
pressures. Other studies have questioned the quantitative relevance of this channel, eg J Ihrig,
S Kamin, D Lindner and J Marquez, “Some simple tests of the globalization and inflation hypothesis”,
International Finance, vol 13, issue 3, Winter 2010; ECB, “Assessing labour market slack”, ECB
Economic Bulletin, issue 3/2017, Box 3; and D Brouillette and L Savoie-Chabot, “Global factors and
inflation in Canada,” Bank of Canada Staff Analytical Note, no 2017–17, October.
4
According to a detailed study of some 800 occupations in 46 advanced and emerging market
economies, about 60% of occupations have at least 30% of their content that can be automated
based on existing technologies (McKinsey Global Institute, A future that works: automation,
employment, and productivity, January 2017).
5
See eg D Andrews, C Criscuolo and P Gal, “Frontier firms, technology diffusion and public policy:
micro evidence from OECD countries”, OECD Productivity Working Papers, no 2, November 2015.
6
For example, the rise of e-commerce (the “Amazon effect”) seems to have lowered retail price
inflation in the United States by at least 0.1% per year between 2011 and 2015, without counting
for the indirect effects; see eg K Kliesen and C Gascon, “An examination of current economic
conditions in the nation and in the Memphis area”, Regional Economic Briefing, Federal Reserve
Bank of St Louis, October 2017.
7
Other factors may have also contributed, including the US Treasury’s decision to shift its issuance
towards shorter maturities and a possible continued strong demand for long maturities by insurers
and pension funds needing to meet regulatory standards and to match their long-term liabilities.
8
See “Volatility is back”, BIS Quarterly Review, March 2018, pp 1–15.
9
The dollar value may have been influenced not only by the expected beginning of normalisation in
the euro area, but also by its expected path. That is, investors seem to have expected that once
normalisation began in the euro area, it would proceed at a more rapid pace than in the United
States.
10
Strong dividends lent some support to US equity valuations. However, dividends per share of US
equities have been growing at a much faster rate since the GFC. High dividends per share have
also been supported by large stock repurchases since the early 2000s. The latter may be further
strengthened by the repatriation of offshore savings following the US tax reform.
11
See eg BIS, 83rd Annual Report, June 2013, Chapter III.
12
See OECD, Going for Growth 2018, March 2018.

13
See BIS, 86th Annual Report, June 2016, Chapter V.
14
Early warning indicators for systemic banking crises also point to the build-up of vulnerabilities in
several EMEs: see I Aldasoro, C Borio and M Drehmann, “Early warning indicators of banking crises:
expanding the family”, BIS Quarterly Review, March 2018, pp 29–45.
15
See also IMF, Global Financial Stability Report, April 2018.
16
These effects are implied by current capitalisation rates (rent-to-price ratios) (ACLI survey) and the
estimates in J Duca, P Hendershott and D Ling, “How taxes and required returns drove commercial
real estate valuations over the past four decades”, National Tax Journal, vol 70, no 3, September
2017, pp 549–83.
17
See C Borio, R McCauley and P McGuire, “FX swaps and forwards: missing global debt?”, BIS
Quarterly Review, September 2017, pp 37–54.
18
One possible mechanism is that when the dollar weakens, the creditworthiness of currency-
mismatched borrowers improves and global banks’ balance sheet constraints are relaxed, thereby
increasing the supply of cross-border lending. In turn, this ends up stimulating real investment (see
S Avdjiev, V Bruno, C Koch and H S Shin, “The dollar exchange rate as a global risk factor: evidence
from investment,” BIS Working Papers, no 695, January 2018). This channel, also known as the risk-
taking channel of the exchange rate, operates in the opposite direction from the textbook trade
channel, which emphasises trade competitiveness resulting from currency movements. The
relevance of the US dollar in influencing financial conditions globally has been documented, using
the broad dollar index, in S Avdjiev, W Du, C Koch and H S Shin, “The dollar, bank leverage and the
deviation from covered interest parity”, BIS Working Papers, no 592, July 2017; S Avdjiev, C Koch
and H S Shin, “Exchange rates and the transmission of global liquidity”, unpublished mimeo, March
2018; and also, using bilateral exchange rates, in V Bruno and H S Shin, “Cross-border banking and
global liquidity”, Review of Economic Studies, vol 82, no 2, April 2015; V Bruno and H S Shin, “Capital
flows and the risk-taking channel of monetary policy”, Journal of Monetary Economics, vol 71, April
2015; and B Hofmann, I Shim and H S Shin, “Sovereign yields and the risk-taking channel of
currency appreciation”, BIS Working Papers, no 538, May 2017.
19
There is some evidence that expectations have become somewhat de-anchored in the aftermath
of the GFC in some countries, especially in the euro area (eg T Lyziak and M Paloviita, “Anchoring
of inflation expectations in the euro area: recent evidence based on survey data”, European Journal
of Political Economy, vol 46, 2017; F Natoli and L Sigalotti, “Tail co-movement in inflation
expectations as an indicator of anchoring”, International Journal of Central Banking, January 2018);
and more generally when inflation is significantly below target or when interest rates are close to
the zero lower bound (eg R Banerjee and A Mehrotra, “Deflation expectations”, BIS Working Papers,
no 699, February 2018). Other studies, however, find no evidence of de-anchoring in most
countries (eg O Blanchard, “The US Phillips curve: back to the ‘60s?“, Peterson Institute for
International Economics Policy Briefs, no PB 16-1, January 2016). De-anchoring of expectations also
creates the risk of a persistent overshoot should inflation pick up.
20
Shale oil production is more responsive to prices than that from other sources, as costs are much
lower, wells can be opened and shut down much more rapidly, and the investment cycle is much
shorter and less uncertain. Despite representing a small fraction of total oil production (6% in
2017), shale oil has accounted for over 60% of its cumulative increase since 2010.
21
See eg M Obstfeld, “Trilemmas and trade-offs: living with financial globalisation”, BIS Working
Papers, no 480, January 2015; B Hofmann and E Takáts, “International monetary spillovers”,
BIS Quarterly Review, September 2105, pp 105–18; and E Kharroubi and F Zampolli, “Monetary
independence in a financially integrated world: what do measures of interest rate co-movement
tell us?”, BIS Papers, no 88, October 2016, pp 193–205.
22
For instance, the share of assets intermediated by other financial intermediaries (OFIs) has steadily
increased, reaching about 30% of total financial system assets in 2016, mainly through collective
investment vehicles and securitisation-based intermediation, which account for about 80% of the
FSB’s narrow measure of shadow banking.

23
Sentiment-driven swings in investment along with fluctuations in credit and financial conditions
are timeless features of business cycles and can explain why economic activity may turn even in the
absence of strong inflationary pressures (see eg V Zarnowitz, “Theory and history behind business
cycles: are the 1990s the onset of a golden age?”, Journal of Economic Perspectives, vol 13, no 2,
1999).
24
For example, IMF, World Economic Outlook: Adjusting to lower commodity prices, October 2015,
considers a scenario in which a greater than anticipated slowing of potential output growth in
EMEs is combined with smaller capital flows to EMEs and tighter financial conditions. After one
year, growth is 0.8 percentage points lower than baseline in the BRICS and 0.4 points lower in
advanced economies. Similarly, P Ollivaud, E Rusticelli and C Schwellnus, “Would a growth
slowdown in emerging markets spill over to high-income countries? A quantitative assessment”,
OECD Economics Department Working Papers, no 1110, 2014, consider a scenario in which EMEs
experience a 2 percentage point decline in domestic demand growth combined with a 10% fall in
equity prices and a 20% currency depreciation (current account deficit countries). The same
scenario also involves a 50 basis point increase in the equity risk premium in OECD economies,
reflecting a negative confidence spillover due to tighter financial conditions in EMEs. As a result,
growth declines by 1 percentage point in Japan and 2⁄3 percentage point in the United States and
Germany, close to the average for OECD countries. These estimates may, however, understate the
true effects, especially at times of heightened financial market volatility. Existing structural and
empirical models may not fully capture the (time-varying and non-linear) power of financial
factors.
25
In addition, over time, the continuation of the expansion, especially if supported by low interest
rates and growing financial imbalances, may also be accompanied by worsening imbalances in the
real sector (see also Box I.A). Of particular note is the risk of sectoral resource misallocations: see
eg C Borio, E Kharroubi, C Upper and F Zampolli, “Labour reallocation and productivity dynamics:
financial causes, real consequences”, BIS Working Papers, no 534, January 2016.
26
See C Borio, M Lombardi and F Zampolli, “Fiscal sustainability and the financial cycle”, in L Ódor
(ed), Rethinking fiscal policy after the crisis, Cambridge University Press, 2017, pp 384–413.
27
Empirical studies have documented the potential negative impact of public debt on future average
growth, albeit not conclusively (for an overview, see eg “Is high public debt a drag on growth?”, in
BIS, 83rd Annual Report, June 2013), and the amplifying effects of high public debt following a
financial crisis (see eg O Jorda, M Schularick and A Taylor, “Sovereigns versus banks: credit, crises,
and consequences”, Journal of the European Economic Association, vol 14, no 1, February 2016).

II. Monetary policy: a narrow normalisation path
After the long period of ample and unconventional monetary accommodation that
helped economies recover from the Great Financial Crisis (GFC), the incipient policy
normalisation in the major advanced economies stands out in important respects. It
involves normalising both policy rates and balance sheets; it is highly asynchronous,
with the Federal Reserve raising policy rates while the ECB and the Bank of Japan
continue with large-scale asset purchases and negative rates; and it takes place
against a macro-financial landscape still marked by the preceding era of historically
low interest rates. As a result, central banks face tough challenges ahead.
As an example of the special challenges confronting central banks, domestic
and global financial conditions have not tightened for most of the period since the
United States started to normalise its monetary policy. While conditions would
probably have been even easier had the authorities not acted, the development
nonetheless raises questions about policy transmission. Several factors may have
been at work. The improved economic outlook and short-term fiscal stimulus may
have boosted asset prices. Continued asset purchases by other major central banks
may have partly offset the effects of US policy normalisation. And the gradual and
predictable nature of this normalisation may itself have played a role. Only well into
the second quarter of 2018 were there signs that a significant change could be in
the offing, especially for emerging market economies (EMEs).
This highlights the delicate balance central banks must strike. On the one hand,
moving too slowly could give rise to overheating and financial stability risks. On the
other hand, moving too fast could trigger disruptive market reactions and harm the
economic recovery, not least as global debt levels relative to GDP have continued
to increase and financial market valuations appear stretched. The task is further
complicated by uncertainties about the strength of transmission, the macroeconomic
backdrop, the level of “equilibrium” interest rates, the impact of adjustments in
central bank balance sheets and, above all, the limited room for manoeuvre to
address any future economic downturn.
After taking stock of the global monetary policy landscape, this chapter homes
in on the experience of the central bank that is furthest along the normalisation
path – the Federal Reserve. It compares the current US policy tightening with
previous ones, documenting its special character. The chapter closes with a
discussion of the key policy challenges faced by central banks.
Monetary policy normalisation: where do we stand?
Monetary policy normalisation in the major advanced economies made uneven

progress in the period under review, reflecting different states of recovery from the
GFC. In the United States, policy rate normalisation gathered pace with three
additional hikes of the federal funds target range, yielding an increase in the
effective federal funds rate of about 80 basis points between June 2017 and May
2018 to 1.7% (Graph II.1, left-hand panel). Despite this, as of April 2018 the rate was
still negative in inflation-adjusted (real) terms (centre panel). At the time of writing,
the Federal Open Market Committee (FOMC) expected that economic conditions
would warrant further gradual increases, so that the rate would be likely to remain
below its expected longer-run level for some time. The expected pace of policy rate

An asynchronous policy normalisation Graph II.1
Short-term nominal rates1 Short-term real rates3 Central bank total assets
Per cent Per cent Percentage of GDP4
5 3 90
4 2 75
3 1 60
2 0 45
1 –1 30
0 –2 15
–3 0
2008 2011 2014 2017 2020 2008 2011 2014 2017 2020 2009 2012 2015 2018
US: EA: JP: US: EA: JP:
Actual: US EA JP
Expected:
2
Median SEP (Mar 2018):
1
For actual: effective federal funds rate (US); EONIA (EA); one-month OIS rate (JP); monthly averages. For expected: OIS forward rates.
As of 25 May 2018. 2 Summary of Economic Projections (SEP) of the US Federal Reserve Board members and US Federal Reserve Bank
presidents. 3 Nominal rate less core inflation. For core inflation: price index for personal consumption expenditures excluding food and
energy (US); HICP all items excluding food and energy (EA); CPI all items excluding fresh food and energy (JP). For expected core inflation:
SEP of the US Federal Reserve Board members and US Federal Reserve Bank presidents, March 2018 (US); ECB staff macroeconomic projections
for the euro area, March 2018 (EA); Bank of Japan, Outlook for Economic Activity and Prices (CPI excluding fresh food), April 2018 (JP). For
Japan, core inflation is adjusted for the consumption tax hike, and fiscal year forecasts are linearly interpolated to obtain calendar year
figures. 4 For the last period, latest available GDP.
Sources: Bloomberg; national data; BIS calculations.
normalisation quickened in the second half of 2017 against the backdrop of

strengthening labour market conditions, but remained very gradual. As of late May
2018, forward curves implied a federal funds rate at 2.6% by end-2020, while the
FOMC participants’ March projections were somewhat higher, at 3.4% (median
projection). Thus, the real federal funds rate was expected to rise gradually to only
0.5–1.3% by 2020. In October 2017, the Federal Reserve also began to unwind its
asset holdings by capping reinvestments. Thus, its balance sheet also shrank in
absolute terms, after having declined relative to GDP since 2014 (right-hand panel).
By contrast, central banks in the euro area and Japan continued their large-
scale asset purchases and their negative rate policies. The ECB took a further step
towards normalisation by halving its monthly net asset purchases to €30 billion
from January, but reiterated its commitment to keep rates at prevailing levels well
past the end of the purchases, which would proceed at least until end-September
this year. The Bank of Japan continued its Quantitative and Qualitative Easing with
Yield Curve Control programme. The two major components of the programme are
yield curve control, consisting of a negative short-term policy interest rate and a
near 0% target for 10-year Japanese government bond yields, and a commitment
to overshoot the 2% inflation target. In April this year, the Bank of Japan clarified
that monetary policy would not be tied to a specific time frame for meeting the
inflation target.
As of late May 2018, euro area and Japanese short-term interest rates were
expected to rise only gradually, if at all, in the years ahead. In the euro area, market-
implied short-term rates rose above zero only from 2020; in Japan, there is little
sign of a meaningful increase any time soon (Graph II.1, left-hand panel). Thus, in
real terms, money market rates were negative in both economies during the past
Page - 1 -
Low policy rates and large central bank balance sheets worldwide Graph II.2
Nominal policy rates Real policy rates1 Central bank balance sheets and
foreign exchange reserves3
Per cent Per cent Per cent Per cent Percentage of GDP4
4.5 7.5 3.0 4 40
3.0 5.0 1.5 2 30
1.5 2.5 0.0 0 20
0.0 0.0 –1.5 –2 10
–1.5 Other AEs (lhs) EMEs (rhs) –2.5 –3.0 Other AEs (lhs) EMEs (rhs) –4 0
Dec May May Dec May May Dec May Apr
2
Dec May Apr
2
2009 2012 2015 2018
06 17 18 06 17 18 06 17 18 06 17 18
Other AEs: EMEs:
Mean Median Interquartile range Central bank assets
Total FX reserves
1
Nominal policy rate less core inflation; if not available, headline inflation. 2
Or latest available. 3
Simple averages across
economies. 4 For the last period, latest available GDP.
Sources: IMF; International Financial Statistics; CEIC; Datastream; national data; BIS policy rate statistics; BIS calculations.
year and, at the time of writing, were not expected to enter positive territory in the
foreseeable future (centre panel). At the same time, the ECB’s and the Bank of
Japan’s balance sheets expanded further, albeit at a slowing pace. By April 2018,
assets at the ECB and the Bank of Japan stood at more than 40% and close to 100%
of GDP, respectively (right-hand panel). Reflecting the mix of negative interest rate
policies and large-scale asset purchases, respectively about 40% and more than
50% of euro area and Japanese government bonds traded at negative yields in late
May 2018.
In most other advanced economies, policy rates changed little during the year,
remaining well below pre-crisis levels (Graph II.2, left-hand panel). Most held their
policy rates constant and maintained an accommodative policy stance as inflation
remained low, including Australia, New Zealand and Norway; in the case of
Denmark, Sweden and Switzerland, rates were kept negative. On the other hand,
Canada raised its policy rate by 75 basis points from mid-2017, while the United
Kingdom increased its base rate in November 2017 back to its pre-Brexit vote
level. In real terms, policy rates in the other advanced economies remained
negative across the board (centre panel). Central banks’ balance sheets in those
economies changed little and stood at 30% of GDP on average in April 2018 (right-
hand panel).
In EMEs, policy rates also barely changed on balance in the period under
review (Graph II.2, left-hand panel). The People’s Bank of China continued to signal
a neutral monetary policy stance and kept its key lending and deposit rates
unchanged. The Reserve Bank of India too aimed at a neutral stance of monetary
policy, with a 25 basis point cut in policy rates in August last year and subsequently
unchanged rates through May 2018. In some cases, subdued inflation has led to
more significant rate cuts as central banks extended policy accommodation (Brazil
and South Africa) or sped up a transition to a neutral policy stance (Russia).
Mexico tightened its policy rate to curb inflation risk as its currency depreciated,
petrol prices were liberalised, and uncertainty rose about its trade relations with
Page - 2 -
the United States. In real terms, EME policy rates stayed on average slightly above
zero (centre panel). Central bank balance sheets remained stable vis-à-vis GDP,
standing on average above 40% in April 2018 and reflecting mainly large FX
reserve holdings.
Starting in April 2018, some countries came under pressure as their currencies
depreciated and capital flows reversed. While largely triggered by idiosyncratic
developments, this also reflected a broader change in investor sentiment, linked to
an appreciating US dollar and rising US interest rates (Chapter I). In particular,
Argentina hiked its main interest rate by a total of 12.75 percentage points in April
and May, to 40%. Also in May, Turkey raised its late liquidity window rate by
3 percentage points, to 16.5%, to stem outflows. Both countries stepped up foreign
exchange intervention, and Argentina applied for an IMF programme. Indonesia
raised interest rates twice in May, totalling 50 basis points and reversing the rate
cuts of the third quarter of 2017, to stabilise the exchange rate.
The current backdrop for monetary policy normalisation is unprecedented in a
number of important respects. Historically, interest rates in advanced economies,
real and nominal, have never stayed this low for this long and central bank balance
sheets have never swelled as large in peacetime. The long spell of multi-pronged
policy accommodation may have left lasting marks on the macro-financial
landscape, making policy effects harder to assess. Meanwhile, a broad-based
economic recovery, with several countries close to or even beyond standard
measures of full employment, coincides with subdued inflation in many jurisdictions
(Chapter I). And debt levels in relation to GDP stand near historical highs.
Monetary and financial conditions: imperfect transmission?
One notable development that may be partly linked to this unprecedented picture
concerns the relationship between monetary policy and financial conditions. A
tightening of monetary policy would normally coincide with a tightening of
financial conditions. Short- and long-term capital market rates would be expected
to rise, risk spreads to widen, asset price increases to at least slow down and the
domestic currency to appreciate whenever interest rate differentials widened. A
tightening in major economies would further be expected to be propagated
globally, working through investor portfolio decisions and changes in risk-taking.
Insofar as financial conditions are a key transmission channel for monetary policy,
any weak link raises questions about the effectiveness of policy measures. And
these conditions may also complicate policy by raising the risk of undesirable
market disruptions further down the road if they induce or reflect higher risk-taking
(Chapter I).
In fact, until at least the first quarter of 2018, no tightening of financial conditions
accompanied the normalisation of US monetary policy; it was only well into the
second quarter that any appreciable tightening was seen, particularly in EMEs (see
also Chapter I). From December 2015, when the United States started tightening,
until late May of this year, two-year US Treasury yields rose in line with higher policy
rates, by more than 150 basis points (Graph II.3). But the yield on the 10-year Treasury
note increased by only around 70 basis points, while very long-term yields traded
sideways. Importantly, the S&P 500 surged by over 30%, and corporate credit spreads
narrowed, in the high-yield segment by more than 250 basis points. The Federal
Reserve Bank of Chicago’s National Financial Conditions Index (NFCI) trended down
to a 24-year trough last year before rebounding slightly this year, in line with several
other financial condition gauges. The dollar appreciated slightly, but this reflected
mainly a reversal from late April that undid its previous depreciation. This reversal

A tightening paradox?
Changes during US monetary policy tightening episodes 1 Graph II.3
US policy rate US two-year bond yield US 10-year bond yield US 30-year bond yield
Percentage points Percentage points Percentage points Percentage points
4 2.4 1.6 1.5
3 1.8 1.2 1.0
2 1.2 0.8 0.5
1 0.6 0.4 0.0
0 0.0 0.0 –0.5
S&P 500 US HY spread2 US IG spread3 US NFCI4

Per cent Basis points Basis points Index points
30 0 0 0.2
24 –60 –10 0.1
18 –120 –20 0.0
12 –180 –30 –0.1
6 –240 –40 –0.2
0 –300 –50 –0.3
US effective exchange rate5 EME foreign currency EME local currency spread7 Flows into EME portfolio
spread6 funds8
Per cent Basis points Basis points USD bn
2 750 NA 0 200
0 500 –20 160
–2 250 –40 120
–4 0 –60 80
–6 –250 –80 40
–8 –500 –100 NA 0
Tightening episode: 1994–95 2004–06 Current

1
Tightening episodes are February 1994−February 1995, June 2004−June 2006 and the current tightening episode (starting from December
2015). The reference periods are, respectively, the months preceding the first rate hike. End-of-month data. As of 25 May 2018. 2 High-yield
(HY) option-adjusted spread. 3 Investment grade (IG) option-adjusted spread. 4 Federal Reserve Bank of Chicago’s National Financial
Conditions Index; lower values indicate looser financial conditions. 5 Federal Reserve broad nominal effective exchange rate index; higher
values indicate a stronger US dollar. 6 JPMorgan EMBI+ stripped spread, EMBI stripped spread before December 1997. 7 Spread of
JPMorgan GBI-EM Broad Diversified composite index over US 10-year sovereign yields. 8 Total net bond and equity flows to EMEs.
Sources: Barclays; Bloomberg; Datastream; EPFR; JPMorgan Chase; national data; BIS policy rate statistics; BIS calculations.
Page - 3 -
went hand in hand with a significant tightening in EME financial conditions
(Chapter I). That said, by late May, EME local currency bond spreads were still
90 basis points below their end-November 2015 levels, and cumulative net flows into
EME portfolio funds over this period amounted to more than $200 billion.
Qualitatively, the current tightening cycle has some similarities with its
counterpart in the mid-2000s. At that time, policy rate hikes of more than 400 basis
points coincided with only marginal increases (or even declines) in long-term
government bond yields – Federal Reserve Chairman Alan Greenspan’s famous
“conundrum”. Stock markets also rose and US credit spreads narrowed, albeit by
less than during the current tightening. The NFCI did at least register a small
increase back then. Also, the US dollar fell by more than 6%, while EME spreads
narrowed and portfolio flows rose.
These two episodes contrast markedly with the tightening of 1994–95, when
the Fed’s actions triggered sharply higher long-term yields, somewhat wider US
credit spreads and a tightening of overall US financial conditions, as captured by
the NFCI index. Back then, the dollar appreciated, and EME spreads widened
significantly on the back of large EME currency depreciations.
There are several possible reasons for monetary policy’s limited impact on
financial conditions. These include factors unrelated to the policy itself, large and
growing central bank balance sheets outside the United States, and possibly the
gradual and predictable nature of the normalisation. Consider each in turn.
The improved macroeconomic backdrop and outlook, potentially further
boosted in the near term by the prospect of fiscal expansion, could have
counteracted the effects of monetary policy tightening. Both in the United States
and globally, the growth outlook has strengthened considerably over the past year,
while inflation has remained subdued. In particular, during the current tightening
cycle, economic momentum, reflected in the change in real GDP growth and in
business sentiment, increased both in the United States and globally, while it tended
Factors offsetting monetary tightening Graph II.4
Strong economic momentum1 Political factors

Percentage points Index points Index Index
2 3
Real GDP growth (lhs) PMI (rhs)
1.0 8 –0.9 2,800
0.5 4 –0.8 2,600
0.0 NA 0 –0.7 2,400
–0.5 –4 –0.6 2,200
–1.0 –8 –0.5 2,000
–1.5 –12 –0.4 1,800

US Rest of the US Rest of the 2016 2017 2018
4
world world5 7
Financial Conditions Index (reversed, lhs)
Tightening episode: 1994–95 2004–06 Current6 S&P 500 (rhs)
The vertical line in the right-hand panel indicates 8 November 2016 (US presidential election).
1
Changes in the respective variables during US tightening episodes. 2 Based on quarterly data. 3 Manufacturing sector (if not available,
whole economy). 4 Simple averages across 10 AEs and 16 EMEs based on data availability. 5 Simple averages across nine AEs and
11 EMEs based on data availability. 6 For real GDP growth: up to Q1 2018 for the US; Q4 2017 for the rest of the world. For PMI: up to April
2018. 7 Federal Reserve Bank of Chicago’s National Financial Conditions Index; lower values indicate looser financial conditions.
Sources: Datastream; IHS Markit; national data; BIS calculations.

to slow during previous tightening episodes (Graph II.4, left-hand panel). Stronger
growth has probably translated into lower perceived default risk and higher
expected dividends, at least in the near term. This could have reduced credit
spreads and boosted asset prices. In addition, political factors could have boosted
financial market sentiment. Specifically, the outcome of the US presidential election
in November 2016 may have buoyed expectations for business- and growth-
friendly policies, thereby raising projected corporate profits and hence asset prices.
Indeed, the election ushered in a steep increase in stock prices and an easing of
broader financial conditions (right-hand panel).
The large-scale asset purchase programmes of the major central banks outside
the United States may have offset the impact of the Fed’s monetary policy
normalisation. While the reduction in the Fed’s balance sheet would be expected to
raise the US term premium, continued large-scale asset purchases in Europe and
Japan may have spilled over across borders and compressed it, as investors turned
to higher-yielding US securities. Indeed, foreign holdings of US debt securities have
increased as significantly during the current tightening as they did during the 2004
conundrum episode (Graph II.5, left-hand panel). By contrast, in 1994, foreign
holdings barely rose.
The co-movements of US, euro area and Japanese bond yields, and of their
term premia, support this notion (centre panel). Indeed, time variation in transatlantic
and trans-Pacific interest rate spillovers can be linked to major changes in monetary
policy (right-hand panel). Between 2014 and 2016, when the ECB and the Bank of
Japan, respectively, launched and expanded their asset purchase programmes and
introduced negative policy rates, movements in German and Japanese government
bond yields explained about 40% of those in Treasury yields. Spillovers weakened in
late 2016 but have risen again since late 2017.
Central bank asset purchases weigh on long-term interest rates Graph II.5
Change in foreign holdings of US debt Nominal yields and term premia2 Transatlantic and trans-Pacific
securities1 spillovers to US yields3
USD bn Per cent Per cent
1,250 3 0.6
1,000 2 0.5
750 1 0.4
500 0 0.3
250 –1 0.2
0 –2 0.1
1994–95 2004–06 Current 2013 2014 2015 2016 2017 2018 2013 2014 2015 2016 2017 2018
US: EA: JP:
Yield:
Term premium:
1
Changes during US tightening episodes. 2 Based on 10-year government zero coupon bond yields; see P Hördahl and O Tristani, “Inflation
risk premia in the euro area and the United States”, International Journal of Central Banking, vol 10, September 2014. Euro area is represented
by France. 3 Spillovers from German and Japanese 10-year government bond yields to US 10-year Treasury yield. Estimated following
F Diebold and K Yilmaz, “Measuring financial asset return and volatility spillovers, with application to global equity markets”, Economic Journal,
vol 119, no 534, January 2009. Contributions are calculated from the forecast error variance matrix inferred from generalised identification of
shocks.
Sources: Federal Reserve Financial Accounts of the United States; Bloomberg; Datastream; national data; BIS calculations.

Finally, the gradualism and predictability of the tightening may also have played
a role. Gradualism is especially called for when there is high uncertainty about the
economic context and monetary transmission, as currently.1 In such a situation, this
can help avoid undesirable financial and economic responses. Yet a high degree of
gradualism and predictability may also dilute the impact of policy tightening. More
gradual hiking paths correspond to a flatter trajectory of expected future short rates
while a high degree of predictability implies low uncertainty about that trajectory,
reducing risk premia. Through these effects, gradualism and predictability could
induce search-for-yield and risk-taking behaviour, further compressing risk premia
and boosting asset prices.2 Moreover, market participants could interpret gradualism
and predictability as signalling that central banks wish to prevent sharp market
moves, thereby providing implicit insurance for risky position-taking.3
Since December 2015, the Federal Reserve has been normalising its monetary
policy very gradually and predictably. The average monthly pace of policy rate
increases was just 5 basis points as of late May 2018, compared with a respective
20‑plus and 15-plus basis points during the tightenings of the mid-1990s and mid-
2000s (Graph II.6, left-hand panel). At the same time, the surprise element of policy
rate changes was generally small. Short-term market interest rates changed on
decision days by less than 2 basis points on average, similar to the impact of rate
hikes in 2004, but much less than in 1994 when surprises tended to exceed 10 basis
points. The surprise in medium- and long-term Treasury yields was somewhat
higher than in the 2000s, possibly because of greater reliance on forward guidance
and the additional effect coming from balance sheet normalisation, but still only
around one half of that during the mid-1990s (left-hand panel). Similarly, the
reduction in the Federal Reserve’s asset holdings has followed a preannounced
schedule with moderate reinvestment caps. Consistently with this overall picture,
the volatility of policy rate futures and implied bond market volatility have eased
since the beginning of policy rate normalisation (right-hand panel).
The current tightening has been highly gradual and predictable Graph II.6
Policy gradualism and predictability1 Volatilities

Basis points Percentage points Basis points
18 0.8 100
12 0.6 80
6 0.4 60
0 0.2 40
2
Pace of MP shock: 2015 2016 2017 2018
tightening1 1 month 1 year 3 years 10 years
Fed funds future (lhs)3 MOVE index (rhs)4
Tightening episode: 1994–95 2004–06 Current
The vertical lines in the right-hand panel indicate 16 December 2015 (first rate hike) and 14 December 2016 (second rate hike).
1
Average monthly changes in the US policy rate. 2 Average absolute changes in key interest rates on FOMC meeting dates. For one-month
and one-year maturities, based on OIS and Libor rates; for three-year and 10-year maturities, based on US Treasury yields. 3 Annualised
standard deviation of the daily price change in 12th generic futures contracts over the 90 most recent trading days. 4 Merrill Lynch Option
Volatility Estimate.
Sources: Bloomberg; Datastream; BIS calculations.

Policy challenges
The weak effect of monetary policy tightening on financial conditions is just one
example of the challenges central banks face in the normalisation process. Given
the unprecedented starting conditions and the large array of instruments in use,
there is considerable uncertainty surrounding the transmission mechanism and
policy benchmarks.
Take the impact of interest rate changes. On the one hand, there is evidence
that the link between short rates and long rates has weakened over yearly horizons
since the early 2000s.4 This suggests that monetary policy may have to act more
strongly to achieve a given effect. On the other hand, there is also evidence that
long-term interest rates have become more sensitive to monetary policy surprises at
higher frequencies (daily and intradaily) over the same period.5 This points to the
risk of a snapback in long-term rates should policy be tightened more than expected.
Similar questions concern the impact of balance sheet normalisation on long-term
rates. Estimates are very imprecise and vary widely. For example, a Federal Reserve
study found that the announced balance sheet run-off plan would increase the 10-year
Treasury term premium by around 15 basis points in the year 2018, but with a statistical
uncertainty range around the level of the term premium of as much as 70 basis points.6
Questions also relate to the end-point for interest rates – sometimes referred
to as the natural or equilibrium rate. This is conventionally defined as the real
interest rate consistent with output at potential and inflation at target. Most
estimates point to a notable decline in the natural rate over the past few decades,
with a further drop post-crisis, but the ranges are very wide (Graph II.7). This decline
has been linked to real developments that reduce investment and raise saving, such
as demographic shifts and slowing potential growth. At the same time, while this
notion draws strength from studies that focus on the past three decades, some
recent work based on longer historical data finds less support for this hypothesis,
including by pointing to a role for monetary policy (Box II.A).
Against this backdrop, central banks must strike a delicate balance in
determining the timing and pace of any normalisation.
On the one hand, there is a risk of moving too early and too rapidly. The
upswing may prove fragile, given the uncertainty over how financial markets and
the economy might respond after the long period of ultra-low rates. Too fast an
increase in interest rates may trigger an abrupt repricing in financial markets if it
prompts an outsize revision in the expected level of risk-free interest rates or a
decompression in risk premia. Such a snapback could be amplified by market
dynamics (Chapter III) and possibly have adverse macroeconomic consequences
(Chapter I and below). It could compromise the economic recovery or spill across
borders in the case of international funding currencies, with broader repercussions.
A particular concern is that the macroeconomic impact of tighter monetary policy
could turn out to be larger than in the past, since debt has continued to rise
globally as interest rates have sagged (Graph II.8, left-hand panel). There is evidence
that the impact of monetary policy on the economy is significantly larger when
debt is high, reflecting in part a much higher short-term impact on debt service
ratios (right-hand panel).7
Other considerations too would support a very patient strategy. By testing how far
the expansion can be accommodated, central banks may partly reverse some of the
crisis-induced loss in production potential.8 This could entice discouraged workers back
into the labour force and boost investment and productivity. Such a strategy would
also allow central banks to test the true extent of slack in the economy, premised on
the view that inflation reliably signals excess capacity. Indeed, it is common practice to
adjust measures of full employment and potential output according to the behaviour

The natural interest rate is measured with considerable uncertainty
In per cent Graph II.7
US natural rate measures Euro area natural rate measures
4 6
3 4
2 2
1 0
0 –2
–1 –4
–2 –6
01 03 05 07 09 11 13 15 17 01 03 05 07 09 11 13 15 17
1
Holston et al Standard error band Holston et al Standard error band1
Lubik and Matthes Real 5-yr 5-yr forward rate Fries et al Real 5-yr 5-yr forward rate3
Johanssen and Mertens FOMC long-run expectations2
1
One standard error bands around natural rate estimates of Holston et al (2016), based on sample averages. 2 Longer-run median
projection from the SEP for the federal funds rate less 2% inflation target. 3 Based on French government bond yields, supplemented by
German government bond yields to interpolate missing data.
Sources: S Fries, J Mésonnier, S Mouabbi and J Renne, “National natural rates of interest and the single monetary policy in the euro area”,
Bank of France, Working Papers, no 611, October 2017; K Holston, T Laubach and J Williams, “Measuring the natural rate of interest:
international trends and determinants”, Federal Reserve Bank of San Francisco, Working Papers, November 2016; B Johannsen and E Mertens,
“A time series model of interest rates with the effective lower bound”, BIS Working Papers, no 715, April 2018; T Lubik and C Matthes,
“Calculating the natural rate of interest: a comparison of two alternative approaches”, Federal Reserve Bank of Richmond, Economic Brief,
October 2015; Bloomberg; national data; BIS calculations.
of inflation, raising them if inflation fails to increase. Moreover, as long as inflation is

not very responsive to demand and expectations remain well anchored – a flattening
of the Phillips curve – the risk of a significant overshoot would be low, allowing the
central bank to remain patient. Indeed, concerns with a de-anchoring of expectations
and the associated loss of credibility have been a key motive for central banks’ efforts
to push inflation towards target and prevent deflation.
On the other hand, running a high-pressure economy raises risks too. The
possibility of a stronger than expected rise in inflation cannot be ruled out
(Chapter I). And given the hypersensitivity of overstretched financial markets, any
disproportionate reaction could potentially damage the economy. The sharp
stock market correction in response to slightly higher than expected wage
increases in the United States underlines this risk. In fact, postponing and/or
slowing down normalisation could further encourage risk-taking, paradoxically
amplifying the likelihood of such a market response. Moreover, one might
conjecture that a central bank risks a larger loss of credibility from overshooting
the inflation target than from undershooting it. After all, inflation targeting was
adopted to fight high inflation, and political economy pressures generally tend to
push for an easing bias.
Even if inflation does not loom, risk would not disappear. Since the mid-1980s,
unsustainable economic expansions appear to have manifested themselves mainly
in the shape of unsustainable increases in debt and asset prices (Chapter I).9 Thus,
even in the absence of any near-term market disruptions, keeping interest rates too
low for too long could raise financial and macroeconomic risks further down the
road. In particular, there are reasons to believe that the downward trend in real
rates and the upward trend in debt over the past two decades are related and even
Page - 7 -
Higher debt raises vulnerabilities Graph II.8
Interest rates sank as debt soared Peak impacts of 100 basis point policy rate increase4
Per cent Percentage of GDP Percentage points Per cent
Lhs Rhs
3.0 250 0.3 1.5
1.5 225 0.1 0.5
0.0 200 –0.1 –0.5
–1.5 175 –0.3 –1.5
–3.0 150 –0.5 –2.5

2000 2003 2006 2009 2012 2015 2018 Short-term Long-term Real GDP Real house Real
Debt service ratio GDP deflator price credit
Lhs: Rhs:
Real policy rates1 Debt3 High-debt countries Low-debt countries
2
Long-term real rates
1
Nominal rate less headline consumer price inflation. Simple average of DE, JP and US. 2 Simple average of index-linked 10-year
government bond yields of FR, JP and US. 3 Total credit to non-financial sectors. Weighted average of the G7 economies plus China based
on GDP and PPP exchange rates. 4 Peak impacts from impulse responses to a 100 bp shock to the policy rate based on mean group panel
VAR estimations using quarterly data for 18 AEs and EMEs. For details, see B Hofmann and G Peersman, “Is there a debt service channel of
monetary transmission?”, BIS Quarterly Review, December 2017.
Sources: Bloomberg; Datastream; national data; BIS calculations.
mutually reinforcing. True, lower equilibrium interest rates may have increased the
sustainable level of debt. But, by reducing the cost of credit, they also actively
encourage debt accumulation. In turn, high debt levels make it harder to raise
interest rates, as asset markets and the economy become more interest rate-
sensitive – a kind of “debt trap” (Graph II.8, right-hand panel).10
A further complication in calibrating normalisation relates to the need to build
policy buffers for the next downturn. Indeed, the room for policy manoeuvre is
much narrower than it was before the crisis: policy rates are substantially lower
and balance sheets much larger. While some central banks have shown that
interest rates can be lowered below zero, this is probably possible only to a limited
extent. And while central banks have field-tested unconventional tools in the wake
of the crisis, their side effects set limits on how far they can be used. Hence, all else
equal, if room for manoeuvre is valuable, it would make sense to adjust the
normalisation trajectory to expand it. How far this is the case depends on the
perceived likelihood of a downturn occurring before normalisation is complete, on
the perceived impact of low rates on debt accumulation and on the perceived
costs of raising rates.
The policy normalisation of major central banks will also affect EMEs and other
advanced economies through spillovers. Specifically, as a result of global investor
arbitrage, there is a strong positive link between government bond yields in the
core advanced economies and those in EMEs and other advanced economies
(Graph II.9, left-hand panel). An increase in the VIX, a gauge of investor risk appetite,
precedes a significant increase in EME yields and a slight decrease in yields in other
advanced economies, probably reflecting safe haven flows (centre panel). More
importantly, US dollar appreciation, working through foreign currency borrowing
and global investor balance sheets, coincides with portfolio outflows from EMEs,
pushing up bond yields there. Together with lower bond yields in the other advanced
economies, this probably again reflects a flight to safety (right-hand panel).11
Page - 8 -
All this amplifies changes in financial conditions globally. During phases in which
interest rates remain low in the main international funding currencies, especially
the US dollar, EMEs in particular tend to benefit from easy financial conditions.
These effects then play out in reverse once interest rates rise. A reversal could occur,
for instance, if bond yields snapped back in core advanced economies, and
especially if this went hand in hand with a rise in stock market volatility and a US
dollar appreciation, as EME borrowers sought to hedge their positions and capital
inflows turned into outflows. A clear case in point is the change in financial
conditions experienced by EMEs since the US dollar started appreciating in the first
quarter of 2018.
Such spillovers have posed a major challenge for central banks in EMEs and
other advanced economies in the past, and will continue to do so in the future. On
the one hand, a further prolongation of easy global financial conditions would
worsen the policy trade-offs for economies that face concerns about appreciating
currencies and the build-up of domestic financial imbalances. In small open
advanced economies that do not rely on foreign currency borrowing and where
inflation is already below target, any easing of domestic monetary policy to prevent
excessive domestic currency appreciation would tend to encourage the further
build-up of financial imbalances. For instance, in Switzerland interest rates have
been negative and inflation very subdued for quite some time while a boom in the
mortgage market has been raising concerns among the authorities. In EMEs that
rely heavily on foreign currency debt, the room for policy manoeuvre is even
narrower. This is because financial conditions in that debt segment depend directly
on the monetary policy of the country issuing the currency of denomination. In
addition, if inflation is above target or the build-up of domestic financial imbalances
is a concern, tightening monetary policy is less effective. The tightening promotes a
Global spillovers
Impulse response of five-year sovereign yields, in basis points1 Graph II.9
100 basis point increase in base 1% increase in the VIX 1% appreciation of the US dollar
currency bond yields2
60 0.3 6
45 0.2 3
30 0.1 0
15 0.0 –3
0 –0.1 –6
5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30
Horizon (days)
Other AEs: EMEs:
Cumulative impact
90% confidence interval
1
Cumulative impact on five-year sovereign yields estimated by fixed effects panel local projections using daily data. The set of control
variables includes the lagged dependent variable and the change in domestic three-month money market rates. 2 For CH, CZ, DK, HU, NO,
PL and SE, the base currency is the euro; for AU, BR, CA, CL, CN, CO, GB, HK, ID, IL, IN, KR, MX, MY, NZ, PH, RU, SG, TH, TR and ZA, the US
dollar.
Sources: ECB; Datastream; BIS calculations.

currency appreciation which, by reducing the foreign currency debt burden, can
further ease domestic financial conditions. Borrowers are induced to borrow more
and lenders to lend more, given the borrowers’ apparent gain in creditworthiness.
To address these trade-offs, the authorities can usefully broaden the set of
instruments deployed – an increasingly common strategy. Macroprudential
measures can address emerging vulnerabilities in a targeted manner. While
extremely useful, the evidence indicates that such measures are more effective in
strengthening the resilience of the financial system than in preventing the build-up
of financial imbalances themselves (Chapter IV). Similarly, FX intervention can offset
some of the undesirable appreciation of the currency while building up a buffer
that could be drawn down when conditions go into reverse.
When conditions do go into reverse, policy trade-offs become especially
difficult, particularly in EMEs. Even if such a reversal is needed to limit the further
build-up of financial imbalances, it could expose financial vulnerabilities in some
countries, especially if it plays out in an abrupt and disorderly fashion. This could
lead to powerful contractionary pressures alongside currency depreciation and
hence, at least in the short run, to higher inflation. Scope for easing monetary
conditions would be severely constrained; in fact, policy has often had to be
tightened to prevent an uncontrolled depreciation. While FX reserves can be drawn
down, experience indicates that their deployability may be smaller than their size
may suggest, as markets can become nervous once the buffer shrinks. And the
evidence suggests that macroprudential tools are more effective in building up
buffers than in cushioning financial busts (Chapter IV).
Global spillovers can also have implications for the core countries at their
origin. The collective size of the countries exposed to the spillovers suggests that
what happens there could also have significant financial and macroeconomic
effects in the economies that originated the spillovers. At a minimum, such
spillbacks argue for enlightened self-interest in the core economies, consistent with
domestic mandates.12 This is an additional policy dimension that complicates the
calibration of the normalisation and which deserves close attention.
To conclude, the normalisation path is a narrow one. Treading it will require a
lot of judgment to evaluate trade-offs as well as a dose of pragmatism in adjusting
to changing conditions. It will also call for flexibility in pursuing inflation objectives.
In particular, since unsustainable expansions may manifest themselves in growing
financial imbalances rather than rising inflation, and since accumulated debt may
greatly constrain future room for manoeuvre, debt and asset price dynamics require
close monitoring and should be factored into policy decisions. The same applies to
the need to regain policy room for manoeuvre, which has narrowed significantly
since the GFC. Given the starting conditions, this journey is bound to be bumpy.
Financial market ructions will no doubt occur, in the manner of withdrawal
symptoms. But as long as financial market disturbances remain contained, central
banks should have no reason to adjust the normalisation pace. Volatility as such is
not a problem. In fact, to the extent that it inhibits unbridled risk-taking, it is healthy
and part of the solution. The challenge will be to normalise with a steady hand,
without overreacting to any transient bouts of volatility.

Box II.A
The decline in natural real interest rates: what do we know?
The global decline in real interest rates in recent decades is often attributed to a lower level of natural real interest
rates, defined as the level that equates desired real saving to investment at full employment. Several factors may
have lowered investment and raised saving over the past few decades, pushing down natural (or equilibrium) real
interest rates. On the investment side, the most prominent candidates are lower productivity and potential growth,
which may reduce the marginal returns to capital and hence investment. The decline in the relative price of capital
(eg computers), which lowers the required investment outlay, is another potential factor. On the saving side,
demographic developments have been highlighted as prompting increased saving, in particular a rising share of the
working age population and increased life expectancy. As life-cycle theory posits, a lower dependency ratio results
in increased saving as the working population tends to save more than retirees. Similarly, greater longevity prompts
increased saving for a longer expected retirement. Greater income inequality also tends to increase aggregate
saving as higher-income households have a higher propensity to save. Lastly, greater demand for safe assets and
higher risk aversion could lead to lower real risk-free interest rates. Possible reasons include the limited global
supply of safe securities, which has not kept pace with the increased saving demand, including from EMEs, and
greater concerns about macroeconomic tail risks more generally.
The pattern seen over the last few decades lends some support to the relevance of these saving-investment
factors. Even a cursory look at the data suggests that saving-investment factors and the real interest rate share
some common trends. For example, the drop in real rates over the last 30 years has coincided with a decline in
dependency ratios and in productivity growth. In addition, life expectancy has moved up, inequality has increased,
and the relative price of capital has fallen, as the hypothesis would postulate. Pairwise correlation between real
interest rates and these variables is therefore high and consistent with theory over this period (Graph II.A, left-hand
panel). Recent research also shows that structural models can explain much of the observed decline in real rates. For
example, studies that emphasise demographics typically use overlapping-generation models to capture the joint
dynamics between the dependency ratio, life expectancy and population growth. These studies find that demographics
may have lowered real interest rates by between 1 and several percentage points over the past few decades. Rachel
and Smith (2017) use pre-existing elasticity estimates and find that potential growth, demographics, the risk
premium and the relative price of capital are the most important factors, together explaining a 3 percentage point
fall in real interest rates since the 1980s.
Another supporting piece of evidence is the fact that inflation has not increased despite the downward trend
in real interest rates. Assuming a stable Phillips curve, a sustained gap between the real interest rate and its natural
counterpart should exert pressure on aggregate demand, ultimately influencing the inflation dynamics. Relatively
stable inflation suggests that real interest rates have merely tracked the natural rates downwards. Indeed, most
“filtered” estimates of the natural rate have relied on the Phillips curve for identification, with most pointing to its
steady decline over the last 30 years (Graph II.7).
While the consensus is that the natural interest rate may have recently declined, there are also reasons to
be more circumspect, at least in practical policymaking. The filtering-based estimates are associated with a
notoriously large degree of statistical uncertainty, not least because the empirical link between inflation and
economic slack has not always been tight (Graph II.7). Additional challenges arise when allowing for possible non-
linearity of the Phillips curve and structural change in the inflation process. Meanwhile, the structural approach,
which focuses on articulating few specific mechanisms at a time, by construction leaves little room for empirically
evaluating different hypotheses. This in turn makes it harder to assess the outlook for the natural rate, as the
future evolution for saving-investment factors may diverge. Ongoing population ageing could finally reverse the
demographic effects and potential growth could trend higher, while inequality and the shortage of safe assets
may be more persistent forces.
There is also a risk that too much emphasis has been placed on the experience over the last 30 years. The
correlation between real interest rates and saving-investment factors either switches sign or becomes substantially
weaker once one extends the sample to cover longer periods (Graph II.A, left-hand panel). Formal empirical studies
using long data series corroborate this observation. Hamilton et al (2015) find that GDP growth, a key determinant
of the natural rate in macro models, bears little relationship to real interest rates, while Lunsford and West (2017)
consider a comprehensive set of factors in the United States, and find only one demographic variable to be
correlated with real rates. Borio et al (2017) study a large set of factors for 19 advanced economies since the late
19th century, and allow these factors to jointly determine real interest rates across various specifications. They find
that none of the saving-investment factors can consistently explain real interest rate movements. The finding
survives various robustness tests and extensions, including a control for the risk premium.

Monetary and non-monetary determinants of the real interest rate Graph II.A
Correlation between real interest rates and saving- Real interest rate and monetary policy regimes2
investment factors1
Correlation coefficient Per cent
Expected
sign: + + + - - +
0.5 4
0.0 0
–0.5 –4
←Wars→ Last 30 years

–1.0 –8
GDP Productivity Dependency Life Inequality Relative 1876 1896 1916 1936 1956 1976 1996 2016
growth growth ratio expectancy price
of capital Real interest rate Contributions from MP regimes
1985–2017 1870–2017
1
Correlation between cross-country median of real long-term interest rate and saving-investment factors. Cross-country median is based on
19 AEs. From 1991 onwards, the dependency ratio includes EMEs. 2 Real interest rate and contributions from monetary policy regimes are
cross-country medians. Contributions from policy regimes for each country are computed using that country’s policy regimes and saving-
investment factors as inputs, with coefficients estimated from a panel regression. Effects of policy regimes are captured via country- and time-
specific dummies, where seven different regimes are identified. War periods are ignored throughout.
Source: C Borio, P Disyatat, M Juselius and P Rungcharoenkitkul, “Why so low for so long? A long-term view of real interest rates”, BIS Working
Papers, no 685, December 2017.
An alternative hypothesis is that monetary factors may have more persistent effects on real interest rates than
usually assumed. There are several possible channels. Inflation expectations may be pinned down more successfully
under certain policy regimes (eg over the last 30 years and during the gold standard), so that changes in the
nominal interest rate are persistently transmitted to the real rate. There is earlier evidence that breaks in mean real
interest rates coincide with those in inflation, suggesting a systematic role for monetary policy (Rapach and Wohar
(2005)). Also, financial boom-bust cycles may in part be driven by monetary policy, leaving a long-lasting imprint
on the real economy, including on real interest rates. Indeed, Borio et al (2017) find that shifts in monetary policy
regimes matter for the levels of real interest rates, even after accounting for the influence of saving-investment
variables. The right-hand-panel of Graph II.A shows the estimated impact of changes in monetary policy regimes on
real interest rates. For example, the shift from post-Bretton Woods in the 1980s to the current policy regime of
inflation targeting is associated with a 1.3 percentage point reduction in the real interest rate. Trends in real rates
also appear to be affected by such regime changes. The persistent effect of monetary policy regimes on real rates
raises deep questions about the real-only saving-investment framework, further highlighting the practical limitations
of the natural interest rate in policymaking.
 Equivalently, in a canonical macro model, it is defined as the level of the real interest rate that is neither expansionary nor contractionary
for output. A distinction is sometimes made between short- and long-run natural interest rates. The short-run natural rate is influenced by
transitory shocks, such as potential growth or productivity shocks; the long-run natural rate prevails once their effects wane. This rate is
smoother, but may still vary over time owing to permanent shocks and structural breaks in economic relationships.      For a detailed literature
review and the references mentioned in this box, see Borio et al (2017) (for full reference, see source line of Graph II.A).      L Rachel and
T Smith, “Are low real interest rates here to stay?”, International Journal of Central Banking, vol 13, issue 3, September 2017, pp 1–42.    
  J Hamilton, E Harris, J Hatzius and K West, “The equilibrium real funds rate: past, present and future”, IMF Economic Review,
vol 64, issue 4, 2016, pp 660–707; K Lunsford and K West, “Some evidence on secular drivers of US safe real rates”, Federal Reserve Bank of
Cleveland, Working Papers, 17-23, 2017.      For full reference, see source line of Graph II.A.      Borio et al (2017) use higher moments of
GDP growth and inflation as proxies for macroeconomic risk.      D Rapach and M Wohar, “Regime changes in international real interest
rates: are they a monetary phenomenon?”, Journal of Money, Credit and Banking, vol 37, issue 5, 2005 pp 887–906.
BIS Annual Economic Report 2018Page - 10 - 39

Endnotes
1
Gradualism can be described as a policy approach where the central bank “tends to adjust interest
rates incrementally, in a series of small or moderate steps in the same direction” (B Bernanke,
“Gradualism”, remarks at an economics luncheon co-sponsored by the Federal Reserve Bank of San
Francisco and the University of Washington, Seattle, 20 May 2004). One rationale for gradualism is
that a more cautious policy approach is called for when there is high uncertainty about how the
economy responds to changes in the monetary policy stance; see W Brainard, “Uncertainty and the
effectiveness of policy”, American Economic Review, vol 57, 1967, pp 411–25; and B Sack, “Does the
Fed act gradually? A VAR analysis”, Journal of Monetary Economics, 2000, pp 229–56. Another
rationale is that a commitment to act gradually could give the central bank more leverage over
long-term interest rates; see M Woodford, “Optimal interest-rate smoothing”, Review of Economic
Studies, vol 70, 2003, pp 861–86.
2
See C Borio and H Zhu, “Capital regulation, risk-taking and monetary policy: a missing link in the
transmission mechanism?”, Journal of Financial Stability, December 2012, for a comprehensive
discussion of the link between monetary policy and the perception and pricing of risk, ie the risk-
taking channel of monetary policy. See T Adrian and H S Shin, “Financial intermediaries, financial
stability and monetary policy”, in Maintaining stability in a changing financial system, proceedings
of the Federal Reserve Bank of Kansas City Jackson Hole Economic Symposium, August 2008, for
the argument on predictability and gradualism being an enabling factor in the build-up of leverage
before the GFC.
3
The consequence could be a “whisper equilibrium”, where the central bank whispers more and
more in order not to upset markets while market participants lean in to hear better and better. As
markets react more, central banks’ efforts to avoid stirring up the market are partially undone and
the signalling value of financial market prices is impaired. See J Stein, “Challenges for monetary
policy communication”, speech at the Money Marketeers of New York University, 6 May 2014; and
H S Shin, “Can central banks talk too much”, speech at the ECB conference on Communications
challenges for policy effectiveness, accountability and reputation, 14 November 2017, for more
detailed discussions of the whisper equilibrium.
4
See S Hanson, D Lucca and J Wright, “Interest rate conundrums in the twenty-first century”, Federal
Reserve Bank of New York, Staff Reports, no 810, March 2017.
5
See Hanson et al (2017), op cit.
6
See B Bonis, J Ihrig and M Wei, “Projected evolution of the SOMA Portfolio and the 10-year
Treasury term premium effect”, Board of Governors of the Federal Reserve System, FEDS Notes,
September 2017.
7
For a more detailed discussion and empirical analysis of the debt service channel of monetary
transmission, see B Hofmann and G Peersman, “Is there a debt service channel of monetary
transmission?”, BIS Quarterly Review, December 2017, pp 23–37, and the references therein.
8
There is evidence for demand-driven recessions inducing long-lasting effects on output via
hysteresis effects; see O Blanchard, E Cerutti and L Summers, “Inflation and activity – two
explorations and their monetary policy implications”, IMF Working Papers, WP/15/230, 2015; and
R Martin, T Munyan and B Wilson, “Potential output and recessions: are we fooling ourselves?”, Board
of Governors of the Federal Reserve System, International Finance Discussion Papers, no 1145,
2015. The argument for running a high-pressure economy is premised on such a hysteresis effect
working in reverse.
9
The concept of finance-neutral output gaps is one way of incorporating information about
financial imbalances in gauging economic slack. These measures have been shown to outperform
traditional output gap measures as real-time indicators of output sustainability, including in the
run-up to the GFC. See BIS, 86th Annual Report, June 2016; and C Borio, P Disyatat and M Juselius,
“Rethinking potential output: embedding information about the financial cycle”, Oxford Economic
Papers, vol 69, no 3, 2017, pp 655–77.

10
The debt trap refers to a situation in which the debt accumulation that coincides with
accommodative monetary policy makes it progressively harder to raise rates. For a more detailed
discussion of the notion of a debt trap and empirical evidence to that effect, see C Borio and
P Disyatat, “Low interest rates and secular stagnation: is debt a missing link?”, VOX, June 2014; and
M Juselius, C Borio, P Disyatat and M Drehmann, “Monetary policy, the financial cycle, and ultra-
low interest rates”, International Journal of Central Banking, vol 13, no 3, 2017, pp 55–90.
11
For an overview of the mechanisms operating through banking flows and capital market financing,
respectively, see V Bruno and H S Shin, “Global dollar credit and carry trades: a firm-level analysis”,
BIS Working Papers, no 510, August 2015; and B Hofmann, I Shim and H S Shin, “Sovereign yields
and the risk-taking channel of currency appreciation”, BIS Working Papers, no 538, January 2016,
revised May 2017. See also BIS, 85th Annual Report, June 2015, Chapter V, for a discussion of global
spillover effects.
12
See BIS (2015), op cit, for a discussion of the policy implications of global spillover effects.

III. The financial sector: post-crisis adjustment and
pressure points
The Basel III reforms are finalised, completing a key part of the regulatory overhaul
in the wake of the Great Financial Crisis (GFC). Given the favourable near-term
economic outlook (Chapter I) and the prevailing easy financial conditions even as
monetary policies are gradually tightened (Chapter II), the window of opportunity
is wide open – for most banks – to finalise their adjustment to the post-crisis
environment. Substantial progress has already been made, with most banks
meeting the more stringent capital requirements and new liquidity standards. Yet
compressed equity valuations indicate that banks’ efforts to fully reap the benefits
of the reforms and ensure sustainable profitability are not yet complete. Meanwhile,
non-bank intermediaries have been gaining ground, pointing to important
structural trends in financial markets that bear on market dynamics, particularly
under stress. This calls for prompt and consistent implementation of all Basel III
standards, along with tight regulation and supervision of both banks and non-
banks, to guard against risks that may have built up during past years of unusually
low interest rates and compressed volatility.
This chapter starts with a review of the rationale and key elements of the
Basel III reforms, including the final package agreed in December 2017. It then
discusses Basel III implementation and banks’ adjustment to the post-crisis
environment, highlighting areas that warrant attention. The last section examines
changing bank/non-bank interactions and their impact on market dynamics under
stress.
Basel III: key elements of the completed framework
The GFC laid bare the vulnerabilities of the international banking system. Major
banks entered the crisis with excessive, mismeasured levels of leverage and
insufficiently stable funding sources. Crisis-related losses accumulated rapidly,
contagiously spreading across markets and countries, and forcing public sector
intervention. What started as strains in US subprime mortgage markets turned into
a full-blown financial crisis (Graph III.1).
Ten years on, the post-crisis reforms of the regulatory framework for
internationally active banks – Basel III – have been completed.1 In addressing the
previous framework’s weaknesses, the reforms have taken a two-stage approach
(Table III.1). Stage 1, beginning in 2010, focused primarily on raising the size and
quality of banks’ capital buffers, while enhancing the robustness of the existing
risk-weighted capital requirements (RWRs) through new capital and liquidity
constraints. Stage 2 focused on the comparability and reliability of the internal
model-based parts of the RWR framework, which allow banks to calculate their
own risk weights. Most of the Basel III elements will be fully implemented as of
2022. Other reforms, such as minimum requirements for global systemically
important banks’ (G‑SIBs’) total loss-absorbing capacity, enhanced bank resolution
regimes and the central clearing of all standardised derivatives contracts, are being
implemented in parallel.2

Unsustainable risk-taking required decisive post-crisis adjustments by banks Graph III.1
Run-up in leverage results in post- Banks return to more stable funding2 Rapid accumulation of crisis-related
crisis contraction1 losses 3
Ratio Ratio Ratio USD trn
61 2.2 1.2 2.0
47 1.9 0.9 1.5
33 1.6 0.6 1.0
19 1.3 0.3 0.5
5 1.0 0.0 0.0

02 05 08 11 14 17 02 05 08 11 14 17 2008 2009 2010 2011
United States United Kingdom Europe Rhs: United States Worldwide North America
Europe Japan (lhs) United Kingdom Europe
Japan
The vertical lines in the left-hand and centre panels indicate August 2007 (interbank market turmoil in the early stages of the Great Financial
Crisis) and December 2010 (the beginning of stage 1 Basel III reforms).
1
Total assets divided by total equity; asset-weighted averages. Based on a sample of major internationally active banks. 2 Loans made by
the banking sector to the private sector divided by banks’ customer deposits; deposit-weighted averages. 3 Banks’ cumulated losses and
writedowns from Q2 2007 to Q2 2011 (series discontinued in Q2 2011).
Sources: IMF, International Financial Statistics; Bloomberg; S&P Capital IQ; national data; BIS calculations.
Stage 1 reforms: raising bank capital and liquidity buffers

A key concern behind the Basel III stage 1 reforms was insufficient loss-absorbing
capital buffers (Graph III.1, left-hand panel). The Basel I standards had established
minimum capital requirements through ratios that weighted assets by their riskiness
– risk-weighted assets (RWAs): the higher the measured riskiness, the higher the
weight. Then, under Basel II, in order to improve this risk sensitivity, banks had been
given the option – subject to supervisory approval – to set the risk weights themselves
through their own internal risk models, such as the internal ratings-based (IRB)
approach for credit risk. Alternatively, they could apply the simpler risk weights set
by supervisors – under the so-called standardised approaches (SAs).
In response to the crisis, the stage 1 reforms substantially tightened the
definition and quality of bank capital as well as the required minimum capital ratio.
Banks now had to comply with a minimum ratio of 4.5% of RWAs, defined in terms
of a stricter Common Equity Tier 1 (CET1) capital definition, and a 6% Tier 1 capital
ratio. They also had to maintain an additional CET1 capital conservation buffer of
2.5%. The resulting 7%–8.5% minimum ratio compares with what was effectively a
4% baseline under Basel II, based on a much weaker capital definition that included
various instruments with limited loss-absorbing capacity.3 In addition, the computation
of RWAs was broadened, based on revised frameworks for securitisations and
trading book positions (Table III.1).
These enhanced RWRs were complemented with four new requirements
covering risks inadequately addressed in the pre-crisis standards. The resulting
“multiple metrics” setup is intended to increase the framework’s robustness by
guarding more explicitly against the inherent uncertainties of risk management and
measurement.4 First, a simple minimum leverage ratio seeks to contain build-ups of

Basel III phase-in arrangements: key standards1 Table III.1
Adoption Requirement Phase-in from Full implementation

Standard year year year
Stage 1: Capital and liquidity
Capital definition 2010 CET1; deductions 2013 2022
Minimum CET1 ratio 2010 4.5% 2013 2015
Capital conservation buffer 2010 2.5% 2016 2019
Countercyclical buffer 2010 0–2.5% 2016 2019
G-SIB capital surcharge 2010 0–3.5% 2016 2019
Leverage ratio (LR) 2010 3% 2015 (disclosure) 2018
Securitisation framework 2014 Revised framework 2018
Market risk framework 2016 Revised framework 2022
Liquidity Coverage Ratio 2010 100% 2015 2019
Net Stable Funding Ratio 2010 100% 2018
Stage 2: Tackling RWA variability
Output floor 2017 72.5% 2022 2027
LR revisions/G-SIB surcharge 2017 50% scaling factor 2022
Credit risk framework 2017 Revised framework 2022
Operational risk framework 2017 Revised framework 2022
1
  The Basel framework distinguishes three pillars: (i) minimum capital requirements, (ii) supervisory review and (iii) market discipline, based on
standardised disclosures. Complementary reforms, such as enhanced bank resolution regimes, are implemented in parallel.
Sources: BCBS; BIS.
excessive leverage in the banking sector, providing a backstop to the RWRs and a
degree of protection against model risk, under both the SAs and internal models.
Second, a countercyclical capital buffer and G-SIB capital surcharges address
macroprudential considerations (Chapter IV). Finally, two liquidity standards (ie the
Liquidity Coverage Ratio (LCR) and the Net Stable Funding Ratio) incentivise greater
reliance on more stable funding sources (Graph III.1, centre panel) and limit
maturity transformation risks.5
Stage 2 reforms: completing the framework
The recently finalised stage 2 of the Basel III reforms completes the framework by
focusing mainly on enhancing the consistency and comparability of banks’ RWAs
(Table III.1). In the past, differences in the design and choice of parameters across
banks’ internal models had resulted in large differences in risk weights and
corresponding capital requirements (“RWA variability”), providing ample leeway for
banks to raise their leverage (Graph III.1, left-hand panel). From a prudential
perspective, RWA variability is welcome if it reflects legitimate differences in
underlying risks or their measurement.6 However, there is evidence that it has also
reflected unwarranted factors, such as “gaming” (ie choosing model assumptions to
reduce measured risks).7
Such unwarranted RWA variability can be material. Assuming a benchmark
capital ratio of 10%, a study by the Basel Committee on Banking Supervision (BCBS)
found that two banks with identical banking book assets might report capital ratios
differing by up to 4 percentage points (Graph III.2, left-hand panel).8 Moreover, in

Basel III floors – addressing unwarranted RWA variability Graph III.2
Use of internal models hampers the …and reduces the risk weights for The Basel III floors better align IRB
comparability of capital ratios…1 corporate exposures2 and SA risk weights
Percentage points Per cent
2 20 15
Change in IRB risk weight (%)

1 0 0
0 –20 –15
–1 –40 –30
–2 –60 –45
Europe North America Asia and Australia 0.4 0.6 0.8 1.0 1.2
Ratio of current IRB risk weight
to hypothetical SA risk weight
1
Change from 10% benchmark capital ratio if banks’ own model-implied (IRB) risk weights were adjusted to the median risk weight reported
by all banks. Based on risk assessments by 32 major financial institutions of an identical (hypothetical) portfolio of sovereign, bank and
corporate exposures; grossed up to overall RWA level, holding all other RWA components stable. 2 Percentage difference from standardised
approach (SA) risk weights. Positive (negative) values indicate average IRB risk weights based on banks’ own probability-of-default and loss-
given-default estimates that are higher (lower) than SA risk weights for an identical exposure.
Sources: BCBS, “Analysis of risk-weighted assets for credit risk in the banking book”, Regulatory Consistency Assessment Programme (RCAP),
July 2013; BCBS, Basel III monitoring report, December 2017; BIS calculations.
many cases, internally modelled risk weights were substantially lower than those
under the SAs – for corporate exposures, by up to more than 60% (Graph III.2,
centre panel). The observed wedge and capital relief are difficult to justify.
Stage 2 sought to address this unwarranted RWA variability through a range of
measures, complementing the leverage ratio introduced with the stage 1 reforms.9
Key among them are constraints on banks’ internal modelling practices, known as
“input and output floors”. These constraints are especially important where model
risk is high, eg when data are scarce or modelling techniques untested or not
robust (ie for operational risk and various low-default credit portfolios).10
Input floors introduce a measure of conservatism in model parameter choice.
They do so by either disallowing the use of internal models for particular exposures
or setting minima for model parameters (such as the probability of default (PD)).
Input floors address specific sources of RWA variability in a targeted manner. Yet, by
design, they have to be set at low levels to avoid penalising some activities (ie by
imposing minimum PDs that may be too high for certain low-risk exposures). As
such, they do not constrain “aggressively” estimated RWAs for riskier exposures.
The output floor provides an additional degree of protection, by ensuring that
a bank’s RWA cannot fall below 72.5% of the RWA amount that would result from
applying the SA to the same portfolio. In contrast to input floors, the output floor
can thus provide a degree of protection against unwarranted RWA variability across
the entire risk spectrum. And, unlike the leverage ratio, it limits the capital relief
banks can obtain by opting for internal models rather than the SAs.
Recent BCBS data illustrate the effect of the new constraints on RWA variability.
Average risk weights tend to change most for the banks that reported risk weights
furthest below those implied by the SA (Graph III.2, right-hand panel). Thus,
assuming that any differences in the two approaches reflect primarily unwarranted
RWA variability, the output floor closes at least part of the gap.

Banks’ adjustment: the long path to sustainability
Reaping Basel III’s financial stability benefits requires timely and consistent
implementation of the new standards, backed by sustainable bank profitability.
Supervisory metrics suggest that banks have already completed most of the
adjustment to the standards ahead of schedule. Some market-based measures,
however, cast doubt on the extent to which banks have managed to transition to
viable new business models. Several levers are available to manage this transition.
The public sector can provide additional support by removing legal or structural
impediments to banks’ own adjustment efforts.
Implementation and bank business models
The Basel III standards are being phased in over extended timelines to help banks
adjust (Table III.1 above). By now, legal implementation is generally well advanced,
with core stage 1 components, such as the new RWRs and the LCR, operational
across all BCBS member – and many other – jurisdictions (Graph III.3, left-hand
panel). National implementation of other elements, such as the leverage ratio, is
progressing, and the stage 2 additions are due to follow mostly by 1 January 2022.
Yet experience suggests that agreed implementation schedules may be difficult to
maintain and that progress may slow. Therefore, progress monitoring is important
– for example, via the BCBS’s Regulatory Consistency Assessment Programme
(RCAP).
Regardless of national implementation, most banks have already adjusted their
balance sheets ahead of time to meet the new standards (Graph III.3, centre panel).
One reason is market expectations. The fully loaded (ie completely phased-in)
Implementation of new requirements and banks’ adjustments are progressing Graph III.3
Continued progress in national Capital and liquidity shortfalls G-SIB balance sheets reflecting
implementation of Basel III1 coming down2 changing business models 4
Per cent EUR bn EUR bn USD trn
2008 2017 2008 2017

100 1,650 875 50
80 1,320 700 40
60 990 525 30
40 660 350 20
20 330 175 10
0 0 0 0
11 12 13 14 15 16 17 11 12 13 14 15 16 17 Stage 23 Assets Liabilities & equity

impact
Risk-based capital ratio (Jan 2013) Cash & equiv Total equity
Lhs: Rhs:
Leverage ratio (Jan 2018) Liquidity Common Trading securities Deposits
Liquidity Coverage Ratio (Jan 2015) Coverage Ratio Equity Tier 1 Other securities Senior debt
Net Stable Funding Ratio (Jan 2018) Net Stable Additional Tier 1 Loans Subordinated debt
Funding Ratio Tier 2 Other assets Other liabilities
1
Percentage of BCBS member jurisdictions in which each standard is in force; agreed implementation dates in parentheses. 2 The height
of each bar shows the aggregated capital shortfall considering requirements for each tier (ie CET1, Additional Tier 1 and Tier 2) of capital for
the major internationally active banks monitored by the BCBS (BCBS (2018)). 3 Estimates based on end-2015 bank balance sheet information
(BCBS (2017), Table 3). 4 Total values; based on a balanced sample of 28 G-SIBs. Cash & equiv = cash and cash equivalents.
Sources: BCBS; BCBS, Basel III monitoring report, December 2017 and March 2018; SNL; BIS calculations.

requirements have become the investor benchmark; and banks with regulatory
shortfalls risk facing market valuation pressures. Another, related reason is increased
supervisory use of stress testing and corresponding disclosures, which often include
fully loaded regulatory metrics.11 Thus, looking ahead, one should expect banks to
front-load their adjustments also to the latest stage 2 revisions. The amounts
involved are small. Resulting CET1 capital shortfalls among the larger, BCBS-
monitored internationally active banks are estimated at €27.9 billion (based on
end-2015 balance sheet information), less than 1% of these banks’ combined CET1
capital. And the estimate probably overstates the true shortfall, as it does not
account for banks’ business model or portfolio adjustments in response to the
regulations.
Indeed, the impact of the reforms is already evident. Trends in aggregate G‑SIB
balance sheets, for example, tally quite closely with the reform’s objectives
(Graph III.3, right-hand panel): more and higher-quality capital; less reliance on
short-term wholesale funding; bigger high-quality liquid asset (HQLA) buffers; and
a shift away from business lines such as proprietary trading, apparent from the
shedding of trading assets. This reflects a broader shift towards more retail-oriented
business models, with relatively stable funding and income sources.12
Bank resilience: moving ahead
With bank balance sheet adjustment to the new regulatory standards mostly
completed, a key question concerns the degree to which tighter regulation
translates into increased bank resilience – Basel III’s ultimate objective.
One way to measure progress is to assess the impact of changes in different
capitalisation metrics on indicators of bank distress.13 For example, simple logistic
regressions – run on data covering 77 banks – provide estimates of the combined
marginal predictive power of two key Basel III metrics (Tier 1 capital/RWAs and the
leverage ratio) for a credit rating downgrade to “distress level” (Graph III.4, left-hand
panel). Subject to the usual caveats, this analysis suggests that the likelihood of a bank
facing distress within a two-year period decreases as the Tier 1 capital ratio increases
(ie shifts along the horizontal axis). And importantly, for a given Tier 1 capital ratio,
higher leverage ratio requirements tend to further reduce the distress probability
(eg shifts from the yellow to the red line). This highlights the complementarity of the
two ratios and supports the framework’s multiple metrics setup (see above).
In the aggregate, higher capital and resilience have been achieved with little
sign of an adverse impact on bank lending.14 Bank lending to the private non-
financial sector as a share of GDP has remained stable in many jurisdictions –
meeting or exceeding pre-crisis averages.15 That said, there are at least two areas
where more action is needed to further increase resilience.
The first area concerns the link between resilience and regulatory reporting
requirements, which can raise the risk of regulatory arbitrage. One such example
relates to banks’ “window-dressing” around regulatory reporting dates. The
incentive arises in part because of differences in how authorities implement the
leverage ratio across jurisdictions. Some, such as in the United States, require the
ratio to be fulfilled on the basis of period averages, while others, such as in the
euro area, do so on the basis of quarter-end values.
There is evidence that banks without averaging requirements markedly
contract their balance sheets at quarter-ends relative to those subject to averaging
(Box III.A). This can influence market functioning and monetary policy implementation,
for instance by hindering access for those market participants that need to transact
at quarter-ends. And it reduces the prudential usefulness of the leverage ratio,
which may end up being met only four times a year.

Box III.A
Banks’ window-dressing: the case of repo markets
Hidden liquidityrefers
Window-dressing riskstonot
the covered
practice ofby additional
adjusting buffers?
balance Graph
sheets around regular reporting dates, such as year-III.7
or
quarter-ends. Window-dressing can reflect attempts to optimise a firm’s profit and loss for taxation purposes. For
Cost
banks, of however,
institutional trades
it may rises
also as responses
reflect Mutual funds’ liquid assets
to regulatory have not especially
requirements, Demandif for volatility-linked
combined with end-period
spreads generally decline1 adjusted to higher liquidity risk2 exposures rises 3
reporting. One example is the Basel III leverage ratio. This ratio is reported based on quarter-end figures in some
Per cent of par Per cent ‘000 contracts
jurisdictions, but is calculated based on daily averages during the quarter in others. The former case can provide
strong incentives to compress exposures around regulatory reporting dates – particularly at year-ends, when
3.2 24 1,200
incentives are reinforced by other factors (eg taxation).
Banks can most easily unwind positions around key reporting dates if markets are both short-term and liquid.
2.4 18 900
Repo markets generally meet these criteria. As a form of collateralised borrowing, repos allow banks to obtain short-
term funding against some of their assets – a balance sheet-expanding operation. The cash received can then be
1.6 12 600
onlent via reverse repos, and the corresponding collateral may be used for further borrowing. At quarter-ends,
banks can reverse the increase in their balance sheet by closing part of their reverse repo contracts and using the
0.8 6 300
cash thus obtained to repay repos. This compression raises their reported leverage ratio.
The data indicate that window-dressing in repo markets is material. Data from US money market mutual
0.0 0 0
funds (MMMFs) point to pronounced cyclical patterns in banks’ US dollar repo borrowing, especially for jurisdictions
with leverage ratio reporting based on quarter-end figures (Graph III.A, left-hand panel). Since early 2015, with
06 08 10 12 14 16 06 09 12 15 18 06 09 12 15 18
the beginning of Basel III leverage ratio disclosure, the amplitude of swings in euro area banks’ repo volumes
Micro (under $100,000) High-yield International bonds Trading volume (daily)
has been rising – with total contractions by major banks up from about $35 billion to more than $145 billion at
Odd-lot (from $100,000 to $1m) Open interest (30-day average)
year-ends. While similar patterns are apparent for Swiss banks (which rely on quarter-end figures), they are
1
Bid-ask
less spreads for for
pronounced US corporate
UK and bonds as estimated
US banks (whichin use
Adrian et al (2017);Banks’
averages). by tradetemporary
size. 2
Liquid asset holdings
withdrawal from as repo
a percentage
marketsshare
is
of funds’ total net assets; by fund category. 3 CBOE S&P 500 Volatility Index (VIX) futures.
also apparent from MMMFs’ increased quarter-end presence in the Federal Reserve’s reverse repo (RRP) operations,
which allows
Sources: T Adrian,them to place
M Fleming, excess
O Schachar and cash
E Vogt,(right-hand panel,
“Market liquidity after black line).crisis”,
the financial Despite
Annualthe implicit
Review floorEconomics,
of Financial providedvolby 9,
2017, pp 43–83;
the rates Federal
on the RRPReserve Bankline),
(yellow of New York;are
there CBOE Global
signs ofMarkets;
volatilityICI;spikes
BIS calculations.
in key repo rates around quarter-ends (blue
line). Such spikes may complicate monetary policy implementation and affect repo market functioning in ways
that can generate spillovers to other major funding markets, especially if stress events coincide with regulatory
reporting dates.
Banks’ window-dressing through the lens of US repo markets Graph III.A
Global banks’ repo borrowing from US MMFs Repo rates and MMF placements at the Federal Reserve
USD bn USD bn Per cent
300 375 1.8
200 250 1.2
100 125 0.6
0 0 0.0
2011 2012 2013 2014 2015 2016 2017 2018 2015 2016 2017 2018
1 2
Banks from: EA US GB CH RRP total (lhs) Rhs: GCF RRP
1
Reverse repo. 2
DTCC General Collateral Financing (GCF) Repo Index (Treasury weighted average).
Sources: Federal Reserve Bank of St Louis (FRED); Office of Financial Research; Crane Data; DTCC; BIS calculations.
  The end-2017 contraction is equivalent to about 1.4% of the sample banks’ total leverage ratio exposure measure.      See CGFS,
“Repo market functioning”, CGFS Papers, no 59, April 2017; and I Aldasoro, T Ehlers and E Eren, “Business models and dollar funding of
global banks”, BIS Working Papers, no 708, March 2018.

Regulatory vs market-based metrics – resilience is up, but by how much?
In per cent Graph III.4
RWRs and LR reinforce each other1 Regulatory LRs rise2 Market-based LRs lag behind2
10 9.0 20
8 7.5 16
Distress probability
6 6.0 12
4 4.5 8
2 3.0 4
0 1.5 0
6 7 8 9 10 02 05 08 11 14 17 02 05 08 11 14 17
Tier 1 capital/RWAs
United States United Kingdom
Leverage ratio (%): 1 3 5 Euro area Other AEs
The dashed vertical line in the left-hand panel indicates the 8.5% Tier 1/RWA minimum capital requirement. The vertical lines in the centre
and right-hand panels indicate August 2007 (interbank market turmoil in the early stages of the Great Financial Crisis) and December 2010
(the beginning of stage 1 Basel III reforms).
1
Estimated probability of distress within two years for a given level of risk-weighted Tier 1 capital (horizontal axis) at three different LRs.
Estimates based on a logistic regression of a distress indicator denoting a bank’s individual rating dropping below D within the next two years
on the variables indicated in each panel and a control variable for total assets as well as a dummy variable to flag observations in the post-
2007 period. The sample is an unbalanced panel of annual observations for 77 banks over the period 1995–2013. The vertical axis measures
the estimated probability of distress for different values of the explanatory variable. 2 Asset-weighted averages of simplified regulatory LRs,
based on the ratio of common equity to total assets (centre panel), and market value-based LRs (right-hand panel) by economy; based on 73
banks and not adjusted for national accounting differences.
Sources: I Fender and U Lewrick, “Calibrating the leverage ratio”, BIS Quarterly Review, December 2015; Bankscope; Datastream; Moody’s;
national data; BIS calculations.
Prudential authorities can help reduce or prevent these types of effect in

various ways. Options include aligning national implementation on the basis of
period averages, stepping up supervisory responses, and requiring banks to disclose
both metrics to enhance market discipline.
The second area concerns the outlook for bank profitability. Bank profitability
is critical for resilience, as it affects the speed with which banks can recover from
losses. Despite the progress made in terms of balance sheet and business model
adjustments, market valuations for many banks point to continued investor
scepticism about profitability prospects. Average bank price-to-book ratios (PBRs)
hovered around a level of two times book value right before the GFC – admittedly,
a level inconsistent with the risks revealed by the GFC (see regulatory discussion
above). They then plummeted to values below one in 2008–09, and recovered only
recently, while generally remaining lower than pre-crisis, especially for European
banks.
Thus, once adjusted for depressed PBRs, conventional resilience measures look
less solid.16 One example is market-based leverage ratios, which have improved by
less than their book value counterparts (compare the right-hand and centre panels,
Graph III.4). Bank credit default swap (CDS) spreads and stand-alone credit ratings
(which seek to abstract from official support, known to have been cut) tell a similar
story (Graph III.5, left‑hand and centre panels).17 Even though pre-crisis levels are
unlikely to be an appropriate benchmark, this suggests that reduced bank
profitability has at least partly offset the stabilising effect of reduced leverage and
maturity transformation. At the same time, there are signs that banks may have

Box III.B
The widening of Libor-OIS spreads
Spreads between short-term US dollar Libor and overnight indexed swap rates (Libor-OIS), a common indicator of
funding stress, widened substantially in early 2018 (Graph III.B, left-hand panel). Yet, unlike during previous episodes,
the surge did not reflect rising bank riskiness, as gauged from credit default swap spreads. Nor did it coincide with
signs of stress in US dollar funding markets, as indicated by cross-currency basis spreads. What might explain it?
Two likely drivers are increased issuance of short-term US Treasury securities (T-bills) and repatriation flows due
to the 2017 US tax reform. T-bill issuance jumped by more than $300 billion in Q1 2018 (centre panel). As a result,
short-term yields increased, with the associated rise in the T-bill–OIS spread accounting for up to 40% of the change
in Libor-OIS (left-hand panel). Tax reform is likely to explain part of the remainder, with US corporates repatriating
some of the foreign profits previously held abroad. As part of these was invested in non-US bank commercial
paper (CP), such flows tend to lower the supply of offshore US dollar funding for banks. This comes on top of
reduced supply from money market mutual funds (MMMFs), which has not returned to pre-October 2016 US MMMF
reform levels. Bank funding costs thus rose amid strong CP issuance in early 2018 (centre panel), adding to the rise
in Libor-OIS.
The tightening of the cross-currency swap basis (left-hand panel) contrasts with previous episodes of Libor-OIS
widening. One explanation put forward is that the US base erosion and anti-abuse tax raised foreign bank US
affiliates’ funding costs. Those affiliates would have to issue more debt of their own while cutting back on inter-
office funding. This would lower demand for FX hedging, contributing to a tighter basis. Yet, contrary to this
explanation, US affiliates’ issuance declined, whereas net inter-office positions rose as Libor-OIS widened (centre
panel). An alternative explanation of tight cross-currency spreads relates to portfolio rebalancing decisions. FX-
hedged returns on long-term US bonds declined relative to those on euro area sovereign debt, amid expectations
of rising US interest rates (right-hand panel). Non-US investors may thus have reduced their dollar securities
holdings and, as a result, their demand for FX hedging. Indeed, according to official data, Japanese investors cut
their US bond holdings by about $50 billion while investing $30 billion in German and French sovereign bonds in
early 2018.
Deciphering the recent surge in Libor-OIS spreads Graph III.B
T-bill yields drive Libor-OIS spreads 1 T-bill and CP issuance on the rise FX-hedged returns diverging4
Basis points Basis points USD trn USD bn Per cent
45 –90 2.2 350 1.25
30 –75 2.0 300 1.00
15 –60 1.8 250 0.75
0 –45 1.6 200 0.50
–15 –30 1.4 150 0.25
–30 –15 1.2 100 0.00
–45 0 1.0 50 –0.25

Q2 17 Q4 17 Q2 18 Q2 17 Q4 17 Q2 18 Q2 17 Q4 17 Q2 18
Lhs: Rhs (reversed): Lhs: Rhs: US Germany France
2 2
Libor-OIS Cross-currency T-bills Foreign financial CP
T-bill–OIS basis USD/JPY US affiliate CP
2
3
Net due to foreign office
1
Based on three-month tenors. 2 Outstanding amounts of T-bills and US commercial paper (CP) of foreign financials and US financials
with foreign bank parent. 3 US financials’ net liabilities vis-à-vis their related foreign offices. 4 Spread between the return on 10-year
sovereign bonds, swapped into Japanese yen (adjusting for three-month rolling hedging cost), and the return on 10-year Japanese
government bonds.
Sources: Federal Reserve Bank of St Louis (FRED); US Department of the Treasury; Bloomberg; Datastream; BIS calculations.
  See BIS, 87th Annual Report, June 2017, Chapter II.
Price-to-book ratios (PBRs) are closely approximated by the valuation equation1 Graph III.C
PBRs: full sample PBRs: actual vs valuation equation3 Decomposition of valuation change3
Ratio Ratio 2007–15 changes
4 4 0.5
Bank credit risk and return-on-equity – further room for improvement Graph III.5
CDS spreads1, 2 Credit ratings1, 3 Lower leverage, higher RoE?4

Basis points Ratings
400 A+ 14
2005
2009
2013
2017
300 A– 7
RoE (%)
200 BBB 0
100 BB+ –7
0 BB– –14
08 10 12 14 16 18 United Euro Other EMEs 4 6 8 10

States area AEs Leverage ratio (%)
United States Other AEs
Euro area EMEs Stand-alone External support G-SIBs Other large banks
1
Asset-weighted averages. Based on a sample of about 50 large banks. Five-year on-the-run credit default swap (CDS) spreads; monthly
2
averages of daily data. 3 Based on Fitch ratings; end-of-year data. 4 Based on a sample of about 100 large banks. Leverage ratio and
return-on-equity (RoE) at end-2017.
Sources: Fitch Solutions; IHS Markit; SNL; BIS calculations.
become less susceptible to adverse funding shocks. One example is the surge in
Libor-OIS spreads in early 2018, which drove up bank funding costs, but left bank
CDS spreads
Exposure to broadly
valuation unaffected
losses and (Boxconcentration
III.B and GraphinIII.5,
asset left-hand
management panel). rising Graph III.6
Banks could seek to raise profitability and valuations in time-tested ways, such
Funds
as hold larger
cutting costs shares of riskier balance
and repairing Rising duration
sheets of IG corporate
by eliminating bond Rising concentration
non-performing loans. in asset
corporate bonds 1 indices 2 management industry
Bank valuations are not generally out of line with what is predicted by simple
Per cent Years USD trn Per cent
valuation models that assign importance to those variables (Box III.C). In addition,
80 7 80 60
there is evidence that a stronger capital base can help too. Even though lower
leverage mechanically reduces banks’ return-on-equity (RoE), better-capitalised
institutions tend to exhibit a similar 60 or even higher RoE than their 6more 60 leveraged 50
peers (Graph III.5, right-hand panel). 18
Looking ahead, a key challenge 40 is that these adjustments have5to take 40 place at 40
a time of rapid technological change in the financial sector (various types of “fintech”
innovation). On the one hand,20many of these innovations allow 4banks 20 to better 30
exploit scale economies and – ultimately – reduce costs. One example is the use of
distributed ledger technology to improve back office functions (Chapter V). On the
0 3 0 20
other
08 10
hand, 12
client
14
expectations
16 18
are changing
08 10
–
12
and 14
with them
16
the
18
nature 08of bank
10 12 14 16
competitors.
Credit rating:
Clients,
AAA to A
in particular those on the retail side,
United States
increasingly 3ask for a
Euro area Non-bank/
Lhs:
“seamless customer BBB toexperience”.
B While this may help segment the customer Top 20base non-insurers
among top 20 (rhs)4
and support price discrimination, the corresponding shift to multipurpose internet Others
platforms
1
Share of totalinvites new competitors.
bond holdings for an unbalancedHere, so-called
sample “big
of more than tech”
1,600 players
US mutual and – dominant bond funds. 2 Effective
exchange-traded
duration of investment grade (IG) corporate bond indices; US Corporate
technology firms from the online sales or messaging sector – loom large. Master Index for the United States and EMU Corporate Index for the
These
euro area. 3 Assets under management. 4 Percentage share of non-bank/non-insurers among the top 20 asset managers.
already have the necessary IT infrastructure, analytical skills, financial resources and
Sources: Datastream; ICE BofAML Indices; Lipper; Willis Towers Watson; 19
established client base to erode banks’ market share. BIS calculations.
Further public sector initiatives could act as a catalyst for banks to make the
necessary adjustments. These include efforts to tighten banks’ provisioning policies
(eg via asset quality reviews) and to tackle impediments to the reduction of
overcapacity and banking sector consolidation.20 The arrival of big tech competitors,
in turn, may require cooperation among regulators from different fields (data
protection agencies, competition authorities and others) and jurisdictions to preserve

Deciphering the recent surge in Libor-OIS spreads Graph III.B
T-bill yields drive Libor-OIS spreads 1 T-bill and CP issuance on the rise FX-hedged returns diverging4Box III.C
Factors driving bankBasis
Basis points equity
points valuations
USD trn USD bn Per cent
45 –90 2.2 350 1.25
30 price-to-book ratios (PBRs), defined

Low –75 2.0as the market value of a bank’s 300 equity to its accounting – or book – value, 1.00
have been a persistent sign of post-crisis –60 1.8 challenges in the banking sector. Having hovered at around two times
0.75
15 250
book value, on average, in the run-up to the Great Financial Crisis (GFC), they plummeted to values below unity by
0
2009 and recovered only recently –45– while
1.6 remaining below pre-crisis 200 levels (Graph III.C, left-hand panel). As 0.50a
measure of the market premium (or discount) applied to a bank’s book value, PBRs are a key indicator of banks’
–15 –30 1.4 150 0.25
expected underlying profitability. Hence, there is a strong interest in understanding what drives these market
premia.
–30 –15 1.2 100 0.00
Recent BIS research sheds some light on this question by estimating a valuation equation (VE) for a sample of
–45 0 1.0 50 –0.25
72 banks from 14 jurisdictions using annual data over the 2000–16 period. The panel regression includes five (sets
Q2 17 Q4 17 Q2 18 Q2 17 Q4 17 Q2 18 Q2 17 Q4 17 Q2 18
of) explanatory variables known to affect PBRs, both directly and indirectly: (i) loans (including non-performing
loans Rhs (reversed):
Lhs: (NPLs)), (ii) deposits, (iii) expenses,Lhs: Rhs:
(iv) other2 bank-specific factors (for US
instance, leverage Germany
or France
dividend payments)
2
Libor-OIS Cross-currency T-bills Foreign financial CP
and (v) return-on-equity (RoE; a
basis USD/JPYproxy of investors’ return expectations).
2
T-bill–OIS US affiliate CP
The estimated VE tracks the evolution of bank PBRs closely
Net due to across both3 time and countries (centre panel). PBRs
foreign office
are generally in line with VE-implied valuations, suggesting that investors’ valuation benchmarks do not seem to
1
Based on three-month tenors. 2 Outstanding amounts of T-bills and US commercial paper (CP) of foreign financials and US financials
have changed materially3 post-crisis. Despite the relatively large number of explanatory variables, just four key drivers
with foreign bank parent. US financials’ net liabilities vis-à-vis their related foreign offices. 4 Spread between the return on 10-year
sovereign bonds, swapped into Japanese yen (adjusting for three-month rolling hedging cost), and the2015,
explain around three quarters of the VE-implied change in bank PBRs between 2007 and returnwith NPLs and
on 10-year RoE
Japanese
the most important
government bonds. ones (right-hand panel).
These findings suggest that banks are well placed to enhance their market value by focusing on a few key
Sources: Federal Reserve Bank of St Louis (FRED); US Department of the Treasury; Bloomberg; Datastream; BIS calculations.
profitability drivers under direct management control, such as proactively addressing NPLs and other legacy assets,
tight control of non-interest expenses, and reducing overcapacity in the sector.
Price-to-book ratios (PBRs) are closely approximated by the valuation equation1 Graph III.C
PBRs: full sample PBRs: actual vs valuation equation3 Decomposition of valuation change3
Ratio Ratio 2007–15 changes
4 4 0.5
3 3 0.0
2 2 –0.5
1 1 –1.0
0 0 –1.5
02 05 08 11 14 17 02 05 08 11 14 17 IT ES US All
Percentiles: Average2 Actual: Valuation equation: PBR: Contribution of changes in:
25th–75th All Actual Dividends
10th–90th United States VE Non-interest expense
Italy Return-on-equity
Spain Non-performing loans
1
The sample covers 72 banks in advanced economies; end-of-quarter data. 2
Asset-weighted average. 3
Based on valuation equations
(VE) in Bogdanova et al (2018).
Sources: Bogdanova et al (2018); Datastream; Fitch Solutions; BIS calculations.
  For details, see B Bogdanova, I Fender and E Takáts, “The ABCs of bank PBRs”, BIS Quarterly Review, March 2018. This analysis takes that
of C Calomiris and D Nissim, “Crisis-related shifts in the market valuation of banking activities”, Journal of Financial Intermediation, vol 23,
no 3, 2014, pp 400–35, and extends it to an international setting.

a level playing field (“same risk, same regulation”), without unduly constraining
technological innovation.21 One such example is aligning constraints on the
accumulation, use and sharing of client data for both banks and non-banks. The
favourable macroeconomic environment (Chapter I), rising term spreads and
reduced legacy problems (eg crisis-related litigation costs) provide further support.
Banks that fail to seize this opportunity may see their resilience tested before their
adjustment to the post-crisis environment is fully complete (see also the discussion
on snapback risk below).
Bank/non-bank interactions: new pressure points?
The adjustment of banks’ business models under way raises a number of questions
at the current juncture. One such question concerns the system-level impact of the
rising share of non-bank intermediaries in financial markets. Their interaction with
banks and other parts of the financial system is changing the dynamics of markets
in response to shocks. A key example relates to the process of monetary policy
normalisation in major advanced economies and how increases in longer-term
interest rates, including the possibility of rapid snapback (Chapter I), could
propagate through the financial system.
Institutional asset managers: bigger footprint, changing dynamics
Non-bank institutional asset managers, ranging from investment management

companies to pension funds and insurers, have grown strongly over the past
decade. Their total assets are estimated at nearly $160 trillion, exceeding those of
banks worldwide.22 Several drivers have contributed to this growth. They include an
increasing demand for long-term investments, such as on the part of pension funds,
and search for yield in an environment of unusually low interest rates, which has
boosted the growth of open-end mutual funds and exchange-traded funds (ETFs).
Past years of exceptionally low interest rates have raised a variety of challenges
for those institutional asset managers that are major fixed income investors.23 These
firms benefited initially from the decline in interest rates, which generated valuation
gains on their bond holdings. Yet persistently low rates and compressed term
premia reduced the yields on new investments, driving down future returns. Pension
funds and insurance companies faced additional pressures owing to the rising
mark-to-market value of their long-term liabilities. This provided strong incentives
for institutional asset managers to extend the duration of their portfolio or to invest
in riskier assets – a trend borne out in the available data (Graph III.6, left-hand and
centre panels). Taken together, these factors suggest that sensitivity to snapback in
both interest rates and volatilities has increased.
Various structural features of the asset management industry may contribute
to magnifying this vulnerability. One is a high concentration of assets under
management, which can result in a clustering of risks within a limited number of
large asset management companies (Graph III.6, right-hand panel). True, investment
activity is typically spread out over a large number of separately managed funds
within these firms, mitigating the risk of concerted trading activity. But there is
evidence that fund families exhibit correlated return and investor flow patterns.24 In
addition, reliance on common service providers (eg for IT infrastructures, risk
management and pricing tools, or custody services) suggests common exposures
to operational risks. And, rising investor demand for lower-cost passive asset
management products may have increased the risk of crowded trades in leading
market indices.25

Exposure to valuation losses and concentration in asset management are rising Graph III.6
Funds hold larger shares of riskier Rising duration of IG corporate bond Rising concentration in asset
corporate bonds 1 indices 2 management industry
Per cent Years USD trn Per cent
80 7 80 60
60 6 60 50
40 5 40 40
20 4 20 30
0 3 0 20
08 10 12 14 16 18 08 10 12 14 16 18 08 10 12 14 16
Credit rating: AAA to A United States Euro area Lhs:3 Non-bank/
BBB to B Top 20 non-insurers
4
Others among top 20 (rhs)
1
Share of total bond holdings for an unbalanced sample of more than 1,600 US mutual and exchange-traded bond funds. 2 Effective
duration of investment grade (IG) corporate bond indices; US Corporate Master Index for the United States and EMU Corporate Index for the
Sources: Datastream; ICE BofAML Indices; Lipper; Willis Towers Watson; BIS calculations.
Hidden liquidity risks not covered

Snapback and by additional buffers?
interlinkages Graph III.7
Cost of institutional trades

Institutional Mutual
rises as asset funds’ liquid
managers assets have
and banks not Demand
are interlinked in afor volatility-linked
variety of ways, opening
upPerboth direct and indirect channels forPerthe
cent of par cent
propagation of snapback‘000 risks and
contracts
similar shocks. Open-end funds are especially relevant in this context. Not only are
they exposed 3.2 to valuation losses, like any other 24rate-sensitive investor, but they1,200 also
face investor redemption risk (Box III.D). Funds’ credit lines and deposits at banks
thus represent2.4 a key direct link. In a snapback 18 scenario, with widespread investor 900
redemptions triggering a run-down of funds’ cash balances, depository banks could
1.6 12 600
be exposed to large deposit withdrawals by fund managers and rising credit
exposures to funds. Granted, direct bank exposures to such risks are perhaps less of
0.8 6 300
a challenge than indirect ones, given the supervisory attention that interest rate
risks have 0.0 received in recent years.26 Even so, there 0 is a need for banks to monitor 0
and manage these risks particularly carefully.
06 08 10 12 14 Additional
16 pressure
06 points12could
09 15 magnify
18 these06effects.09 For 12 one, 15 large-scale
18
Micro (under $100,000) redemptions could force
High-yield open-end funds to
International bonds sell relatively illiquid assets
Trading volume (daily) at short
Odd-lot (from $100,000 notice and, hence, at large discounts, further depressing valuations. Indeed, funds
to $1m) Open interest (30-day average)
1
that promise
Bid-ask spreads for US corporate daily redemptions
bonds as estimated – iebythose
in Adrian et al (2017); offering
trade size. 2
deposit-like
Liquid asset holdingsinstruments
as a percentage –of have
funds’
total net assets; by fund category.
increased
3
CBOE S&P footprint.
their 500 Volatility Index
In the(VIX)United
futures. States, for example, they now hold more
Sources: T Adrian, M Fleming,than 16% of

O Schachar and corporate
E Vogt, “Marketdebt according
liquidity to financial
after the financial accounts
crisis”, Annual Review ofdata – up
Financial from vol
Economics, less
9,
2017, pp 43–83; Federal Reserve Bank of New York; CBOE Global Markets; ICI; BIS calculations.
than 7% in 2005. This is bound to have changed market dynamics.
Funds’ liquidity management thus plays an important role in assessing the
broader market impact of investor redemptions. Funds that invest in relatively illiquid
assets, such as corporate bonds or some emerging market economy (EME) debt,
need to strike a difficult balance between selling illiquid assets at potentially large
discounts and running down their cash buffers, which may leave them vulnerable to
future outflows. In addition, investors may be tempted to redeem their shares to
front-run others, anticipating the liquidation cost-induced dilution of fund portfolios.

Box III.D
When yields “snap back” – funds’ exposures and amplification effects
Interest rate risk is inherent in the regular activities of bond market investors and therefore actively managed. Even
so, past episodes of snapbacks in long-term rates are useful reminders of the potential vulnerabilities of some
segments of the fund industry. Historically, interest rate shocks have been linked to monetary policy decisions. As
such, they coincided with rising short-term rates and flattening yield curves. More recently, however, long-term
rates have occasionally snapped back without notable changes in short-term rates (Graph III.D, left-hand panel).
Drivers of market dynamics may thus have changed, possibly giving rise to more abrupt market adjustments than in
the past.
Open-end bond funds and exchange-traded funds (ETFs), key buyers of corporate bonds and other fixed
income instruments in recent years, are particularly exposed to episodes of rapidly rising rates. This reflects both the
induced valuation losses and the redemption pressures caused by declining fund returns (centre panel). Such
redemptions may force sales at large discounts, exacerbating the downward pressure on fund returns and triggering
further redemptions. Likewise, ETF investors may find it difficult to sell their shares in secondary markets, with bid-
ask spreads often widening as fund returns deteriorate (right-hand panel).
Several factors may amplify such dynamics. For one, credit spreads are already quite compressed. Bond
investors are thus unlikely to benefit from any offsetting effect of tighter spreads during snapbacks. In addition,
portfolio duration has increased for many funds, amplifying the valuation impact of rate changes. Persistently low
market volatility, notwithstanding recent increases, may have further sustained fixed income positions at low yields,
increasing the scope for abrupt sell-offs. Finally, funds may amplify market adjustments by shedding assets in excess
of redemptions to increase cash buffers – especially if other liquidity management tools (eg swing pricing) fail to
discourage investors from redeeming.
Fund flows and performance during recent episodes of rising long-term yields Graph III.D
Recent episodes of rising yields... …cut returns and triggered outflows 1 ETF liquidity strained3
Per cent Per cent of total net assets Per cent Mn shares/day Basis points
4 4 4 24 24
3 2 2 18 18
2 0 0 12 12
1 –2 –2 6 6
0 –4 –4 0 0
2010 2012 2014 2016 2018 2010 2012 2014 2016 2018 Ma
y
Ma
y Jun Jun Jul
13 27 10 24 08
2
US yields: 1-month 10-year Flows (lhs) Returns (rhs) 2013
Trading volume (lhs) Bid-ask (rhs)
The shaded areas in the left-hand and centre panels indicate episodes during which 10-year US Treasury yields increased by at least 80 basis
points before falling again.
1
Active US open-end mutual bond funds. 2 Three-month rolling average of nominal fund returns; weighted by funds’ total net
assets. 3 Five-day rolling averages of volume-weighted bid-ask spreads and daily volume of the 10 largest bond ETFs by total assets during
the US taper tantrum.
Sources: Bloomberg; Lipper; national data; BIS calculations.
  See S Morris, I Shim and H S Shin, “Redemption risk and cash hoarding by asset managers”, Journal of Monetary Economics, 89, 2017,
pp 88–91; and U Lewrick and J Schanz, “Is the price right? Swing pricing and investor redemptions”, BIS Working Papers, no 664, October 2017.

Other institutional investors, for instance insurance companies and pension
funds, could further intensify such market reactions. One issue is increasingly similar
portfolio holdings in an environment of low rates (see above), making concerted
selling more likely. Another is the use of dynamic hedging strategies, which implies
that these investors would respond to any rise in long-term yields by selling long-
term bonds in order to contain duration mismatches, adding to the risk of abrupt
interest rate adjustments.27
Indirect, market-based interlinkages represent an additional channel of shock
propagation. Any initial, snapback-induced position adjustment by mutual funds or
other rate-sensitive investors is likely to be exacerbated by deteriorating market
liquidity conditions, which would tend to spill over into other markets (eg via
collateral valuations).
A key point is that structural changes in the provision of immediacy services
may not be visible in standard measures of market liquidity, masking the risks
associated with holding assets that may turn out to be illiquid in some scenarios.
For example, many banks and other market-makers have cut back the amount of
risk capital they allocate to trading activities.28 To some extent, the reduction in
market-making has been compensated by increased agency-based trading, in
which the intermediary matches offsetting client orders with limited commitment
of own balance sheet capacity. One implication of this trend is that the execution of
large orders, particularly during stressed market conditions, has become more
difficult (Graph III.7, left-hand panel). The associated risks have shifted from market-
makers to investors, especially in the less liquid segments of the fixed income
markets, such as EME debt or corporate bonds. On top of that, low yields and
increasing competition have discouraged funds from raising low-yielding liquidity
buffers, affecting their ability to manage redemption risks (Graph III.7, centre panel).
Alternative liquidity providers, such as proprietary trading firms (PTFs), have
increased their market share in some fixed income markets. Their activities, however,
have typically been limited to the most liquid segments, for example those for
major advanced economy sovereign bonds. In addition, many PTFs trade with
limited commitment of risk capital and lack the balance sheet capacity to maintain
large inventories – as is generally necessary for market-making in infrequently traded
assets, such as corporate bonds. This suggests that, while increasing competition is
likely to further reduce the transaction costs of relatively liquid assets, funds
invested in relatively illiquid ones remain exposed to high liquidity risks.
Developments in the ETF sector illustrate how these different factors can
interact during abrupt interest rate moves and volatility spikes. ETFs are index-
tracking investment funds. Yet, in contrast to traditional open-end mutual funds,
their shares trade on secondary markets, and their creation and redemption are
exclusively settled between designated financial intermediaries (“authorised
participants”, APs) and the ETF sponsor, usually an asset management company.29
While ETFs are thus not directly exposed to investor redemption risk, other pressure
points may arise.
One concerns APs’ capacity to support secondary market liquidity in a snapback
scenario. Under normal market conditions, APs arbitrage any difference in the ETF
share price and that of the securities in the underlying index. But large selling
pressure from ETF investors could overwhelm the APs’ capacity to fund such
arbitrage. Corporate bond ETFs provide one such example. While APs can redeem
the shares they acquire from investors with the ETF sponsor in exchange for the
underlying bonds, APs may hesitate to build up large bond inventories at a time of
high risk and strained market liquidity in the underlying bond markets. This would
drive a wedge between ETF prices and those of the underlying securities, which
could in turn trigger further position adjustments and cross-market spillovers.

1
Share of total bond holdings for an unbalanced sample of more than 1,600 US mutual and exchange-traded bond funds. 2 Effective
duration of investment grade (IG) corporate bond indices; US Corporate Master Index for the United States and EMU Corporate Index for the
Sources: Datastream; ICE BofAML Indices; Lipper; Willis Towers Watson; BIS calculations.
Hidden liquidity risks not covered by additional buffers? Graph III.7
Cost of institutional trades rises as Mutual funds’ liquid assets have not Demand for volatility-linked
Per cent of par Per cent ‘000 contracts
3.2 24 1,200
2.4 18 900
1.6 12 600
0.8 6 300
0.0 0 0
06 08 10 12 14 16 06 09 12 15 18 06 09 12 15 18
Micro (under $100,000) High-yield International bonds Trading volume (daily)
Odd-lot (from $100,000 to $1m) Open interest (30-day average)
1
Bid-ask spreads for US corporate bonds as estimated in Adrian et al (2017); by trade size. 2
Liquid asset holdings as a percentage of funds’
total net assets; by fund category. 3 CBOE S&P 500 Volatility Index (VIX) futures.
Sources: T Adrian, M Fleming, O Schachar and E Vogt, “Market liquidity after the financial crisis”, Annual Review of Financial Economics, vol 9,
2017, pp 43–83; Federal Reserve Bank of New York; CBOE Global Markets; ICI; BIS calculations.
Concentration risk could amplify such tensions, since major APs also provide
immediacy services in other markets and to other investors (eg for open-end funds).
Another issue concerns the procyclical trading activity in new ETF structures.
The past few years have seen a growing demand for financial instruments that
allow volatility trading – among others, ETFs (Graph III.7, right-hand panel). These
products are designed to maintain a target exposure to a given volatility index, ie
buying when the index rises and selling when it declines in a rather mechanical
way. As a result, bouts of volatility prompt procyclical trading, reinforcing the initial
volatility shock. Indeed, recent episodes of volatility spikes in equity markets have
uncovered such dynamic feedback effects, highlighting the need for effective
market backstops and prudent management of volatility risk.30
Implications for prudential policy
Structural changes in the asset management industry suggest that shock

propagation can work through new, market-based channels that may amplify price
movements relative to pre-crisis. Thus, even though banks and other intermediaries
have become more resilient, snapback and similar shocks could lay bare new
vulnerabilities. This has implications for prudential policy, adding to those in other
policy areas (Chapters I and II).
For the banking sector, changes in market dynamics generate new exposures
that may be insufficiently covered by current risk management practices. This
underlines the need for tight supervision. In addition to guidance clarifying
supervisory expectations regarding the management of interest rate risk, stress
tests represent a key tool. In this context, adverse scenarios may need to be adapted
to better reflect snapback-related exposures of banks’ clients and counterparties
and any associated knock-on effects.
Furthermore, supervisory attention may need to shift more in the direction of
non-bank players and how these would perform in snapback and other stress

scenarios. One key issue is investment funds’ liquidity management, which is their
first line of defence in response to redemption pressures. Various tools are available
to improve the management of these risks. These include additional liquidity
buffers to mitigate fire sale risks as well as efforts to further develop investment
funds’ liquidity stress testing capacity (Chapter IV).31 More work may be needed,
however, to assess the effectiveness of these measures in different redemption
scenarios in order to inform regulatory calibration decisions.

Endnotes
1
See BCBS, Governors and Heads of Supervision finalise Basel III reforms, press release, 7 December
2017; and Basel III: finalising post-crisis reforms, December 2017.
2
Under the new total loss-absorbing capacity (TLAC) standard, as of 1 January 2022 all G-SIBs will
be required to have eligible TLAC instruments equal to a minimum of at least 18% of their risk-
weighted assets (RWAs), not including any applicable regulatory capital buffers. TLAC will also
need to be equivalent to at least 6.75% of the Basel III leverage ratio exposure measure. For details,
see FSB, Summary of findings from the TLAC impact assessment studies, November 2015.
3
Only half of the 8% Basel II minimum requirement was defined in terms of Tier 1 capital
instruments, which included a range of hybrid structures and intangibles. See S Cecchetti, “The jury
is in”, CEPR Policy Insights, no 76, December 2014.
4
For a motivation of this multiple metrics setup in a general equilibrium context, see F Boissay and
F Collard, “Macroeconomics of bank capital and liquidity regulations”, BIS Working Papers, no 596,
December 2016.
5
In addition, through revised guidance principles, the framework stresses the importance of prudent
corporate governance (eg by promoting effective control functions).
6
For details, see BIS, 83rd Annual Report, June 2013, Box V.B.
7
See M Behn, R Haselmann and V Vig, “The limits of model-based regulation”, ECB Working Papers,
no 1928, July 2016, for estimates of the extent of such gaming. Similar evidence based on more
recent data is provided in F Niepmann and V Stebunovs, “Modeling your stress away”, mimeo, 2018.
8
See BCBS, “Analysis of risk-weighted assets for credit risk in the banking book”, Regulatory
Consistency Assessment Programme (RCAP), July 2013, for details.
9
See BCBS, Basel III leverage ratio framework and disclosure requirements, January 2014; and M Brei
and L Gambacorta, “Are bank capital ratios pro-cyclical? New evidence and perspectives”, Economic
Policy, vol 31, no 86, 2016, pp 357– 403.
10
The scarcity (or outright lack) of reliable data can prevent supervisors from validating model
outputs with sufficient confidence. This suggests that the use of internally modelled approaches
may have to be withdrawn or restricted. Under Basel III, operational risk and various low-default
credit portfolios now fall into this category. Exposures to large and mid-sized corporates, for
example, are migrated away from the advanced IRB (which allows for modelling of loss-given-
default (LGD)) to the foundation IRB or SA (which do not).
11
See eg European Banking Authority, 2016 EU-wide stress test: results, July 2016.
12
See BIS, 87th Annual Report, June 2017, Chapter V. For more analysis, see R Roengpitya, N Tarashev,
K Tsatsaronis and A Villegas, “Bank business models: popularity and performance”, BIS Working
Papers, no 682, December 2017; and CGFS, “Structural changes in banking after the crisis”, CGFS
Papers, no 60, January 2018.
13
See I Fender and U Lewrick, “Calibrating the leverage ratio”, BIS Quarterly Review, December 2015,
pp 43–58.
14
Discussions of the macroeconomic impact of higher bank capital often presume that higher bank
capital increases funding costs, which then translates into higher lending spreads and less lending.
Recent research suggests that the opposite may be true, in that higher bank capital goes hand in
hand with higher lending. See eg L Gambacorta and H S Shin, “Why bank capital matters for
monetary policy”, Journal of Financial Intermediation, 2018 (forthcoming).
15
In Q3 2017, bank lending-to-GDP ratios in the euro area and the United Kingdom (at around 90%
in each case) as well as the United States (about 45%) remained broadly unchanged from the
average levels in 2002–06. Major EMEs, such as Brazil, China and India, even saw rising ratios
relative to pre-crisis levels. See BIS, 86th Annual Report, June 2016, Chapter VI.

16
For a discussion, see N Sarin and L Summers (2016), “Understanding bank risk through market
measures”, Brookings Papers on Economic Activity, Fall, 2016, pp 57–127.
17
See C Borio, “The banking industry: struggling to move on”, keynote speech, Fifth EBA Research
Workshop, 28–29 November 2016.
18
Econometric analysis suggests that, post-GFC, higher capitalisation levels relative to total assets
coincide with higher bank valuations. Accordingly, investors appear to have shifted from viewing
leverage as a mechanism primarily for increasing RoE to a greater focus on ways in which excessive
leverage can threaten solvency. See B Bogdanova, I Fender and E Takáts (2018), “The ABCs of bank
PBRs”, BIS Quarterly Review, March 2018, pp 81–95; and C Calomiris and D Nissim, “Crisis-related
shifts in the market valuation of banking activities”, Journal of Financial Intermediation, vol 23, no 3,
2014, pp 400–35.
19
In a recent survey by Bain & Company, more than half of all US respondents – and 74% of those
aged 18–24 – indicated that they expect to buy a financial product from a technology firm in the
next five years. See Bain & Company, “Banking’s Amazon moment”, Bain Brief, March 2018.
20
See C Borio, B Vale and G von Peter, “Resolving the financial crisis: are we heeding the lessons from
the Nordics?”, BIS Working Papers, no 311, June 2010; and European Systemic Risk Board, “Is
Europe overbanked?”, Reports of the Advisory Scientific Committee, no 4, June 2014.
21
See A Carstens, “A level playing field in banking”, keynote address, Institute of International Finance
Board of Directors dinner, 21 January 2018.
22
See FSB, Global shadow banking monitoring report 2017, March 2018.
23
For a discussion, see BIS, 86th Annual Report, June 2016, Chapter VI.
24
See eg E Elton, M Gruber and C Green, “The impact of mutual fund family membership on investor
risk”, Journal of Financial and Quantitative Analysis, vol 42, no 2, 2007, pp 257–78.
25
For a discussion of risks related to passive asset management, see V Sushko and G Turner, “The
implications of passive investing for securities markets”, BIS Quarterly Review, March 2018,
pp 113–31.
26
For example, since early 2010, US bank and depository supervisors have had explicit inter-agency
guidance in place to alert market participants to the importance of interest rate risk and remind
them of supervisory expectations regarding sound risk management practices. Similar guidance
has been issued in other jurisdictions.
27
See D Domanski, H S Shin and V Sushko, “The hunt for duration: not waving but drowning?”, IMF
Economic Review, vol 65, no 1, 2017, pp 113–53, for a discussion of potential amplification effects
arising from insurance companies.
28
For a discussion, see CGFS, “Market-making and proprietary trading: industry trends, drivers and
policy implications”, CGFS Papers, no 52, November 2014; and “Fixed income market liquidity”,
CGFS Papers, no 55, January 2016.
29
For a discussion of ETFs and the associated risks, see V Sushko and G Turner, “What risks do
exchange-traded funds pose?”, Bank of France, Financial Stability Review, April 2018, pp 133–44.
30
See eg BIS, “Volatility is back”, BIS Quarterly Review, March 2018, pp 1–15.
31
FSB, Policy recommendations to address structural vulnerabilities from asset management activities,
January 2017.

IV. Moving forward with macroprudential frameworks
The Great Financial Crisis (GFC) revealed the inadequacy of pre-crisis prudential
requirements and the limitations of the then existing tools to preserve financial
stability. In response, authorities around the world have strengthened financial
regulation and supervision (Chapter III) and adopted a macroprudential orientation
to financial stability. The new macroprudential frameworks focus on the stability of
the financial system as a whole and how it affects the real economy, rather than just
on the stability of individual institutions.1 This is important because the GFC and
previous crises have shown that vulnerabilities may build up across the system even
though individual institutions may look stable on a standalone basis. Indeed, many
systemic financial crises of recent decades, the GFC included, resulted from the
financial system’s procyclicality – its tendency to amplify financial expansions and
contractions, often with serious macroeconomic costs.2
Experience indicates that substantial progress has been made, but more needs
to be done. Macroprudential frameworks have been very useful as a complement to
the other financial reforms put in place after the GFC. Macroprudential measures build
buffers, discourage risky lending and strengthen the financial system’s resilience.
They can also slow credit growth but, as deployed so far, their restraining impact on
financial booms has not always prevented the emergence of the familiar signs of
financial imbalances. And, as with any medicine, they come with side effects. This
suggests that these measures would be most effective if embedded in a broader
macro-financial stability framework that includes other policies, notably monetary,
fiscal and structural.
The chapter is organised as follows. The first section describes the key elements
of macroprudential frameworks and the main implementation challenges. The
second discusses how authorities have dealt or could deal with some of these
challenges, such as risk identification, instrument choice, policy communication and
governance. The third section reviews evidence on the impact of macroprudential
measures. A final section explores the role of macroprudential measures in a
broader macro-financial stability framework and coordination with other policies.
Two boxes discuss, respectively, macroprudential approaches to capital market
activities and the use of FX intervention to reduce systemic risk.
Macroprudential frameworks: elements and challenges
Although the term dates back to the 1970s, it languished for the most part in
obscurity until the turn of this century, when BIS General Manager Andrew Crockett
called for a “macroprudential” approach to financial stability.3 In the same speech,
he differentiated the macroprudential dimension of financial stability – the stability
of the financial system – from the microprudential dimension – the stability of
individual institutions. What distinguishes the two perspectives is less the specific
instruments – they are often the same – than why they are used and how they are
calibrated.
It took the GFC to expose the limitations of a microprudential perspective.
After the crisis, as these limitations were recognised in policy circles, more and more
countries adopted financial stability mandates and implemented macroprudential
measures. As a result, the term “macroprudential” has entered the mainstream

vocabulary of central banks (Graph IV.1, left-hand panel) and also of other
policymakers. In addition, the average number of macroprudential measures
adopted post-crisis has significantly increased for both advanced economies (AEs)
and emerging market economies (EMEs) (Graph IV.1, right-hand panel).4 In
particular, AEs have stepped up their use of macroprudential measures in recent
years.
The FSB, IMF and BIS set out the key elements of a macroprudential framework
in a series of notes prepared for the G20.5 These identified three intermediate
objectives: (i) to increase the financial system’s resilience to aggregate shocks by
building and releasing buffers; (ii) to constrain financial booms; and (iii) to reduce
structural vulnerabilities in the financial system that arise from common exposures,
interlinkages and the critical role of individual intermediaries. This chapter focuses
mainly on the first two objectives, which refer to the “time” dimension of systemic
risk. By pursuing these objectives, macroprudential measures can build resilience
and moderate financial cycles.
Adopting a macroprudential orientation to financial stability comes with a
number of challenges. First, the ultimate objective – financial stability – is hard to
define. For this reason, policymakers often resort to intermediate objectives, such
as improving lending or risk management standards, strengthening banks’ resilience
and reducing fluctuations in credit.6 Such intermediate objectives can help
communicate macroprudential measures and improve the coordination between
different policymakers responsible for financial stability. But even they may be too
vague when it comes to assessing the impact of particular measures.
Second, macroprudential goals may conflict with other policy objectives. This is
an issue because macroprudential authorities typically resort to instruments that
may also be used for other purposes or from different perspectives. For instance, in
a boom, bank supervisors may see no need to tighten regulatory requirements
since individual institutions look solid when viewed in isolation, while
macroprudential authorities might be more worried about procyclicality in the
financial system and aggregate risk-taking and thus wish to tighten prudential
instruments. Conversely, in a generalised downturn, macroprudential authorities
may wish to release buffers to smooth the impact on the real economy, while bank
Macroprudential orientation moves to the mainstream Graph IV.1
Central bank speeches mentioning “macroprudential” Increasing use of macroprudential measures over time1
Number of speeches
200 12.5
160 10.0
120 7.5
80 5.0
40 2.5
0 0.0
99 01 03 05 07 09 11 13 15 17 1995–2000 2001–06 2007–12 2013–18
AEs EMEs
1
The bars show the average number of macroprudential measures per year and per 10 economies in each group of economies.
Sources: BIS central bankers‘ speeches; BIS calculations based on macroprudential measures recorded in Table IV.1.

supervisors may prefer that institutions preserve as much capital as possible so as
to better weather their losses. Tensions may also arise between macroprudential
and monetary or fiscal authorities. Resolving them puts a premium on appropriate
governance arrangements.
Third, it is typically difficult to identify financial vulnerabilities early enough
and with sufficient certainty to take action. In some cases, it may be hard to
disentangle the development of financial imbalances from welcome financial
deepening and innovation. Vulnerabilities may also build up over many years,
without leading to acute stress. The system may appear stable in the interim,
especially since signs of low risk (eg compressed spreads) may in fact simply reflect
high risk-taking. And, tightening measures when the financial system is already
vulnerable could trigger the very instability one seeks to prevent.
Fourth, a bias towards inaction could result from the political economy costs of
taking preventive measures, combined with difficulties in the timely identification
of systemic risk. For one, policymakers are usually wary of sounding a false alarm,
preferring to wait and see whether a development is actually harmful – even though
prompt intervention is typically more effective. Further, the near-term costs of
preventive actions are quite visible but their long-term benefits, while large,7 are
harder to discern, since policymakers and their policies rarely get any credit for a
crisis that did not happen. Indeed, preventive measures may be quite unpopular, as
they may hamper access to credit precisely when the general picture looks good. In
this context, the temptation can be strong to argue that this time really is different,
and that no action is needed.8
Finally, the impact of macroprudential measures can be hard to measure, given
the plethora of potential instruments, their complex interactions and, frequently,
the scantiness of evidence about their effectiveness. And this may be the case even
when the objective is well defined. The way in which monetary policy and
macroprudential measures may interact only adds to these challenges. In fact,
despite recent progress, models that link the financial sector to the real economy
tend to be highly stylised.9 Thus, it could be argued that the calibration of
macroprudential measures is more art than science.
When and how to act?
Ten years after the concept “macroprudential” entered the vocabulary of

policymakers, there is a growing body of analytical research and practical
experience on how some of these issues may be addressed. This section considers,
in turn, the challenges in identifying risks, selecting and calibrating the instruments,
communication and governance.
Identifying risks
Early warning indicators (EWIs) serve as a useful starting point for identifying
systemic risks. Typically, they are calibrated on whether they would have been able
to predict past crises. Many studies find that when credit and asset prices start
deviating from long-run trends and breach certain critical thresholds, they can help
to identify unsustainable booms with reasonable accuracy several years before a
full-blown crisis actually develops.10 Even so, such indicators can also sound a false
alarm, not least because their critical thresholds are based on averages across a
wide range of countries and over extended periods. As a result, they may not
sufficiently take into account country-specific features or how financial systems
evolve over time, including in response to changing regulation.

On balance, EWIs are a useful first step in identifying risks, but need to be
complemented by more in-depth analysis. For instance, the distribution of
exposures across borrowers may matter: even if the average borrower is solid, the
failure of a critical mass of fragile ones could propagate through the system and
cause a systemic event. Yet, on its own, analysis at the level of individual institutions
will not suffice, since it cannot measure the impact of vulnerabilities on the financial
system and the macroeconomy.11
A popular method for gauging financial system resilience is aggregate, or
macro, stress testing. For example, major AE central banks use it, and all EME central
banks responding to a recent BIS survey either used it or were planning to do so.12
The tool helps assess resilience in response to hypothetical low-probability but
high-impact macroeconomic and financial shocks. Stress tests have the advantage
that they are forward-looking and can cover various scenarios. This makes them a
valuable instrument for assessing specific systemic vulnerabilities – for instance, to
assess the amount of capital required during an ongoing financial crisis. But it
makes them less useful as a tool to identify risks when the range of scenarios is
more open. Stress tests have other shortcomings too. They can help assess, say, the
immediate impact of declines in house prices and increased mortgage defaults on
bank balance sheets. But they are less good at capturing second-round effects
arising from fire sales, lower market liquidity or the weaker spending and rising
unemployment that follow such financial shocks.13 Indeed, the stress tests carried
out prior to the GFC gave little indication of any significant risks in the banking
sector.14
Given the difficulties in identifying risks early and sufficiently surely, authorities
often wait to see whether a development will have adverse consequences. This
tendency may affect the timing of tightening actions during a credit boom. For
example, the frequency of tightening actions increases as the credit gap crosses
the 2 percentage point lower threshold in the Basel Committee guidelines on
the calibration of the countercyclical capital buffer (CCyB) (Graph IV.2, all panels).
As the total credit gap and general bank credit gap increase further above the
Tightening actions are used more frequently as credit booms build up1 Graph IV.2
Total credit-to-GDP gap2 Bank credit-to-GDP gap3 Housing credit-to-GDP gap4
0.10 0.10 0.10
0.08 0.08 0.08
0.06 0.06 0.06
0.04 0.04 0.04
0.02 0.02 0.02
0.00 0.00 0.00

0–2 2–10 >10 0–2 2–10 >10 0–2 2–10 >10
Gap ranges (% pts)
1
The values on the vertical axis are the number of economy-quarters with at least one tightening action taken by the economy divided by
the total number of economy-quarters for which the gap is in the specified range. The respective credit-to-GDP gap is calculated by a one-
sided Hodrick-Prescott filter with lambda 400,000. 2 Total credit refers to the total amount of credit to an economy’s private non-financial
sector extended by banks and non-banks in the form of loans and debt securities. 3 Bank credit refers to the amount of bank credit to the
private non-financial sector. 4 Housing credit refers to the amount of housing loans extended by banks.
Sources: National data; BIS.

10 percentage point upper threshold in the guidelines, tightening actions tend to
be used more frequently (left-hand and centre panels). By contrast, as the housing
credit gap increases above the 10 percentage point threshold, tightening actions
tend to be used slightly less frequently (right-hand panel). One possible reason is
that national authorities can more easily identify financial imbalances building up
in the housing market and thus more promptly deploy tightening actions before
the gap exceeds the threshold.
Use of macroprudential measures by targeted credit, instrument type and region

Number of policy actions, 1995–2018 Table IV.1
Region1
Asia- Central Latin Middle North Western All
Pacific and eastern America East and America Europe economies
Targeted credit Europe Africa
Instrument type [11] [14] [6] [4] [2] [18] [55]
General credit 31 156 68 5 – 56 316
Countercyclical capital buffers 3 4 – – – 6 13
Limits on FX mismatch, position
8 32 15 1 – 7 63
or liquidity
Capital inflow- or FX liability-based RR2 5 44 17 4 – – 70
Credit growth- or asset-based marginal
– 24 25 – – 6 55
RR2
Others3 5 1 – – – 5 11
Housing/consumer/household credit 168 125 24 13 13 114 457
LTV4 limits and loan prohibitions 76 37 9 4 7 35 168
DSTI, DTI5 limits and other lending
49 34 4 3 6 23 119
criteria
Risk weights 17 40 8 4 0 42 111
Loan loss provisioning rules 15 3 3 2 0 10 33
Others6 11 9 – – – 1 21
Corporate credit (including CRE loans) 7 18 19 2 – – 24 63
Credit to financial institutions 8 2 2 2 – – 3 9
Total 9 219 302 96 18 13 197 845
(1.00) (1.02) (0.88) (0.29) (0.31) (0.49) (0.75)
Memo items: Total 158 219 52 18 – 66 513
(0.72) (0.74) (0.48) (0.29) (0.17) (0.46)
General liability-based average RR 2 115 159 50 17 – 34 375
Liquidity requirements 10 43 60 2 1 – 32 138
Asia-Pacific = AU, CN, HK, ID, IN, KR, MY, NZ, PH, SG and TH; central and eastern Europe = BG, CZ, EE, HR, HU, LT, LV, PL, RO, RS, RU, SI, SK and
TR; Latin America = AR, BR, CL, CO, MX and PE; Middle East and Africa = AE, IL, SA and ZA; North America = CA and US; western Europe = AT,
BE, CH, DE, DK, ES, FI, FR, GB, GR, IE, IS, IT, LU, NL, NO, PT and SE.
1
  The figures in square brackets indicate the number of economies in each region.    2  Reserve requirements.    3  Structural capital surcharges,
other capital surcharges and loan loss provisioning rules on general credit.    4  Loan-to-value.    5  DSTI = debt service-to-income; DTI = debt-
to-income.    6  Exposure limits on the housing sector and limits on FX loans to households.     7  Comprising LTV limits, DSTI limits, risk weights,
loan loss provisioning rules and exposure limits. CRE = commercial real estate.    8  Comprising limits on interbank exposure, exposure limits
on non-bank financial institutions and risk weights on exposure to financial institutions.    9  The figures in parentheses indicate the average
number of actions per country per year for each region.    10  Liquidity Coverage Ratio, Net Stable Funding Ratio and liquid asset ratio.
Sources: Budnik and Kleibl (2018); Reinhardt and Sowerbutts (2016); Shim et al (2013); national data; BIS calculations.

Selecting and calibrating instruments
A broad array of tools can potentially be used to reduce systemic risk, although in
some jurisdictions legal impediments or coordination issues may significantly
restrict those that can actually be deployed. Essentially all prudential tools, such as
restrictions on particular types of lending and capital or liquidity requirements, can
be used from a macroprudential perspective as well as in the more traditional
microprudential sense. In addition, monetary policy tools may also be used
macroprudentially, for instance in the form of reserve requirements or even foreign
exchange interventions.15
In practice, a wide range of tools has been deployed, primarily targeting various
types of bank credit (Table IV.1). Authorities in both Asia-Pacific and central and
eastern Europe have been the most active. Many economies have also introduced
measures targeting commercial real estate mortgages and property developer loans.
In particular, most EU member states have adjusted risk weights for loans
collateralised with commercial property, while some EMEs have changed loan-to-
value (LTV), debt service-to-income (DSTI) and exposure limits as well as loan loss
provisioning rules on commercial real estate loans (Table IV.A1). Although the bulk of
the measures focus on bank credit, authorities have reacted to the growing
importance of market finance by also taking a macroprudential perspective on the
capital market activities of asset managers and other institutional investors (Box IV.A).
The tools operate through different mechanisms. Some instruments refer to
borrower characteristics, even though they are enforced on the lenders’ side.
Examples are caps on LTV, debt-to-income (DTI) and DSTI ratios. These increase the
borrowers’ resilience to house price or income fluctuations, in turn limiting the
lenders’ credit risk. By constraining effective credit demand, they may also put a
brake on credit growth and, indirectly, on house prices too. Other tools work
directly on the lender side. Examples are countercyclical capital requirements,
provisioning rules and credit growth limits. Capital tools, in particular, increase
banks’ buffers to absorb losses, provided that they can actually be drawn down in
case of stress. In addition, capital and provisioning requirements increase the cost
of providing housing credit, which should slow credit growth.
The wide variety of potential tools lets authorities target specific exposures or
activities.16 For example, the Central Bank of Brazil imposed restrictions on auto
loans that it deemed particularly risky, but not on other types of auto loan.17 Such
targeted actions can reduce the costs of intervention, but they also have drawbacks.
First, they tend to have more immediate distributional consequences, which could
result in greater political pressures. Second, they are more vulnerable to leakages –
defined as the migration of the targeted activity outside the scope of the tool’s
application and enforcement.
Leakages can take many forms. At one end of the spectrum are evasive ploys
that merely shift the targeted activity into a new guise, without changing the nature
of its risks. For example, in Malaysia tighter LTV limits on mortgages to individuals
led to a surge in home purchases by firms set up specifically to circumvent the
restrictions.18 Exposures may also migrate to lending institutions that are not subject
to the specific measure – for instance, to shadow banks or foreign intermediaries.
Some evidence suggests that macroprudential measures implemented on bank
credit have led to an expansion in the credit provided by non-banks, and that
measures targeting external bank borrowing have boosted offshore corporate bond
issuance.19 Such leakages may reduce the direct risk exposures of the domestic
banking system but not the likelihood of corporate sector stress as such.
Partly in response to leakages, the authorities have in several cases progressively
broadened the scope of the measures employed, for instance by expanding the set

Box IV.A
Macroprudential approaches to capital market activities
As current macroprudential measures focus mainly on banks, they may be less effective in dealing with risks arising
from the market-based financing that has become more prevalent post-GFC. Similarly, financial innovation and the
application of new technology to the financial industry may shift the nature of risk, requiring a new set of policy
responses and an expanded arsenal of instruments (Chapter III). In this context, how can macroprudential approaches
help address systemic risk arising from asset management funds and other institutional investors such as insurance
companies and pension funds?
Correlated and procyclical trading by asset management funds could destabilise asset markets, resulting in
large losses that could propagate through the financial system. Such effects are possible even if each market
participant acts prudently on a standalone basis, given the interactions between market dynamics and the collective
actions of individual market participants. However, current regulation on the asset management fund industry is
geared mainly towards microprudential and consumer protection objectives and thus fails to fully incorporate how
actions by one player can affect the health of others via changes in asset prices, exchange rates and market liquidity.
The macroprudential perspective should be extended to asset management funds to address these concerns.
Authorities have a number of options to address these risks. For example, minimum liquidity requirements for
asset management funds may allow them to meet redemptions without selling relatively illiquid assets. If so, such
requirements could help increase the resilience of market liquidity. In January 2017, the US Securities and Exchange
Commission (SEC) implemented new rules requiring open-end mutual funds and exchange-traded funds to establish
liquidity risk management programmes. Among other measures, the rules require these funds to consider current
market conditions and establish appropriate liquidity risk management policies and procedures in light of both
normal and reasonably foreseeable stressed market conditions. Such requirements incorporate a macroprudential
perspective in that they recognise that liquidity is adversely affected by market stress.
Liquidity stress tests for asset management funds have also been implemented by a few other national
authorities. For example, in 2015 the Bank of Mexico assessed liquidity risk in domestic mutual funds. The French
market supervisory authority has also published a guidance document on stress testing for asset management
funds. But, in these exercises, the authorities took a mainly microprudential approach, by focusing on fund-level
liquidity risks. By contrast, in February 2018 the European Systemic Risk Board published a recommendation on
action to address systemic risks related to liquidity mismatches. In particular, it explicitly considered an amplification
channel whereby mismatches between the liquidity of open-end investment funds’ assets and their redemption
profiles could lead to fire sales to meet redemption requests in times of market stress, potentially affecting other
financial market participants holding the same or correlated assets.
To deal effectively with systemic risks stemming from asset management funds and other institutional investors,
close cooperation among the various authorities involved is crucial – central banks, bank regulators, insurance
regulators and securities regulators. Here, differences in perspectives can complicate matters. For instance, securities
regulators with responsibility for asset managers put prime emphasis on investor protection, while central banks and
bank regulators focus more on financial stability and hence are more inclined to apply macroprudential approaches.
National authorities are currently making the very first steps towards a macroprudential perspective on capital
market activities, as compared with the progress already made in introducing macroprudential frameworks to the
banking sector. The growing importance of asset managers and other institutional investors in both domestic and
cross-border financial intermediation requires national authorities to monitor potential systemic risks from these
activities at both the national and global levels and to consider how best to employ macroprudential approaches to
deal with such risks.
  See Borio (2004) for further details of this interaction.      FSB (2017) provides specific policy recommendations for dealing with liquidity
risks in the asset management sector.      For details of the initial proposed rules, comments received and the final rules, see SEC (2016).
of activities targeted. In other cases, they have taken a relatively broad approach,
applying a portfolio of measures with the aim of reducing possible channels for
evasion.20
While a broad approach using many instruments may be more effective in
targeting risks, it also has its drawbacks. It can easily become complex and difficult
to communicate. In the extreme, it could result in the macroprudential authority

effectively taking on the role of credit allocation.21 And it can complicate the
calibration of individual instruments, by making it harder to assess their interaction
and overall effect.
Macroprudential authorities have addressed the difficulties of calibrating
instruments in a variety of ways. In some cases, central banks have relied on
econometric estimates linking actions and outcomes. In others, they have used
micro data to gauge the effects of specific measures on credit.22 In the more
common complex situations, they have resorted to stress tests,23 calibrated models24
or plain judgment.
A fairly common reaction to the uncertainty about how instruments work has
been to start cautiously and then increase the intensity, scope and frequency of the
measures until the effects become more apparent.25 Such a gradual approach is
consistent with the view that uncertainty calls for caution. But the heavy costs of a
crisis could tip the scale towards more decisive action early on.26
Communication
As in many other policymaking areas, effective communication is critical for success.

Communication can explain the objectives, strategy and policy process to the
public, and thus build political support. In addition, it can help the authorities share
their risk assessment with both the affected parties and the broader public, which
can enhance effectiveness.27
For communication to achieve the desired effects, the message needs to be
delivered effectively to the right audience. The primary audience largely determines
the content, sophistication and channels used. Most central banks communicate
financial stability risks in speeches, press conferences and their regular financial
stability reviews. While such reviews are useful in communicating with specialists,
they can easily prove opaque for the broader public, not least homeowners.28 Thus,
several authorities also use more targeted channels. For instance, the Reserve Bank
of India issues short and simplified press releases for an audience with limited
financial literacy. Establishing links with the media, such as through background
briefings, is another common tool.
In one sense, communication might even be viewed as a macroprudential tool
in its own right.29 In theory, central bank warnings might head off adverse
developments, obviating the need for any subsequent remedial action. In practice,
examples of warnings that appear to have taken effect without subsequent concrete
actions (or at least the threat thereof) are rather few. In Chile, warnings from the
central bank in its Financial Stability Report between June and December 2012
appear to have affected bank lending practices, inducing a shift towards lower-LTV
mortgages.30
Governance
The multiple purposes of the instruments, the scope for strong political pressure
and the mismatch between the mandate and tools put an onus on adequate
governance arrangements. This involves several aspects: having a clear operational
objective; providing incentives to act and tools commensurate with that objective;
ensuring accountability and transparency;31 and ensuring effective coordination
across the policy areas that have a bearing on financial stability.32
The institutional arrangements governing macroprudential frameworks vary
across countries. The most common is to allocate macroprudential functions to
several bodies that coordinate through a committee (Graph IV.3, left-hand panel).
The second most common one is to vest both macroprudential and microprudential

Macroprudential institutions and powers Graph IV.3
Who is responsible for macroprudential policy? Inter-agency committees have mostly soft powers 3
Percentage of 51 countries and territories
Discussion/coordination
80 Oversight/macroprudential supervision/analysis
1
With IAC
Recommend macroprudential actions

60 General policy formulation
Warnings
40 Comply or explain
1
Policy decision on macroprudential matters

Without IAC
20 Power to request data/information

Power to designate
0 Appeals body
2
2008 2017 2008 2017 2008 2017 2008 2017 10 20 30 40
Multiple Central bank Central bank Integrated
agencies with micro- without micro- microprudential Number of committees
prudential prudential supervisor Soft powers
Medium-strength powers
Hard powers
1
IAC = inter-agency committee. 2
Data not available. 3
One agency can have several attributes.
Source: BIS surveys.
responsibilities in the central bank. Far less frequently adopted are other possible
arrangements, such as the sharing of responsibilities without a formal coordinating
committee, or giving macroprudential responsibilities to an integrated microprudential
supervisor.
The jury is still out on the effectiveness of these arrangements. In particular,
many of them do not fully align financial stability responsibilities with decision-
making powers over the necessary instruments. Notably, many of the inter-agency
committees set up after the GFC lack hard decision-making powers (Graph IV.3,
right-hand panel). Moreover, very few of the post-GFC financial stability mandates
explicitly mention trade-offs between different policy objectives, let alone how to
resolve them. In response to a BIS survey, only six out of 14 EME central banks that
participated in inter-agency committees said that these had helped coordinate
policies.33 Several respondents stressed that decision-making powers remained with
individual authorities, raising questions about the effectiveness of coordination. In
some cases, the very inclusiveness of such committees can complicate decision-
making.34 In the United Kingdom, the tripartite system that comprised the Treasury,
the central bank and the supervisory authority was abandoned, with most financial
stability-related tasks and responsibilities shifting to the Bank of England.
Impact: the experience so far
Ultimately, macroprudential measures are effective if they ensure that the financial
system is stable. But this benchmark is too general to be useful when assessing the
effectiveness of individual tools. Narrower criteria focus on more specific objectives,
such as curbing the growth of a particular form of credit or increasing the resilience
of the financial system to the unwinding of financial booms or adverse shocks.
Effectiveness can be measured by the change in the rate of credit growth or the
increase in the banking system’s capital or liquidity buffers.

Empirical evidence on the impact of macroprudential measures overall
suggests that macroprudential measures have been generally successful in
strengthening the financial system’s resilience. By construction, capital and liquidity
requirements increase the buffers available to, respectively, absorb future losses
and bridge periods of illiquidity. That said, they can only do so if they can actually
be drawn down when needed. By changing the relative price of different forms of
credit, capital or reserve requirements on particular types of lending can also affect
the composition of credit, reducing the riskiness of loan books.35 All of this should
result in a more resilient and thus stable financial system, as suggested by the small
number of econometric studies that measure the impact of macroprudential
measures on bank risk.36
There is also evidence that certain macroprudential measures have moderated
financial booms. Panel regressions across a broad set of countries indicate that
tighter LTV and DSTI limits may help curb housing credit growth (Graph IV.4, middle
left-hand panel).37 Tighter DSTI caps also appear to dampen house price growth
(bottom left-hand panel). These results are largely in line with the empirical
literature38 and many central banks’ own assessments.39 Maximum LTV and DSTI
ratios as well as limits on credit growth and foreign currency lending also appear to
have moderating effects on bank leverage and asset growth.40 Likewise, there is
some evidence that FX position limits have a measurable impact on credit growth
(top left-hand panel).41 But other measures have less discernible effects or even
work in the wrong direction.42
Interestingly, loosening LTV or DSTI requirements does not appear to have any
effect on credit or house prices (Graph IV.4, right-hand panels). Again, such
asymmetries are in line with the empirical literature. It appears that tightening
measures can help discourage credit expansion but loosening does not encourage
it much during financial downturns – much like pushing on a string. But looking at
the impact of loosening measures on overall credit may not be the right criterion
for success. A better one could be whether the measures help avoid unnecessary
constraints on the supply of credit, not whether they prevent necessary
deleveraging.43
Still, easing by releasing buffers is not without its problems. The market may
view a discretionary release of capital or liquidity buffers during a downturn as a
signal of worse to come rather than as a tool to stabilise the financial system. Thus,
to be effective in a bust, buffers may need to be sufficiently large to start with and
to be released in a non-discretionary fashion.44
While some macroprudential measures appear to have helped slow credit
growth, their restraining impact on financial booms has not always prevented the
emergence of financial imbalances.45 It remains an open issue whether this reflects
inaction bias, leading to belated and overly timid action, a fear of side effects that
limits the strength of the measures, evasion, or any intrinsic limitations of the
instruments.
Although macroprudential measures tend to target specific activities or
exposures, their effects are often much broader. For example, the activation of the
CCyB on mortgages in Switzerland triggered a rise in corporate lending.46 Such a
spillover is largely unavoidable and differs from evasive ploys that leave risks
essentially unchanged. More generally, the same panel regression analysis on direct
effects also provides evidence of spillovers and leakages. In particular, housing,
consumer and household credit growth significantly increased from the quarter
following the implementation of measures tightening corporate credit including
commercial real estate loans (Graph IV.5). In addition, policy actions tightening
consumer credit appear to have boosted housing credit, and those tightening
housing credit to have encouraged consumer credit, which is likely to indicate

Direct effects of macroprudential measures on bank credit and house prices 1
In percentage points Graph IV.4
Real general bank credit growth
Tightening Loosening 12
–4
–8
RR CG RR CF FX limits CCyB All GC policy RR CG RR CF FX limits CCyB2 All GC policy
Real housing credit growth
Tightening Loosening
4
–2
–4
–6
LTV DSTI RW Provisioning All HC LTV DSTI RW Provisioning All HC
policy policy
Real house price growth

Tightening Loosening 6
–2
–4
–6
LTV DSTI RW Provisioning All HC LTV DSTI RW Provisioning All HC
policy policy
Significance at: No statistical
1% 5% 10% significance
One-quarter impact:
One-year cumulative impact:
All GC policy = all policy actions on general bank credit; All HC policy = all policy actions on housing credit; CCyB = countercyclical capital
buffers; DSTI = maximum debt service-to-income ratios, maximum debt-to-income ratios and other lending criteria; FX limits = limits on FX
mismatch or position; LTV = maximum loan-to-value ratios and loan prohibitions; Provisioning = loan loss provisioning rules on housing
loans; RR CF = capital flow- or FX liability-based reserve requirements; RR CG = credit growth- or asset-based marginal reserve requirements;
RW = risk weights on housing loans.
1
The expected sign of the bars for tightening (loosening) actions is negative (positive). 2
Data not available.
Source: BIS calculations based on Kuttner and Shim (2016).

Effects of targeted macroprudential tightening measures on other types of credit
In percentage points Graph IV.5
*** 5
**
** 4
***
3
** 2
0
Impact of corporate credit measures on: Impact of consumer Impact of housing
Housing credit Consumer credit Household credit credit measures on credit exposure limits
housing credit on consumer credit
One-quarter impact One-year cumulative impact
**/*** indicates statistical significance at the 5/1% level.
Source: BIS calculations based on Kuttner and Shim (2016).
leakages. Whether such behavioural responses should raise concerns will depend
on their systemic risk impact.
Sterilised
SimilarFXto intervention
monetary and policydomestic
measures, creditmacroprudential
growth measures affect
economic activity
In percentage points
by changing the cost of borrowing or modifying households’ or Graph IV.B
firms’ access to finance. A relatively small number of studies find that tightening
macroprudential
Impact on the domesticmeasures tends to
credit-to-GDP reduce
ratio 1
output Impact
growth,onbut
real evidence of their
domestic credit growth2
effect on inflation is rather mixed.47
No analysis of policy impact 0.2
*** would be complete without considering side 0.08
***
effects. These can come in many guises. For instance, the measures may have
0.1 0.04
undesired distributional effects, such as limiting access to finance for those who
need it most and discouraging financial innovation. 0.0 They may also distort credit 0.00
allocation. Unfortunately, the evidence on these issues is so far limited.
–0.1 –0.04
In a financially integrated world, developments in one country may give rise to
systemic risk in another. For example, low –0.2 interest rates and** unconventional –0.08
***
monetary policy actions in the large AEs post-crisis have resulted in large capital
flows to EMEs and small open AEs, fuelling domestic –0.3 financial booms.48 International –0.12
FX reserve accumulation/GDP Net capital flows/GDP
spillovers may also result from macroprudential measures. For instance, recent FX purchase/GDP Net capital flows/GDP
studies
**/*** findstatistical
indicates that bank regulation
significance at the 5/1%oflevel.
multinational banks in their home country
affects their lending standards elsewhere. 49
1
This panel shows the coefficient of the variables on the horizontal axis from a BIS panel regression analysis for 20 EMEs from 2000 to 2017,
where the dependent variable is the change in the ratio of domestic credit-to-GDP and the control variables are the lagged dependent
variable, the US dollar exchange rate, the real domestic money market rate, country fixed effects and time fixed effects. 2 This panel shows
Towards an integrated macro-financial stability framework
the coefficient of the variables on the horizontal axis from a panel regression analysis for 45 EMEs from 2005 to 2013 reported in
specification (7) in Table 9.2 of Ghosh et al (2017).
Sources: Ghosh et al (2017); BIS calculations.

The adoption of a more macroprudential approach to financial regulation and
supervision represents an important step forward, in both identifying and
mitigating financial stability risks. It entails a major cultural shift in the concept of
risk, by acknowledging the limitations of market prices as risk indicators as well as
recognising the importance of self-reinforcing financial booms and busts (“financial
cycles”) and financial system-wide considerations. Moreover, the implementation of
macroprudential frameworks has helped strengthen the financial system’s resilience
and moderate financial excesses.

At the same time, such frameworks are no panacea. The task of ensuring
sustainable financial stability is clearly complex. Reliance on one set of tools alone,
even when based on solid arrangements, is unlikely to be enough. For instance,
while macroprudential measures can mitigate credit expansion, those employed so
far, in some cases quite actively, have not necessarily prevented the familiar signs of
financial imbalances from manifesting themselves. Given the economic and social
costs of financial crises and the macroeconomic costs of financial cycles more
generally, it would be imprudent to look exclusively to macroprudential frameworks
to deliver the desired results.
All this suggests that macroprudential frameworks should be embedded in a
more holistic, comprehensive and balanced macro-financial stability framework.
Alongside more micro-oriented financial regulation and supervision, such a
framework would also encompass monetary, fiscal and structural policies. The
ultimate goal would be to have the various policies work alongside each other to
ensure macroeconomic and financial stability while raising long-term sustainable
growth. At a minimum, such a framework would also reduce the risk that different
policies work at cross purposes.
Designing such a framework raises difficult analytical and practical issues. Some
general observations are offered here.
An important element of a macro-financial stability framework is monetary
policy. Interest rates directly affect both asset prices and borrowers’ willingness and
ability to take on leverage. In addition, and partly for the same reasons, they appear
to affect economic agents’ risk-taking.50 As a result, monetary policy influences the
financial cycle and systemic risk and, through these, macroeconomic fluctuations.
This is the case whether or not it operates through interest rates, balance sheet
policies or foreign exchange intervention (Box IV.B).51 At the very least, therefore,
monetary and macroprudential authorities need to take into account each other’s
actions when making decisions.
How far monetary policy should go in taking financial stability considerations
into account is controversial. The answer depends on a range of factors, including
the degree to which monetary policy affects risk-taking, debt and asset prices; the
effectiveness of macroprudential actions; the particular nature of the risks; and the
secondary effects of taking action.52 For instance, macroprudential measures can
advantageously target more granular risks, such as in the mortgage sector, while
monetary policy has a more pervasive impact, thus limiting leakage and regulatory
arbitrage.53 Similarly, in more open economies, higher interest rates have the
disadvantage of encouraging more capital inflows and exchange rate appreciation,
which could offset at least in part their restraining influence on the build-up of
financial imbalances. By contrast, macroprudential measures do not suffer from this
limitation. Moreover, it is precisely in this context that foreign exchange intervention
can contribute to the design of a more balanced policy response. It can do so by
building up buffers for use when the tide turns and by dampening the expansionary
impact of an exchange rate appreciation on capital inflows and the build-up of
imbalances (Box IV.B and Chapter II).54 That said, there is always some tension when
operating macroprudential and monetary policy measures in opposite directions, as
when interest rates are reduced to address inflation concerns while macroprudential
measures are tightened to restrain the build-up of financial imbalances.
These are just some of the many considerations that need to be taken into
account when evaluating the role of monetary policy and macroprudential
measures in a macro-financial stability framework. They help explain the range of
possible roles monetary policy can in principle play, from serving simply as a
backstop for more targeted macroprudential measures to playing a much more
prominent part. What is clear is that a more active role for monetary policy requires

a flexible interpretation of inflation objectives, so as to better reconcile near-term
stabilisation objectives with longer-run financial and hence macroeconomic
stability. This, in turn, calls for longer horizons than the widely adopted two-year
ones. These refinements would provide necessary room for manoeuvre to address
the slower-moving build-up of financial vulnerabilities linked to the financial cycle
(Chapter I).55
Fiscal policy is another key element of a broader macro-financial stability
framework.56 Relatively tight fiscal policy may help restrain the build-up of
imbalances during financial booms. More importantly, it will build buffers that can
be drawn upon to dampen the real effects of financial downturns. For example, the
loosening of fiscal policy was crucial in supporting output during the GFC.
Sufficiently large fiscal buffers can also help prevent the sovereign itself being a
source of financial instability, as has happened recently in several European
economies. But running a sound fiscal policy during a financial boom is challenging
for a number of reasons.57 First, financial booms create revenues that will not be
there when the boom stops. Second, financial booms also result in an overestimation
of potential output and growth. It is therefore very important, when assessing
structural deficits, to allow for the flattering effect of financial booms on public
finances. Third, financial booms may create hidden contingent liabilities, including
the fiscal costs of dealing with financial instability, or lower revenues and higher
expenditures from an ensuing recession.
Often financial imbalances result at least in part from distortions in the tax
system or the real sector. For example, the tax codes of most economies favour
debt over equity, creating incentives for leverage.58 Changing the tax code to
reduce this bias could result in less vulnerable funding structures. In the case of
housing booms, measures that expand the supply of land or encourage construction
could have a more lasting impact on property prices and, indirectly, mortgage
credit than higher interest rates or tighter macroprudential measures. Similarly,
authorities may limit speculative activity in the shape of rapid turnover (or “flipping”)
in the housing market by raising transaction taxes or stamp duties, rather than
through macroprudential or monetary policy measures.59
Putting together a comprehensive macro-financial stability framework naturally
raises coordination issues. In some cases, it might be enough to merely take into
account decisions in other policy spheres, just as monetary policymakers take the
fiscal stance as given when deciding on interest rates. In other instances, the case
for coordination is stronger. Coordination between the monetary authority and
other authorities, such as macroprudential and fiscal, may also broaden political
and social support.
Coordination may become even more difficult when it has to take place across
borders. While the literature rightly cautions that not all forms of cross-border
spillover call for policy intervention,60 they sometimes do. For macroprudential
measures, the case for international coordination is perhaps easiest to make in
preventing cross-border arbitrage, ie “leakages”. The Basel agreement on reciprocity
in the implementation of the Basel III CCyB is one such example. Designed to
overcome the problems associated with global banks bypassing national regulations
on capital requirements, this agreement stipulates that when the CCyB is activated
in any given country, all countries are expected to apply the same buffer on
exposures to that country from banks in their jurisdiction.61 But there may be other
cases too. For example, when a country introduces prudential measures to reduce
systemic risks related to FX borrowing, capital flows may be diverted to another
country in the same region. In this case, coordination would mean that both
countries tighten such prudential measures, ie that their actions complement each
other.62

Box IV.B
FX interventions in a macroprudential context
The frequency and size of capital flow surges and reversals in EMEs have increased over the past three decades.
Such surges and reversals pose macro-financial stability risks by significantly raising the volatility of exchange rates
and interest rates as well as the risk of financial crises. This raises the questions of how to respond, and how best to
combine policies as part of a holistic macro-financial stability framework. This box considers what role foreign
exchange intervention can play.
FX intervention can help underpin financial stability in two ways. First, intervening in response to capital inflows
can help build international reserves that can be deployed when tides turn. Second, intervention may constrain the
build-up of financial imbalances. All else equal, an appreciating exchange rate tends to improve the creditworthiness
of domestic borrowers and thus open the door for more borrowing. This is most obvious if debt is denominated in
foreign currency and assets are denominated in the domestic currency. In this case, an appreciation of the exchange
rate reduces the value of this debt relative to domestic assets and income. But the effect may also be felt even in the
absence of currency mismatches. An exchange rate appreciation tilts the relative value of domestic versus foreign assets
that could serve as collateral, thus making international banks and institutional investors more willing to lend.
EMEs have frequently used FX intervention to mitigate the effects of external conditions on the domestic
economy, especially those of exchange rate and capital flow volatility. Many cross-country studies on the
effectiveness of sterilised FX intervention in EMEs find evidence that it has tempered exchange rate appreciation in
response to gross inflows. By doing so, intervention can also weaken the impact of foreign financial conditions on
domestic credit and thus reduce systemic risk. Indeed, Graph IV.B shows that sterilised FX intervention tends to
offset the impact of capital inflows on domestic credit growth.
In contrast to restrictions on capital flows, FX intervention works directly on the source of shocks, ie the
exchange rate, rather than directly discouraging inflows. However, FX intervention does not always work well. While
it helps build buffers and neutralise the exchange rate channel, it does not offset the direct effect of inflows on
debt. In general, intervention works better when the inflow is less persistent and less sensitive to return differentials.
Therefore, FX intervention could be best regarded as a complement to other policies, such as interest rate policy
and domestic macroprudential measures that EMEs can use to maintain macro-financial stability.
FX intervention to smooth a depreciation of the domestic currency in the face of capital outflows has to be
communicated properly in order to be effective. In particular, national authorities should emphasise the macroprudential
Sterilised FX intervention and domestic credit growth

In percentage points Graph IV.B
Impact on the domestic credit-to-GDP ratio1 Impact on real domestic credit growth2
*** 0.2 0.08

***
0.1 0.04
0.0 0.00
–0.1 –0.04
**
–0.2 –0.08
***
–0.3 –0.12
FX reserve accumulation/GDP Net capital flows/GDP FX purchase/GDP Net capital flows/GDP
**/*** indicates statistical significance at the 5/1% level.

1
This panel shows the coefficient of the variables on the horizontal axis from a BIS panel regression analysis for 20 EMEs from 2000 to 2017,
where the dependent variable is the change in the ratio of domestic credit to GDP and the control variables are the lagged dependent variable,
the US dollar exchange rate, the real domestic money market rate, country fixed effects and time fixed effects. 2 This panel shows the
coefficient of the variables on the horizontal axis from a panel regression analysis for 45 EMEs from 2005 to 2013 reported in
specification (7) in Table 9.2 of Ghosh et al (2017).
Sources: Ghosh et al (2017); BIS calculations.

objective of such actions, making sure that they are not interpreted as a lack of conviction on the part of the authorities
to make other, more fundamental policy adjustments when they are required, as in the monetary or fiscal space.
  See Bruno and Shin (2015a,b).      See Blanchard et al (2015) and Daude et al (2016). Fratzscher et al (2017) examine foreign exchange
intervention based on daily data covering 33 AEs and EMEs from 1995 to 2011, and find that intervention works well in terms of smoothing
the path of exchange rates, and stabilising the exchange rate in countries with narrow band regimes.      This is in line with recent studies
using a variety of methodologies, For cross-country evidence, see Ghosh et al (2017). Hofmann et al (2018) look at the micro data of
Colombia and find that sterilised FX interventions counter the procyclical effects of capital inflows on bank lending. Using Korean bank-
level data, Yun (2018) finds that, facing reserve accumulation, primary dealer banks and foreign bank branches reduced lending more than
non-primary dealer banks and domestic banks, respectively.      For details, see Ghosh et al (2017).

Endnotes
1
In line with agreed terminology, this chapter defines macroprudential frameworks as those that use
(primarily) prudential tools to target specifically systemic risk and thus mitigate the impact of
systemic events on the macroeconomy. The operational objectives of existing macroprudential
frameworks have been to strengthen the resilience of financial systems and dampen the financial
booms and busts at the heart of much of the financial instability seen historically. For more details,
see FSB-IMF-BIS (2011a,b, 2016).
2
For an early in-depth analysis of the concept of procyclicality and its implications, see Borio et al
(2001).
3
See Crockett (2000). Clement (2010) traces the term “macroprudential” back to a submission of the
Bank of England to the Cooke Committee, the precursor of the Basel Committee on Banking
Supervision. Borio (2003) sought to clarify its contours more precisely.
4
This graph is based on 845 macroprudential measures taken by 55 economies over 1995–2018.
5
At their meeting in Seoul in November 2010, G20 leaders asked the FSB, the IMF and the BIS to
undertake further work on macroprudential policy. See FSB-IMF-BIS (2011a,b, 2016) for summaries
of this work.
6
For examples, see Table 3 in CGFS (2016).
7
See Fender and Lewrick (2016) for a recent review of estimates of the costs of financial distress.
8
See Reinhart and Rogoff (2009).
9
For an overview, see Claessens and Kose (2018).
10
See Aldasoro et al (2018) for a recent contribution and further references. While credit gaps also
have predictive power for EMEs, the case where credit grows exponentially over an extended
period starting from a very low level may not be comparable with the deviation of credit in an
advanced economy, which tends to behave more cyclically.
11
Moreover, the complexity and interconnections that give rise to systemic risk are often the result of
financial intermediation having grown large. See Shin (2017).
12
See Anderson et al (2018) for details on the macroprudential stress tests conducted by major
advanced economy central banks and international organisations; and Arslan and Upper (2017) for
the BIS survey on practices in EMEs.
13
Over time, testing methodologies have started to incorporate feedback effects through contagion
between firms, or through the interaction between the economy’s financial and real sectors. But
these second-round effects tend to be mechanistic, failing to capture the behaviour of firms or
banks.
14
For a critical assessment of stress tests, see Borio et al (2014).
15
Non-prudential instruments need to be specifically targeted at systemic risk and underpinned by
governance arrangements that prevent any slippage in order to be considered macroprudential.
See FSB-IMF-BIS (2011b).
16
See CGFS (2010), especially Table 1, Crowe et al (2013) and Claessens (2015) for mappings from
particular vulnerabilities to tools.
17
The restrictions were applied to auto loans with long maturities and high loan-to-value ratios. See
Costa de Moura and Martins Bandeira (2017) for more details.
18
The central bank responded by introducing tighter loan-to-value caps on housing loans to firms
too. See Central Bank of Malaysia (2017).
19
See Cizel et al (2016) and Bruno et al (2017).

20
A multi-instrument approach is also consistent with the theory of policymaking under uncertainty
developed by Brainard (1967), which suggests that policymakers should use all available tools –
but cautiously – to mitigate the effects of uncertainty.
21
Indeed, some of the measures used for macroprudential purposes, for instance some credit
restrictions, were originally introduced to allocate credit.
22
See eg Allen et al (2017).
23
For example, the Bank of England set the countercyclical capital buffer (CCyB) so that the sum of
the 2.5% Basel III capital conservation buffer and the CCyB was equivalent to the average loss of
3.5% of banks’ risk-weighted assets as revealed by the Bank’s stress test. See Bank of England
(2017).
24
For example, the Bank of France uses dynamic stochastic general equilibrium models with several
macro-financial variables to calibrate a rule that links the CCyB to macroeconomic developments.
25
The country studies in BIS Papers, no 94, provide many examples.
26
See Bahaj and Foulis (2017), who relax Brainard’s (1967) assumptions that the costs of missing the
target are symmetrical.
27
See CGFS (2016) and Patel (2017).
28
The inaccessibility is only partly due to the nature of the issues. Textual analysis finds that many
central banks use overly complex language. See Patel (2017).
29
CGFS (2016) provides extensive discussion of communication as an instrument, including many
practical examples.
30
See Alegría et al (2017). For counterexamples, see CGFS (2016).
31
See Powell (2018) for discussions on the role of public transparency and accountability for both
financial stability and monetary policy.
32
See FSB-IMF-BIS (2011b).
33
See Villar (2017).
34
At one extreme, the European Systemic Risk Board (ESRB) has 78 member institutions and three
observers, although the ESRB has formal procedures for conducting macroprudential policies.
35
For examples, see Gambacorta and Murcia (2017) and the country studies in BIS Papers, no 94.
36
See eg Aguirre and Repetto (2017), Altunbas et al (2018) and Gómez et al (2017).
37
The analysis uses the sample of macroprudential measures described in Table IV.A1. In line with
most other cross-country studies, it defines dummy variables for tightening (+1) and loosening
(–1) actions. Recently, a small number of papers have attempted to capture the intensity of policy
actions considering the size (and sometimes even the scope) of changes in regulatory ratios. See
Glocker and Towbin (2015), Vandenbussche et al (2015) and Richter et al (2018). See Galati and
Moessner (2017) for a recent review of the effectiveness of macroprudential measures.
38
See eg Cerutti et al (2017), Gambacorta and Murcia (2017), Kuttner and Shim (2016) and Lim et al
(2011). Many country-level studies also reach similar conclusions. For example, see Igan and Kang
(2011) for Korea and Wong et al (2011) for Hong Kong SAR.
39
See Arslan and Upper (2017).
40
For example, Claessens et al (2013) use a sample of around 2,800 banks in 48 countries over the
period 2000–10 and find that maximum LTV and DSTI ratios as well as limits on credit growth and
foreign currency lending have reduced bank leverage and asset growth during booms. By contrast,
they find that few policies have helped to stop declines in bank leverage and assets during
downturns.

41
Lim et al (2011) obtain similar results.
42
In particular, the top left-hand panel of Graph IV.4 shows that policy actions which tighten capital
flow- or FX liability-based reserve requirements or credit growth- or asset-based marginal reserve
requirements significantly increased real general bank credit growth. Empirical studies on the
impact of reserve requirements also show mixed results.
43
See Takáts and Upper (2013).
44
For the discussion on the appropriate criteria, see Borio (2014). Jiménez et al (2017) find that the
ability of Spanish banks to keep lending during the GFC depended on how much capital they had
put aside under the automatic dynamic provisioning programme.
45
See Aldasoro et al (2018).
46
See Auer and Ongena (2016).
47
For country-specific studies, see Kelber and Monnet (2014), Aikman et al (2016) and Monnet
(2014). For cross-country studies, see Sánchez and Röhn (2016), Boar et al (2017), Kim and
Mehrotra (2018) and Richter et al (2018).
48
See Agénor and Pereira da Silva (2018) for a review.
49
See eg Buch and Goldberg (2017), Hoggarth et al (2013), Ongena et al (2013), Reinhardt and
Sowerbutts (2015) and Tripathy (2017).
50
This is the risk-taking channel of monetary policy first introduced by Borio and Zhu (2012). For
further evidence, see Jiménez et al (2012). For a critique, see Svensson (2017).
51
To illustrate the use of two monetary policy tools – the policy rate and sterilised FX intervention –
under imperfect capital mobility to stabilise inflation, the output gap and the exchange rate, see
Blanchard (2012).
52
The strength of the risk-taking channel of monetary policy is controversial. This is important
because models in which this channel is strong tend to indicate that monetary policy should
include a financial stability objective, while models in which it is absent tend to suggest that it
should not (Adrian and Liang (2018)). For an overview of the arguments of whether monetary
policy should lean against the development of financial imbalances, see IMF (2015), Filardo and
Rungcharoenkitkul (2016) and Adrian and Liang (2018) as well as references therein.
53
As succinctly put by former US Federal Reserve Governor Jeremy Stein, “monetary policy gets into
all the cracks” (Stein (2013)). See also Crockett (2000), Borio and Lowe (2002), Crowe et al (2013)
and Blanchard et al (2013).
54
Capital flow management (CFM) tools used for prudential purposes can complement FX
intervention in dealing with capital flows and thus financial imbalances. Recent empirical studies
generally show that CFM tools are sometimes effective in slowing down targeted flows but that
the effects tend to be temporary and leakages abound. Such CFM tools are often used when other
types of tool do not successfully moderate capital flows. Moreover, there is no consensus on which
types of CFM tool are macroprudential and which are not.
55
For instance, Drehmann et al (2012) document how the equity price crashes in 1987 and 2001, and
the associated economic slowdowns or mild recessions, did not stop the expansion of the financial
cycle, as credit growth and property price increases continued. When the financial cycle turned a
few years later, it ushered in financial stress and a more severe recession – what the authors term
the “unfinished recession” phenomenon. Presumably, the monetary policy easing in response to
the equity crashes and economic slowdowns contributed to the financial cycle expansion at the
time. See also eg Borio and White (2004) and Beau et al (2014). In turn, Juselius et al (2017), by
estimating a model of the economy that embeds an articulated version of the financial cycle
(Juselius and Drehmann (2015)), find that an augmented Taylor rule which also includes a financial
cycle proxy could have improved both output and inflation performance over longer horizons
since the 1990s.

56
See Chapter V of BIS (2016).
57
See eg Borio et al (2016).
58
See Box V.C in BIS (2016).
59
Crowe et al (2013) and Kuttner and Shim (2016) find that such measures tend to have a sizeable
impact on both housing credit and house prices.
60
Korinek (2017) sets out three conditions of which at least one needs to be violated to generate
inefficiency and scope for cooperation: (i) policymakers act competitively in the international
market; (ii) they have sufficient external policy instruments; and (iii) international markets are free
of imperfections. If one of these conditions is violated, then international cooperation can improve
welfare. For a discussion of the need for international cooperation on monetary policy, see BIS
(2015).
61
A special case is coordination in multilayered jurisdictions such as the European Union, where
multilateral institutions such as the European Systemic Risk Board and the Single Supervisory
Mechanism have some directive powers over national bodies.
62
See Agénor and Pereira da Silva (2018) and Agénor et al (2017).

References
Adrian, T and N Liang (2018): “Monetary policy, financial conditions, and financial stability”, International
Journal of Central Banking, vol 14, no 1, January, pp 73–131.
Agénor, P-R, E Kharroubi, L Gambacorta, G Lombardo and L Pereira da Silva (2017): “The international
dimensions of macroprudential policies”, BIS Working Papers, no 643, June.
Agénor, P-R and L Pereira da Silva (2018): “Financial spillovers, spillbacks and the scope for international
macroprudential policy coordination”, BIS Papers, no 97, April.
Aguirre, H and G Repetto (2017): “Capital and currency-based macroprudential policies: an evaluation
using credit-registry data”, BIS Working Papers, no 672, November.
Aikman, D, O Bush and A Taylor (2016): “Monetary versus macroprudential policies: causal impacts of
interest rates and credit controls in the era of the UK Radcliffe Report”, Bank of England, Staff Working
Papers, no 610.
Aldasoro, I, C Borio and M Drehmann (2018): “Early warning indicators of banking crises: expanding the
family”, BIS Quarterly Review, March, pp 29–45.
Alegría, A, R Alfaro and F Córdova (2017): “The impact of warnings published in a financial stability
report on loan-to-value ratios”, BIS Working Papers, no 633, May.
Allen, J, T Grieder, T Roberts and B Peterson (2017): “The impact of macroprudential housing finance
tools in Canada”, BIS Working Papers, no 632, May.
Altunbas, Y, M Binici and L Gambacorta (2018): “Macroprudential policy and bank risk”, Journal of
International Money and Finance, vol 81, pp 203–20.
Anderson, R, C Baba, J Danielsson, U Das, H Kang and M Segoviano (2018): Macroprudential stress tests
and policies: searching for robust and implementable frameworks, London School of Economics Systemic
Risk Centre.
Arslan, Y and C Upper (2017): “Macroprudential frameworks: implementation and effectiveness”, BIS
Papers, no 94, December, pp 25–47.
Auer, R and S Ongena (2016): “The countercyclical capital buffer and the composition of bank lending”,
BIS Working Papers, no 593, December.
Bahaj, S and A Foulis (2017): “Macroprudential policy under uncertainty”, International Journal of Central
Banking, vol 13, no 3, pp 119–54.
Bank for International Settlements (2015): 85th Annual Report, June, Chapter V.
——— (2016): 86th Annual Report, June.
Bank of England (2017): “Overview of risks to UK financial stability and UK countercyclical capital
buffer”, Financial Stability Review, November, pp 1–7.
Beau, D, C Cahn, L Clerc and B Mojon (2014): “Macro-prudential policy and the conduct of monetary
policy”, in S Bauducco, L Christiano and C Raddatz (eds), Macroeconomic and financial stability:
challenges for monetary policy, Central Bank of Chile.
Blanchard, O (2012): “Monetary policy in the wake of the crisis”, in O Blanchard, D Romer, M Spence and
J Stiglitz (eds), In the wake of the crisis: leading economists reassess economic policy, MIT Press.
Blanchard, O, G Adler and I de Carvalho Filho (2015): “Can foreign exchange intervention stem exchange
rate pressures from global capital flow shocks?”, NBER Working Papers, no 21427, July.
Blanchard, O, G Dell’Ariccia and P Mauro (2013): “Rethinking macro policy II: getting granular”, IMF Staff
Discussion Notes, no 13/03, April.

Boar, C, L Gambacorta, G Lombardo and L Pereira da Silva (2017): “What are the effects of
macroprudential policies on macroeconomic performance?”, BIS Quarterly Review, September, pp 71–88.
Borio, C (2003): “Towards a macroprudential framework for financial supervision and regulation?”, BIS
Working Papers, no 128, February.
——— (2004): “Market distress and vanishing liquidity: anatomy and policy options”, BIS Working
Papers, no 158, July.
——— (2014): “Macroprudential frameworks: (too) great expectations?”, Central Banking Journal, 25th
anniversary issue, August.
Borio, C, M Drehmann and K Tsatsaronis (2014): “Stress-testing macro stress testing: does it live up to
expectations?”, Journal of Financial Stability, vol 12, pp 3–15.
Borio, C, C Furfine and P Lowe (2001): “Procyclicality of the financial system and financial stability: issues
and policy options”, BIS Papers, no 1, March, pp 1–57.
Borio, C, M Lombardi and F Zampolli (2016): “Fiscal sustainability and the financial cycle”, BIS Working
Papers, no 552, March.
Borio, C and P Lowe (2002): “Asset prices, financial and monetary stability: exploring the nexus”, BIS
Working Papers, no 114, July.
Borio, C and W White (2004): “Whither monetary and financial stability: the implications for evolving
policy regimes”, BIS Working Papers, no 147, February.
Borio, C and H Zhu (2012): “Capital regulation, risk-taking and monetary policy: a missing link in the
transmission mechanism?”, Journal of Financial Stability, vol 8, no 4, pp 235–51.
Brainard, W (1967): “Uncertainty and the effectiveness of policy”, American Economic Review Papers and
Proceedings, vol 57, pp 411–25.
Bruno, V, I Shim and H S Shin (2017): “Comparative assessment of macroprudential policies”, Journal of
Financial Stability, vol 28, pp 183–202.
Bruno, V and H S Shin (2015a): “Cross-border banking and global liquidity”, Review of Economic Studies,
vol 82, no 2, pp 535–64.
——— (2015b): “Capital flows and the risk-taking channel of monetary policy”, Journal of Monetary
Economics, vol 71, pp 119–32.
Buch, C and L Goldberg (2017): “Cross-border prudential policy spillovers: how much? how important?
Evidence from the international banking research network”, International Journal of Central Banking,
vol 13, no 2, pp 505–58.
Budnik, K and J Kleibl (2018): “Macroprudential regulation in the European Union in 1995–2014:
introducing a new data set on policy actions of a macroprudential nature”, ECB Working Papers,
no 2123, January.
Central Bank of Malaysia (2017): “Macroprudential frameworks: implementation, and relationship with
other policies – Malaysia”, BIS Papers, no 94, December, pp 231–8.
Cerutti, E, S Claessens and L Laeven (2017): “The use and effectiveness of macroprudential policies: new
evidence”, Journal of Financial Stability, vol 28, pp 203–24.
Cizel, J, J Frost, A Houben and P Wierts (2016): “Effective macroprudential policy: cross-sector
substitution from price and quantity measures”, IMF Working Papers, no 16/94.
Claessens, S (2015): “An overview of macroprudential policy tools”, Annual Review of Financial
Economics, vol 7, no 1, December, pp 397–422.
Claessens, S, S Ghosh and R Milhet (2013): “Macro-prudential policies to mitigate financial system
vulnerabilities”, Journal of International Money and Finance, vol 39, pp 153–85.

Claessens, S and M Kose (2018): “Frontiers of macro-financial linkages”, BIS Papers, no 95, January.
Clement, P (2010): “The term ‘macroprudential’: origins and evolution”, BIS Quarterly Review, March,
pp 59–67.
Committee on the Global Financial System (2010): “Macroprudential instruments and frameworks: a
stocktaking of issues and experiences”, CGFS Papers, no 38.
——— (2016): “Objective-setting and communication of macroprudential policies”, CGFS Papers, no 57.
Costa de Moura, M and F Martins Bandeira (2017): “Macroprudential policy in Brazil”, BIS Papers, no 94,
December, pp 77–86.
Crockett, A (2000): “In search of anchors for financial and monetary stability”, speech at the SUERF
Colloquium in Vienna, 27–29 April.
Crowe, C, G Dell’Ariccia, D Igan and P Rabanal (2013): “How to deal with real estate booms: lessons from
country experiences”, Journal of Financial Stability, vol 9, pp 300–19.
Daude, C, E Levy-Yeyati and A Nagengast (2016): “On the effectiveness of exchange rate interventions
in emerging markets”, Journal of International Money and Finance, vol 64, pp 239–61.
Drehmann, M, C Borio and K Tsatsaronis (2012): “Characterising the financial cycle: don’t lose sight of
the medium term!”, BIS Working Papers, no 380, June.
Fender, I and U Lewrick (2016): “Adding it all up: the macroeconomic impact of Basel III and outstanding
reform issues”, BIS Working Papers, no 591, November.
Filardo, A and P Rungcharoenkitkul (2016): “A quantitative case for leaning against the wind”, BIS
Working Papers, no 594, December.
Financial Stability Board (2017): Policy recommendations to address structural vulnerabilities from asset
management activities, 12 January.
Financial Stability Board, International Monetary Fund and Bank for International Settlements (2011a):
Macroprudential policy tools and frameworks: update to G20 Finance Ministers and Central Bank
Governors, February.
——— (2011b): Macroprudential policy tools and frameworks, progress report to the G20, October.
——— (2016): Elements of effective macroprudential policies: lessons from international experience,
report to the G20, August.
Fratzscher, M, O Gloede, L Menkhoff, L Sarno and T Stöhr (2017): “When is foreign exchange intervention
effective? Evidence from 33 countries”, CEPR Discussion Papers, no 12510.
Galati, G and R Moessner (2017): “What do we know about the effects of macroprudential policy?”,
Economica, February, pp 1–36.
Gambacorta, L and A Murcia (2017): “The impact of macroprudential policies and their interaction with
monetary policy: an empirical analysis using credit registry data”, BIS Working Papers, no 636, May.
Ghosh, A, J Ostry and M Qureshi (2017): Taming the tide of capital flows: a policy guide, MIT Press.
Glocker, C and P Towbin (2015): “Reserve requirements as a macroprudential instrument: empirical

evidence from Brazil”, Journal of Macroeconomics, vol 44 (Supplement C), pp 158–76.
Gómez, E, A Lizarazo, J Mendoza and A Murcia (2017): “Evaluating the impact of macroprudential
policies on credit growth in Colombia”, BIS Working Papers, no 634, May.
Hofmann, B, H S Shin and M Villamizar-Villegas (2018): “Sterilised foreign exchange intervention as

reverse QE”, unpublished manuscript.

Hoggarth, G, J Hooley and Y Korniyenko (2013): “Which way do foreign branches sway? Evidence from
the recent UK domestic credit cycle”, Bank of England, Financial Stability Papers, no 22, June.
Igan, D and H-D Kang (2011): “Do loan-to-value and debt-to-income limits work? Evidence from Korea”,
IMF Working Papers, no 11/297.
International Monetary Fund (2015): “Monetary policy and financial stability”, Staff Report, August.
Jiménez, G, S Ongena, J-L Peydró and J Saurina (2012): “Credit supply and monetary policy: identifying
the bank balance-sheet channel with loan applications”, American Economic Review, vol 102, no 5,
pp 2301–26.
——— (2017): “Macroprudential policy, countercyclical bank capital buffers, and credit supply: evidence
from the Spanish dynamic provisioning experiments”, Journal of Political Economy, vol 125, no 6,
pp 2126–77.
Juselius, M, C Borio, P Disyatat and M Drehmann (2017): “Monetary policy, the financial cycle, and ultra-
low interest rates”, International Journal of Central Banking, vol 13, no 3, September, pp 55–89.
Juselius, M and M Drehmann (2015): “Leverage dynamics and the real burden of debt”, BIS Working
Papers, no 501, May.
Kelber, A and E Monnet (2014): “Macroprudential policy and quantitative instruments: a European
historical perspective”, Bank of France, Financial Stability Review, April, pp 165–74.
Kim, S and A Mehrotra (2018): “Effects of monetary and macroprudential policies – evidence from four
inflation targeting economies”, Journal of Money, Credit and Banking, forthcoming.
Korinek, A (2017): “Currency wars or efficient spillovers? A general theory of international policy
cooperation”, BIS Working Papers, no 615, March.
Kuttner, K and I Shim (2016): “Can non-interest rate policies stabilize housing markets? Evidence from a
panel of 57 economies”, Journal of Financial Stability, vol 26, pp 31–44.
Lim, C, F Columba, A Costa, P Kongsamut, A Otani, M Saiyid, T Wezel and X Wu (2011): “Macroprudential
policy: what instruments and how to use them”, IMF Working Papers, no 11/238.
Monnet, E (2014): “Monetary policy without interest rates: evidence from France’s Golden Age (1948 to
1973) using a narrative approach”, American Economic Journal: Macroeconomics, vol 6, no 4, pp 137–69.
Ongena, S, A Popov and G Udell (2013): “When the cat’s away the mice will play: does regulation at
home affect bank risk-taking abroad?”, Journal of Financial Economics, vol 108, no 3, pp 727−50.
Patel, N (2017): “Macroprudential frameworks: communication”, BIS Papers, no 94, December, pp 49–56.
Powell, J (2018): “Financial stability and central bank transparency”, speech at 350 years of central
banking: the past, the present and the future, Sveriges Riksbank anniversary conference, Stockholm,
25 May.
Reinhardt, D and R Sowerbutts (2015): “Regulatory arbitrage in action: evidence from banking flows and
macroprudential policy”, Bank of England, Staff Working Papers, no 546, September.
——— (2016): “Macroprudential policies: a granular database”, unpublished manuscript.
Reinhart, C and K Rogoff (2009): This time is different: eight centuries of financial folly, Princeton
University Press.
Richter, B, M Schularick and I Shim (2018): “Output effects of macroprudential policy”, unpublished
manuscript.
Sánchez, A and O Röhn (2016): “How do policies influence GDP tail risks?”, OECD Economics Department
Working Papers, no 1339.

Securities and Exchange Commission (2016): Final Rule: Investment Company Liquidity Risk Management
Programmes, release no 33-10233, 13 October.
Shim, I, B Bogdanova, J Shek and A Subelyte (2013): “Database for policy actions on housing markets”,
BIS Quarterly Review, September, pp 83–91.
Shin, H S (2017): “Leverage in the small and in the large”, panel remarks at the IMF Annual Meeting
seminar on systemic risk and macroprudential stress testing, Washington DC, 10 October.
Stein, J (2013): “Overheating in credit markets: origins, measurement and policy responses”, speech at
the research symposium on Restoring household financial stability after the Great Recession: why
household balance sheets matter sponsored by the Federal Reserve Bank of St Louis, St Louis, Missouri,
7 February.
Svensson, L (2017): “The relation between monetary policy and financial stability policy”, paper
presented at the XXI Annual Conference of the Central Bank of Chile on Monetary policy and financial
stability: transmission mechanisms and policy implications, November.
Takáts, E and C Upper (2013): “Credit and growth after financial crises”, BIS Working Papers, no 416, July.
Tripathy, J (2017): “Cross-border effects of regulatory spillovers: evidence from Mexico”, Bank of
England, Staff Working Papers, no 684.
Vandenbussche, J, U Vogel and E Detragiache (2015): “Macroprudential policies and housing prices: a
new database and empirical evidence for central, eastern, and southeastern Europe”, Journal of Money,
Credit and Banking, vol 47 (S1), pp 343–77.
Villar, A (2017): “Macroprudential frameworks: objectives, decisions and policy interactions”, BIS Papers,
no 94, December, pp 7–24.
Wong, E, T Fong, K Li and H Choi (2011): “Loan-to-value ratio as a macroprudential tool: Hong Kong’s
experience and cross-country evidence”, Hong Kong Monetary Authority, Working Papers, no 01/2011.
Yun, Y (2018): “Reserve accumulation and bank lending: evidence from Korea”, Bank of Korea, Working
Papers, no 2018–15.

Use of macroprudential measures by targeted credit, instrument type and region
Number of policy actions, 1995–2018 Table IV.A1
Region1
Asia- Central Latin Middle North Western All
Pacific and eastern America East and America Europe economies
Targeted credit Europe Africa
Instrument type [11] [14] [6] [4] [2] [18] [55]
General credit 31 156 68 5 – 56 316
Countercyclical capital buffers 3 4 – – – 6 13
Non-cyclical (structural) systemic risk
1 5 2 – – 11 19
capital surcharges (D-SIB, O-SII, SRB)2
Other capital surcharges3 – 14 4 – – 15 33
Loan loss provisioning rules (general,
9 32 5 – – 6 52
specific, dynamic, statistical, FX loans)
Limits on FX mismatch, position or
8 32 15 1 – 7 63
liquidity
Capital inflow- or FX liability-based
5 44 17 4 – – 70
reserve requirements
Credit growth- or asset-based marginal
– 24 25 – – 6 55
reserve requirements
Credit growth limits4 5 1 – – – 5 11
Housing/consumer/household credit 168 125 24 13 13 114 457
Countercyclical capital buffers
– – – – – 2 2
(housing credit)
LTV limits and loan prohibitions 76 37 9 4 7 35 168
DSTI, DTI limits and other lending
49 34 4 3 6 23 119
criteria
Risk weights 17 40 8 4 0 42 111
Loan loss provisioning rules 15 3 3 2 0 10 33
Exposure limits 11 9 – – – 1 21
Limits on FX mismatch or FX loans – 2 – – – 1 3
Corporate credit (including CRE loans) 18 19 2 – – 24 63
LTV limits and loan prohibitions 2 3 – – – 4 9
DSTI, DTI limits and other lending
3 2 – – – – 5
criteria
Risk weights 2 12 – – – 18 32
Loan loss provisioning rules 5 1 1 – – – 7
Exposure limits 1 1 – – – 2 4
Limits on FX mismatch or FX loans 5 – 1 – – – 6
Credit to financial institutions 5
2 2 2 – – 3 9
Total 219 302 96 18 13 197 845
Memo items: Total 158 219 52 18 – 66 513
General liability-based average reserve
115 159 50 17 – 34 375
requirements
Liquidity requirements (LCR, NSFR,
43 60 2 1 – 32 138
liquid asset ratio, loan-to-deposit ratio) 6

Asia-Pacific = AU, CN, HK, ID, IN, KR, MY, NZ, PH, SG and TH; central and eastern Europe = BG, CZ, EE, HR, HU, LT, LV, PL, RO, RS, RU, SI, SK and
TR; Latin America = AR, BR, CL, CO, MX and PE; Middle East and Africa = AE, IL, SA and ZA; North America = CA and US; western Europe = AT,
BE, CH, DE, DK, ES, FI, FR, GB, GR, IE, IS, IT, LU, NL, NO, PT and SE.
1
  The figures in square brackets indicate the number of economies in each region.    2  D-SIB = domestic systemically important bank; O-SII =
other systemically important institution; SRB = systemic risk buffer.    3  On trading activities, FX exposures and others.    4  Growth limit on
general, housing or consumer credit.    5  Including limits on interbank exposure, exposure limits on non-bank financial institutions and risk
weights on exposure to financial institutions.    6  LCR = Liquidity Coverage Ratio; NSFR = Net Stable Funding Ratio.
Sources: Budnik and Kleibl (2018); Reinhardt and Sowerbutts (2016); Shim et al (2013); national data; BIS calculations.

V. Cryptocurrencies: looking beyond the hype
Less than 10 years after their inception, cryptocurrencies1 have emerged from
obscurity to attract intense interest on the part of businesses and consumers, as
well as central banks and other authorities. They garner attention because they
promise to replace trust in long-standing institutions, such as commercial and
central banks, with trust in a new, fully decentralised system founded on the
blockchain and related distributed ledger technology (DLT).
This chapter evaluates whether cryptocurrencies could play any role as money:
looking beyond the hype, what specific economic problems, if any, can current
cryptocurrencies solve? The chapter first reviews the historical context. Many
episodes of monetary instability and failed currencies illustrate that the institutional
arrangements through which money is supplied matter a great deal. This review
shows that the essence of good money has always been trust in the stability of its
value. And for money to live up to its signature property – to act as a coordination
device facilitating transactions – it needs to efficiently scale with the economy and
be provided elastically to address fluctuating demand. These considerations call for
specific institutional arrangements – hence the emergence of today’s independent
and accountable central banks.
The chapter then gives an introduction to cryptocurrencies and discusses the
economic limitations inherent in the decentralised creation of trust which they
entail. For the trust to be maintained, honest network participants need to control
the vast majority of computing power, each and every user needs to verify the history
of transactions and the supply of the cryptocurrency needs to be predetermined
by its protocol. Trust can evaporate at any time because of the fragility of the
decentralised consensus through which transactions are recorded. Not only does
this call into question the finality of individual payments, it also means that a
cryptocurrency can simply stop functioning, resulting in a complete loss of value.
Moreover, even if trust can be maintained, cryptocurrency technology comes with
poor efficiency and vast energy use. Cryptocurrencies cannot scale with transaction
demand, are prone to congestion and greatly fluctuate in value. Overall, the
decentralised technology of cryptocurrencies, however sophisticated, is a poor
substitute for the solid institutional backing of money.
That said, the underlying technology could have promise in other applications,
such as the simplification of administrative processes in the settlement of financial
transactions. Still, this remains to be tested. As cryptocurrencies raise a host of
issues, the chapter concludes with a discussion of policy responses, including
regulation of private uses of the technology, the measures needed to prevent
abuses of cryptocurrencies and the delicate questions raised by the issuance of
digital currency by central banks.
Putting the rise of cryptocurrencies into perspective
A good way to examine whether a new technology can be a truly useful addition to
the existing monetary landscape is to step back and review the fundamental roles
of money in an economy and what history teaches us about failed attempts to
create new private moneys. Then one can ask whether money based on this new
technology can improve upon the current monetary landscape in any way.2

A brief history of money
Money plays a crucial role in facilitating economic exchange. Before its advent
millennia ago, goods were primarily exchanged for the promise to return the favour
in the future (ie trading of IOUs).3 However, as societies grew larger and economic
activity expanded, it became harder to keep a record of ever more complex IOUs,
and default and settlement risks became concerns. Money and the institutions
issuing it came into existence to address this growing complexity and the associated
difficulty in maintaining trust.
Money has three fundamental and complementary roles. It is: (i) a unit of
account – a yardstick that eases comparison of prices across the things we buy, as
well as the value of promises we make; (ii) a medium of exchange: a seller accepts it
as a means of payment, in the expectation that somebody else will do the same;
and (iii) a store of value, enabling users to transfer purchasing power over time.4
To fulfil these functions, money needs to have the same value in different
places and to keep a stable value over time: assessing whether to sell a certain
good or service is much easier if one is certain that the received currency has a
guaranteed value in terms of both current and future purchasing power. One way
to achieve this is by pure commodity moneys with intrinsic value, such as salt or
grain. But commodity money by itself does not effectively support exchange: it
may not always be available, is costly to produce and cumbersome in exchange,
and may be perishable.5
The expansion of economic activity required more convenient moneys that
could respond to increasing demand, be efficiently used in trade and have a stable
value. However, maintaining trust in the institutional arrangements through which
money is supplied has been the biggest challenge. Around the world, in different
settings and at different times, money started to rely on issuance by centralised
authorities. From ancient times, the stamp of a sovereign certified a coin’s value in
transactions. Later, bills of exchange intermediated by banks developed as a way for
merchants to limit the costs and risks of travelling with large quantities of coinage.6
However, historical experience also made clear an underlying trade-off, for
currencies that are supplied flexibly can also be debased easily.7 Sustained episodes
of stable money are historically much more of an exception than the norm. In fact,
trust has failed so frequently that history is a graveyard of currencies. Museums
around the world devote entire sections to this graveyard – for example, room 68
of the British Museum displays stones, shells, tobacco, countless coins and pieces of
paper, along with many other objects that lost their acceptability as exchange and
found their way to this room. Some fell victim to the expansion of trade and
economic activity, as they were rendered inconvenient with a larger scale of use.
Some were discarded when the political order that supported them weakened or
fell. And many others fell victim to the erosion of trust in the stability of their value.
History proves that money can be fragile whether it is supplied through private
means, in a competitive manner, or by a sovereign, as a monopolist supplier. Bank-
issued money is only as good as the assets that back it. Banks are meant to
transform risks, and therefore, under certain extreme scenarios, confidence in
privately issued money can vanish overnight. Government-backed arrangements,
where assuring trust in the instrument is a centralised task, have not always worked
well either. Far from it: a well known example of abuse is the competitive
debasement of coins issued by German princes in the early 17th century, known as
the Kipper- und Wipperzeit (clipping and culling times).8 And there have been
many others, up to the present-day cases of Venezuela and Zimbabwe. Avoiding
abuse by the sovereign has thus been a key consideration in the design of monetary
arrangements.

The quest for solid institutional underpinning for trust in money eventually
culminated in the emergence of today’s central banks. An early step was the
establishment of chartered public banks in European city-states during the period
1400–1600. These emerged to improve trading by providing a high-quality, efficient
means of payment and centralising a number of clearing and settlement operations.
Such banks, set up in trading hubs such as Amsterdam, Barcelona, Genoa, Hamburg
and Venice, were instrumental in stimulating international trade and economic
activity more generally.9 Over time, many of these chartered banks functioned in
ways similar to current central banks. Formal central banks, as we know them today,
also often emerged in direct response to poor experiences with decentralised
money. For example, the failures of wildcat banking in the United States eventually
led to the creation of the Federal Reserve System.
The current monetary and payment system
The tried, trusted and resilient way to provide confidence in money in modern
times is the independent central bank. This means agreed goals: clear monetary
policy and financial stability objectives; operational, instrument and administrative
independence; and democratic accountability, so as to ensure broad-based political
support and legitimacy. Independent central banks have largely achieved the goal
of safeguarding society’s economic and political interest in a stable currency.10 With
this setup, money can be accurately defined as an “indispensable social convention
backed by an accountable institution within the state that enjoys public trust”.11
In almost all modern-day economies, money is provided through a joint
public-private venture between the central bank and private banks, with the central
bank at the system’s core. Electronic bank deposits are the main means of payment
between ultimate users, while central bank reserves are the means of payment
between banks. In this two-tiered system, trust is generated through independent
and accountable central banks, which back reserves through their asset holdings
and operational rules. In turn, trust in bank deposits is generated through a variety
of means, including regulation, supervision and deposit insurance schemes, many
ultimately emanating from the state.
As part of fulfilling their mandate to maintain a stable unit of account and
means of payment, central banks take an active role in supervising, overseeing and
in some cases providing the payments infrastructure for their currency. The central
bank’s role includes ensuring that the payment system operates smoothly and
seeing to it that the supply of reserves responds appropriately to shifting demand,
including at intraday frequency, ie ensuring an elastic money supply.12
Thanks to the active involvement of central banks, today’s diverse payment
systems have achieved safety, cost-effectiveness, scalability and trust that a
payment, once made, is final.
Payment systems are safe and cost-effective, handling high volumes and
accommodating rapid growth with hardly any abuse and at low costs. An important
contributor to safety and cost-effectiveness is scalability. In today’s sophisticated
economies, the volume of payments is huge, equal to many multiples of GDP. Despite
these large volumes, expanding use of the instrument does not lead to a proportional
increase in costs. This is important, since an essential feature of any successful money
and payment system is how widely used it is by both buyers and sellers: the more
others connect to a particular payment system, the greater one’s own incentive to use it.
Users not only need to have trust in money itself, they also need to trust that a
payment will take place promptly and smoothly. A desirable operational attribute is
thus certainty of payment (“finality”) and the related ability to contest transactions
that may have been incorrectly executed. Finality requires that the system be largely

The money flower: a taxonomy of money Graph V.1
Central bank-
Digital
issued
Virtual
Widely currency
Peer-to-peer
accessible
Central bank
Bank reserves and
deposits settlement
accounts
Central bank
Central bank
deposited
digital currencies
currency
(wholesale)
accounts
Central
bank
digital
currencies Cryptocurrency
(retail) (permissioned DLT)
Cash
Commodity
money Cryptocurrency
(permissionless DLT)
Source: Adapted from M Bech and R Garratt, “Central bank cryptocurrencies”, BIS Quarterly Review, September 2017, pp 55–70.
free of fraud and operational risks, at the level of both individual transactions and
the system as a whole. Strong oversight and central bank accountability both help
to support finality and hence trust.
While most modern-day transactions occur through means ultimately supported
by central banks, over time a wide range of public and private payment means has
emerged. These can be best summarised by a taxonomy characterised as the
“money flower” (Graph V.1).13
The money flower distinguishes four key properties of moneys: the issuer, the
form, the degree of accessibility and the payment transfer mechanism. The issuer
can be a central bank, a bank or nobody, as was the case when money took the
form of a commodity. Its form can be physical, eg a metal coin or paper banknote,
or digital. It can be widely accessible, like commercial bank deposits, or narrowly so,
like central bank reserves. A last property regards the transfer mechanism, which
can be either peer-to-peer, or through a central intermediary, as for deposits.
Money is typically based on one of two basic technologies: so called “tokens” or
accounts. Token-based money, for example banknotes or physical coins, can be
exchanged in peer-to-peer settings, but such exchange relies critically on the
payee’s ability to verify the validity of the payment object – with cash, the worry is
counterfeiting. By contrast, systems based on account money depend fundamentally
on the ability to verify the identity of the account holder.
Cryptocurrencies: the elusive promise of decentralised trust
Do cryptocurrencies deliver what they promise? Or will they end up as short-lived

curiosities? In order to answer these questions, it is necessary to define them more

precisely, to understand their supporting technology and to examine the associated
economic limitations.
A new petal in the money flower?
Cryptocurrencies aspire to be a new form of currency and promise to maintain trust

in the stability of their value through the use of technology. They consist of three
elements. First, a set of rules (the “protocol”), computer code specifying how
participants can transact. Second, a ledger storing the history of transactions. And
third, a decentralised network of participants that update, store and read the ledger
of transactions following the rules of the protocol. With these elements, advocates
claim, a cryptocurrency is not subject to the potentially misguided incentives of
banks and sovereigns.
In terms of the money flower taxonomy, cryptocurrencies combine three key
features. First, they are digital, aspiring to be a convenient means of payment and
relying on cryptography to prevent counterfeiting and fraudulent transactions.
Second, although created privately, they are no one’s liability, ie they cannot be
redeemed, and their value derives only from the expectation that they will continue
to be accepted by others. This makes them akin to a commodity money (although
without any intrinsic value in use). And, last, they allow for digital peer-to-peer
exchange.
Compared with other private digital moneys such as bank deposits, the
distinguishing feature of cryptocurrencies is digital peer-to-peer exchange. Digital
bank accounts have been around for decades. And privately issued “virtual
currencies” – eg as used in massive multiplayer online games like World of Warcraft
– predate cryptocurrencies by a decade. In contrast to these, cryptocurrency
transfers can in principle take place in a decentralised setting without the need for
a central counterparty to execute the exchange.
Distributed ledger technology in cryptocurrencies
The technological challenge in digital peer-to-peer exchange is the so-called

“double-spending problem”. Any digital form of money is easily replicable and can
thus be fraudulently spent more than once. Digital information can be reproduced
more easily than physical banknotes. For digital money, solving the double-spending
problem requires, at a minimum, that someone keep a record of all transactions.
Prior to cryptocurrencies, the only solution was to have a centralised agent do this
and verify all transactions.
Cryptocurrencies overcome the double-spending problem via decentralised
record-keeping through what is known as a distributed ledger. The ledger can be
regarded as a file (think of a Microsoft Excel worksheet) that starts with an initial
distribution of cryptocurrency and records the history of all subsequent transactions.
An up-to-date copy of the entire ledger is stored by each user (this is what makes it
“distributed”). With a distributed ledger, peer-to-peer exchange of digital money is
feasible: each user can directly verify in their copy of the ledger whether a transfer
took place and that there was no attempt to double-spend.14
While all cryptocurrencies rely on a distributed ledger, they differ in terms of
how the ledger is updated. One can distinguish two broad classes, with substantial
differences in their operational setup (Graph V.2).
One class is based on “permissioned” DLT. Such cryptocurrencies are similar to
conventional payment mechanisms in that, to prevent abuse, the ledger can only
be updated by trusted participants in the cryptocurrency – often termed “trusted
nodes”. These nodes are chosen by, and subject to oversight by, a central authority,

Centralised ledger and permissioned/permissionless decentralised ledgers Graph V.2
One copy
Centralised ledger $
All parties reconcile their local databases with a

How many copies centralised electronic ledger that is maintained
of the ledger are and controlled by a trusted central party.
there?
$
Many
copies
Only selected nodes

can participate
Distributed ledger (permissioned)
In a permissioned system, nodes need
$
permission from a central entity to access the
network and make changes to the ledger.
A c c e s s c o n t ro l s c a n i n c l u d e i d e n t i t y
verification.
$
Distributed
ledgers
Distributed ledger (permissionless)

Anybody can access Each node in a peer-to-peer network stores a
and participate full and up-to-date copy of the entire ledger.
Every proposed local addition to the ledger by
a network participant is communicated across
$
the network to all nodes. In principle, nodes
collectively attempt to validate the addition
through an algorithmic consensus mechanism.
If validation is accepted, the new addition is
added to all ledgers to ensure data consistency
$
across the entire network.
Private electronic money based on Privately issued cryptocurrencies

fiat system Permissioned Permissionless
1 Storage of Ledger (accounts) stored centrally by
Decentralised storage of ledger
balances/holdings banks and other financial institutions
2 Verification to avoid Identity-based concept Peer-to-peer concept: distributed ledger can be checked to see
double-spending whether a specific unit of a currency has already been spent
3 Processing of Accounts updated by bank Updating of ledger via trusted Updating of ledger via proof-of-
transactions nodes work
Rule to follow longest chain
4 Finality/settlement Settlement ultimately via central bank Settlement in cryptocurrency Probabilistic concept of finality
concept itself via rule to follow longest chain
5 Elasticity of supply Central bank policy, eg regarding Protocol can be changed by Protocol-determined
intraday credit trusted nodes
6 Trust-creating Reputation of banks and central banks, Reputation of issuing firm and Proof-of-work requires honest
mechanisms banking supervision, lender of last nodes computing majority
resort, legal tender laws, central bank Trusted nodes, some of which
independence and accountability, may be subject to regulation
AML/CFT checks, cyber-security
Sources: Adapted from H Natarajan, S Krause and H Gradstein, “Distributed ledger technology (DLT) and blockchain”, World Bank Group,
FinTech Note, no 1, 2017; BIS.

eg the firm that developed the cryptocurrency. Thus, while cryptocurrencies based
on permissioned systems differ from conventional money in terms of how
transaction records are stored (decentralised versus centralised), they share with it
the reliance on specific institutions as the ultimate source of trust.15
In a much more radical departure from the prevailing institution-based setup,
a second class of cryptocurrencies promises to generate trust in a fully decentralised
setting using “permissionless” DLT. The ledger recording transactions can only be
changed by a consensus of the participants in the currency: while anybody can
participate, nobody has a special key to change the ledger.
The concept of permissionless cryptocurrencies was laid out for the case of
Bitcoin16 in a white paper by an anonymous programmer (or group of programmers)
under the pseudonym Satoshi Nakamoto, who proposed a currency based on a
specific type of distributed ledger, the “blockchain”. The blockchain is a distributed
ledger that is updated in groups of transactions called blocks. Blocks are then
chained sequentially via the use of cryptography to form the blockchain. This
concept has been adapted to countless other cryptocurrencies.17
Blockchain-based permissionless cryptocurrencies have two groups of
participants: “miners” who act as bookkeepers and “users” who want to transact in
the cryptocurrency. At face value, the idea underlying these cryptocurrencies is
simple: instead of a bank centrally recording transactions (Graph V.3, left-hand
panel), the ledger is updated by a miner and the update is subsequently stored by
all users and miners (right-hand panel).18
Valid transactions in a centralised ledger/bank account and in a permissionless

cryptocurrency Graph V.3
Centralised ledger Distributed ledger
Ship goods Ship goods

Sees
Buyer authorises Bank informs Buyer new
payment seller of announces consensus
payment payment
Miner
Miner
Miner
$
Miner
Miner Miner
Miner
Bank Miner Multiple Miner

Buyer account balance –1 Ledger update: verifications
Seller account balance +1 Buyer holdings –1
Miner
Miner
Seller holdings +1
A buyer purchases a good from the seller, who initiates shipment upon perceived confirmation of the payment. If the payment takes place
via bank accounts – ie via a centralised ledger (left-hand panel) – the buyer sends the payment instruction to their bank, which adjusts account
balances debiting the amount paid from the buyer’s account and crediting it to the seller’s account. The bank then confirms payment to the
seller. In contrast, if payment takes place via a permissionless cryptocurrency (right-hand panel), the buyer first publicly announces a payment
instruction stating that the cryptocurrency holdings of the buyer are reduced by one, while those of the seller are increased by one. After a
delay, a miner includes this payment information in a ledger update. The updated ledger is subsequently shared with other miners and users,
each verifying that the newly added payment instruction is not a double-spend attempt and is authorised by the buyer. The seller then
observes that the ledger including the payment instruction emerges as the one commonly used by the network of miners and users.
Source: Adapted from R Auer, “The mechanics of decentralised trust in Bitcoin and the blockchain”, BIS Working Papers, forthcoming.

Underlying this setup, the key feature of these cryptocurrencies is the
implementation of a set of rules (the protocol) that aim to align the incentives of all
participants so as to create a reliable payment technology without a central trusted
agent. The protocol determines the supply of the asset in order to counter
debasement – for example, in the case of Bitcoin, it states that no more than 21
million bitcoins can exist. In addition, the protocol is designed to ensure that all
participants follow the rules out of self-interest, ie that they yield a self-sustaining
equilibrium. Three key aspects are the following.
First, the rules entail a cost to updating the ledger. In most cases, this cost
comes about because updating requires a “proof-of-work”. This is mathematical
evidence that a certain amount of computational work has been done, in turn
calling for costly equipment and electricity use. Since the proof-of-work process
can be likened to digging up rare numbers via laborious computations, it is often
referred to as mining.19 In return for their efforts, miners receive fees from the users
– and, if specified by the protocol, newly minted cryptocurrency.
Second, all miners and users of a cryptocurrency verify all ledger updates, which
induces miners to include only valid transactions. Valid transactions need to be initiated
by the owners of funds and must not be attempts to double-spend. If a ledger update
includes an invalid transaction, it is rejected by the network and the miner’s rewards
are voided. The verification of all new ledger updates by the network of miners and
users is thus essential to incentivise miners to add only valid transactions.20
Third, the protocol specifies rules to achieve a consensus on the order of
updates to the ledger. This is generally done by creating incentives for individual
miners to follow the computing majority of all other miners when they implement
updates. Such coordination is needed, for example, to resolve cases where
communication lags lead to different miners adding conflicting updates – ie updates
that include different sets of transactions (Box V.A).
With these key ingredients, it is costly – though not impossible – for any
individual to forge a cryptocurrency. To successfully double-spend, a counterfeiter
would have to spend their cryptocurrency with a merchant and secretly produce a
forged blockchain in which this transaction was not recorded. Upon receipt of the
merchandise, the counterfeiter would then release the forged blockchain, ie reverse
the payment. But this forged blockchain would only emerge as the commonly
accepted chain if it were longer than the blockchain the rest of the network of
miners had produced in the meantime. A successful double-spend attack thus
requires a substantial share of the mining community’s computing power. Conversely,
in the words of the original Bitcoin white paper, a cryptocurrency can overcome the
double-spending problem in a decentralised way only if “honest nodes control a
majority of [computing] power”.21
Assessing the economic limitations of permissionless cryptocurrencies
Cryptocurrencies such as Bitcoin promise to deliver not only a convenient payment

means based on digital technology, but also a novel model of trust. Yet delivering on
this promise hinges on a set of assumptions: that honest miners control the vast
majority of computing power, that users verify the history of all transactions and that
the supply of the currency is predetermined by a protocol. Understanding these
assumptions is important, for they give rise to two basic questions regarding the
usefulness of cryptocurrencies. First, does this cumbersome way of trying to achieve
trust come at the expense of efficiency? Second, can trust truly and always be achieved?
As the first question implies, a key potential limitation in terms of efficiency is
the enormous cost of generating decentralised trust. One would expect miners to
compete to add new blocks to the ledger through the proof-of-work until their

anticipated profits fall to zero.22 Individual facilities operated by miners can host
computing power equivalent to that of millions of personal computers. At the time
of writing, the total electricity use of bitcoin mining equalled that of mid-sized
economies such as Switzerland, and other cryptocurrencies also use ample
electricity (Graph V.4, left-hand panel). Put in the simplest terms, the quest for
decentralised trust has quickly become an environmental disaster.23
But the underlying economic problems go well beyond the energy issue. They
relate to the signature property of money: to promote “network externalities”
among users and thereby serve as a coordination device for economic activity. The
shortcomings of cryptocurrencies in this respect lie in three areas: scalability,
stability of value and trust in the finality of payments.
First, cryptocurrencies simply do not scale like sovereign moneys. At the most
basic level, to live up to their promise of decentralised trust cryptocurrencies require
each and every user to download and verify the history of all transactions ever
made, including amount paid, payer, payee and other details. With every transaction
adding a few hundred bytes, the ledger grows substantially over time. For example,
at the time of writing, the Bitcoin blockchain was growing at around 50 GB per year
and stood at roughly 170 GB. Thus, to keep the ledger’s size and the time needed to
verify all transactions (which increases with block size) manageable, cryptocurrencies
have hard limits on the throughput of transactions (Graph V.4, centre panel).
A thought experiment illustrates the inadequacy of cryptocurrencies as an
everyday means of payment (Graph V.4, right-hand panel). To process the number
of digital retail transactions currently handled by selected national retail payment
systems, even under optimistic assumptions, the size of the ledger would swell well
beyond the storage capacity of a typical smartphone in a matter of days, beyond
that of a typical personal computer in a matter of weeks and beyond that of servers
in a matter of months. But the issue goes well beyond storage capacity, and extends
Energy consumption and scaling issues Graph V.4
Energy usage of select Number of transactions per second2 Hypothetical ledger size for nation-
cryptocurrencies 1 wide retail cryptocurrency3
Terawatt hours/year Transactions/second Gigabytes
3,526
2,061
241
60 200 100,000
45 150 75,000
--
-
--
-
--
-
--
-
30 100 50,000
15 50 25,000
3.30
3.18
0.26
0 0 0
Mastercard
Visa
PayPal
Bitcoin
Ether
Litecoin
2017 2018 2018 2019 2020

Bitcoin Ether China United States
Euro area4
1
Estimated. 2 2017 data. 3 The displayed hypothetical size of the blockchain/ledger is calculated assuming that, starting from 1 July
2018, all non-cash retail transactions of either China, the United States or the euro area are processed via a cryptocurrency. Calculations are
based on information on non-cash transaction numbers from CPMI (2017) and assume that each transaction adds 250 bytes to the
ledger. 4 BE, FR, DE, IT and NL.
Sources: Committee on Payments and Market Infrastructures, Statistics on payment, clearing and settlement systems in the CPMI countries,
December 2017; www.bitinfocharts.com; Digiconomist; Mastercard; PayPal; Visa; BIS calculations.
Transaction fees over time and in relation to transaction throughput Graph V.5
to processing capacity: only supercomputers could keep up with verification of the
incoming transactions. The associated communication volumes could bring the
internet to a halt, as millions of users exchanged files on the order of magnitude of
a terabyte.
Another aspect of the scalability issue is that updating the ledger is subject to
congestion. For example, in blockchain-based cryptocurrencies, in order to limit the
number of transactions added to the ledger at any given point in time, new blocks
Energy
can onlyconsumption
be added atand scaling issues
pre-specified intervals. Once the number of incoming Graph V.4
transactions is such that newly added blocks are already at the maximum size
Energy usage
permitted byofthe
selectprotocol, the system Number of transactions
congests and many per second 2
transactions Hypothetical
go into aledger size for nation-
cryptocurrencies 1 wide retail cryptocurrency3
queue. With capacity capped, fees soar whenever transaction
Terawatt hours/year
demand reaches the
Transactions/second Gigabytes
capacity limit (Graph V.5). And transactions have at times remained in a queue for
3,526
2,061
several hours, interrupting the payment process. This limits 200 cryptocurrencies’
241
60 100,000
usefulness for day-to-day transactions such as paying for a coffee or a conference
fee, not to mention for wholesale 45 payments.24 Thus, the more 150 people use a 75,000
--
-
--
-
--
-
--
-
cryptocurrency, the more cumbersome payments become. This negates an essential
property of present-day money:30the more people use it, the stronger 100 the incentive 50,000
to use it.25
The second key issue with15cryptocurrencies is their unstable 50 value. This arises 25,000
from the absence of a central issuer with a mandate to guarantee the currency’s
3.30
3.18
0.26
stability. Well run central banks0succeed in stabilising the domestic 0 value of their 0
sovereign currency by adjusting the supply of the means of payment 2018 in line 2019
with
Mastercard
Visa
PayPal
Bitcoin
Ether
Litecoin
2017 2018 2020
transaction
Bitcoin demand. EtherThey do so at high frequency, in particular during times China of United States
market stress but also during normal times. Euro area4
1 This contrasts
Estimated. 2
2017 data.with3 aThecryptocurrency, where
displayed hypothetical generating
size of some confidence
the blockchain/ledger in its that, starting from 1 July
is calculated assuming
value
2018, allrequires thattransactions
non-cash retail supply be predetermined
of either China, the Unitedby a protocol.
States Thisareprevents
or the euro area it from
processed via a cryptocurrency. Calculations are
being supplied elastically. Therefore, any fluctuation in demand translates into adds 250 bytes to the
based on information on non-cash transaction numbers from CPMI (2017) and assume that each transaction
ledger. 4 BE, FR, DE, IT and NL.
changes in valuation. This means that cryptocurrencies’ valuations are extremely
volatile (Graph on
Sources: Committee V.6,Payments
left-hand panel).
and Market And the Statistics
Infrastructures, inherent instability
on payment, is unlikely
clearing to be
and settlement systems in the CPMI countries,
December 2017; www.bitinfocharts.com; Digiconomist; Mastercard; PayPal; Visa; BIS calculations.
fully overcome by better protocols or financial engineering, as exemplified by the
experience of the Dai cryptocurrency. While engineered to be fixed to the US dollar
Transaction fees over time and in relation to transaction throughput Graph V.5
Transaction fees spike… …when blocks are full and the system congests
USD/transaction USD/transaction
48 4 60
1
Transaction fee (USD)
36 3 45
24 2 30
12 1 15
0 0 0
Q1 16 Q3 16 Q1 17 Q3 17 Q1 18 0.5 MB 1 MB
≈ 1.67 transactions/second ≈ 3.3 transactions/second
Lhs: Bitcoin Rhs: Ether Litecoin
Block size
1
Transaction fee paid to miners over the period 1 August 2010–25 May 2018; daily averages.
Sources: www.bitinfocharts.com; BIS calculations.

at a rate of one to one, it reached a low of $0.72 just a few weeks after its launch in
late 2017. Other cryptocurrencies designed to have a stable value have also
fluctuated substantially (centre panel).
This outcome is not coincidental. Keeping the supply of the means of payment
in line with transaction demand requires a central authority, typically the central
bank, which can expand or contract its balance sheet. The authority needs to be
willing at times to trade against the market, even if this means taking risk onto its
balance sheet and absorbing a loss. In a decentralised network of cryptocurrency
users, there is no central agent with the obligation or the incentives to stabilise the
value of the currency: whenever demand for the cryptocurrency decreases, so does
its price.
Further contributing to unstable valuations is the speed at which new
cryptocurrencies – all tending to be very closely substitutable with one another –
come into existence. At the time of writing, several thousand existed, though
proliferation makes reliable estimates of the number of outstanding cryptocurrencies
impossible (Graph V.6, right-hand panel). Recalling the private banking experiences
of the past, the outcome of such liberal issuance of new moneys is rarely stability.
The third issue concerns the fragile foundation of the trust in cryptocurrencies.
This relates to uncertainty about the finality of individual payments, as well as trust
in the value of individual cryptocurrencies.
In mainstream payment systems, once an individual payment makes its way
through the national payment system and ultimately through the central bank
books, it cannot be revoked. In contrast, permissionless cryptocurrencies cannot
guarantee the finality of individual payments. One reason is that although users can
verify that a specific transaction is included in a ledger, unbeknownst to them there
can be rival versions of the ledger. This can result in transaction rollbacks, for
example when two miners update the ledger almost simultaneously. Since only one
of the two updates can ultimately survive, the finality of payments made in each
ledger version is probabilistic.
Volatility of select cryptocurrencies and number of cryptocurrencies Graph V.6
Major cryptocurrencies are “Stable coins” fluctuate in value2 Number of cryptocurrencies is

comparatively volatile1 growing rapidly3
Per cent US dollars Number
9 1.1 4,000
6 1.0 3,000
3 0.9 2,000
0 0.8 1,000
–3 0.7 0
2016 2017 2018 Jan 18 Mar 18 May 18 13 14 15 16 17 18

Gold Ether BitUSD Dai CoinMarketCap CoinLib
Bitcoin Litecoin
1
Thirty-day moving averages of daily returns. 2 Daily price minimum. 3 Based on monthly snapshots from two different providers.
CoinMarketCap includes only cryptocurrencies with a minimum 24-hour trading volume of $100,000; CoinLib does not use a threshold.
Sources: www.bitinfocharts.com; www.coinlib.io; www.coinmarketcap.com; Datastream.
Mining concentration and Economic

BIS Annual bitcoin Report
value2018
during a temporary fork Graph V.7
101
Mining is highly concentrated across all Value of bitcoin during a 2013 temporary fork2
1
The lack of payment finality is exacerbated by the fact that cryptocurrencies
can be manipulated by miners controlling substantial computing power, a real
possibility given the concentration of mining for many cryptocurrencies (Graph V.7,
left-hand panel). One cannot tell if a strategic attack is under way because an
attacker would reveal the (forged) ledger only once they were sure of success. This
implies that finality will always remain uncertain. For cryptocurrencies, each update
of the ledger comes with an additional proof-of-work that an attacker would have
to reproduce. Yet while the probability that a payment is final increases with the
number of subsequent ledger updates, it never reaches 100%.26
Not only is the trust in individual payments uncertain, but the underpinning of
trust in each
Volatility of cryptocurrency is also fragile.
select cryptocurrencies and This
numberis dueoftocryptocurrencies
“forking”. This is a process Graph V.6
whereby a subset of cryptocurrency holders coordinate on using a new version of
Majorledger
the cryptocurrencies are
and protocol, “Stablestick
while others coins”tofluctuate in value2one. InNumber
the original of cryptocurrencies
this way, a is
comparatively volatile1 growing rapidly3
cryptocurrency can split into two subnetworks of users. While there are many recent
Per cent US dollars Number
examples, an episode on 11 March 2013 is noteworthy because – counter to the
9 1.1 4,000
idea of achieving trust by decentralised means – it was undone by centralised
coordination of the miners. On6 that day, an erroneous software 1.0
update led to 3,000
incompatibilities between one part of the Bitcoin network mining on the legacy
protocol and another part mining 3
using an updated one. For several 0.9
hours, two 2,000
separate blockchains grew; once news of this fork spread, the price of bitcoin
tumbled by almost a third (Graph V.7, 0 right-hand panel). The fork 0.8 was ultimately 1,000
rolled back by a coordinated effort whereby miners temporarily departed from
protocol and ignored the longest –3 chain. But many transactions were 0.7 voided hours 0
after users had believed them to be final. This episode shows just how easily
cryptocurrencies
2016 can split, leading
2017 2018 toJansignificant
18 valuation
Mar 18 losses.
May 18 13 14 15 16 17 18
An even
Gold more worrying
Ether aspect underlying
BitUSD such episodes
Dai is that forking may
CoinMarketCap CoinLib
only be symptomatic
Bitcoin of a fundamental shortcoming: the fragility of the decentralised
Litecoin
consensus
1 involved
Thirty-day moving in updating
averages the ledger
of daily returns. 2
Dailyand,
price with it, of 3the
minimum. underlying
Based on monthlytrust in the
snapshots from two different providers.
cryptocurrency. Theoretical
CoinMarketCap includes analysis with
only cryptocurrencies (Box V.A) suggests
a minimum that coordination
24-hour trading on CoinLib
volume of $100,000; how thedoes not use a threshold.
ledger is updated could break

Sources: www.bitinfocharts.com; down at
www.coinlib.io; any time, resultingDatastream.
www.coinmarketcap.com; in a complete loss of value.
Mining concentration and bitcoin value during a temporary fork Graph V.7
Mining is highly concentrated across all Value of bitcoin during a 2013 temporary fork2
cryptocurrencies1
Per cent US dollars
100 48
75 45
50 42
25 39
0 36
Bitcoin Bitcoin Cash Ether Litecoin 18:00 00:00 06:00 12:00
Mining pools: #1 #2 #3 11 March 2013 12 March 2013
#4 #5 Others
Bitcoin price
1
Data for the largest mining pools as of 28 May 2018. 2
Bitcoin price dynamics during Bitcoin fork on 11–12 March 2013.
Sources: www.btc.com; www.cash.coin.dance; CoinDesk; www.etherchain.org; www.litecoinpool.org.

Box V.A
Forking and the instability of decentralised consensus in the blockchain
Forking has contributed to the explosive growth in the number of cryptocurrencies (Graph V.6, right-hand panel).
For example, the month of January 2018 alone brought to the fore the Bitcoin ALL, Bitcoin Cash Plus, Bitcoin Smart,
Bitcoin Interest, Quantum Bitcoin, BitcoinLite, Bitcoin Ore, Bitcoin Private, Bitcoin Atom and Bitcoin Pizza forks. There
are many different ways in which such forks can arise, some permanent and others temporary. One example is
termed a “hard fork” (Graph V.A). It arises if some of the miners of a cryptocurrency coordinate to change the
protocol to a new set of rules that is incompatible with the old one. This change could involve many aspects of the
protocol, such as the maximum permitted block size, the frequency at which blocks can be added to the blockchain
or a change to the proof-of-work required to update the blockchain. The miners who upgrade to the new rules start
from the old blockchain, but subsequently add blocks that are not recognised by the miners who have not upgraded.
The latter continue to build on the existing blockchain following the old rules. In this way, two separate blockchains
grow, each with its own transaction history.
Example of a hard fork Graph V.A
Follows Follows Follows Follows Follows

old old old old old
rules rules rules rules rules
Follows Follows Follows

new new new
rules rules rules
Source: BIS.
Frequent episodes of forking may be symptomatic of an inherent problem with the way consensus is formed in
a cryptocurrency’s decentralised network of miners. The underlying economic issue is that this decentralised
consensus is not unique. The rule to follow the longest chain incentivises miners to follow the computing majority,
but it does not uniquely pin down the path of the majority itself. For example, if a miner believes that the very last
update of the ledger will be ignored by the rest of the network of miners, it becomes optimal for the miner to also
ignore this last update. And if the majority of miners coordinates on ignoring an update, this indeed becomes a new
equilibrium. In this way, random equilibria can arise – and indeed frequently have arisen, as indicated by forking and
by the existence of thousands of “orphaned” (Bitcoin) or “uncle” (Ethereum) blocks that have retroactively been
voided. Additional concerns regarding the robustness of the decentralised updating of the blockchain relate to
miners’ incentives to strategically fork whenever the block added last by a different miner includes high transaction
fees that can be diverted by voiding the block in question via a fork.
  For an analysis of the uniqueness of the updating of the blockchain, see B Biais, C Bisière, M Bouvard and C Casamatta, “The blockchain
folk theorem”, TSE Working Papers, no 17–817, 2017. For an analysis of strategic motives to create a fork, see M Carlsten, H Kalodner,
S M Weinberg and A Narayanan, “On the instability of Bitcoin without the block reward”, Proceedings of the 2016 ACM SIGSAC Conference
on Computer and Communications Security.
Overall, decentralised cryptocurrencies suffer from a range of shortcomings.

The main inefficiencies arise from the extreme degree of decentralisation: creating
the required trust in such a setting wastes huge amounts of computing power,
decentralised storage of a transaction ledger is inefficient and the decentralised

consensus is vulnerable. Some of these issues might be addressed by novel
protocols and other advances.27 But others seem inherently linked to the fragility
and limited scalability of such decentralised systems. Ultimately, this points to the
lack of an adequate institutional arrangement at the national level as the
fundamental shortcoming.
Beyond the bubble: making use of distributed ledger technology
While cryptocurrencies do not work as money, the underlying technology may have
promise in other fields. A notable example is in low-volume cross-border payment
services. More generally, compared with mainstream centralised technological
solutions, DLT can be efficient in niche settings where the benefits of decentralised
access exceed the higher operating cost of maintaining multiple copies of the
ledger.
To be sure, such payment solutions are fundamentally different from
cryptocurrencies. A recent non-profit example is the case of the World Food
Programme’s blockchain-based Building Blocks system, which handles payments
for food aid serving Syrian refugees in Jordan. The unit of account and ultimate
means of payment in Building Blocks is sovereign currency, so it is a “cryptopayment”
system but not a cryptocurrency. It is also centrally controlled by the World Food
Programme, and for good reason: an initial experiment based on the permissionless
Ethereum protocol resulted in slow and costly transactions. The system was
subsequently redesigned to run on a permissioned version of the Ethereum
protocol. With this change, a reduction of transaction costs of about 98% relative
to bank-based alternatives was achieved.28
Permissioned cryptopayment systems may also have promise with respect to
small-value cross-border transfers, which are important for countries with a large
share of their workforce living abroad. Global remittance flows total more than
$540 billion annually (Graph V.8, left-hand and centre panels). Currently, forms of
Indicators of the volume and cost of remittances Graph V.8
Remittance volumes are on the rise, …a large volume of low-value …at high average costs 2
resulting in… payments between often illiquid
currency pairs…1
USD bn USD bn US dollars
540 60 20
360 40 15
180 20 10
0 0 5
92 95 98 01 04 07 10 13 16 IN PH FR DE SA 13 14 15 16 17 18
CN MX CN CH US
World inflows IN SA G20
Inflows Outflows CN US
1
Data for 2016. 2 Average total cost for sending $200 with all remittance service providers worldwide. For CN and IN, receiving country
average total cost; for G20, SA and US, sending country average total cost.
Sources: World Bank, Remittance Prices Worldwide, remittanceprices.worldbank.org; World Bank; BIS calculations.
The
104 shutdown of an illegal marketplace and the legitimacy ofEconomic
BIS Annual ICOs Report 2018 Graph V.9
Cryptocurrency prices react strongly to shutdowns of A large share of initial coin offerings is thought to be
1
international payments involve multiple intermediaries, leading to high costs (right-
hand panel). That said, while cryptopayment systems are one option to address
these needs, other technologies are also being considered, and it is not clear which
will emerge as the most efficient one.
More important use cases are likely to combine cryptopayments with
sophisticated self-executing codes and data permission systems. Some decentralised
cryptocurrency protocols such as Ethereum already allow for smart contracts that
self-execute the payment flows for derivatives. At present, the efficacy of these
products is limited by the low liquidity and intrinsic inefficiencies of permissionless
cryptocurrencies. But the underlying technology can be adopted by registered
exchanges in permissioned protocols that use sovereign money as backing,
simplifying settlement execution. The added value of the technology will probably
derive from the simplification of administrative processes related to complex
financial transactions, such as trade finance (Box V.B). Crucially, however, none of
the applications require the use or creation of a cryptocurrency.
Policy implications
The rise of cryptocurrencies and related technology brings to the fore a number of
policy questions. Authorities are looking for ways to ensure the integrity of markets
and payment systems, to protect consumers and investors, and to safeguard overall
financial stability. An important challenge is to combat illicit usage of funds. At the
same time, authorities want to preserve long-run incentives for innovation and, in
particular, maintain the principle of “same risk, same regulation”.29 These are largely
recurrent objectives, but cryptocurrencies raise new challenges and potentially call
for new tools and approaches. A related question is whether central banks should
issue their own central bank digital currency (CBDC).
Regulatory challenges posed by cryptocurrencies
A first key regulatory challenge is anti-money laundering (AML) and combating the
financing of terrorism (CFT). The question is whether, and to what extent, the rise of
cryptocurrencies has allowed some AML/CFT measures, such as know-your-
customer standards, to be evaded. Because cryptocurrencies are anonymous, it is
hard to quantify the extent to which they are being used to avoid capital controls
or taxes, or to engage in illegal transactions more generally. But events such as
Bitcoin’s strong market reaction to the shutdown of Silk Road, a major marketplace
for illegal drugs, suggest that a non-negligible fraction of the demand for
cryptocurrencies derives from illicit activity (Graph V.9, left-hand panel).30
A second challenge encompasses securities rules and other regulations
ensuring consumer and investor protection. One common problem is digital theft.
Given the size and unwieldiness of distributed ledgers, as well as high transaction
costs, most users access their cryptocurrency holdings via third parties such as
“crypto wallet” providers or “crypto exchanges”. Ironically – and much in contrast
to the original promise of Bitcoin and other cryptocurrencies – many users who
turned to cryptocurrencies out of distrust in banks and governments have thus
wound up relying on unregulated intermediaries. Some of these (such as Mt Gox
or Bitfinex) have proved to be fraudulent or have themselves fallen victim to
hacking attacks.31
Fraud issues also plague initial coin offerings (ICOs). An ICO involves the
auctioning of an initial set of cryptocurrency coins to the public, with the proceeds
sometimes granting participation rights in a startup business venture. Despite

Box V.B
Distributed ledger technology in trade finance
The World Trade Organization estimates that 80–90% of global trade relies on trade finance. When an exporter and
an importer agree to trade, the exporter often prefers to be paid upfront due to the risk that the importer will not
make a payment after receiving the goods. Conversely, the importer prefers to reduce their own risk by requiring
documentation that the goods have been shipped before initiating payment.
Trade financing offered by banks and other financial institutions aims to bridge this gap. Most commonly, a
bank in the importer’s home country issues a letter of credit guaranteeing payment to the exporter upon receipt of
documentation of the shipment, such as a bill of lading. In turn, a bank in the exporter’s country might extend credit
to the exporter against this pledge, and collect the payment from the importer’s bank to complete the transaction.
In its current form (Graph V.B, left-hand panel), trade finance is cumbersome, complex and costly. It involves
multiple document exchanges between the exporter, the importer, their respective banks, and agents making
physical checks of shipped goods at each checkpoint, as well as customs agencies, public export credit agencies or
freight insurers. The process often involves paper-based administration. DLT can simplify the execution of the
underlying contracts (right-hand panel). For example, a smart contract might automatically release payment to the
exporter upon the addition of a valid bill of lading to the ledger. And the better availability of information on which
shipments have already been financed could also reduce the risk that exporters illegally obtain credit multiple times
for the same shipment from different banks.
How trade finance on a distributed ledger works Graph V.B
Current DLT-based
Buyer/applicant Beneficiary/seller Buyer/applicant Beneficiary/seller

Buy/sell agreement
Ship goods
Encrypted
documents
Distributed ledger
$ $ $ $
Issuing bank Advising bank Issuing bank Advising bank
• Letter of credit
• Documents
• Payment
Source: Adapted from www.virtusapolaris.com.
warnings by authorities, investors have flocked to ICOs even though they are often
linked to opaque business projects for which minimal and unaudited information is
supplied. Many of these projects have turned out to be fraudulent Ponzi schemes
(Graph V.9, right-hand panel).
A third, longer-term challenge concerns the stability of the financial system. It
remains to be seen whether widespread use of cryptocurrencies and related self-
executing financial products will give rise to new financial vulnerabilities and
systemic risks. Close monitoring of developments will be required. And, given their
novel risk profiles, these technologies call for enhanced capabilities of regulators

1
Data for 2016. 2 Average total cost for sending $200 with all remittance service providers worldwide. For CN and IN, receiving country
average total cost; for G20, SA and US, sending country average total cost.
Sources: World Bank, Remittance Prices Worldwide, remittanceprices.worldbank.org; World Bank; BIS calculations.
The shutdown of an illegal marketplace and the legitimacy of ICOs Graph V.9
Cryptocurrency prices react strongly to shutdowns of A large share of initial coin offerings is thought to be
illegal marketplaces 1 fraudulent
US dollars Per cent
120 22.5
105 15.0
90 7.5
75 0.0
12:00 18:00 00:00 06:00 12:00 Flagged by newspapers (1)
2 October 2013 3 October 2013 Website discontinued after ICO (2)
Based on white paper assessment (3)
Bitcoin price
At least one of (1)–(3)
1
Bitcoin price during October 2013 Silk Road shutdown.
Sources: C Catalini, J Boslego and K Zhang, “Technological opportunity, bubbles and innovation: the dynamics of initial coin offerings”, MIT
Working Papers, forthcoming; CoinDesk.
and supervisory agencies. In some cases, such as the execution of large-value, high-
volume payments, the regulatory perimeter may need to expand to include entities
using new technologies, to avoid the build-up of systemic risks.
The need for strengthened or new regulations and monitoring of cryptocurrencies
and related cryptoassets is widely recognised among regulators across the globe. In
particular, a recent communiqué of the G20 Finance Ministers and Central Bank
Governors highlights issues of consumer and investor protection, market integrity,
tax evasion and AML/CFT, and calls for continuous monitoring by the international
standard-setting bodies. It also calls for the Financial Action Task Force to advance
global implementation of applicable standards.32
However, the design and effective implementation of strengthened standards
are challenging. Legal and regulatory definitions do not always align with the new
realities. The technologies are used for multiple economic activities, which in many
cases are regulated by different oversight bodies. For example, ICOs are currently
being used by technology firms to raise funds for projects entirely unrelated to
cryptocurrencies. Other than semantics – auctioning coins instead of shares – such
ICOs are no different from initial public offerings (IPOs) on established exchanges,
so it would be natural for securities regulators to apply similar regulation and
supervision policies to them. But some ICOs have also doubled as “utility tokens”,
which promise future access to software such as games. This feature does not
constitute investment activity and instead calls for the application of consumer
protection laws by the relevant bodies.33
Operationally, the main complicating factor is that permissionless
cryptocurrencies do not fit easily into existing frameworks. In particular, they lack a
legal entity or person that can be brought into the regulatory perimeter.
Cryptocurrencies live in their own digital, nationless realm and can largely function
in isolation from existing institutional environments or other infrastructure. Their
legal domicile – to the extent they have one – might be offshore, or impossible to
establish clearly. As a result, they can be regulated only indirectly.

How can authorities implement a regulatory approach? Three considerations
are relevant.
First, the rise of cryptocurrencies and cryptoassets calls for a redrawing of
regulatory boundaries. The boundaries need to fit a new reality in which the lines
demarcating the responsibilities of different regulators within and across jurisdictions
have become increasingly blurred.34 Since cryptocurrencies are global in nature,
only globally coordinated regulation has a chance to be effective.35
Second, the interoperability of cryptocurrencies with regulated financial entities
could be addressed. Only regulated exchanges can provide the liquidity necessary
for DLT-based financial products to be anything but niche markets, and settlement
flows ultimately need to be converted into sovereign currency. The tax and capital
treatment rules for regulated institutions wanting to deal in cryptocurrency-related
assets could thus be adapted. Regulators could monitor whether and how banks
deliver or receive cryptocurrencies as collateral.
Third, regulation can target institutions offering services specific to cryptocurrencies.
For example, to ensure effective AML/CFT, regulation could focus on the point at
which a cryptocurrency is exchanged into a sovereign currency. Other existing laws
and regulations relating to payment services focus on safety, efficiency and legality
of use. These principles could also be applied to cryptocurrency infrastructure
providers, such as “crypto wallets”.36 To avoid leakages, the regulation would ideally
be broadly similar and consistently implemented across jurisdictions.
Should central banks issue digital currencies?
A related medium-term policy question concerns the issuance of CBDCs, including

who should have access to them. CBDCs would function much like cash: the central
bank would issue a CBDC initially, but once issued it would circulate between banks,
non-financial firms and consumers without further central bank involvement.37 Such
a CBDC might be exchanged between private sector participants bilaterally using
distributed ledgers without requiring the central bank to keep track and adjust
balances. It would be based on a permissioned distributed ledger (Graph V.2), with
the central bank determining who acts as a trusted node.
While the distinction between a general purpose CBDC and existing digital
central bank liabilities – reserve balances of commercial banks – may appear
technical, it is actually fundamental in terms of its repercussions for the financial
system. A general purpose CBDC – issued to consumers and firms – could
profoundly affect three core central banking areas: payments, financial stability
and monetary policy. A recent joint report by the Committee on Payments and
Market Infrastructures and the Markets Committee highlights the underlying
considerations.38 It concludes that the strengths and weaknesses of a general
purpose CDBC would depend on specific design features. The report further notes
that, while no leading contenders have yet emerged, such an instrument would
come with substantial financial vulnerabilities, while the benefits are less clear.
At the moment, central banks are closely monitoring the technologies while
taking a cautious approach to implementation. Some are evaluating the pros and
cons of issuing narrowly targeted CBDCs, restricted to wholesale transactions
among financial institutions. These would not challenge the current two-tier system,
but would instead be intended to enhance the operational efficiency of existing
arrangements. So far, however, experiments with such wholesale CBDCs have not
produced a strong case for immediate issuance (Box V.C).

Box V.C
Wholesale central bank digital currencies
In recent decades, central banks have harnessed digital technologies to improve the efficiency and soundness of
payments and the broader financial system. Digital technology has enabled central banks to economise on liquidity
provision to real-time gross settlement (RTGS) systems. Linking these systems through Continuous Linked Settlement
(CLS), commercial banks around the world settle trillions of dollars of foreign exchange around the clock every day.
CLS helps to remove Herstatt risk – the risk that a correspondent bank in a foreign exchange transaction runs into
financial trouble before paying the equivalent foreign currency to the designated recipient – which had previously
posed a significant financial stability risk. More recently, faster retail payments have spread across the world, and
central banks are actively promoting and facilitating this trend.
As part of their broader ventures into new payment technology, central banks are also experimenting with
wholesale CBDCs. These are token-based versions of traditional reserve and settlement accounts. The case for
wholesale DLT-based CBDCs depends on the potential for these technologies to improve efficiency and reduce
operational and settlement costs. The gains could be substantial, to the extent that many current central bank-
operated wholesale payment systems rely on outdated and costly-to-maintain technologies.
There are two key challenges for the implementation of wholesale CBDCs. First, the limitations of permissionless
DLT also apply to CBDCs, meaning that they need to be modelled on permissioned protocols. Second, the design
choices for the convertibility of central bank reserves in and out of the distributed ledger need to be implemented
carefully, so as to sustain intraday liquidity while minimising settlement risks.
A number of central banks, including the Bank of Canada (Project Jasper), the ECB, the Bank of Japan (Project
Stella) and the Monetary Authority of Singapore (Project Ubin), have already run experiments operating DLT-based
CBDC wholesale RTGS systems. In most cases, the central banks have chosen a digital depository receipt (DDR)
approach, whereby the central bank issues digital tokens on a distributed ledger backed by and redeemable for
central bank reserves held in a segregated account. The tokens can then be used to make interbank transfers on a
distributed ledger.
Central banks are now publishing the results. In their initial stages, each of the experiments largely succeeded
in replicating existing high-value payment systems. However, the results have not been clearly superior to existing
infrastructures.
  See M Bech and R Garratt, “Central bank cryptocurrencies”, BIS Quarterly Review, September 2017, pp 55–70; and Committee on Payments
and Market Infrastructures and Markets Committee, Central bank digital currencies, March 2018.

Endnotes
1
Terminology on this topic is fluid and evolving, with related legal and regulatory ambiguities. The
use of the term “cryptocurrencies” in this chapter is not meant to indicate any particular view of
what the underlying protocol-based systems are; typically, they have some, but not all, of the
characteristics of a sovereign currency and their legal treatment varies across jurisdictions. In some
cases, the chapter refers to specific cryptocurrencies or cryptoassets as examples. These examples
are not exhaustive and do not constitute any endorsement by the BIS or its shareholders of any
cryptocurrency, firm, product or service.
2
On this issue, see also Carstens (2018a,c).
3
Graeber (2011) argues that money only became widespread with the invention of coinage, which
appeared in China, India and Lydia almost simultaneously around 600–500 BCE. He further shows
that, contrary to popular belief, prior to the use of money, exchange took place mostly through
bilateral IOUs rather than barter.
4
These functions of money have been studied extensively in the literature. A few key examples are
the following: Kiyotaki and Wright (1989) show how money, when used as a medium of exchange,
can improve on barter. Kocherlakota (1996) shows that when perfect record-keeping and
commitment are not possible, money improves outcomes by serving as “memory”. Samuelson
(1958) shows in an overlapping generations model that money can improve efficiency when used
as store of value. Doepke and Schneider (2017) show how using a common unit of account
improves outcomes and why government money is the unit of account and the medium of
exchange at the same time.
5
Examples of items used as commodity money include shells in Africa, cocoa beans in the Aztec
civilisation and wampum in North American colonies. Even in these cases, credit relationships no
doubt coexisted with these arrangements. See eg Melitz (1974) for a more detailed discussion.
6
On the evolution of letters of credit and the pivotal role they have played in the development
of monetary systems in general, and the financing of trade in particular, see De Roover (1948,
1953). For a detailed analysis and history, see Kindleberger (1984) for a general treatment and
Santarosa (2015) for the importance of the introduction of joint liability.
7
Commodity-backed government money, such as the gold standard, was another attempt to strike
a balance. While offering stability in normal times, its constraints have tended to limit the central
bank’s ability to elastically supply currencies at times of financial and economic strains. In extreme
circumstances, these constraints have often simply been discarded, with a shift to inconvertibility.
For example, under the gold standard, one could regard the function of convertibility into gold as
constraining the sovereign’s ability to overissue and debase the currency. The constraint was
credible precisely because the commodity has a market value in non-monetary uses, ie other than
as a means of payment. This prevented the sovereign from keeping the holders hostage to its
monopoly powers. See Giannini (2011) for further discussion.
8
For a recent treatment, including an analysis of incentives to debase the money, see Schnabel and
Shin (2018).
9
See Van Dillen (1964), Roberds and Velde (2014) and Bindseil (2018). For the link with central
banking, see Ugolini (2017), Bindseil (2018) and Schnabel and Shin (2018).
10
Moreover, central banks have generally had the flexibility to act as lenders of last resort. The recent
Great Financial Crisis was yet another reminder of the both the fragility and the adaptability of the
current monetary arrangements, even in the most advanced economies. While the crisis laid bare
the shortcomings of the prevailing regulatory framework, the increased focus post-crisis on bank
supervision and regulation highlights how institutional arrangements can evolve to maintain trust
in money within the broad framework of the two-tiered system.
11
See Carstens (2018a). Giannini (2011) also highlights the importance of institutional arrangements
through which money is supplied: “The evolution of monetary institutions appears to be above all
the fruit of a continuous dialogue between economic and political spheres, with each taking turns
to create monetary innovations … and to safeguard the common interest against abuse stemming
from partisan interests.”

12
Indeed, central banks these days oversee payment systems and provide large amounts of intraday
credit to secure precisely this outcome, notably in wholesale payment systems. Depending on the
specificities of the arrangements, this credit may also be extended overnight or at longer
maturities. For a further description of the arrangements, operating procedures and other issues,
see BIS (1994) and Borio (1997).
13
See Bech and Garratt (2017) and CPMI-MC (2018) for a detailed discussion.
14
Much like with banknotes and other physical tokens, each transaction is verified with reference to
the payment object, ie the respective ledger entry. This differs from other forms of electronic
money, where verification is based on the identity of the account holder. Cryptocurrencies are
hence token-based digital money.
15
Current or planned examples of cryptocurrencies employing a permissioned model with designated
trusted nodes include the coin to be issued by the SAGA Foundation, Ripple and Utility Settlement
Coin.
16
We use “Bitcoin” to denote the protocol and network of users and miners of the cryptocurrency,
and “bitcoin” to denote the unit of currency.
17
Examples include Ethereum, Litecoin and Namecoin.
18
Auer (2018) presents a detailed description of the technological elements of Bitcoin and other
blockchain-based cryptocurrencies such as digital signatures, hashing and the cryptographic chaining
of blocks. See also Berentsen and Schär (2018).
19
Technically, this is implemented via the use of cryptographic hash functions (such as SHA-256 in
Bitcoin). These have the property that results are unpredictable, and a specific result can thus only
be generated by trial and error.
20
For a permissionless cryptocurrency to function in an entirely trustless environment, all miners and
users need to store an up-to-date copy of the entire ledger. However, in practice many users trust
the information provided by others. Some users only verify summary information of the ledger via
a process called simplified payment verification. And, much in contrast to the original idea
underlying Bitcoin, an even larger number of users can only access their funds through a third-
party website. In these cases, the third party alone is in control of its clients’ cryptocurrency
holdings.
21
Nakamoto (2009), p 8.
22
This is achieved by self-calibration of the proof-of-work, which increases the required level of
mathematical difficulty up to the point where the combined computing power of all miners just
suffices to update the ledger at the speed pre-set by the protocol.
23
See Carstens (2018a).
24
While congestion could be removed by allowing for bigger block sizes, this might actually be even
more destructive. Block rewards aside, having some congestion is essential to induce users to pay
for transactions, for if the system operates below its limit, all transactions will be processed and
rational users will thus post almost no transaction fees. The miners would not receive any benefits
for updating the transactions, and the equilibrium could break down. See in particular Hubermann
et al (2017) and Easley et al (2017), as well as Abadi and Brunnermeier (2018).
25
In technical terms, the interaction between the users is that of strategic substitutes, not strategic
complements. Cryptocurrencies are hence a congestion, rather than a coordination, game.
26
The probabilistic nature of finality could in particular create aggregate risks if cryptocurrencies
were used in wholesale settings, where funds tend to be reinvested without delay. In fact, this
would create an entirely new dimension of aggregate risk, as exposures would be linked to each
other via the probability of non-finality of the entire transaction history.

27
There is no shortage of proposed solutions, but most have yet to be proved in practice. On the
one hand, future cryptocurrency protocols might do away with costly proof-of-work by replacing it
with “proof-of-stake”, the underlying idea of which is to achieve credibility by staking cryptocurrency
holdings rather than doing costly computational work. Proposed solutions for the scaling problem
include the Lightning Network, which essentially shifts small transactions off the main blockchain
and into a separate pre-funded environment. There are also new cryptocurrencies, such as IOTA,
that aim to replace the blockchain with a more complex ledger and verification structure.
28
See Juskalian (2018).
29
See Carstens (2018a,b).
30
Government officials are also not immune from the lure of cryptocurrencies: two US government
agents have been charged with theft of bitcoins confiscated during the closing of Silk Road.
31
For example, most bitcoin payments made via smartphone are most likely made indirectly via third
party, since the current blockchain size exceeds the storage capacity of most smartphones. Reuters
(2017) and Moore and Christin (2013) list some of the cases in which such third parties have
proved to be fraudulent or have fallen victim to hacking attacks. For an analysis of illicit uses of
cryptocurrencies, see Fanusie and Robinson (2018) and Foley et al (2018).
32
See G20 Finance Ministers and Central Bank Governors (2018).
33
Clayton (2017), discussing the regulation of ICOs as opposed to IPOs from a US perspective, states
that a “change in the structure of a securities offering does not change the fundamental point that
when a security is being offered, our securities laws must be followed”. FINMA (2018) sets out a
regulatory framework in Switzerland that classifies ICOs according to the eventual use of the
tokens issued: in payments, as assets or as utility tokens.
34
Technically, all that is needed for protocol-based cryptocurrencies to operate is for at least one
country to allow access. The authorities’ difficulties in shutting down illegal download sites such as
Napster or The Pirate Bay and download protocols such as BitTorrent underline the associated
enforcement problems.
35
Financial Action Task Force (2015) argues that treating similar products and services consistently
according to their function and risk profile across jurisdictions is essential for enhancing the
effectiveness of the international AML standards.
36
One complication is that payments are regulated by a set of authorities and laws with very different
goals, such as payment system oversight, prudential supervision, consumer protection and AML/CFT.
For example, US-based institutions must adhere to, among others, the Bank Secrecy Act, the USA
PATRIOT Act and Office of Foreign Assets Control regulations. Another complication has to do with
the applicability of existing legislation to the new instruments. For instance, in the European Union
the legal definition of electronic money includes the requirement that balances should represent a
claim on the issuer. As cryptocurrencies do not represent any claim, they cannot be considered
electronic money and are thus by default not covered by the respective legislation.
37
There are many potential technical implementations of token-based CBDCs. They could be based
on DLT, with similar characteristics to cryptocurrencies, with the difference being that the central
bank rather than the protocol itself would be in control of the amount issued and would guarantee
the token’s value.
38
CPMI-MC (2018).

References
Abadi, J and M Brunnermeier (2018): “Blockchain economics”, Princeton University, mimeo, May.
Auer, R (2018): “The mechanics of decentralised trust in Bitcoin and the blockchain”, BIS Working Papers,
forthcoming.
Bank for International Settlements (1994): 64th Annual Report, June.
Bech, M and R Garratt (2017): “Central bank cryptocurrencies”, BIS Quarterly Review, September 2017,
pp 55–70.
Berentsen, A and F Schär (2018): “A short introduction to the world of cryptocurrencies”, Federal Reserve
Bank of St Louis, Review, vol 100, no 1.
Bindseil, U (2018): “Pre-1800 central bank operations and the origins of central banking”, Mannheim
University, mimeo.
Borio, C (1997): “The implementation of monetary policy in industrial countries: a survey”, BIS Economic
Papers, no 47, July.
Carstens, A (2018a): “Money in the digital age: what role for central banks?”, lecture at the House of
Finance, Goethe University, Frankfurt, 6 February.
——— (2018b): “Central banks and cryptocurrencies: guarding trust in a digital age”, remarks at
Brookings Institution, Washington DC, 17 April.
——— (2018c): “Technology is no substitute for trust”, Börsen-Zeitung, 23 May.
Catalini, C, J Boslego and K Zhang (2018): “Technological opportunity, bubbles and innovation: the
dynamics of initial coin offerings”, MIT Working Papers, forthcoming.
Clayton, J (2017): “Statement on cryptocurrencies and initial coin offerings”, www.sec.gov/news/public-

statement/statement-clayton-2017-12-11, 11 December.
Committee on Payments and Market Infrastructures and Markets Committee (2018): Central bank
digital currencies, March.
De Roover, R (1948): Money, banking and credit in mediaeval Bruges: Italian merchant-bankers, Lombards
and money changers – a study in the origins of banking, Mediaeval Academy of America.
——— (1953): L’évolution de la lettre de change: XIVe-XVIIIe siècle, Armand Colin.
Doepke, M and M Schneider (2017): “Money as a unit of account”, Econometrica, vol 85, no 5, pp 1537–74.
Easley, D, M O’Hara and S Basu (2017): ”From mining to markets: The evolution of Bitcoin transaction
fees”, papers.ssrn.com/sol3/papers.cfm?abstract_id=3055380.
Fanusie, Y and T Robinson (2018): “Bitcoin laundering: an analysis of illicit flows into digital currency
services”, Center on Sanctions & Illicit Finance memorandum, January.
Financial Action Task Force (2015): Guidance for a risk-based approach to virtual currencies, June.
Financial Market Supervisory Authority (FINMA) (2018): Guidelines for enquiries regarding the regulatory
framework for initial coin offerings (ICOs), 16 February.
Foley, S, J Karlsen and T Putniņš (2018): “Sex, drugs, and bitcoin: how much illegal activity is financed
through cryptocurrencies?”, dx.doi.org/10.2139/ssrn.3102645.
G20 Finance Ministers and Central Bank Governors (2018): Buenos Aires Summit communiqué, 19–20 March.
Giannini, C (2011): The age of central banks, Edward Elgar.

Graeber, D (2011): Debt: the first 5,000 years, Melville House.
Huberman, G, J Leshno and C Moellemi (2017): “Monopoly without a monopolist: an economic analysis
of the Bitcoin payment system“, Columbia Business School Research Papers, no 17–92.
Juskalian, R (2018): “Inside the Jordan refugee camp that runs on blockchain”, MIT Technology Review,
online edition, 12 April.
Kindleberger, C (1984): A financial history of western Europe, Allen & Unwin.
Kiyotaki, N and R Wright (1989): “On money as a medium of exchange”, Journal of Political Economy, vol 97,
no 4, pp 927–54.
Kocherlakota, N (1996): “Money is memory”, Journal of Economic Theory, vol 81, issue 2, pp 232–51.
Melitz, J (1974): Primitive and modern money: an interdisciplinary approach, Addison-Wesley.
Moore, T and N Christin (2013): “Beware the middleman: empirical analysis of Bitcoin-exchange risk”, in
A-R Sadeghi (ed), Lecture Notes in Computer Science, vol 7859.
Nakamoto, S (2009): “Bitcoin: a peer-to-peer electronic cash system”, white paper.
Reuters (2017): “Cryptocurrency exchanges are increasingly roiled by hackings and chaos”, 29 September.
Roberds, W and F Velde (2014): “Early public banks”, Federal Reserve Bank of Chicago Working Papers,
no 2014–03.
Samuelson, P (1958): “An exact consumption-loan model of interest with or without the social
contrivance of money”, Journal of Political Economy, vol 66, no 6, pp 467–82.
Santarosa, V (2015): “Financing long-distance trade: the joint liability rule and bills of exchange in
eighteenth-century France“, The Journal of Economic History, vol 75, no 3, pp 690–719.
Schnabel, I and H S Shin (2018): “Money and trust: lessons from the 1620s for money in the digital age”,
BIS Working Papers, no 698, February.
Ugolini, S (2017): The evolution of central banking: theory and history, Palgrave-Macmillan.
Van Dillen, J G (1964): History of the principal public banks, Frank Cass & Co.

Promoting global monetary
and financial stability
Bank for International Settlements (BIS)

www.bis.org
[email protected]
ISSN 2616-9436
ISBN 978-92-9259-174-8

BISar 2018 e

Uploaded by

Copyright:

Available Formats

BISar 2018 e

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

BISar 2018 e

Uploaded by

Copyright:

Available Formats

BIS

Promoting global monetary

ISSN 2616-9428 (print)

Annual Economic Report 2018: Editorial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix

The global economy: developments, prospects and risks . . . . . . . . . . . . . . . . . . . . . . . . . . ix

I. A stronger expansion: how to make it last . . . . . . . . . . . . . . . . . . . . . . . . 1

The global expansion strengthens amid low inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

II. Monetary policy: a narrow normalisation path . . . . . . . . . . . . . . . . . . 25

Monetary policy normalisation: where do we stand? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

III. The financial sector: post-crisis adjustment and pressure

Basel III: key elements of the completed framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

BIS Annual Economic Report 2018 iii

Macroprudential frameworks: elements and challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

V. Cryptocurrencies: looking beyond the hype . . . . . . . . . . . . . . . . . . . . . . 91

Putting the rise of cryptocurrencies into perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91

iv BIS Annual Economic Report 2018

E.1 The global economy strengthens further as inflation converges . . . . . . . . . . . . . . . x

I.1 GDP growth improves, investment strengthens and unemployment declines . . . 2

II.1 An asynchronous policy normalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

III.1 Unsustainable risk-taking required decisive post-crisis adjustments by banks . . . 44

IV.1 Macroprudential orientation moves to the mainstream . . . . . . . . . . . . . . . . . . . . . . 64

V.1 The money flower: a taxonomy of money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

BIS Annual Economic Report 2018 v

III.1 Basel III phase-in arrangements: key standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

Conventions used in the Annual Economic Report

$ US dollar unless specified otherwise

Components may not sum to totals because of rounding.

vi BIS Annual Economic Report 2018

Global: All AEs and EMEs, as listed.

Commodity exporters (countries whose average share of commodities in export

BIS Annual Economic Report 2018 vii

The global economy: developments, prospects and risks

BIS Annual Economic Report 2018 ix

GDP growth1, 2 Unemployment rates 2, 3 Inflation1, 4

BIS Annual Economic Report 1

Global debt continues to rise1 Graph E.2

USD trn % of GDP

Global total (lhs) Rhs: General government Non-financial corporates Households

BIS Annual Economic Report 2018 xi

xii BIS Annual Economic Report 2018

BIS Annual Economic Report 2018 xiii

Government debt1 Interest rates and central bank Structural reforms6

BIS Annual Economic Report 3

The implementation of macroprudential frameworks represents a major and

Cryptocurrencies promise to replace trust in long-standing institutions, such as

BIS Annual Economic Report 2018 xv

xvi BIS Annual Economic Report 2018

The global expansion strengthens amid low inflation

BIS Annual Economic Report 2018 1

Real GDP Contributions: AEs EMEs Global

2 BIS Annual Economic Report 2018

Inflation remains low and wage growth subdued Graph I.2

Inflation1 Real wage growth1, 2 Real compensation and labour

Sources: OECD; Datastream; national data; BIS calculations.

BIS Annual Economic Report 2018 3

Financial conditions remained very easy for most of the year

4 BIS Annual Economic Report 2018

Long-term view of Term spread2 US financial conditions3 Corporate spreads 4

12 1.25 2.25 1,000