Journal of Economic Perspectives—Volume 33, Number 1—Winter 2019—Pages 107–130

Would Macroprudential Regulation Have

Prevented the Last Crisis?

David Aikman, Jonathan Bridges, Anil Kashyap,

and Caspar Siegert

A
key response of official sectors around the world to the financial and
economic crises of ten years ago has been the formation of financial stability
committees. Such committees now exist in over 40 countries worldwide
(Edge and Liang 2017). The remits of these committees are “macroprudential.”
Macroprudential policy focuses on potential system-wide risks and amplification
mechanisms, complementing the detailed firm-specific risk assessments of micro-
prudential regulators. In addition, it has the explicit objective to ensure that the
financial system does not amplify a downturn in the real economy—for example,
by being forced to cut back on the supply of credit in a stress (Borio 2003; in this
journal, Hanson, Kashyap, and Stein 2011).
This paper asks whether macroprudential authorities, as they have been
designed over the past decade, could prevent—or materially dampen—a rerun of
the last crisis. To be clear at the outset, macroprudential regulation does not seek to

■ David Aikman is Technical Head of the Macroprudential Strategy and Support Division,
Bank of England, London, United Kingdom. Jonathan Bridges is a Senior Economist, Macro-
prudential Strategy and Support Division, Bank of England, London, United Kingdom. Anil
Kashyap is the Edward Eagle Brown Professor of Economics and Finance at the University
of Chicago Booth School of Business, Chicago, Illinois. Since 2016, he has been a Member of
the Financial Policy Committee, Bank of England. He is also a Research Associate, National
Bureau of Economic Research, Cambridge, Massachusetts, and a Research Fellow, Centre for
Economic Policy Research, London, England. Caspar Siegert is a Senior Economist, Macro-
prudential Strategy and Support Division, Bank of England, London, United Kingdom.
Kashyap is the corresponding author at [email protected].
†
For supplementary materials such as appendices, datasets, and author disclosure statements, see the
article page at
https://doi.org/10.1257/jep.33.1.107 doi=10.1257/jep.33.1.107
108 Journal of Economic Perspectives

eliminate recessions. Instead, it is aimed at ensuring that the financial system does
not create shocks that trigger recessions or amplify other shocks to make recessions
materially worse. With this in mind, the first part of our paper provides an account
of the amplifying factors that made the last crisis so severe. Our diagnosis centers
on two overlapping but distinct vulnerabilities: the increase in leverage and short-
term funding at financial intermediaries, and the build-up in indebtedness in the
household sector. These factors, we argue, can account for around two-thirds to
three-quarters of the fall in US GDP that followed the financial crisis. We describe
and calibrate the policy interventions required to address these vulnerabilities.
We then contrast how well-equipped two prominent macroprudential regulators
are to make these interventions. We argue that the US Financial Stability Oversight
Council would likely make little difference were we to experience a rerun of the
factors that caused the last crisis. It has no macroprudential levers under its direct
control, and not all of its members have mandates to protect financial stability. A
macroprudential regulator modeled on the UK’s Financial Policy Committee stands
a better chance because it has many of the necessary powers. But spotting build-ups
in vulnerabilities in real-time is challenging. And given the role played by loosely
regulated nonbank financial institutions prior to the last crisis—and the continuing
evolution of the financial system—a successful macroprudential intervention would
likely require political backing to be nimble in widening the perimeter of ­regulation
to capture such institutions. More generally, such a regulator would have to be fairly
aggressive in using its powers. Given the novelty of these powers, there is no clear
evidence on whether such forceful interventions would be realistic were risks to
escalate again. Our conclusion distils some key challenges and priorities for the
development of a successful macroprudential framework.
The test we pose is really not very tough. Today’s macroprudential frameworks
were created in response to the scenario we are revisiting, whereas the challenges
facing macroprudential regulators in the future will likely be new. But while our
essay explores how today’s macroprudential regimes might respond if vulnerabili-
ties similar to those that caused the last crisis were to reoccur, we also invite readers
to use this thought-experiment to consider how macroprudential committees might
respond if other “resilience gaps” opened up in the future.

Fault-Lines That Led to the 2008 Financial Crisis

In Aikman, Bridges, Kashyap, and Siegert (2018), we describe the competing

factors that contributed to the 2008 financial crisis and discuss the dimensions of
their relative contributions. Here we summarize the evidence regarding the two
dominant contributors: 1) the fragilities in the financial system associated with
excessive leverage and the use of potentially flighty short-term funding; and 2)
the unprecedented (by US standards) lending boom to the household sector that
began in the mid-2000s. Bernanke (2018) also identifies these two channels to be of
primary importance, with particular emphasis on the former.
 David Aikman, Jonathan Bridges, Anil Kashyap, and Caspar Siegert 109

Table 1
Size and Structure of the Leveraged Financial System

2001:Q4 2007:Q4

Assets Liquid Short-term Assets Liquid Short-term

($bn) Leverage assets funding ($bn) Leverage assets funding

Commercial banks 6,552 11.0 6.6% 26.5% 11,182 9.8 4.6% 33.2%
of which: large institutions 2,291 12.2 6.7% 32.9% 5,422 11.8 4.6% 37.5%
Savings institutions 1,317 11.6 3.0% 18.2% 1,852 9.1 2.3% 22.6%
Broker-dealers 2,376 28 2.4% 57.3% 4,686 45 0.4% 63.4%
Government-sponsored 1,417 42.3 0.2% 1,677 23.7 0.7%
enterprises
Total 11,662 19,397

Source: Financial Accounts of the United States; Call Reports; FDIC; Adrian, Fleming, Shachar, and Vogt
(2017); and Annual Reports of Fannie Mae and Freddie Mac.
Note: “Leverage” is defined as total assets divided by (book) equity. “Liquid assets” refers to the ratio of
cash and Treasury securities to total assets. For brokers, “short-term funding” refers to repo funding
relative to total assets. For deposit-takers, it refers to (estimated) uninsured domestic deposits and
foreign deposits relative to total assets. While deposits are typically short-term liabilities, many types
of deposits, including insured deposits in particular, are “behaviorally stable” and were not withdrawn
during the crisis (Martin, Puri, and Ufier 2018). “Large commercial banks” are defined as banks with at
least $150 billion in total assets. For 2007, this is adjusted using the Consumer Price Index ($180 billion).
Government-sponsored enterprises include Fannie Mae and Freddie Mac.

Our thesis is that these two factors amplified the initial losses that occurred
when house prices fell. The fragilities in the financial system meant that lenders had
to cut back lending as they struggled to absorb losses, which led to a credit crunch
that reduced investment and employment. As households also struggled to deal
with excessive debt, they cut spending, amplifying the downturn further.

Fragilities in the Financial System

Vulnerabilities in the financial system built significantly in the years leading up
to the global financial crisis (Brunnermeier 2009; Acharya, Philippon, Richardson,
and Roubini 2009; Duffie 2018). As a result, even relatively small losses on financial
institutions’ mortgage exposures were sufficient to trigger stability concerns for the
entire financial system that ultimately spilled over into the real economy.
Table 1 documents the solvency, liquidity, and funding positions of different
classes of US financial institutions at two points prior to the global financial crisis:
end-2001, a period when the US economy was recovering from the strains caused
by the collapse of the dot-com bubble; and end-2007, the beginning of the financial
crisis. The total assets of the institutions we capture here increased from $12 tril-
lion to almost $20 trillion between these dates. Clearly, the use of debt finance, or
“leverage,” varied significantly across institutions. For the largest commercial banks,
leverage changed little in the years leading up to the crisis. Commerical banks
did, however, reduce the amount of assets that could easily be sold without price
110 Journal of Economic Perspectives

concessions (liquid assets) and expanded their reliance on short-term funding,

which can rapidly disappear during times of stress.
The most extreme vulnerabilities developed for the parts of the financial
system that did not take traditional deposits. Consider, for instance, the changes
for ­broker-dealers, a category that includes specialised investment banks and the
investment banking subsidiaries of larger banking groups. The assets of these enti-
ties increased from 28 to 45 times their equity between 2001 and 2007, meaning that
a roughly 2 percent decline in the value of broker-dealers’ assets would have been
sufficient to wipe out all of their equity. In addition, these firms were traditionally
highly reliant on short-term wholesale funding (Rosengren 2014), and became even
more so during this period.
Much of this short-term funding took the form of repurchase agreements, or
“repos.” Repos are a form of borrowing in which the broker-dealer sells securities
that it holds, receives the value of those securities in cash, and a few days later repur-
chases the securities at a predetermined price that includes an additional interest
payment. The repo liabilities of broker-dealers increased from $1.4 trillion in 2001
to $3.0 trillion in 2007.1 Moreover, an increasing fraction of repos were backed by
low-quality securities.
Figure 1 shows the rise in repo funding, along with commercial paper, another
form of funding that experienced rapid growth over this period. Traditional commer-
cial paper is short-term debt issued by companies to fund operations. However, by
the end of 2006, 60 percent of outstanding commercial paper consisted of so-called
“asset-backed commercial paper” that had been issued to fund the purchase of specific
securities such as credit card receivables, auto loans, or mortgage-backed securities.
The growth in repos and commercial paper coincided with an increase in
the size of money market mutual funds, which purchased much of the repos and
commercial paper issued. Regulators allowed money market mutual funds to invest
in assets with a weighted average maturity of up to 90 days, but these funds offered
investors the ability to withdraw their money at a day’s notice. Moreover, money
market mutual funds did not have any capital that would shield these short-term
investors from losses. In a crisis, investors in money market mutual funds who with-
drew their funds first were certain to be fully paid, while later claims might not be
fully paid, providing incentives to “run” on the fund.
In summary, nonbanks became an increasingly important source of credit for
the real economy in the years preceding the crisis: between 2001 and 2007, nonbank
financials accounted for over 70 percent of the total growth in home mortgage
credit (according to the Financial Accounts of the United States). This growth was
accompanied by an increased reliance on debt financing of the nonbank system.
Short-term borrowing became more important, with the belief that it could be

1
Total repo liabilities for all types of institutions recorded in the Financial Accounts of the US data for
end-2001 and 2007 were $2.2 trillion and $4.8 trillion, respectively. None of these numbers were readily
available in the run-up to the crisis, as broker-dealers repo liabilities were only reported on a netted basis
(Eichner, Kohn, and Palumbo 2013; Holmquist and Gallin 2014).
 Would Macroprudential Regulation Have Prevented the Last Crisis? 111

Figure 1
Reliance on Short-Term Funding
($ billions)
4,500

4,000
Money market funds
3,500

3,000
Repo liabilities of broker-dealers
2,500

2,000

1,500

1,000
Commercial paper
500

0
19
19 Q1
19 Q1
19 Q1
19 Q1
19 Q1
19 Q1
19 Q1
19 Q1
19 Q1
20 Q1
20 Q1
20 Q1
20 Q1
20 Q1
20 Q1
20 Q1
20 Q1
20 Q1
80
82
84
86
88
90
92
94
96
98
00
02
04
06
08
10
12
14
16
Q
1
Source: March 2018 release of the Financial Accounts of the United States, based on Adrian, de
Fontnouvelle, Yang, and Zlate (2017). The size of money-market funds is measured as outstanding
money market fund shares (liabilities) in table L.121. Commercial paper refers to commercial paper
(liabilities) issued by any sector (table L.2019), which includes asset-backed commercial paper. Repo
liabilities of broker-dealers are based on security repurchase agreements (liabilities) in table L.130.

rolled over continually. These observations suggest that macroprudential regulators’

­arsenals must include tools to affect the overall propensity to rely on debt financing
and also affect the maturity of the funding, and macroprudential ­regulators must
have scope to apply these tools both to banks and to nonbanks.

The Build-up in Household Debt

Alongside the pronounced build-up in leverage and short-term funding in the
financial system, there was a rapid build-up in debt in the real economy, concen-
trated in household mortgages. Mortgage debt doubled in the six years before the
crisis, and by 2007 reached 72 percent of GDP. Two aspects of this debt build-up are
noteworthy and will inform our later macroprudential analysis.
First, the increase in mortgage debt was accompanied by a house price boom,
shown in Figure 2. House prices rose by two-thirds in the five years to their peak in
early 2006 (according to the S&P Case-Shiller US National Home Price Index), and
ongoing rapid house price appreciation was embedded in expectations (Gennaioli
and Shleifer 2018). The aggregate loan-to-value ratio on the stock of US housing
remained broadly flat during this period, meaning that for each 1 percent increase
in house values, homeowners also increased their mortgage debt by around
1 percent. In part, this reflected new homeowners taking out larger mortgages in
order to purchase more expensive homes. But in addition, existing homeowners
112 Journal of Economic Perspectives

Figure 2
Mortgage Debt and House Prices
(annual percentage change)
15
Mortgage debt growth

10

Percent (annual)
0

−5

House price growth −10

−15
1990 1994 1998 2002 2006 2010 2014 2018

Source: Financial Accounts of the United States and S&P Case-Shiller US National Home Price Index.

also extracted housing equity by taking out additional debt. Mian and Sufi (2011)
estimate that existing homeowners borrowed $0.25 on average for every $1 increase
in home-equity value during the housing boom, enough to account for over half of
the increase in debt for homeowners between 2002 and 2006.
Second, there were clear signs in the years before the financial crisis that lending
standards were being loosened and borrower quality was deteriorating. The Federal
Reserve Board’s Senior Loan Officer Opinion Survey on Bank Lending Practices
reported easing standards between 2004Q1 and 2006Q3. The expansion of credit to
the most risky borrowers was particularly pronounced. For example, according to the
Federal Reserve’s Survey of Consumer Finances, the share of the stock of mortgagors
with debt of over four times their income more than doubled between 2001 and 2007
from 6 percent to 13 percent. The number of new subprime mortgages nearly doubled
between 2003 and 2005, 80 percent of which were made with short-term “teaser” interest
rates (in this journal, Mayer, Pence, and Sherlund 2009). “Near-prime” mortgages also
increased rapidly. The private-label securitization market, in which these mortgages
were bundled into tranched financial securities and resold, was an important driver of
these frothy credit supply conditions (Keys, Mukherjee, Seru, and Vig 2010).
In summary, the years running up to the Great Recession saw an unprece-
dented surge in US household debt. That boom was accompanied and reinforced
by soaring property prices. Aggressive credit supply expansion, compounded by
financial innovation, provided the undercurrent for an unsustainable cycle. House-
hold balance sheets became increasingly vulnerable to a shock as more credit was
extended to highly indebted households.
 David Aikman, Jonathan Bridges, Anil Kashyap, and Caspar Siegert 113

Quantifying the Effects of the Credit Crunch and Deleveraging

Extrapolating the 20-year average growth rate from 2007Q4 through 2010Q4
suggests that the level of GDP per capita was 8.5 percent below trend by 2010Q4.
How much of this might plausibly be attributed to the credit crunch and the
deleveraging by overly indebted households? Estimating precise contributions is
challenging. Nonetheless, triangulating across a range of studies, it seems very likely
that these factors account for a large part of that gap. In Aikman, Bridges, Kashyap,
and Siegert (2018), we provide the details of how we arrive at this conclusion; here,
we simply summarize the main points.
A variety of studies, using a variety of methods, find that the fragility of lenders
resulted in a credit crunch that had the potential to materially affect real economic
activity (for example, Chodorow-Reich 2014; Greenlaw, Hatzius, Kashyap, and
Shin 2008; Bassett, Chosak, Driscoll, and Zakrajšek 2014; Guerrieri et al. 2015).
Translating the estimates from any one of these studies into an impact on the GDP
shortfall requires a number of assumptions. For example, Chodorow-Reich (2014)
identifies the impact of the credit crunch on employment in enterprises with less
than 1000 employees. For our purposes, this finding needs to be extrapolated to the
entire economy and then translated into an impact on GDP. Averaging across our
five preferred studies, in Aikman, Bridges, Kashyap, and Siegert (2018), we find that
about 35 percent of the 2010 GDP gap can be attributed to the abrupt tightening of
credit conditions. That is, around 3 percentage points of the overall GDP shortfall
can be explained by the fragilities in the financial system, which meant that the
economy was prone to suffering a credit crunch.
There is also convincing evidence that a strong relationship exists between
household debt growth in the years preceding economic downturns and the severity
of the subsequent downturn (for example, Jordà, Schularick, and Taylor 2013, 2016;
Bridges, Jackson, and McGregor 2017; Mian and Sufi 2012; Mian, Sufi, and Verner
2017; Gertler and Gilchrist 2018). Converting the various estimates into an impact
on the fall in GDP during the Great Recession is subject to the same basic chal-
lenges as with scaling the impact of the credit crunch. In Aikman, Bridges, Kashyap,
and Siegert (2018), we average across a number of approaches and find that the
household debt boom can account for about one-half of the GDP gap, or just over
4 percentage points of the overall fall in GDP.
Our reading of the existing evidence is therefore that, taken together, these two
effects account for around three-quarters of the contraction in output that occurred
during the Great Recession. That is, absent the credit crunch and the deleveraging
by households, the cumulative fall in GDP growth during the recession would have
been three-quarters smaller.
It is obviously an oversimplification to treat the effects of the credit crunch
and the household deleveraging channels as if they were clear and distinct events.
Rather, there was two-way feedback between these phenomena: tight credit condi-
tions intensified households’ need to deleverage, and the reduction in spending
by highly indebted households led to an economic contraction that made it harder
for all borrowers to service their debts, generating larger losses for banks and other
114 Journal of Economic Perspectives

financial intermediaries. However, disentangling the precise impact of these chan-

nels is not material to our argument. Rather, our claim is that each channel had
a very substantial bearing on the costs of the crisis. Should we see a rerun of the
factors that caused the last crisis, therefore, macroprudential policy would have
to address vulnerabilities associated with excessive debt-financing and short-term
funding in the financial system, and excessive debt levels in the household sector.

What Could a Macroprudential Regulator Have Done to Address the

Build-Up in These Vulnerabilities?

We start this section by asking whether it was possible to spot the vulnerabilities
in both the financial system and in household balance sheets documented above in
real time. We then consider the policies a macroprudential regulator could intro-
duce in response. In the next section, we discuss the institutional frictions that actual
macroprudential regulators would face in implementing such policies in practice.

Identifying the Build-up in Risk in Real Time

Identifying macroprudential policy interventions requires spotting emerging
risks and accumulating vulnerabilities prior to a crisis. Identifying the debt build-up
in the household sector was relatively straightforward. The Bank for International
Settlements was sounding alarms about the risks from credit build-ups in 2004 (for
example, Borio and White 2004). The IMF’s Global Financial Stability Report in April
2005 had a chapter on the state of household balance sheets in advanced economies.
Table 2 summarizes how often some key words associated with building fragilities
were mentioned in Federal Open Market Committee (FOMC) transcripts through
the 2000s. Again, the build-up in household debt was clearly evident from the early
2000s. The house price bubble was also observed well ahead of the crisis. By its
meeting in June 2005, the FOMC was discussing evidence that houses may be up to
20 percent overvalued, leading to the spike in transcript references to “house price,”
“bubble,” and words associated with mortgage lending. This assessment turned out
to be pretty accurate: by the end of the recession in June 2009, house prices were
13 percent below their June 2005 level, and the peak-to-trough fall during the crisis
was 20 percent (based on the Case-Shiller US National Home Price Index). Between
the end of the recession in June 2009 and February 2012, house prices fell further,
bringing the overall peak-to-trough fall to 26 percent. However, the extent to which
the build-up in debt was being concentrated at riskier, heavily indebted borrowers was
not being adequately picked up (Eichner, Kohn, and Palumbo 2013). It is striking
that the word “subprime” was mentioned 314 times in the FOMC’s 2007 transcripts,
but only 27 times in all the transcripts from 2000 to 2006. Commercial paper and
“securitization” were also rarely mentioned before 2007.
More broadly, policymakers did not understand the effects that a sharp fall
in house prices would have on the financial system. This lack of resilience might
have been identified via stress tests—since the crisis, such tests have become a key
 Would Macroprudential Regulation Have Prevented the Last Crisis? 115

Table 2
Financial Stability Terms Appearing in Discussions of the Federal Open Market
Committee

2001 2002 2003 2004 2005 2006 2007

General
“Financial stability” 0 1 2 9 5 9 13
Financial System:
“Bank” 502 429 449 302 284 309 1,024
“Capital”/“Leverage” 454 308 340 208 183 177 402
“Shadow”/“Broker”/“Money market” 17 21 40 10 17 28 59
“Fund”/“Liquid”/“Repo” 1,226 962 1,150 1,058 932 1,110 1,779
“Commercial paper”/“Securitization” 23 22 15 3 14 2 133
Housing Market
“House price” 2 23 4 41 160 85 83
“Bubble” 6 15 14 19 114 4 8
“Loan”/“Lend”/“Debt”/“Credit”/“Borrow” 413 442 452 269 409 251 1,563
“Mortgage” 84 100 96 67 176 118 481
“Subprime” 0 3 1 0 8 15 314
“LTV”/“Heloc”/“Teaser”/“Alt-A” 2 1 1 0 40 0 45

Note: For each year, transcripts of the eight FOMC meetings and any Conference Calls were searched. All
transcripts available here: https://www.federalreserve.gov/monetarypolicy/fomc_historical_year.htm. A
simple count of all words containing the stem words listed in the table above was conducted.

component of macroprudential regulators’ toolkits. Hirtle, Kovner, Vickery, and

Bhanot (2016) demonstrate that a top-down stress test based on a macroeconomic
scenario like the one that played out in 2007–08 would have predicted a significant
capital shortfall in the US banking system as early as 2004. But even with a severe
stress test, it would have been difficult to understand the fragility of funding flows
across the system prior to the crisis, which led to fragilities in the nonbank sector
and amplified the macroeconomic downturn. To reveal the full extent of the vulner-
abilities that existed, stress tests would have had to cover the entire financial system:
broker-dealers; commercial paper, repo, and derivative markets; specialized invest-
ment vehicles (SIVs); and other conduits. Building a complete map of funding
interconnections between these markets and entities is challenging even today.
Thus, while some warning signs were clearly present in the lead-up to the crisis,
we are cautious about the ability of macroprudential regulators to understand the
nature of systemic financial risks as they emerge in real time. One implication is
that policymakers should seek to develop systematic frameworks within which to
monitor emerging risks and their potential implications for macroeconomic tail
events.2 Another implication is that we should be humble (see also Tarullo 2014).
The fact that even in hindsight we believe it would have been hard to diagnose fully
the risks in the run-up to the crisis suggests that macroprudential policy frameworks

2
Analysis of “GDP-at-risk” and its link to financial indicators is one promising avenue here—see for
example Adrian, Boyarchenko, and Giannone (2016), IMF (2018b), and Aikman et al. (2018).
116 Journal of Economic Perspectives

should be calibrated with some built-in “slack” to account for the inherent difficulty
of risk assessment, particularly in real time.

Tools and Actions to Reduce Leverage

How much additional capital would US banks have needed to be resilient given
the extent of the credit bubble that was building in this period? For a sense of the
necessary scale, consider the government capital support that occurred at the height
of the crisis via the Troubled Asset Relief Program (TARP). Under this scheme, the
US Treasury invested $200 billion in the preferred stock of 15 large US banks to
enhance market confidence in the banking system and to increase its capacity to
lend.3 While establishing cause and effect is difficult, there is evidence that this
intervention led to dramatic improvements in how market participants viewed the
solvency of the US banking system. For example, the interest rate spreads for banks’
unsecured borrowing—often measured by looking at the difference between the
three-month interbank borrowing rate and the risk-free Treasury bill yield—fell
sharply almost immediately after the TARP was announced on October 14, 2008.
One means by which the authorities could increase system-wide levels of capital
in the banking system in response to an emerging “resilience gap” is through a
macroprudential tool called the countercyclical capital buffer (CCyB). The CCyB
allows regulators to increase capital requirements according to the aggregate risk
environment. What level of the countercyclical capital buffer would have delivered
a level of resilience equivalent to the TARP injections? The countercyclical capital
buffer is typically expressed as a percentage of a firm’s assets, weighted by the riski-
ness. It is then adjusted by a “domestic lending conversion factor,” which accounts
for the fact that large banks operate across international boundaries. An estimate of
the necessary countercyclical capital buffer rate is hence:
Required countercyclical capital buffer
$200 billion
= ​​ __________________
  
   ​​* Domestic lending conversion factor
Risk-weighted assets
As of 2005, the 15 TARP recipients on which we focus had total risk-weighted assets
of approximately $8.4 trillion—the denominator of the expression above.4 The average
“domestic lending conversion factor” was around 75 percent (Avraham, Selvaggi,
and Vickery 2012). That is, because large US banks have substantial global assets,

3
The 15 bank holding companies and broker-dealers that received the largest injections in the Troubled
Asset Relief Program (TARP) were Citigroup, Bank of America, JP Morgan Chase, Wells Fargo, Goldman
Sachs, Morgan Stanley, PNC Financial Services Group, U.S. Bancorp, SunTrust, Capital One Financial,
Regions Financial Corporation, Fifth Third Bancorp, BB&T, Bank of New York Mellon, and Key Corp. The
estimates we report in the text do not include capital provided by this program to other, smaller banks.
4
This number is estimated using published accounts and an average risk-weight of 67.5 percent (based
on the New York Fed Quarterly Trends for Consolidated U.S. Banking Organizations). This total includes
the assets of firms that did not themselves receive TARP assistance, but that in the course of 2008 were
acquired by one of the 15 TARP-recipients on which we focus. These acquired firms include Countrywide
Financial, Merrill Lynch, Wachovia, Bear Stearns, Washington Mutual, and National City Corp.
 David Aikman, Jonathan Bridges, Anil Kashyap, and Caspar Siegert 117

Table 3
Countercyclical Capital Buffer Rate (CCyB) That Would Have Been Necessary to
Avoid the Troubled Asset Relief Program

Calculation

Baseline: Replacing bail-outs

Total capital injections $198bn
Total risk-weighted assets (RWAs) $8,409bn
Bailout in percent of RWAs $198bn/$8,409bn 2.4%
Domestic assets in percent of total assets 76%
Required CCyB rate 2.4%/76% 3.1%
Variant 1: Replacing bail-outs and private sector capital raising
Additional private sector capital raising $70bn
Required CCyB rate 3.1% × ($198bn + $70bn)/$198bn 4.2%
Variant 2: Replacing bail-outs, and supporting additional lending
Additional RWAs if credit growth had continued
   along pre-crisis trend $1,050bn
Assumed stressed target capital ratio 10%
Additional capital to support credit growth $1,050bn × 10% $105bn
Required CCyB rate 3.1% + $105bn/($8,409bn × 76%) 4.7%

Source: US Treasury, Published Accounts; New York Fed Quarterly Trends for Consolidated US Banking
Organizations; Financial Accounts of the United States; Avraham et al. (2012).
Note: For variant 2, we assume that banks balance sheets had grown by 7 percent rather than 1 percent
per year over two consecutive years. This is in line with the difference in the commercial bank credit
growth rate between the 20 years before the crisis and the crisis (Q4 2007 to Q4 2009).

an increase in the US countercyclical capital buffer rate will not pass through one-
for-one into their capital requirements. Using these parameters and the calculations
in Table 3, we estimate that a countercyclical capital buffer of 3 percent would have
provided an equivalent level of resilience as the $200 billion TARP injection. Had a
countercyclical capital buffer of 3 percent been built-up in the run-up to the crisis,
it would have, in effect, brought the capital raising that ultimately proved necessary
forward in time, substituting public provision of capital for private sector resources.
The approach above does not account for the capital that banks raised privately
after the results of the stress tests through the Supervisory Capital Assessment
Program were published in May 2009. Banks had six months to raise any required
capital in private markets, with an explicit backstop option to obtain capital from the
US Treasury if necessary. The banks in our sample raised approximately $70 billion
of capital in order to meet these requirements and did not turn to additional
government funding. If this $70 billion is added to the $200 billion of TARP, our
thought experiment suggests a countercyclical capital buffer of around 4.2 percent
would have been required to bring forward the public and private capital raising
that occurred during the heights of the crisis.
While TARP significantly reduced stress in the banking system, it was not
fully sufficient to restart the provision of credit to the economy.5 Thus, a second

5
Chavaz and Rose (forthcoming) show that the effect of TARP on the provision of mortgage credit differed
across regional markets and that TARP recipients reduced mortgage lending in the average county.
118 Journal of Economic Perspectives

s­ ensitivity check is to estimate what size of the countercyclical capital buffer would
have allowed banks to continue lending in line with historical credit growth rates. As
shown by the calculations in Table 3, a countercyclical capital buffer of 4.7 percent
would have ensured that banks would have had sufficient capital to avoid applying
for TARP and to continue growing their balance sheets in line with the long-run
average growth rate.
Given its profitability in the years preceding the crisis, the banking system
had ample capacity to meet increases in the countercyclical capital buffer rate of
this magnitude through a combination of new issuance of equity and additional
retentions. Hirtle (2016) finds that between 2005 and the collapse of Lehman
Brothers in 2008, dividend payments of large bank holding companies amounted to
$162 billion, and total share buy-backs amounted to an additional $131 billion.
Indeed, dividend payments and share buy-backs amounted to $49 billion and
$18 billion, respectively, between mid-2007 and the failure of Lehman in September
2008. By mid-2007, New Century Financial Corporation, a leading subprime mort-
gage lender, had already failed, and Bear Stearns and BNP Paribas had started
halting redemptions on a number of their investment funds.
Finally, we note that among the 15 institutions that we consider are some
broker-dealers. These institutions were not subject to standard prudential require-
ments.6 As a first step, a macroprudential authority would have needed to bring
these firms inside the regulatory perimeter. As illustrated in Table 1, bringing all
US broker-dealers to the same capital standards that commercial banks had in 2007
would already have added a substantial amount of capital to the system.

Tools and Actions to Reduce Funding Mismatches

During the financial crisis, the Federal Reserve set up a number of new liquidity
facilities (Fleming 2012).7 These facilities, which were phased out within a few years
of the end of the crisis, provided around $1.5 trillion of short-term liquidity to the
financial system, an amount equivalent to 9 percent of commercial banks’ and
broker-dealers’ assets. We posit that requiring firms to replace $1.5 trillion of their
short-term funding with longer-term debt would have reduced liquidity outflows in
the crisis in a way that would have avoided a need for extraordinary central bank
liquidity facilities. This is likely to be an overestimate of the scale of appropriate
policy intervention because some public provision of liquidity in a crisis is likely to
be efficient (Holmström and Tirole 1998).

6
In 2004, the Securities and Exchange had created the voluntary “Consolidated Supervised Entities”
program to regulate large investment bank holding companies. However, this regime was primarily
intended to satisfy foreign regulators (Financial Crisis Inquiry Commission 2011). It was generally seen
as being insufficiently robust and was terminated following the failure of Lehman Brothers in 2008.
7
The facilities included the Discount Window Funding, the Term Auction Facility, the Primary Dealer
Credit Facility, the Term Securities Lending Facility, the Term Asset-Backed Securities Loan Facility, the
Commercial Paper Funding Facility, and the Asset-Backed Commercial Paper Money Market Mutual
Fund Liquidity Facility.
 Would Macroprudential Regulation Have Prevented the Last Crisis? 119

As a method of reducing the risks that entail when banks fund long-term,
illiquid assets with short-term funding, the Basel Committee on Banking Supervision
proposed a “net stable funding ratio” standard that took effect in 2018 (for an over-
view see, https://www.bis.org/fsi/fsisummaries/nsfr.pdf). Data provided in Lallour
and Mio (2016) suggest that prior to the crisis, applying the net stable funding ratio
to 12 of the largest US banking and investment banking groups at a consolidated level
would have led to an increase in long-term funding of $1.4 trillion by end-2007.8
What impact might such an intervention have on the real economy? While a
$1.5 trillion increase in the supply of long-term debt would probably affect equilib-
rium yields, we think such an intervention would increase firms’ average funding
costs by less than 10 basis points. Conservative estimates suggest this in turn might
increase lending spreads by less than 20 basis points, reducing the level of GDP by
less than 0.2 percent.9

Tools and Actions to Reduce Build-Up in Household Debt

Higher capital and liquidity requirements might also reduce household debt
growth and house prices by increasing the cost of credit for borrowers. However, the
impact of implementing such measures in a boom may be small (for evidence, see
Bahaj, Bridges, Malherbe, and O’Neill 2016). Thus, a macroprudential regulator
determined to reduce a rapid build-up in household debt might wish to take addi-
tional actions.
Here, we consider the potential impact on the household debt boom of
imposing loan-to-income limits and accompanying affordability criteria on new
mortgages. Adelino, Schoar, and Severino (2016) document the widespread
nature of the household credit boom: mortgage originations—and subsequent
­delinquencies—increased across the income distribution and across credit scores.
By constraining unsustainable borrowing choices across the spectrum, macropru-
dential loan-to-income limits and affordability tests would therefore have helped to
reduce the build-up of household debt vulnerabilities in the run-up to the crisis. We
focus on these potential interventions rather than minimum down payment (“loan-
to-value”) restrictions because the impact of the latter may have been limited by
the twin nature of the household debt and house price booms. As discussed above,
the aggregate loan-to-value ratio on the stock of US housing remained broadly flat

8
The estimates in Lallour and Mio (2016) are based on end-2006 balance sheet data. To make them
comparable to the size of the Fed’s liquidity interventions, we scaled them up to reflect the average
growth of the relevant groups’ balance sheets between 2006 and 2007.
9
The estimate of 10 basis points is based on the conservative assumption of a 100 basis point difference
between the spreads on short-term funding and long-term (five-year) debt. To put this into context, the
average difference between the cost of repo funding and five-year corporate bond spreads in 2006 was
around 70 basis points. The estimated impact on lending spreads is based on the assumption that the
increase in funding costs is fully passed on to borrowers by increasing spreads on loans (which represented
about 50 percent of total assets), and that financial institutions’ cost of equity remains the same despite
the more stable funding base. Given that around 10 percent of banks’ liabilities had to be replaced by
long-term debt, this translates into an increase in lending spreads of 20 basis points. The impact of higher
lending spreads on GDP is estimated based on multipliers in Firestone, Lorenc, and Ranish (2017).
120 Journal of Economic Perspectives

Table 4
Impact of Different Loan-to-Income Limits on Gross Mortgage Lending for
Owner-Occupier House Purchase (First Lien Loans Only)

Loans Number of loans (millions) Loans Impact on value ($ billions) of mortgages

granted impacted by loan-to-income granted originated assuming all impacted loans
(number, limit of: (value, reduced in size:
millions) 2x 3x 4x 5x $ billions) 2x 3x 4x 5x

2003 4.1 2.9 1.3 0.4 0.1 755.8 189.6 59.0 16.1 6.0
2004 4.6 3.2 1.5 0.5 0.1 906.6 245.8 81.9 21.7 6.1
2005 4.8 3.4 1.6 0.5 0.1 1,031.5 288.6 95.8 23.7 6.0
2006 4.2 2.9 1.4 0.4 0.1 939.5 245.9 75.0 18.7 6.0
2007 3.4 2.4 1.2 0.4 0.1 755.9 204.4 67.8 17.7 4.8
Cumulative
total:
2003–2007 21.1 14.8 7.0 2.2 0.6 4,389 1,174 379 98 29

Source: Home Mortgage Disclosure Act (HDMA) data.

Note: The left panel identifies the number of mortgage originations for owner-occupier house purchase
that would have been affected by loan-to-income limits set at the levels labelled. The right panels give the
value reduction in gross lending that would have resulted if all those affected mortgages were reduced in
size such that they just met the listed loan-to-income limit.

in the run-up to the crisis, whereas household debt rose sharply relative to income.
Adelino, Schoar, and Severino (2017) also document the relatively stable distribu-
tion of combined loan-to-value ratios at origination between 2001 and 2007.10
As a simple illustration, consider how loan-to-income limits would have affected
loan-level mortgage originations for owner-occupier house purchases in the run-up
to the crisis.11 For example, a loan-to-income limit of four times income applied
from 2003 to 2007 would have meant that 2.2 million of the 21 million mortgages
originated would have had to be reduced in size (see Table 4). Assuming all of
these affected loans were still originated at the largest size possible given the limit,
this intervention would have left the mortgage stock on the eve of the crisis around
$100 billion or about 1 percent lower than the $10.6 trillion observed.
However, this naive experiment is likely to understate significantly the effect of
loan-to-income limits on the mortgage stock. First, the data sample excludes second
lien or “piggyback” mortgages, whereas well-designed loan-to-income limits would
take into account the combined value of first and additional loans. To the extent that
these additional loans pushed households’ overall indebtedness above the loan-
to-income limit, some would have been curtailed. This could have had a material

10
Mayer, Pence, and Sherlund (2009) document an increase in the median combined loan-to-value
ratio for non-prime purchase loans between 2003 and 2007. We argue below that income and affordability
limits would have been effective in moderating the boom in non-prime mortgage lending.
11
With thanks to Matthieu Chavaz for assistance, we use annual data resulting from the 1975 Home
Mortgage Disclosure Act (HMDA), which covers the vast majority of mortgages originated. From this
dataset, we analyse first lien mortgages for house purchase by owner-occupiers where both loan size and
income is reported.
 David Aikman, Jonathan Bridges, Anil Kashyap, and Caspar Siegert 121

effect: the total flow of second lien mortgage originations from 2003 to 2007 was
over $200 billion. Second, this experiment focuses only on owner-occupier house
purchase loans, with no impact assumed on investor loans or refinancing, both of
which were important features of the household credit boom. Third, our calcula-
tion assumes that all affected borrowers still receive a loan and at the largest size
possible given the limit—in reality, some borrowers would likely be shifted further
below the limit and some might be excluded altogether. As an upper bound of
the impact, if all originations constrained by a loan-to-income limit of four times
income were excluded altogether—rather than just reduced in size—the impact on
gross lending would rise from around $100 billion to $620 billion.
Borrowers at the riskier end of the spectrum that could either not certify their
income or that had particularly stretched affordability characteristics would have
perhaps been the most likely to have been excluded altogether. We can attempt
to quantify these considerations. Between 2003 and mid-2007, about half of non-
prime mortgage originations for house purchase had low or no documentation
of income, assets, or both (Mayer, Pence, and Sherlund 2009). That amounted to
around 1.7 million loans or 8 percent of the total captured in Table 4 over the same
period.12 These loans performed significantly worse than those with full documen-
tation—by 2008 serious delinquencies on low or no documentation subprime loans
had risen to 25 percent (Mayer, Pence, and Sherlund 2009). Many of these loans
would likely have been curtailed by a loan-to-income limit, given the requirement
that some actual documentation of income is provided in order to implement a
­loan-to-income policy. To give a sense of scale, had all of these loans been excluded
by a loan-to-income policy between 2003 and mid-2007, lending could have been
reduced by around $360 billion (next to last line of Table 5).13
As a complement to loan-to-income limits, a macroprudential authority
aiming to enhance the resilience of household sector balance sheets could also
recommend the introduction of affordability tests, which require lenders to assess
borrowers’ capacity to service debts in different circumstances (as discussed in Bank
of England 2017). Evidence from the Survey of Consumer Finances suggests that by
2007, 20 percent of the total stock of mortgagors had debt service burdens of over
40 percent of their income—a situation that might reasonably have been flagged by
an affordability test. The impact of such tests on the rapid growth in ­non-prime mort-
gage borrowing could have been significant. For example, 76 percent of subprime
mortgages for house purchase between 2003 and mid-2007 were short-term hybrid

12
The Mayer, Pence, and Sherlund (2009) sample is based on data from First American Loan Perfor-
mance. These data capture the vast majority of securitized, non-prime (that is subprime or near-prime
“Alt-A”) first lien mortgage originations. The sample covers a total of 9.7 million originations, including
investor loans and refinances. For comparability with Table 4, we focus on the subsample of about
3.4 million loans to owner-occupiers for house purchase. When comparing to Table 4, this subsample
underestimates the total share of non-prime loans, since it does not capture mortgages retained by the
lender rather than securitized.
13
This assumes low- or no-documentation loans had a proportionate share in the total value of origina-
tions in Table 4.
122 Journal of Economic Perspectives

Table 5
Potential Impact of a 4× Loan-to-Income Limit and Accompanying Affordability
Test on Household Debt Boom

Mortgage debt stock

Total mortgage debt stock (2007)a $10,638bn
Gross flow of new mortgages (for owner-occupier house purchase)
Total value of loans granted (2003 to 2007)b $4,389bn
Direct impact of 4× loan-to-income limit (2003 to 2007)b
Lower-bound estimate: all loans still originated at maximum size within limit: - $98bn
Upper-bound estimate: all loans with loan-to-income > 4× excluded altogether: - $622bn
Potential upper-bound impacts on non-prime lending (2003 to 2007)c
If income requirement excluded all low- or no-documentation subprime loans - $359bn
If affordability test excluded all non-prime originations on teaser rates - $366bn

a
Financial accounts of the United States.
b
Home Mortgage Disclosure Act (HDMA) data; gross flow of first lien owner-occupier purchase loans.
Impact estimates do not include potential reduction in second lien loans, investor loans, or loans for
refinance.
c
Number of non-prime first lien owner-occupier purchase loans estimated based on Mayer et al. (2009)
and share in total value of loans assumed to be a proportionate. Exclusions for low- or no-documentation
loans and teaser-rate loans would have overlapped; taken together these borrowers accounted for about
$580bn of lending from 2003 to 2007.

loans with initial “teaser” rates, which were low introductory interest rates that
would last for the first year or two of the mortgage (Mayer, Pence, and Sherlund
2009). By 2008, serious delinquency rates on these loans exceeded 30 percent. An
affordability test that required mortgagors to demonstrate resilience to an interest
rate stress up front would likely have curtailed non-prime lending on teaser rates
significantly. If the test had prevented these loans altogether, it could have reduced
mortgage lending by around $370 billion (last line of Table 5).
Taken together, actions to restrict the borrowing of those that either could not
certify their income or had stretched affordability characteristics would likely have
materially dampened the surge in non-prime lending prior to the crisis. As Table 5
shows, we are left with a wide range of possible effects, but it seems plausible that
combining loan-to-income and affordability rules would have moderated the scale
of the household debt boom in the run-up to the crisis.
The macroeconomic benefit of any limits would incorporate the fact that
they would have targeted the most highly indebted borrowers. Bunn and Rostom
(2015), Andersen, Duus, and Jenson (2016), and Fagereng and Halvorsen (2016)
show a correlation between pre-crisis household leverage and subsequent nega-
tive consumption responses. Although these studies do not demonstrate causality,
they suggest that limiting leverage of the most highly indebted households could
have a stronger aggregate effect on spending in a downturn than reducing debt
uniformly across households. Targeted macroprudential interventions of this kind
could therefore have been particularly effective in dampening the macroeconomic
fallout from the crisis.
 Would Macroprudential Regulation Have Prevented the Last Crisis? 123

Table 6
Survey Data on Usage of Macroprudential Tools

Use of bank-focused tools

(positive countercyclical capital buffer, Use of household-focused tools
forward-looking provisions, caps on (loan-to-income or
credit growth) debt-service-to-income limits) Both

All advanced economies (18) 44% of countries 33%of countries 22% of countries

Advanced economies with 60% of countries 40% of countries 40% of countries

financial stability committee
with formal powers (5)
Other advanced economies (13) 38% of countries 31% of countries 15% of countries

Note: We consider the 19 advanced economies covered in Edge and Liang (2017), minus South Korea,
for which no data on tool usage is available. Numbers are based on country classification in Edge
and Liang (2017) and survey responses on tool usage in IMF (2018a) that consider tools in use at the
date of the survey. “Formal powers” refers to powers to act unilaterally or to issue “comply-or-explain”
recommendations. Results for the United Kingdom have been adjusted to account for measures that had
been agreed but were not yet binding at the date of the survey.

Could the Macroprudential Frameworks Set Up Since the Crisis

Implement Such Policies?

Of the 58 countries surveyed in Edge and Liang (2017) that have created
macroprudential frameworks since the crisis, 41 have set up multi-agency finan-
cial stability committees. Perhaps surprisingly, only 11 of these have formal powers,
including either direct controls over macroprudential policy tools or the right to
issue “comply or explain” recommendations to which other authorities are formally
obliged to respond. The remaining cases rely on the voluntary cooperation of other
regulators to achieve their policy aims.14
There is tentative evidence that financial stability committees with formal
powers are more likely to act than government agencies exposed to short-term
political pressures. Table 6, for instance, presents evidence from a sample of 18
advanced economies that 60 percent of countries with high-powered financial
stability committees have taken bank-focused policy actions; this compares to 38
percent for countries where financial stability policy requires interagency coop-
eration. Similarly, countries with high-powered financial stability committees are
also more likely to have used household-focused macroprudential tools. While the
sample is small and causation is likely to run both ways, this evidence suggests that
institutional frameworks do matter for mitigating biases towards inaction in the
application of macroprudential policies.
If we line up the multi-agency financial stability committees in order of the
powers at their direct disposal, two cases stand out at opposite ends of the spec-
trum. The UK Financial Policy Committee has a wide-ranging toolkit to achieve its

14
Forbes (2018) provides a detailed summary of some of the macroprudential measures that different
countries have put in place after the crisis.
124 Journal of Economic Perspectives

mandate to protect and enhance the resilience of the UK financial system. At the
other end of the spectrum, the US Financial Stability Oversight Council has few
powers under its direct control. In this section, we compare and contrast these polar
cases, and assess what they could do if faced with a rerun of the factors that led to
the last financial crisis.

The US Financial Stability Oversight Council

The Financial Stability Oversight Council was set up in 2010 as part of the Dodd–
Frank Act, and its ten voting members include the heads of existing regulatory agencies,
including the Federal Reserve, Federal Deposit Insurance Corporation, Securities
Exchange Commission, alongside the Secretary of the Treasury and one independent
member with insurance expertise. It also has five non-voting members, including the
Director of the Office of Financial Research, who serve in an advisory capacity. It is
chaired by the Secretary of the Treasury, and most decisions are taken by majority rule.
Its overall mandate is to identify and respond to risks to US financial stability
that could arise from the distress or failure of large, interconnected bank holding
companies or nonbank financial companies. It is also charged with promoting
market discipline by removing expectations that investors will be shielded from
losses by the government. It is designed to facilitate information sharing between
relevant regulatory agencies. Its only binding tool is the power to designate nonbank
financial institutions deemed to be systemically important for enhanced supervi-
sion, a decision that has to be backed by two-thirds of the voting members. It has
no other macroprudential powers. For other policy interventions, the Financial
Stability Oversight Council can only issue recommendations to other regulators,
not all of whom have an explicit financial stability objective (Kohn 2014).
We argue that the Financial Stability Oversight Council in its current form does
not have sufficient powers to ensure financial stability in the face of a credit boom. The
Council’s track record to date supports this pessimistic assessment. First, consider the
efforts to reform money market mutual funds. The runs on the Reserve Primary Fund
in September 2008, followed by large outflows from money market mutual funds in
general, revealed the fundamental vulnerability of these institutions to runs (Squam
Lake Group 2011). Four years later in November 2012, the Financial Stability Over-
sight Council (unanimously) suggested three options for reform: making the shares
in money market mutual funds have a floating rather than fixed value; mandating a
1 percent capital requirement along with a requirement that large withdrawals could
be delayed; or mandating a 3 percent capital requirement that might be combined
with other options. All of these actions would have reduced incentives to run a fund
before it was no longer able to redeem shares at their face value.
Following heavy resistance, including criticisms that money market mutual funds
were outside of the FSOC’s remit, no final recommendation was issued (Cochran,
Freeman, and Clark 2015). The Securities and Exchange Commission, as primary
regulator for money market mutual funds, rejected the idea of capital requirements
and ultimately passed a floating value option (on a 3 to 2 vote) that only applied to a
subset of money market mutual funds. This took effect in October 2016.
 David Aikman, Jonathan Bridges, Anil Kashyap, and Caspar Siegert 125

A second case involves the long-running debate over attempts to reform

US housing finance. There has been bipartisan political support for using
­government-sponsored enterprises like Fannie Mae and Freddie Mac to support the
housing market, with some emphasis on making housing more affordable for lower-
income borrowers (Rajan 2010). However, Alan Greenspan (2005), while Fed Chair
and at the peak of his influence on public policy, repeatedly testified in favor of
restraining the ability of government-sponsored enterprises to purchase private-label
mortgage-backed securities on financial stability grounds, but to no avail. After the
crisis, the Dodd–Frank Act did ban certain types of mortgages, such as interest-only
mortgages or those with negative amortization. But it left the question of minimum
down-payment restrictions to a group of six regulators involved in housing, which
ultimately opted against introducing such a requirement. While risks in the housing
market have significantly declined since the crisis, average loan-to-value ratios on
mortgages are not lower than they were in the early 2000s. Furthermore, no US regu-
lator has the ability to impose loan-to-income requirements, even if the Financial
Stability Oversight Council wished to recommend this action.
Problems associated with lack of power of the Financial Stability Oversight
Council would be mitigated if other authorities had the tools and incentives to
act. For example, maintaining the stability of the financial system and containing
the systemic risk that may arise in financial markets has long been central to the
mission of the Federal Reserve (Kohn 2006). The Fed has a recognized set of powers
including: conducting annual stress tests for large bank holding companies and the
nonbank financial companies it supervises; setting countercyclical capital buffers
for bank holding companies; imposing liquidity requirements on the largest and
most complex financial institutions; and setting minimum margin requirements.
But the Fed’s powers are limited. For example, the Dodd–Frank Act curtailed
the Fed’s ability to provide emergency lending to nonbanks. Moreover, the Fed
lacks authority over many parts of the financial system and has no tools that can
be used to tackle household debt vulnerabilities. A June 2015 “war game” exercise
conducted by four Reserve Bank presidents concluded that the Fed had insufficient
macroprudential powers to address a build-up in risks that resembled the earlier
financial crisis (Adrian, de Fountnouvelle, Yang, and Zlate 2017). Also, Fed officials
have cast doubt on whether its mandate permits it to use monetary policy to act
against a build-up in financial stability risks.

The UK Financial Policy Committee

The UK Financial Policy Committee was established in 2013. It has twelve voting
members: the Governor of the Bank of England, four Deputy Governors, the Execu-
tive Director for Financial Stability, the head of the UK Financial Conduct Authority,
and five independent external members. A representative of the Treasury attends the
meetings as a non-voting member. Most of the Governors of the Bank of England also
sit on the UK’s monetary policy and microprudential policy committees, which facili-
tates policy coordination. The large external membership, and the fact that decisions
are taken primarily by consensus, means that external members have a strong voice.
126 Journal of Economic Perspectives

The Financial Policy Committee is the most muscular macroprudential regu-

lator in the world. It unilaterally sets the countercyclical capital buffer for all banks,
building societies, and large investment firms operating in the United Kingdom,
along with a countercyclical leverage buffer for large banks. Despite more than
70 countries reporting that they have put a countercyclical capital buffer frame-
work in place, the UK Financial Policy Committee is one of only around 10 to
have implemented it at positive rates to date and the only country to have released
it in response to risks crystallizing. The Financial Policy Committee can also vary
risk weights by sector for certain types of risk. It can impose limits on house-
hold borrowing via loan-to-income and loan-to-value restrictions. It advises on
the scenarios used in the annual stress tests of the largest UK banking groups.
It has successfully petitioned the government for additional powers. It can issue
“comply or explain” recommendations to other regulators. In the past, it has made
18 recommendations to other regulators, all of which have been implemented.
Finally, it makes an annual assessment on whether the perimeters of prudential
regulation are drawn appropriately.
These powers are accompanied by a strong accountability framework. All
members of the Financial Policy Committee are personally accountable to Parlia-
ment and typically provide testimony at least once a year. These testimonies follow
the release of a biannual Financial Stability Report, which is increasingly designed
to reach a wide audience to enhance public accountability.
If confronted by a rerun of the events leading to the financial crisis, the Finan-
cial Policy Committee would have the direct power to increase the resilience of the
banking system by raising capital requirements via the countercyclical capital buffer
rate, sectoral capital requirements, and countercyclical leverage buffers. While it
does not have powers to direct changes in banks’ liquidity or funding requirements,
it could issue comply-or-explain recommendations to the microprudential regu-
lator to implement such changes. It seems plausible to us that, faced with evidence
of mounting vulnerabilities throughout the early-to-mid 2000s, the Financial Policy
Committee would have commissioned a stress test of the largest UK banking groups
that would have assumed severe falls in house prices. Such an exercise might not
have uncovered all the channels via which losses eventually transpired (for example,
we doubt it would have been feasible to understand the full extent of losses that
materialised on the seemingly very safe “AAA” tranches of collateralized debt obliga-
tions backed by mortgage securities15). Nevertheless, such tests would have exposed
the fragile solvency and liquidity position of the largest banking groups (including
their broker-dealer subsidiaries) at this point—thus signalling the need for higher
capital and liquidity standards.

15
That said, a paper written prior to the collapse in valuations of the senior AAA tranches of collateral-
ized debt obligations argued that these assets were significantly overvalued given the likely states of the
economy when defaults might occur—that is, given the systematic nature of the risks being borne by
investors (Coval, Jurek, and Stafford 2009).
 Would Macroprudential Regulation Have Prevented the Last Crisis? 127

In addition, a regulator modeled on the Financial Policy Committee could

have used its annual assessment to recommend changes to the regulatory perim-
eter to include other parts of the financial system like stand-alone broker-dealers
that were not part of wider banking groups. Finally, the Financial Policy Committee
could have guarded against vulnerabilities associated with household indebtedness
by limiting the extension of certain mortgages with high loan-to-income ratios.
Indeed, a loan-to-income limit and accompanying affordability test, similar to those
considered above, was put in place by the Financial Policy Committee in 2014.

Conclusion

Could macroprudential policy frameworks have prevented the last crisis?

Perhaps. There would have been challenges in spotting and responding to build-
ups of risk in real time. But our analysis suggests that a macroprudential regime with
a suitably strong mandate, coupled with powers to adjust financial system leverage
and maturity/liquidity transformation and to limit household sector indebtedness,
could have significantly ameliorated the macroeconomic fall-out from the collapse
of the real estate bubble.
Are today’s macroprudential regimes sufficiently well-equipped to do this? The
US Financial Stability Oversight Council is not. The circumscribed structure of the
Financial Stability Oversight Council reflects a political choice to limit the remit of
financial regulation and, notably, to limit its ability to respond to financial sector
developments outside the commercial banking system. While this may be deemed
politically desirable, it would severely restrict the ability of US regulators to prevent
a rerun of the crisis in the future. A macroprudential regulator modeled on the UK
Financial Policy Committee would have the necessary mandate and powers, including
in relation to household indebtedness. But a similar regime in a rerun of the crisis
would still have required political backing to widen the perimeter of regulation to
capture loosely regulated nonbank financial institutions and then to act aggressively.
This raises the important question of how much direct authority a macropru-
dential regulator requires. Many macroprudential regulators must rely on making
nonbinding recommendations to other regulators. The evidence presented above
suggests that one obvious risk of this arrangement is that the recipient of these
recommendations does not share the macroprudential regulator’s objectives, and
no action will be taken. A less obvious risk is that a financial stability committee
that lacks the authority to address risks will be tempted to see risks e­ verywhere—
after all, warning of such risks is costless and a useful way to hedge one’s bets. For
this reason, the warnings of a more powerful financial stability committee might be
more targeted and informative.16

16
We would like to thank Sir Jon Cunliffe for suggesting this point to us.
128 Journal of Economic Perspectives

A related question is how wide the remit of a macroprudential regulator should

be. The evidence summarized above suggests that it may be necessary to take actions
to ensure the resilience of both lender and borrower balance sheets, including by
taking targeted action to prevent build-ups in household debt. While many coun-
tries have implemented policies aimed at preventing excessive levels of household
debt, they remain controversial. Preventing a willing borrower and lender from
consummating a mortgage contract where neither party is likely to default, for fear
of the macroeconomic spillovers such contracts might create, is a more interven-
tionist conception of macroprudential policy than one focused solely on resilience
of the banking and financial system. It is not clear whether such interventions
should be left to technical committees or to democratically elected governments
(Balls, Howat, and Stansbury 2016; Tucker 2018).
A key challenge in making macroprudential policy effective is therefore to give
the relevant financial stability committees clear powers and an appropriately wide
remit, but also to put in place robust governance arrangements that ensure macro-
prudential policymakers are accountable for the way in which they use their tools.

■ We thank Mathieu Chavez, Sir Jon Cunliffe, Beverly Hirtle, Don Kohn, Richard Sharp,
Martin Taylor, Paul Tucker, Skander Van den Heuvel, and Andrei Zlate for helpful
conversations about these issues. Kashyap’s research has been supported by a grant from the
Alfred P. Sloan Foundation to the Macro Financial Modeling (MFM) project at the University
of Chicago and by the Chicago Booth Initiative on Global Markets and Fama-Miller Center.
The views in this paper are those of the authors, and not necessarily those of the Bank of
England or its policy committees.

References

Acharya, Viral V., Thomas Philippon, Matthew Yang, and Andrei Zlate. 2017. “Macroprudential
Richardson, and Nouriel Roubini. 2009. “The Policy: A Case Study from a Tabletop Exercise.”
Financial Crisis of 2007–2009: Causes and Reme- Economic Policy Review 23(1): 1–30.
dies.” Financial Markets, Institutions & Instruments Adrian, Tobias, Michael Fleming, Or Shachar,
18(2): 89–137. and Erik Vogt. 2017. “Market Liquidity after
Adelino, Manuel, Antoinette Schoar, and Felipe the Financial Crisis.” Annual Review of Financial
Severino. 2016. “Loan Originations and Defaults in Economics 9: 43–83.
the Mortgage Crisis: The Role of the Middle Class.” Aikman, David, Jonathan Bridges, Stephen
Review of Financial Studies 29(7): 1635–70. Burgess, Richard Galletly, Iren Levina, Cian
Adelino, Manuel, Anoinette Schoar, and Felipe O’Neill, and Alexandra Varadi. 2018. “Measuring
Severino. 2017. “Dynamics of Housing Debt in the Risks to UK Financial Stability.” Bank of England
Recent Boom and Great Recession.” NBER Macro- Staff Working Paper 738
economics Annual 2017, 32(1): 265–311. Aikman, David, Jonathan Bridges, Anil Kashyap,
Adrian, Tobias, Nina Boyarchenko, and and Caspar Siegert. 2018. “Would Macroprudential
Domenico Giannone. 2016. “Vulnerable Growth.” Regulation Have Prevented the Last Crisis?” Bank
Federal Reserve Bank of New York Staff Report of England Staff Working Paper 747.
794. Andersen, Asger Lau, Charlotte Duus, and
Adrian, Tobias, Patrick de Fontnouvelle, Emily Thais Lækholm Jensen. 2016. “Household Debt
 David Aikman, Jonathan Bridges, Anil Kashyap, and Caspar Siegert 129

and Spending during the Financial Crisis: Evidence American Economic Review 99(3): 628–66.
from Danish Mico Data.” European Economic Review Duffie, Darrell. 2018. Prone to Fail: The Pre-Crisis
89(C): 96–115. Financial System. Forthcoming.
Avraham, Dafna, Patricia Selvaggi, and James Edge, Rochelle M., and Nellie Liang. 2017.
Vickery. 2012. “A Structural View of US Bank “Who is in Charge of Financial Stability, Why, and
Holding Companies.” Federal Reserve Bank of What They Can Do.” Hutchins Center Working
New York Economic Policy Review 7: 65–81. Paper 32. https://www.brookings.edu/research/
Bahaj, Saleem, Jonathan Bridges, Frederic who-is-in-charge-of-financial-stability-why-and-what-
Malherbe, and Cian O’Neill, C. 2016. “What Deter- they-can-do/.
mines How Banks Respond to Changes in Capital Eichner, Matthew J., Donald L. Kohn, and
Requirements?” Bank of England Staff Working Michael Palumbo. 2013. “Financial Statistics for
Paper 593. the United States and the Crisis: What Did They
Balls, Ed, James Howat, and Anna Stansbury. Get Right, What Did They Miss, and How Could
2016. “Central Bank Independence Revisited: They Change?” In Measuring Wealth and Financial
After the Financial Crisis, What Should a Model Intermediation and Their Links to the Real Economy,
Central Bank Look Like?” M-RCBG Associate edited by Charles R. Hulten and Marshall B. Reins-
Working Paper 67. dorf, 39–66. University of Chicago Press.
Bank of England. 2017. Financial Stability Report: Fagereng, Andreas, and Elin Halvorsen. 2016.
June 2017. “Debt and Household Consumption Responses.”
Bassett, William F., Mary Beth Chosak, John C. Norges Bank Staff Memo No. 1.
Driscoll, and Egon Zakrajšek. 2014. “Changes in Financial Crisis Inquiry Commission. 2011. The
Bank Lending Standards and the Macroeconomy.” Financial Crisis Inquiry Report: Final Report of the
Journal of Monetary Economics 62: 23–40. National Commission on the Causes of the Financial
Bernanke, Ben. 2018. The Real Effects of and Economic Crisis in the United States. Submitted
the Financial Crisis. BPEA Conference Drafts, pursuant to Public Law 111-21, January. Available
September 13–14. at: https://www.gpo.gov/fdsys/pkg/GPO-FCIC/
Borio, Claudio. 2003. “Towards a Macropru- content-detail.html.
dential Framework for Financial Supervision Firestone, Simon, Amy Lorenc, and Benjamin
and Regulation?” CESifo Economic Studies 49(2): Ranish. 2017. “An Empirical Economic Assessment
181–215. of the Costs and Benefits of Bank Capital in the
Borio, Claudio, and William R. White. 2004. US.” Finance and Economics Discussion Series
“Whither Monetary and Financial Stability? The Working Paper no. 2017-034.
Implications of Evolving Policy Regimes.” BIS Fleming, Michael J. 2012. “Federal Reserve
Working Paper 147. Liquidity Provision during the Financial Crisis of
Bridges, Jonathan, Chris Jackson, and Daisy 2007–2009. Annual Review of Financial Economics 4:
McGregor. 2017. “Down in the Slumps: The Role 161–77.
of Credit in Five Decades of Recessions.” Bank of Forbes, Kristin. 2018. “Macroprudential Policy
England Staff Working Paper 659. after the Crisis: Forging a Thor’s Hammer for
Brunnermeier, Markus K. 2009. “Deciphering Financial Stability in Iceland.” Report prepared at
the Liquidity and Credit Crunch 2007–2008.” the request of the Task Force to review monetary
Journal of Economic Perspectives 23(1): 77–100. and currency policies for Iceland.
Bunn, Philip, and May Rostom. 2015. “House- Gennaioli, Nicola, and Andrei Shleifer. 2018.
hold Debt and Spending in the United Kingdom.” A Crisis of Beliefs: Investor Psychology and Financial
Bank of England Staff Working Paper 554. Fragility. Princeton University Press.
Chavaz, Matthieu, and Andrew K. Rose.Forth- Gertler Mark, and Simon Gilchrist. 2018. “What
coming. “Political Borders and Bank Lending in Happened: Financial Factors in the Great Reces-
Post-Crisis America.” Review of Finance. sion.” Journal of Economic Perspectives 32(3): 3–30.
Chodorow-Reich, Gabriel. 2014. “The Employ- Greenlaw, David, Jan Hatzius, Anil K Kashyap,
ment Effects of Credit Market Disruptions: and Hyun Song Shin. 2008. “Leveraged Losses:
Firm-Level Evidence from the 2008–9 Financial Lessons from the Mortgage Market Meltdown. In
Crisis.” Quarterly Journal of Economics 129(1): 1–59. Proceedings of the US Monetary Policy Forum, pp. 7–59.
Cochran, Martha L., David F. Freeman, and Greenspan, Alan. 2005. “Regulatory Reform
Helen Mayer Clark. 2015. “Money Market Fund of the Government-Sponsored Enterprises.” Testi-
Reform: SEC Rulemaking in the FSOC Era.” mony before the Committee on Banking, Housing,
Columbia Business Law Review 861. and Urban Affairs, U.S. Senate. Speech 63, Board
Coval, Joshua D., Jakub W. Jurek, and Erik of Governors of the Federal Reserve System, US.
Stafford. 2009. “Economic Catastrophe Bonds.” April 6.
130 Journal of Economic Perspectives

Guerrieri, Luca, Matteo Iacoviello, Francisco 307–62.

Covas, John C. Driscoll, J., Michael T. Kiley, Kohn, Donald L. 2006. “The Evolving Nature
Mohammad Jahan-Parvar, Albert Queralto Olive, of the Financial System: Financial Crises and the
and Jae W. Sim. 2015. “Macroeconomic Effects of Role of the Central Bank.” Speech given at the
Banking Sector Losses across Structural Models.” Conference on New Directions for Understanding
Finance and Economics Discussion Series Working Systemic Risk, New York, New York. May 18.
Paper 2015-044. Kohn, Donald L. 2014. “Institutions for Macro-
Hanson, Samuel G., Anil K Kashyap, and Jeremy prudential Regulation: The UK and the U.S.”
C. Stein. 2011. “A Macroprudential Approach to Speech at the Kennedy School of Government,
Financial Regulation.” Journal of Economic Perspec- Harvard University. April 17.
tives 25(1): 3–28. Lallour, Antoine, and Hitoshi Mio. 2016. “Do
Hirtle, Beverly. 2016. “Bank Holding Company We Need a Stable Funding Ratio? Banks’ Funding
Dividends and Repurchases during the Financial in the Global Financial Crisis.” Bank of England
Crisis.” Federal Reserve Bank of New York Staff Staff Working Paper 602.
Report 666 Martin, Christopher, Manju Puri, and Alex-
Hirtle, Beverly, Anna Kovner, James Vickery, ander Ufier. 2018. “Deposit Inflows and Outflows
and Meru Bhanot. 2016. “Assessing Financial in Failing Banks: The Role of Deposit Insurance.”
Stability: The Capital and Loss Assessment under NBER Working Paper 24589.
Stress Scenarios (CLASS) Model.” Journal of Mayer, Christopher, Karen Pence, and Shane M.
Banking & Finance 69(Supplement 1): S35–S55. Sherlund. 2009. “The Rise in Mortgage Defaults.”
Holmquist, Elizabeth, and Josh Gallin. 2014. Journal of Economic Perspectives 23(1): 27–50.
“Repurchase Agreements in the Financial Accounts Mian, Atif, and Amir Sufi. 2011. “House Prices,
of the United States.” FEDS Notes, Board of Gover- Home Equity-Based Borrowing, and the US House-
nors of the Federal Reserve System, June 30. hold Leverage Crisis.” American Economic Review
Homlström, Bengt, and Jean Tirole. 1998. 101(5): 2132–56.
“Private and Public Supply of Liquidity.” Journal of Mian, Atif, and Amir Sufi. 2012. “What Explains
Political Economy 106(1): 1–40. High Unemployment? The Aggregate Demand
International Monetary Fund (IMF). 2005. Channel.” NBER Working Paper 17830.
Global Financial Stability Report, April 2005: Market Mian, Atif, Amir Sufi, and Emil Verner. 2017.
Developments and Issues. “Household Debt and Business Cycles Worldwide.”
International Monetary Fund (IMF). 2018a. Quarterly Journal of Economics 132(4): 1755–1817.
The IMF’s Annual Macroprudential Policy Survey— Rajan, Raghuram G. 2010. “Fault Lines:
Objectives, Design, and Country Responses. How Hidden Fractures Still Threaten the World
International Monetary Fund (IMF). 2018b. Economy.” Princeton University Press.
Global Financial Stability Report, April 2018: A Bumpy Rosengren, Eric S. 2014. “Broker-Dealer
Road Ahead. Finance and Financial Stability.” Speech at the
Jordà, Òscar, Moritz Schularick, and Alan M. Conference on the Risks of Wholesale Funding in
Taylor. 2013. “When Credit Bites Back.” Journal of New York, New York. August 13, 2014.
Money, Credit and Banking 45(S2): 3–28. Squam Lake Group. 2011. “Reforming Money
Jordà, Òscar, Moritz Schularick, and Alan M. Market Funds: A Proposal by the Squam Lake
Taylor. 2016. “The Great Mortgaging: Housing Group.”
Finance, Crises and Business Cycles.” Economic Tarullo, Daniel K. 2014. “Macroprudential
Policy 31(85): 107–152. Regulation.” Yale Journal on Regulation 31(3):
Keys, Benjamin J., Tanmoy Mukherjee, Amit 505.
Seru, and Vikrant Vig. 2010. “Did Securitization Tucker, Paul. 2018. Unelected Power: The Quest
Lead to Lax Screening? Evidence from Subprime for Legitimacy in Central Banking and the Regulatory
Loans.” Quarterly Journal of Economics 125(1): State. Princeton University Press.