The Interaction Between Domestic Monetary Policy and Macroprudential Policy in Israel
The Interaction Between Domestic Monetary Policy and Macroprudential Policy in Israel
The Interaction Between Domestic Monetary Policy and Macroprudential Policy in Israel
Economic Modelling
journal homepage: www.journals.elsevier.com/economic-modelling
A R T I C L E I N F O A B S T R A C T
JEL classification: The global financial crisis (GFC) triggered the use of macroprudential policies imposed on the banking sector.
E51 Using bank-level panel data for Israel for the period 2004–2019, we find that domestic macroprudential measures
E52 changed the composition of bank credit growth but did not affect the total credit growth rate. Specifically, we
E58
show that macroprudential measures targeted at the housing sector moderated housing credit growth but tended
G01
G21
to increase business credit growth. We also find that accommodative monetary policy surprises tended to increase
G28 bank credit growth before the GFC. We show that accommodative monetary policy surprises increased consumer
Keywords:
credit when interacting with macroprudential policies targeting the housing market. Accommodative monetary
Financial stability policy interacted with nonhousing macroprudential measures to increase total credit.
Policy evaluation
Banking sector
Credit markets
Regulation
Global financial crisis
1. Introduction effect of these measures on total bank credit was weakened due to sub-
stitution effects. We show that the MaP measures targeted at housing
House prices in Israel have more than doubled since 2008, alongside a credit reduced its growth rate, but tended to increase the growth rate of
substantial increase in housing credit (mortgages). This development, business credit. MaP measures targeted at reducing banks’ credit con-
which may pose a significant risk to the financial sector and particularly centration did not significantly affect consumer or business credit, but
to banks, together with an understanding of the importance of the contributed to the growth of housing credit. We find that other general
financial sector stability as a whole, triggered the imposition of macro- MaP measures (capital requirements from the banks) that were expected
prudential (MaP) measures on the banks. This paper analyzes the effects to slow the growth of business credit influenced business credit with a
of MaP policy and domestic monetary policy on domestic bank credit in significant lag, apparently due to the time frame given to the banks to
Israel between 2004 and 2019. We chose to investigate the effects of MaP complete the adjustment of capital and risk assets. While the interaction
policy on housing, consumer (nonhousing), and business sector credit between accommodative monetary policy and MaP measures targeting
separately in order to understand the direct effect of the MaP measures the housing market did not have any additional effect on housing credit,
on the market at which they were targeted and possible effects on the we find weak support for its expansionary effect on consumer credit.
other bank credit sectors and on total credit. Overall, MaP measures did not change the development of total credit
We find that alongside the desired direct, moderating effect of the significantly, but they did change its composition. Monetary policy was
MaP measures on the credit markets at which they were targeted, the generally found to be effective, with less accommodative policy slowing
☆
Any views expressed in the paper are those of the authors and do not necessarily reflect those of the Bank of Israel. The authors are grateful to Sushanta Mallick
(Editor), the associate editor, anonymous referees, Steven Laufer, and participants at the Bank of Israel Research Department seminar and the International Banking
Research Network (IBRN) Meeting, Federal Reserve Bank of New York (2019), for their helpful comments. This research was initiated as part of the 2018 IBRN project
examining the interaction between macroprudential policy and monetary policy.
* Corresponding authors.
E-mail addresses: [email protected] (I. Gamrasni), [email protected] (S. Ribon).
https://doi.org/10.1016/j.econmod.2022.105872
Received 14 June 2021; Received in revised form 7 April 2022; Accepted 16 April 2022
Available online 20 April 2022
0264-9993/© 2022 Elsevier B.V. All rights reserved.
J. Benchimol et al. Economic Modelling 112 (2022) 105872
housing and consumer credit growth only before the global financial measures that reinforce monetary policy (i.e., in the same direction of
crisis (GFC). In addition, we find that higher foreign policy rates tended influence) are more effective.
to reduce housing and consumer credit growth, similar to the impact of Everett et al. (2021) contributed to the literature with Dutch and Irish
domestic policy, suggesting their ability to indicate future domestic data.1 They employed a similar methodology to the one we adopt,
monetary policy. examining the effect of monetary policy and MaP policy and their
MaP policy is a regulatory tool that seeks to ensure the financial interaction on domestic mortgage lending. As both Ireland and the
system’s stability, which is necessary for stable economic growth. The Netherlands are small economies in the euro system, monetary policy
overall objective of MaP policy is to limit the buildup of financial vul- shocks may be regarded as exogenous to these two countries. Like Israel,
nerabilities in order to reduce the financial system’s sensitivity to shocks. the banking sector in Ireland and the Netherlands is relatively concen-
MaP policy targets a specific credit sector, such as the consumer, busi- trated and both have employed various MaP measures in the past decade
ness, or housing credit sector, to achieve a particular objective. Specif- against the background of rapid increases in house prices. Everett et al.
ically, MaP measures targeting housing credit are intended to limit (2021) complemented their analysis by using confidential bank-level
banks’ exposure to mortgage defaults. Consequently, most of these data for domestic lending. They found that positive (less accommoda-
measures are expected to increase the mortgage cost, moderating hous- tive) domestic monetary policy shocks reduced mortgage lending growth
ing credit growth. A similar mechanism applies to capital adequacy or in both countries, while prudential regulations weakened this effect in
banking concentration MaP measures. While MaP measures tend to Ireland but not in the Netherlands. They found only weak evidence for an
decrease the growth rate of the targeted credit, substitution effects may international lending channel: a foreign (US, UK) monetary policy shock,
be reflected in the growth of other credit sectors. For instance, MaP with or without interaction with the MaP measures, was found to have no
measures targeting housing credit may increase consumer credit as the significant effect on mortgage lending for Ireland, while for the
latter may serve as an alternative to the former, whether for banks as Netherlands, weak evidence was found that tightening monetary policy
suppliers of credit or for borrowers as demanders of credit. Alternatively, in the UK reduces mortgage lending.
consumer credit may decline if MaP measures decrease the demand for Cao et al. (2021) studied the interaction between MaP policy and
housing, thus reducing the demand for housing-related consumption and monetary policy in Norway and Sweden and found evidence that
consumer credit. Another example relates to MaP measures that limit aggregate MaP policy depresses aggregate lending, although different
credit to large companies. The supply of this credit may move to smaller types of MaP measures do not lead to significantly different effects.
businesses but can also be routed to housing or consumer credit. Bussiere et al. (2021) summarized the main findings of seven papers that
The use of MaP measures has expanded since the GFC (Cerutti et al., related mainly to MaP policies targeted at capital flows and the possible
2017). The theoretical literature based on dynamic stochastic general spillovers from monetary policies in core economies to recipient econo-
equilibrium (DSGE) models supports the view that MaP measures can mies. The findings show that MaP policies in recipient countries can
dampen credit cycles. Using a DSGE model with a housing sector, Kannan partly offset monetary policy spillover effects from the core (large)
et al. (2012) demonstrated that monetary and macroeconomic policies economies. Bussiere et al. (2021) also remarked that the impact differs
can help stabilize the economy, given the shocks that hit the economy. considerably across MaP policy instruments, suggesting the importance
When financial or housing demand shocks drive a credit and housing of a detailed analysis.
boom, MaP measures can temper them and improve welfare. Neverthe- In the absence of any interest rate adjustment, a credit-restricting
less, DSGE models offer limited empirical guidance on the influence of MaP shock could reduce the debt to GDP ratio in the short term, but it
the interaction between MaP measures and monetary policy on credit is unlikely to reduce the ratio of firms’ debt to their internal funds.
cycles. We aim to assess these empirical linkages. Repullo and Suarez Greenwood-Nimmo and Tarassow (2016) suggested that MaP policy
(2013) showed that the credit (financial) and business cycles are not alone is likely to have ambiguous effects on financial stability. Never-
necessarily synchronized. Therefore, monetary policy alone cannot theless, MaP policy is rarely used alone. When monetary policy adjusts in
smooth both cycles. Consequently, MaP policy plays a more significant response to an MaP shock, both ratios decrease substantially, indicating
role in moderating the financial cycle according to their analysis, thus, an increase in financial stability. According to this finding, a combination
promoting financial stability, while monetary policy remains responsible of monetary and MaP measures for achieving financial stability may be
for price and output stability. desirable, as Yellen (2014) advocated. Arena et al. (2020) highlighted the
The effects of the interaction between MaP and monetary policies is possibility that MaP policies can limit riskier mortgages and contribute to
still unsettled (Dell’Ariccia et al., 2012; IMF, 2013; Aiyar et al., 2014). De stronger bank balance sheets, countering the accommodative monetary
Marco et al. (2021) found that stricter monetary policy and MaP mea- policy stance. Following this line of thought, we analyze the interaction
sures reinforced each other for the UK, but only for small banks. Simi- between MaP and monetary policies and their effect on credit growth.
larly, Forbes et al. (2017) found that monetary policy can amplify the Following Altavilla et al. (2020), we show that monetary policy and
impact of regulatory measures. By contrast, De Jonghe et al. (2020) MaP measures are mutually reinforcing in relation to consumer loans
found a trade-off between monetary policy and supervisory capital re- more than to mortgage loans. This result may be related to the
quirements in the case of Belgium: the higher the banks’ capital re- risk-weighted ratio of capital to assets. Part of the banks’ income can be
quirements are, the weaker the effect of monetary accommodation on used to rebuild capital and increase headroom if the latter is below its
credit supply is. Cerutti et al. (2017) found that MaP measures have less normal level, and margins on consumer loans can be raised to speed up
of an effect on credit growth in advanced economies that tend to have the process (Davis et al., 2019).
alternative sources of nonbank credit and enable borrowers to obtain As a small open economy, more than half of Israel’s trades are related to
funds from abroad in more open economies. We seek to understand how the US and the Eurozone. Imported inflation and capital flows from these
MaP policy affects each component of bank credit, and whether the places directly influence domestic monetary policy and thus credit ag-
interaction with monetary policy has an additional effect on these credit gregates. Our findings that higher foreign policy rates moderate housing
sectors. and consumer credit growth while tending to increase business sector
Gambacorta and Murcia (2020) examined the effect of domestic MaP credit growth complement the findings of Niepmann et al. (2020).
measures on domestic credit through an empirical exercise similar to Although foreign credit is not significant in Israel, our results indicate the
ours. Unlike in our estimation, they used granular information on bank importance of foreign monetary policy surprises for local credit aggregates.
loans at the firm level from eight different countries. They found that
MaP measures are successful in easing credit cycles and reducing banking
sector risk. They also found that bank-specific characteristics influence 1
Everett et al. (2021), Cao et al. (2021), and Bussiere et al. (2021) were all
the impact of MaP measures on credit. Finally, they found that MaP part of the 2018 International Banking Research Network (IBRN) project.
2
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Fig. 1. Annual GDP growth in Israel and advanced and emerging economies.
Notes: Annual GDP growth in percent. Sources: Bloomberg and Central Bureau Fig. 2. House starts and house prices in Israel.
of Statistics. Notes: The left axis relates to the housing starts in units, and the right axis relates
to nominal house prices, which are expressed in annual percentage changes.
Source: Central Bureau of Statistics.
The remainder of the paper is organized as follows. Section 2 briefly
presents the economic background and the Israeli environment. Section 3
describes the data used in Section 4 for the estimations and results.
Section 5 presents robustness checks and Section 6 concludes. The Ap-
pendix presents additional results.
3
J. Benchimol et al. Economic Modelling 112 (2022) 105872
4
In addition, the Bank of Israel purchased a pre-announced amount annually
3
Leumi, Hapoalim, Discount, Mizrahi-Tefahot, First International, Union as part of a program intended to offset the effects of natural gas production on
Bank, and Bank of Jerusalem. the exchange rate. This program ceased at the beginning of 2019.
4
J. Benchimol et al. Economic Modelling 112 (2022) 105872
5
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 1
Macroprudential measures.
Stability Measures Decision Date Application Date Influence Date Type
1 The directive regarding limitations on the indebtedness of a borrower and a group of borrowers was 20/08/2003 31/03/2004 20/08/2003 CON
amended.
2 A 100% capital surcharge on groups of borrowers who buy properties collectively. 25/03/2010 01/04/2010 25/03/2010 HSE
3 Banking corporations shall make a supplemental provision at the rate of at least 0.75 percent on account 11/07/2010 01/07/2010 11/07/2010 HSE
of outstanding housing loans that were issued on or after July 1, 2010, and in which the existing ratio in
each case between the debt (prorated to the bank’s share of the mortgage) and the value of the
mortgaged property on the date of loan execution exceeds 60 percent.
4 Risk weights increased on housing loans with LTVs above 60%, a floating component of more than 25%, 28/10/2010 26/10/2010 26/10/2010 HSE
and a mortgage value greater than NIS 800,000. A minimum risk weight of 100% is applied to these
mortgages (banks can lower it to 75%). This is a change from the previous 35–75% range.
5 The core capital ratio target shall be set to a rate no lower than 7.5 percent. 30/06/2010 31/12/2010 30/06/2010 GEN
6 Limiting variable interest to a maximum of one-third of the mortgage loan. 03/05/2011 05/05/2011 05/05/2011 HSE
7 Limiting LTV ratio in housing loans: up to 75% for first-time homebuyers, up to 50% for investors, and 01/11/2012 01/11/2012 01/11/2012 HSE
up to 70% for those upgrading their homes.
8 Risk weights on some housing loans are increased, depending on their LTV. Loans with LTVs between 21/03/2013 01/01/2013 01/01/2013 HSE
45% and 60% have a higher risk weight of 50%, and those with LTVs between 60% and 75% are
weighted at 75%.
9 A bank will not issue a housing loan with a PTI of more than 50%. 29/08/2013 01/09/2013 01/09/2013 HSE
10 Risk weight of 100% imposed on mortgages with payment to income (PTI) between 40% and 50% 29/08/2013 01/09/2013 01/09/2013 HSE
11 The maximum variable rate portion of a mortgage loan cannot exceed 2/3, and the maximum portion of 29/08/2013 01/09/2013 01/09/2013 HSE
a variable rate that can change within 5 years from the date of approval cannot exceed 1/3.
12 A loan shall not be approved or granted if the term to final repayment exceeds 30 years. 29/08/2013 01/09/2013 01/09/2013 HSE
13 A. Banking corporations should increase their Common Equity Tier I Capital target by a rate that 28/09/2014 01/01/2015 28/09/2014 HSE
represents 1 percent of their outstanding housing loans.
B. Banking corporations may reduce the risk weight attributed to variable interest leveraged loans from
100% to 75%.
As the first decision was more influential than the second, ultimately the decisions were declared
to be a stringency.
14 Basel III - Small & Large Banks: Tier 1 Capital target 28/03/2012 01/01/2015 28/03/2012 GEN
15 Basel III - Small & Large Banks: Total Capital Target 28/03/2012 01/01/2015 28/03/2012 GEN
16 Limitations on borrowers’/groups of borrowers’ indebtedness: (1) the indebtedness of a borrower other 09/06/2015 01/01/2016 09/06/2015 CON
than a bank to a banking corporation shall not exceed 15% of the banking corporation’s capital; (2) the
indebtedness of a group of borrowers to a banking corporation shall not exceed 25% of the banking
corporation’s capital; (3) the indebtedness of a banking group of borrowers to a banking corporation
shall not exceed 15% of the banking corporation’s capital; (4) the indebtedness of a controlled group of
borrowers to a banking corporation shall not exceed 50% of the banking corporation’s capital; (5) the
total indebtedness of all “borrowers,” “groups of borrowers,” and “banking groups of borrowers,” whose
net indebtedness to the banking corporation exceeds 10% of the banking corporation’s capital, shall not
exceed 120% of the banking corporation’s capital.
17 Basel III - Large Banks 28/03/2012 01/01/2017 01/01/2015 GEN
Notes: Housing combines two categories: Real Estate Credit (REC) and LTV limits. CON stands for concentration limits; HSE stands for housing macroprudential
regulation; GEN stands for general capital requirements. Application Date is the date of the MaP measure’s application, Decision Date is the communication’s date, and
Influence Date is the date since the MaP measures were assumed to influence the credit aggregates.
Moreover, an additional leverage ratio surcharge also helps to control the Department on a quarterly basis in accordance with Banking Supervision
banking system’s leverage. It may influence the banks’ capital and, as a Directive 831.
result, credit supply. In our analysis, we refer to liquidity and capital tools The data include credit to the public,7 deposits, and other assets and
together as bank stability tools. liabilities, excluding bonds and securities that were borrowed or pur-
chased under repurchase agreements. The credit data is disaggregated to
2.5.3. Concentration measures commercial credit, which we refer to as business credit,8 housing credit,
Our analysis distinguishes between MaP measures targeting lending and other credit to households (consumer credit).
concentration, i.e., measures limiting credit to big borrowers, and gen- Other than the credit data, the source for the rest of the banking
eral capital requirement measures, since each of the two groups may system’s data is the publicly available quarterly reports of the banks,
affect banks’ credit differently. As a result of the big borrowers’ domi- describing the level of liquid assets, liabilities to assets, and deposits to
nance in the business sector, a MaP directive limiting the indebtedness of assets. We include cash, deposits, and securities in the liquid assets
borrowers or groups of borrowers was announced in 2003. This MaP category, provided that they are not encumbered to the lenders.
measure limited the banks’ capital exposures to big borrowers, thus Our panel includes the seven largest banks in the domestic banking
changing the banking sector supply for credit distribution by increasing sector, which account for about 99 percent of the sector (see Section 2.3).
the proportions of small and medium businesses and households. In
2015, more restrictive measures for big borrowers and groups of bor-
rowers were imposed.
7
During the first quarter of 2011, changes were made in the Banking Su-
3. The data
pervision Department directives. This created a break in the banking system’s
credit data. In order to deal with this discontinuity of the series, we included a
3.1. Bank credit dummy variable in the estimation to represent the reporting change.
8
Including agriculture, industry, electricity, water, construction and real es-
The bank credit data are taken from the domestic banks’ confidential tate, commerce, transportation, information and telecommunications, and other
reports to the Bank of Israel, submitted to the Banking Supervision business credit, excluding credit for financial services.
6
J. Benchimol et al. Economic Modelling 112 (2022) 105872
10
Our classification of all MaP measures is based on the IBRN international
9
Until April 2017, interest rate decisions were made by the Bank of Israel 12 dataset of MaP measures.
11
times a year, toward the end of each month. Since April 2017, the Committee Our robustness checks aggregate the general and concentration groups dis-
has decided on the interest rate eight times a year. cussed in Section 5.
7
J. Benchimol et al. Economic Modelling 112 (2022) 105872
types include only three MaP measures that aimed to address specific
reapproval. For the estimation of the business credit sector, we lag the
issues and were not supposed to affect the credit sector we investigated.12
accumulated measures by four quarters, rather than by one quarter as in
General MaP measures were aimed at increasing the bank’s sound-
the other sectors. This is because while shifting credit from the business
ness, which can affect all bank credit sectors or a specific sector.
sector to the other sectors can lead this process, the process of reducing
Reducing the growth rate of business credit may be the preferred reaction
the credit of the business sector can be done only after the reapproval of a
to the general MaP measures, due to their high weight as risk assets. The
credit portfolio under the new concentration restrictions.
concentration MaP measures are aimed at reducing the share of big
Table 2 reveals that, with the exception of one MaP measure, all of the
borrowers in each bank’s assets, i.e., reducing the bank’s business credit
MaP measures were implemented after the GFC. Therefore, there is no
volume.
need to distinguish in the estimations between pre- and post-GFC when
In the estimation of each of the types of credit—housing, consumer,
relating to the MaP measures or their interaction with monetary policy. It
business, and total credit—we include the three groups of MaP measures.
is also evident that the use of housing MaP measures is more intensive
We expect measures that refer to the studied credit sector to moderate its
than other measures due to the rapid increases in house prices during
growth rate, while MaP measures that refer to other types of credit,
these years (Figs. 3 and 4).
which are substitutes for the banks as suppliers of credit, or for the
borrowers as demanders of credit, may work to increase the growth rate
of the credit studied. 3.5. Other macroeconomic variables
Each of the MaP categories is assigned the value of 1 in periods in
which at least one MaP measure in the category became effective (ac- In order to take into account the effect of the business cycle and
cording to the column “Influence Date” in Table 1). We allow different housing market activity on credit aggregates, we add control variables
time spans for MaP measures to affect the market. For housing MaP such as unemployment rate, change in real wages, business sector GDP
measures, we include the accumulation of measures since the beginning growth, and real change in house prices. A detailed description of the
of the studied period as the indicator variable, lagged one period, as we macroeconomic variables in each specification is presented below in
assume that the implemented measures continue to affect the market as Section 4.
long as they are in place and reflect the MaP policy stance. General MaP
measures are included in the estimation accumulated over eight quarters, 3.6. Sample period
lagged one-quarter relative to the date they became effective. We use this
accumulated period since after the threshold for capital adequacy is The estimation is carried out as a weighted cross-section panel esti-
achieved (plus some buffer), the effect of the restriction on the credit mation, correcting for cross-section heteroscedasticity and contempora-
growth rate ceases. Eight quarters is a period that covers the maximum neous correlation. The estimation is performed for seven banks over 63
number of periods between the time a general MaP measure began periods (2004:Q2–2019:Q4) for a total of 441 observations. The depen-
influencing banks’ behavior and the time it had to be applied (See dent variable is the log difference of the real quarterly volume of credit.
Table 1). All the thresholds were achieved by the banks before the due Due to the substantial impact of the GFC on policy and the market, we
date, and so accumulating over a longer period is unnecessary. Once distinguish monetary policy surprises before and after the GFC, starting
banks achieve the recommended capital adequacy, the MaP policy stance from the third quarter of 2008.
toward tightening banking risks is fulfilled.
The concentration measures were accumulated over four quarters, 4. Estimation and results
lagged one period. We accumulate four quarters backward since the
volume of business credit that a bank extends, on which these measures As described in Section 2, the banking sector in Israel is characterized
are intended to operate, is approved and validated once a year. Imple- by a small number of domestic banks for which lending activity is
mentation of the concentration restrictions may therefore be fulfilled concentrated in the domestic market, and its sources are domestic as
over one year after the credit portfolio of all businesses come up for well. That is, the lending activity abroad of the domestic banks and the
activities of foreign banks in the domestic market are marginal. Natu-
rally, those qualities lead us to focus on estimates of the influence of the
12
domestic monetary policy and domestic MaP measures, and the
In 2011, the Bank of Israel changed definitions related to interbank loans,
which we exclude from our data (we use credit for the nonfinancial sector only),
and raised the reserve requirement imposed on banking corporations for FX
derivative transactions by nonresidents (we use local credit only).
8
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 2 as business credit), either due to a shift in the demand for this credit due
Macroprudential measures by category and quarter imposed. to limitations on borrowers, or due to supply-side considerations on the
Concentration General Housing part of the banks, so that the effect of these measures on total bank credit
may have been small or insignificant. The same may hold for the general
Sep-03 1
Mar-10 1 and concentration MaP measures: they may have induced a possible in-
Jun-10 1 direct substitution effect alongside the direct moderating effect on
Sep-10 1 business, and possibly consumer and housing, credit. In order to analyze
Dec-10 1 the direct and indirect effects of the three classes of MaP measures, we
Jun-11 1
Jun-12 1
choose to include all three in the estimation of each of the separate credit
Dec-12 1 components. In addition, we estimate their effect on the growth rate of
Mar-13 1 total domestic bank credit, which comprises housing, consumer, and
Sep-13 1 business credit.
Sep-14 1
As mentioned above, we include in the estimations each of the three
Mar-15 1*
Jun-15 1 MaP housing, concentration, and general measures accumulated over
different lengths of periods. While we find that the effect of housing MaP
Notes: *Large banks only.
measures on the mortgage sector is best exhibited when considering the
possible effect of accumulated measures in the past on this market, we
find that the general and concentration MaP measures affect credit dy-
interaction between them, on the domestic banks’ lending. Because Israel namics when we consider the accumulation of the MaP measures over a
is a small open economy, we complement the specification with mone- shorter period into account (Fig. 9).
tary policy surprises from the US,13 as described in Section 3.2, in order The interaction between MaP measures and monetary policy is
to allow us to test the effect of the interest rate differential between specified as the product of the MaP measures index of dummies and the
domestic and foreign financing on the growth of domestic credit. lagged one- and two-period monetary policy. Foreign monetary policy is
Our first objective is to assess the effect of domestic MaP poli- included in the estimations with one and two lags.
cies—including their interaction with monetary policy—on the housing For each of the credit types, we present four specifications. Column 1
credit (mortgage) market, which is one of the components of the do- of each of the tables includes the basic specification. Column 3 presents
mestic banks’ total lending activity. the specification including global monetary policy surprises, and Col-
The second component of bank credit is lending to the business sector umns 2 and 4 add the interaction terms between MaP measures and
(excluding the financial sector). We assess the effect of domestic MaP monetary policy to these specifications. In the specifications shown in
policies and of the interactions of these policies with monetary policy on Column 4 of Tables 3–6, we represent global monetary policy by
business sector lending. including monetary policy shocks in the US, as most of the foreign cur-
Nonhousing consumer credit is the third component of bank credit. rency credit is denominated in dollars. We present an alternative speci-
This credit is not backed by the borrower’s housing, and is usually used fication including eurozone monetary policy surprises in the Appendix.
for consumption, but may also serve—for the banks as credit suppliers, Our qualitative results remain robust to this alternative specification.
and for households as credit demanders—as a substitute for mortgages. In the following estimation results, the significance of the sum of the
As a final step, we assess the effect of monetary and MaP policies on total monetary policy lags coefficients and the interaction lags coefficients is
bank credit, which is comprised of the above three components. determined using the Wald test.
The general formulation of our estimation is
4.1. Housing credit estimation results
ΔCrediti;b;t ¼ αi þ β1i MaPt þ β2i MonPolt þ β3i MaPt MonPolt
þ β4i ForeignMonPolt þ β5i Macroi;t þ β6i BankChari;b;t þ εi;b;t Table 3 shows that MaP measures relating to the housing market
succeeded in reducing the growth rate of housing credit according to all
where t spans from 2004:Q2 to 2019:Q4, and b stands for the bank. specifications. General MaP measures contributed to reducing the growth
ΔCrediti;b;t is the log change of domestic bank credit of type i (housing, rate of housing credit, although the results were significant only in the
consumer, business, or total credit) by bank b in quarter t. MaPt are specifications excluding the interaction terms.
indices of MaP measures and MonPolt and ForeignMonPolt are the unex- We also find that MaP measures targeting the concentration of the
pected change in the Bank of Israel and foreign interest rates. The esti- banks’ activity increased housing credit growth (the results are signifi-
mated parameters α and β, macroeconomic control variables (Macroi,t), cant only in the full specification, Column 4). This demonstrates a sub-
and bank-specific control variables (BankChari,b,t) vary by type of credit i. stitution effect on business credit from the supply side of the banks,
We will elaborate on the specific variables and lags included in the es- increasing mortgage lending as a substitute for business lending, partic-
timations in the sections describing the results. εi;b;t are cross-section- ularly to large borrowers. We find that positive monetary policy surprises
weighted and allow for conditional cross-section correlation between (unexpected tightening) significantly decreased the growth rate of
the contemporaneous residuals but are restricted to be uncorrelated be- housing credit, but turned insignificant in the period after the financial
tween different periods. The estimation results are presented in crisis, starting in the third quarter of 2008 (Table 3). For some of the
Tables 3–6. specifications, the interaction between the general or concentration MaP
We choose to separately analyze the components of local bank cred- measures and monetary policy amplifies the substitution effect on
it—housing, business, and consumer credit—in order to study the effect mortgage credit.
of each of the three classes of MaP measures on each credit type. While As Israel is a small open economy, we look for the effect of global
measures targeting housing credit were meant to moderate its growth in interest rates, represented by US monetary policy. We find that higher
order to minimize the risk to the banking sector’s stability, they may have rates abroad tended to moderate local housing credit growth before the
induced the substitution of housing credit with other types of credit (such GFC, rather than increasing it, as would have been if local housing debt
and foreign debt were substitutes. Given that the Israeli mortgage market
is (almost) entirely a local market, the negative effect on the growth rate
13
In an alternative specification, we also include monetary policy surprises of local mortgages may indicate that US surprises are a proxy for ex-
from the eurozone. The qualitative results remain unchanged (see the pected domestic monetary policy, which slowed the growth rate of
Appendix). mortgages before the GFC. Another possible interpretation links the
9
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 3
Housing Credit, Sample 2004:Q2 – 2019:Q4, 7 banks.
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.001** 0.002** 0.002*** 0.002***
MaP_general (-1) to (8) 0.004* 0.002 0.005* 0.004
MaP_concentration (-1) to (4) 0.002 0.006 0.007* 0.010**
Monetary Policy Monetary policy (-1) until the GFC 1.89* 2.23** 1.34 2.08
Monetary policy (-2) until the GFC 4.56*** 4.53*** 2.67** 2.78**
Aggregated before 2008 ¡6.45*** ¡6.76*** ¡4.00** ¡4.86***
Monetary policy (-1) after the GFC 0.82 0.88 0.41 2.70
Monetary policy (-2) after the GFC 0.75 0.56 0.22 0.29
Aggregated after 2008 1.57* 1.43 ¡0.20 ¡2.99
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.14 0.24
Cumulative MaP_housing (-1) Mon. policy (-2) 0.22 0.15
Aggregated MaP_housing £ Mon. policy ¡0.36 0.09
MaP_general (-1) to (8) Mon. policy (-1) 1.04 2.38
MaP_general (-1) to (8) Mon. policy (-2) 2.76 2.83
Aggregated MaP_general £ Mon. policy 3.80 5.21*
MaP_concentr.(-1) to (4) Mon. policy (-1) 4.08 3.90
MaP_concentr.(-1) to (4) Mon. policy (-2) 3.09 2.94
Aggregated MaP_concentr. £ Mon. policy 7.18* 6.85*
Foreign US monetary policy (-1) until the GFC 1.89 1.27
US monetary policy (-2) until the GFC 3.00** 2.93**
US Aggregated before 2008 ¡4.88*** ¡4.20**
US monetary policy (-1) after the GFC 2.17 5.64*
US monetary policy (-2) after the GFC 0.78 2.04
US Aggregated after 2008 2.95 7.68
Banks^ Excess Tier(-1) 0.38*** 0.40*** 0.57*** 0.57***
Macro D (real wage (-2)) 0.12* 0.14** 0.18** 0.19**
D (unemployed 25–64 (-5)) 0.77*** 0.61** 0.71** 0.58*
D (log (house_prices (-1)) 0.07*** 0.06** 0.05* 0.05
Seasonal Dummies Yes Yes Yes Yes
Data Break Dummies Yes Yes Yes Yes
C 0.001 0.001 0.004 0.003
CRISIS2008 0.01** 0.02*** 0.02*** 0.02***
Observations 441 441 441 441
Adjusted R2 0.17 0.17 0.18 0.18
Notes: ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. ^The estimation also includes the change from liabilities to assets, the change in the
banks’ real assets, the change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were
insignificant.
higher foreign interest rate to the actual and expected local currency 4.2. Business credit estimation results
appreciation. Since house prices were generally denominated in US
dollars before the GFC, an appreciation would lead, at least in the short Although the banks’ share of total business credit declined in the past
term, to a moderation of the local currency value of house prices and, decade from about 75 percent to 50 percent (Fig. 5b), business credit is
therefore, of mortgages. still an important sector, accounting for about a 40 percent share of total
The macroeconomic indicators—real wages, unemployment, and bank credit.
house prices—all have the expected effect on the mortgage market. We find that general MaP measures tended to reduce the growth rate
Higher prices indicate excess demand in the market, but also increase the of business credit, as expected, although the results were significant only
nominal value of a given number of transactions, inducing higher de- in some of the specifications (Table 4). In addition, our results show that
mand for mortgages. MaP measures aimed at housing credit tended to increase the supply of
We also control for bank characteristics that may affect the banks’ credit to the business sector, as those are substitutes on the asset side of
tendency to offer credit. An increase in the banks’ capital relative to the the banks’ balance sheets.
supervisory requirements (lagged two periods), i.e., a positive gap, tends We do not find evidence that the concentration MaP measures, which
to increase the growth of housing credit. were implemented only once during our sample period, had any signif-
Fig. 10 presents the average annual contributions of the various fac- icant effect on business credit.14 Similar to our results for housing credit,
tors that affect the dynamics of housing credit. Housing MaP measures higher policy rates tended to reduce business credit growth before the
made a substantial downward contribution to the development of GFC, but this influence diminishes and we find that higher policy rates
housing credit after 2011. Foreign monetary policy had some impact on tend to increase business credit after the GFC, contrary to our basic
local housing credit in the years before the GFC, and did not contribute to intuition. We find that the interactions between monetary policy and the
credit dynamics in the years after the GFC. Macroeconomic variables and MaP measures under the full specification do not significantly affect
bank characteristics contributed to the housing credit increase after business credit.
2009. There is partial evidence that less accommodative foreign monetary
The post-GFC dummy reflects post-2008 factors, such as unconven- policy reduced the growth rate of domestic business credit after 2008.
tional monetary policies, that sustained the demand for housing credit.
Fig. 10 also shows that excess tier 1 capital increased housing credit.
14
We included one measure taken in September 2003, which is before the
sample period but, due to lags in the specification, is also relevant.
10
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 4
Business Credit, Sample 2004:Q2—2019:Q4, 7 banks.
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.001** 0.001** 0.001*** 0.001**
MaP_general (-1) to (8) 0.008*** 0.005* 0.007** 0.005
MaP_concentration (-4) to (7) 0.004 0.006 0.005 0.005
Monetary Policy Monetary policy (-1) until the GFC 0.12 0.44 1.29 0.53
Monetary policy (-2) until the GFC 2.75** 2.17* 3.71*** 3.00*
Aggregated before 2008 ¡2.63* ¡2.61 ¡2.42 ¡2.47
Monetary policy (-1) after the GFC 1.28** 0.22 2.97*** 2.36
Monetary policy (-2) after the GFC 3.32*** 3.80*** 3.46*** 6.74***
Aggregated after 2008 4.60*** 4.02*** 6.43*** 9.10***
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.23 0.04
Cumulative MaP_housing (-1) Mon. policy (-2) 0.43* 0.81**
Aggregated MaP_housing £ Mon. policy ¡0.20 ¡0.86
MaP_general (-1) to (8) Mon. policy (-1) 3.00 2.25
MaP_general (-1) to (8) Mon. policy (-2) 3.57* 2.61
Aggregated MaP_general £ Mon. policy 6.57* 4.86
MaP_concentr.(-4) to (7) Mon. policy (-1) 1.89 1.13
MaP_concentr.(-4) to (7) Mon. policy (-2) 2.66 1.03
Aggregated MaP_concentr. £ Mon. policy ¡0.77 0.1
Foreign US monetary policy (-1) until the GFC 1.97 1.42
US monetary policy (-2) until the GFC 1.54 1.03
US Aggregated before 2008 ¡0.43 ¡0.39
US monetary policy (-1) after the GFC 4.43** 3.14
US monetary policy (-2) after the GFC 0.60 5.40
US Aggregated after 2008 ¡5.02 ¡8.54
Banks^ Excess Tier(-5) 0.40*** 0.29* 0.31* 0.36**
D (Liabilities (-1)/Assets (-1)) 0.73 1.06* 0.88 1.01*
Macro D (BusinessGDP (-2)) 0.22*** 0.20*** 0.21*** 0.21**
D (unemployed 25–64 (-2)) 1.55*** 1.77*** 1.74*** 1.63***
Seasonal Dummies Yes Yes Yes Yes
Data Break Dummies Yes Yes Yes Yes
C 0.03** 0.02** 0.03*** 0.02***
CRISIS2008 0.02*** 0.01*** 0.01** 0.01***
Observations 441 441 441 441
Adjusted R2 0.30 0.30 0.30 0.30
Notes: ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. ^The estimation also includes the change from liabilities to assets, the change in the
banks’ real assets, the change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were
insignificant.
This suggests that the signaling effect of a surprise in foreign interest market MaP measures, we find evidence for housing and consumer credit
rates, assuming one exists, is stronger than the possible substitution effect being substitutes. The interaction term of these measures with monetary
on foreign credit. Similar to our results for housing credit, macroeco- policy is negative, indicating, as for business credit, that when stricter
nomic indicators and bank-specific indicators have the expected effect on housing MaP measures accompany accommodative monetary policy,
the evolution of the business credit aggregates (Table 4). consumer (nonhousing) credit will tend to increase. On the other hand,
coupling a surprising interest rate hike with accumulated housing market
4.3. Consumer credit estimation results MaP measures intensifies the effect of monetary surprises on consumer
credit.
As stated above, consumer credit comprises a relatively small share of We find that until 2008 a positive surprise to the global interest rate
the banks’ total credit, and this share has been relatively constant during tended to decrease the growth rate of consumer credit. After 2008, this
the past two decades. Therefore, our ability to account for changes in effect is not significant. As for housing credit, interest rate surprises
consumer credit due to changes in policy or macroeconomic factors is abroad may be an indication of the expected local monetary policy and
limited. therefore affect the growth rate of consumer credit.
The share of variance explained in our estimation is close to zero.
Therefore, any conclusions referring to the drivers of consumer credit are 4.4. Total credit estimation results
limited. Our results are presented in Table 5 and show that we do not find
any significant effect of MaP measures on consumer credit. Less accom- The previous sections analyzed the effect of MaP measures and
modative monetary policy reduced the growth rate of consumer credit monetary policy on each component of bank credit: housing, business,
only before the GFC, similar to the effect found for business and housing and consumer credit. Although housing MaP measures are naturally
credit. targeted at the housing market, the policy aims to influence the macro-
The expected effect of MaP measures aimed at housing credit on economic conditions and financial stability.
consumer credit is ambiguous. MaP measures limiting housing credit or Therefore, it is important to examine the effect of these measures on
making it more expensive are expected to increase consumer credit due total bank credit. We repeat our estimation for aggregate bank credit,
to substitution effects on both the demand and supply sides. By contrast, which is the sum of the three components mentioned above. Table 6
if housing (mortgage) credit and consumer credit are complements, presents the detailed results.
limiting mortgages will also tend to reduce demand for consumer credit. We find that MaP measures targeted at specific credit sectors, which
Looking at the interaction between monetary policy and housing influence these sectors, do not effect the dynamics of total credit
11
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 5
Consumer Credit, Sample 2004:Q2—2019:Q4, 7 banks.
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.001 0.002 0.002 0.003
MaP_general (-1) to (8) 0.003 0.005 0.000 0.005
MaP_concentration (-1) to (4) 0.003 0.013 0.003 0.013
Monetary Policy Monetary policy (-1) until the GFC 0.52 1.60 8.17* 7.03
Monetary policy (-2) until the GFC 12.37*** 12.73*** 16.86*** 17.77***
Aggregated before 2008 ¡12.89** ¡14.33** ¡8.68 ¡10.74
Monetary policy (-1) after the GFC 0.07 0.77 0.53 1.58
Monetary policy (-2) after the GFC 0.54 3.07 3.23 6.83
Aggregated after 2008 0.47 2.30 ¡3.76 5.25
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.80 -.95
Cumulative MaP_housing (-1) Mon. policy (-2) 2.05** 2.38*
Aggregated MaP_housing £ Mon. policy ¡2.84** ¡3.33*
MaP_general (-1) to (8) Mon. policy (-1) 6.2 9.55
MaP_general (-1) to (8) Mon. policy (-2) 7.21 4.89
Aggregated MaP_general £ Mon. policy 13.41 14.44
MaP_concentr.(-1) to (4) Mon. policy (-1) 7.88 0.40
MaP_concentr.(-1) to (4) Mon. policy (-2) 19.06 18.93
Aggregated MaP_concentr. £ Mon. policy 26.94* 19.33
Foreign US monetary policy (-1) until the GFC 11.84*** 11.10**
US monetary policy (-2) until the GFC 5.67 6.64
US Aggregated before 2008 ¡6.17 ¡4.46
US monetary policy (-1) after the GFC 1.26 1.57
US monetary policy (-2) after the GFC 7.73 6.50
US Aggregated after 2008 6.46 ¡4.92
Banks^ Excess Tier(-1) 0.15 0.04 0.01 0.05
D (Liabilities (-1)/Assets (-1)) 3.98** 4.29** 4.45** 4.49**
Macro D (real wage (-1)) 0.67*** 0.62** 1.00*** 0.87***
D (unemployed 25–64 (-2)) 0.94 0.37 1.87 1.22
Seasonal Dummies Yes Yes Yes Yes
Data Break Dummies Yes Yes Yes Yes
C 0.03** 0.03* 0.02 0.02
CRISIS2008 0.005 0.003 0.007 0.004
Observations 441 441 441 441
Adjusted R2 0.04 0.04 0.05 0.05
Notes: ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. ^The estimation also includes the change from liabilities to assets, the change in the
banks’ real assets, the change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were
insignificant.
according to our estimations. The effect of housing and general MaP general MaP measures. The effect of concentration MaP measures is also
measures on total credit is insignificant, and the accumulated effect of weaker due to their influence in opposite directions on credit
concentration-related measures is also null. As we found in the full components.
specification (Column 4) for some of the specific credit sectors, monetary We may conclude that the effect of MaP measures on total bank credit
policy affected credit growth only before the GFC. A negative sign on the was less evident than on each of the credit components due to the sub-
interaction between monetary policy and housing MaP measures in the stitution effects, probably originating from both the supply and demand
restricted specification (Column 2) indicates that accommodative mon- sides. Housing MaP measures were aimed specifically at the housing
etary policy tended to support total credit growth when coupled with market, and housing credit intended to minimize the risks to financial
restrictive housing MaP measures. stability due to distress in this sector did succeed according to our esti-
Fig. 11 shows that monetary policy is not central in explaining total mation in moderating housing credit growth.
bank credit dynamics. The interaction between monetary policy and Other MaP measures targeted at banks’ capital structure may have
nonhousing MaP measures decreased total credit growth between 2012 achieved their goal of minimizing financial risks by inducing changes in
and 2014. the banks’ balance sheets, while affecting credit growth only marginally.
The housing MaP measures made a significant contribution to
changes in housing and business credit. However, the substitution be- 5. Robustness checks
tween the types of credit offsets these effects, resulting in nearly no
contribution to changes in total credit. As Fig. 11 shows, this influence We perform several robustness checks in order to verify our results.15
was augmented by the interaction of housing MaP measures and mone- First, we check whether aggregating general and concentration measures
tary policy, contributing to total credit mainly through the effect of the into a single index of dummies matters and find that qualitative results
interaction on consumer credit. did not change. The substitution effect is still evident, as housing MaP
Table 7 presents the main qualitative effect of MaP policy on each of measures tend to restrict housing credit growth but contribute to busi-
the different credit components, based on the estimation including both ness credit growth and, on the whole, do not affect the evolution of total
the interaction terms and global monetary policy, presented in the fourth credit. The negative effect of housing MaP measures on consumer credit
column of each table. We find that the effect of the various MaP measures
on total credit is weaker than on some of its components due to the
substitution effect mentioned earlier. Total bank credit is not affected 15
We refer in this section to the results for the full specification (Column 4 of
significantly by the measures taken to curb housing credit or by the each table).
12
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 6
Total Credit, Sample 2004:Q2—2019:Q4, 7 banks.
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.0002 0.0002 0.0006 0.0000
MaP_general (-1) to (8) 0.0002 0.002 0.0001 0.002
MaP_concentration (-1) to (4) 0.006** 0.008** 0.008** 0.010***
MaP_concentration (-4) to (7) 0.008*** 0.009*** 0.007*** 0.008***
Monetary Policy Monetary policy (-1) until the GFC 0.10 0.68 2.44** 1.67
Monetary policy (-2) until the GFC 4.60*** 4.74*** 5.82*** 6.01***
Aggregated before 2008 ¡4.70*** ¡5.42*** ¡3.38** ¡4.34***
Monetary policy (-1) after the GFC 0.59 0.17 1.06 0.82
Monetary policy (-2) after the GFC 1.76*** 2.31*** 1.27* 2.44
Aggregated after 2008 2.35*** 2.48*** 2.32* 1.63
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.12 0.02
Cumulative MaP_housing (-1) Mon. policy (-2) 0.62*** 0.58**
Aggregated MaP_housing £ Mon. policy ¡0.74** ¡0.60
MaP_general (-1) to (8) Mon. policy (-1) 2.69** 3.62**
MaP_general (-1) to (8) Mon. policy (-2) 4.31*** 3.92**
Aggregated MaP_general £ Mon. policy 6.99*** 7.54***
MaP_concentr.(-1) to (4) Mon. policy (-1) 7.83*** 6.27***
MaP_concentr.(-1) to (4) Mon. policy (-2) 0.17 0.50
Aggregated MaP_concentr. £ Mon. policy 7.66** 6.78**
MaP_concentr.(-4) to (7) Mon. policy (-1) 2.04 2.59
MaP_concentr.(-4) to (7) Mon. policy (-2) 1.49 2.16
Aggregated MaP_concentr. £ Mon. policy ¡0.55 ¡0.43
Foreign US monetary policy (-1) until the GFC 3.17*** 2.47**
US monetary policy (-2) until the GFC 1.48 1.40
US Aggregated before 2008 ¡1.68 ¡1.07
US monetary policy (-1) after the GFC 1.38 1.65
US monetary policy (-2) after the GFC 1.08 0.05
US Aggregated after 2008 ¡0.30 1.60
Banks^ Excess Tier(-1) 0.27*** 0.32*** 0.33*** 0.35***
D (log (Real Assets (-1)) 0.08 0.08* 0.06 0.07
Macro YoY(Real Wage (-2)) 0.01 0.04 0.10 0.04
D (Unemployed 25–64 (-2)) 1.15*** 1.01*** 1.42*** 1.26***
YoY(House Prices (-1)) 0.01 0.001 0.002 0.01
Seasonal Dummies Yes Yes Yes Yes
Data Break Dummies Yes Yes Yes Yes
C 0.002 0.002 0.003 0.003
CRISIS2008 0.007 0.009* 0.007 0.01*
Observations 441 441 441 441
Adjusted R2 0.28 0.32 0.29 0.32
Notes: ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. ^The estimation also includes the change from liabilities to assets, the change in the
banks’ real assets, the change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were
insignificant.
13
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 8
Robustness Checks, Sample 2004:Q2–2019:Q4, 7 banks.
1 2 3 4
MaP Cumulative MaP_housing (-1), since 2004 0.0013** 0.001*** 0.003* 0.00
MaP_non_housing (1) to (8) 0.002 0.004** 0.01 0.00
Monetary Policy Monetary policy (-1) until the GFC 2.08 0.07 4.88 0.81
Monetary policy (-2) until the GFC 3.96*** 3.22** 20.25*** 5.91***
Aggregated before 2008 ¡6.04*** ¡3.15 ¡15.37** ¡5.11***
Monetary policy (-1) after the GFC 0.59 2.27 0.43 0.57
Monetary policy (-2) after the GFC 0.59 8.56*** 2.48 3.71**
Aggregated after 2008 0.01 10.83*** 2.91 4.29*
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.24 0.03 1.43 0.04
Cumulative MaP_housing (-1) Mon. policy (-2) 0.09 0.91** 2.28* 0.63**
Aggregated MaP_housing £ Mon. policy 0.15 ¡0.88 ¡3.71** ¡0.68
MaP_non_housing (-1) to (8) Mon. policy (-1) 0.45 1.86 11.07* 1.91
MaP_non_housing (1) to (8) Mon. policy (-2) 1.75 0.45 12.03** 2.24*
Aggregated MaP_non_housing £ Mon. policy 2.20 2.31 23.11*** 4.16**
Foreign US monetary policy (-1) until the GFC 1.06 1.26 10.00** 2.28**
US monetary policy (-2) until the GFC 2.09* 1.25 7.06 1.18
US Aggregated before 2008 ¡3.15* 0.00 ¡2.94 ¡1.10
US monetary policy (-1) after the GFC 2.04 2.94 2.12 0.41
US monetary policy (-2) after the GFC 0.11 8.07** 0.82 2.34
US Aggregated after 2008 1.93 ¡11.01** ¡2.94 ¡2.74
Banks^ Excess Tier(-1) 0.52*** 0.14 0.31***
Excess Tier(-5) 0.36**
Macro D (real wage (-1)) 1.04***
D (real wage (-2)) 0.19** 0.05
D (unemployed 25–64 (-2)) 1.56*** 0.97 1.11***
D (unemployed 25–64 (-5)) 0.72**
(continued on next page)
14
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 8 (continued )
1 2 3 4
Notes: ***, **, and * indicate significance at the 1%, 5%, and 10% level, respectively. ^The estimation also includes the change from liabilities to assets, the change in the
banks’ real assets, the change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were
insignificant.
surprise. The results remain qualitatively unchanged,16 although the not persist after the crisis, at least according to our estimation. We find
interaction terms with MaP policy become insignificant. some evidence for the effect of monetary policy when it interacts with
MaP measures that were taken after 2008. As the share of foreign cur-
6. Concluding remarks rency credit is very small, the effect of monetary policy in the US is in the
same direction as that of domestic monetary policy, indicating that it may
This paper analyzes the effect of monetary and MaP policies on bank be a signal for possible future local policy rather than a substitute.
credit. We present the results of an analysis of the effect of domestic We find that the interaction between monetary policy and MaP policy
monetary policy and MaP policy implemented in Israel during the period is important. We show that accommodative monetary policy tended to
2004–19 on domestic bank credit. We estimate the effect of monetary increase consumer credit and total credit in some specifications, when
policy, MaP measures, and the interaction between them on housing interacting with macroprudential policies targeting the housing market.
credit, consumer credit, credit to the business sector, and total credit. We also find that macroeconomic conditions, i.e., output growth and
We find that the MaP policy affected the development of bank credit. employment growth, are associated with credit growth rates, and that
Our results show that MaP measures targeted at housing credit did bank-specific characteristics, particularly the bank’s capital relative to
reduce the growth rate of this credit while contributing to the increased the supervisory requirements, affect the supply of credit.
growth of business credit. General MaP measures worked to reduce the
growth rate of business credit. We could not find a clear effect of the
single measure targeted at the lending concentration on any of the credit Declaration of competing interest
components.
Monetary policy was found to have a clear and significant effect on On behalf of all authors, the corresponding author states that there is
housing and total credit growth rates before the GFC, but this effect did no conflict of interest.
Table 3a
Housing Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Including ECB Monetary Policy
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.001** 0.002** 0.002*** 0.002***
MaP_general (-1) to (8) 0.004* 0.002 0.004* 0.003
MaP_concentration (-1) to (4) 0.002 0.006 0.005 0.01**
Monetary Policy Monetary policy (-1) until the GFC 1.89* 2.23** 0.81 0.86
Monetary policy (-2) until the GFC 4.56*** 4.53*** 3.13** 3.61***
Aggregated before 2008 ¡6.45*** ¡6.76*** ¡3.94** ¡4.46**
Monetary policy (-1) after the GFC 0.82 0.88 0.35 3.31
Monetary policy (-2) after the GFC 0.75 0.56 0.31 0.64
Aggregated after 2008 1.57* 1.43 0.05 ¡2.67
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.14 0.29
Cumulative MaP_housing (-1) Mon. policy (-2) 0.22 0.24
Aggregated MaP_housing £ Mon. policy ¡0.36 0.05
MaP_general (-1) to (8) Mon. policy (-1) 1.04 3.22*
MaP_general (-1) to (8) Mon. policy (-2) 2.76 2.43
(continued on next page)
16
In this specification, there is no evidence for the substitution effect of
housing MaP measures between housing credit growth and business credit
growth.
15
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 3a (continued )
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
Notes: *p-value < 10%, **p-value < 5%, ***p-value < 1%. ^ The estimation also includes the change from liabilities to assets, the change in the banks’ real assets, the
change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were insignificant.
Table 4a
Business Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Including ECB Monetary Policy
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.001** 0.001** 0.001** 0.001**
MaP_general (-1) to (8) 0.008*** 0.005* 0.006** 0.005
MaP_concentration (-4) to (7) 0.004 0.006 0.004 0.004
Monetary Policy Monetary policy (-1) until the GFC 0.12 0.44 0.45 0.81
Monetary policy (-2) until the GFC 2.75** 2.17* 3.93*** 2.69
Aggregated before 2008 ¡2.63* ¡2.61 ¡3.48* ¡3.50*
Monetary policy (-1) after the GFC 1.28** 0.22 3.00*** 2.68
Monetary policy (-2) after the GFC 3.32*** 3.80*** 3.45*** 6.33**
Aggregated after 2008 4.60*** 4.02*** 6.44*** 9.01**
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.23 0.02
Cumulative MaP_housing (-1) Mon. policy (-2) 0.43* 0.80**
Aggregated MaP_housing £ Mon. policy ¡0.20 ¡0.83
MaP_general (-1) to (8) Mon. policy (-1) 3.00 1.75
MaP_general (-1) to (8) Mon. policy (-2) 3.57* 3.26
Aggregated MaP_general £ Mon. policy 6.57* 5.01
MaP_concentr.(-4) to (7) Mon. policy (-1) 1.89 1.03
MaP_concentr.(-4) to (7) Mon. policy (-2) 2.66 2.85
Aggregated MaP_concentr. £ Mon. policy ¡0.77 ¡1.81
Foreign US monetary policy (-1) until the GFC 1.40 0.50
US monetary policy (-2) until the GFC 1.72 0.84
US Aggregated before 2008 0.33 0.34
US monetary policy (-1) after the GFC 2.40 1.01
US monetary policy (-2) after the GFC 0.05 4.68
US Aggregated after 2008 ¡2.35 ¡5.68
ECB monetary policy (-1) until the GFC 21.19 23.98
ECB monetary policy (-2) until the GFC 13.12 15.72
ECB Aggregated before 2008 ¡34.31* ¡39.70*
ECB monetary policy (-1) after the GFC 2.80 3.69
ECB monetary policy (-2) after the GFC 0.02 0.80
ECB Aggregated after 2008 ¡2.78 ¡2.90
Banks^ Excess Tier(-5) 0.40*** 0.29* 0.33** 0.39**
(continued on next page)
16
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 4a (continued )
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
Notes: *p-value < 10%, **p-value < 5%, ***p-value < 1%. ^ The estimation also includes the change from liabilities to assets, the change in the banks’ real assets, the
change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were insignificant.
Table 5a
Consumer Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Including ECB Monetary Policy
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.001 0.002 0.002 0.003
MaP_general (-1) to (8) 0.003 0.005 0.001 0.005
MaP_concentration (-1) to (4) 0.003 0.013 0.002 0.01
Monetary Policy Monetary policy (-1) until the GFC 0.52 1.60 7.32 6.95
Monetary policy (-2) until the GFC 12.37*** 12.73*** 20.03*** 20.77***
Aggregated before 2008 ¡12.89** ¡14.33** ¡12.71* ¡13.82**
Monetary policy (-1) after the GFC 0.07 0.77 1.34 1.29
Monetary policy (-2) after the GFC 0.54 3.07 1.46 12.52
Aggregated after 2008 0.47 2.30 0.12 13. 81
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.80 0.87
Cumulative MaP_housing (-1) Mon. policy (-2) 2.05** 2.91**
Aggregated MaP_housing £ Mon. policy ¡2.84** ¡3.79**
MaP_general (-1) to (8) Mon. policy (-1) 6.2 7.11
MaP_general (-1) to (8) Mon. policy (-2) 7.21 2.72
Aggregated MaP_general £ Mon. policy 13.41 9.83
MaP_concentr.(-1) to (4) Mon. policy (-1) 7.88 0.22
MaP_concentr.(-1) to (4) Mon. policy (-2) 19.06 21.73*
Aggregated MaP_concentr. £ Mon. policy 26.94* 21.95
Foreign US monetary policy (-1) until the GFC 10.65** 11.07**
US monetary policy (-2) until the GFC 9.45** 10.05**
US Aggregated before 2008 ¡1.20 ¡1.02
US monetary policy (-1) after the GFC 2.65 3.55
US monetary policy (-2) after the GFC 14.05* 6.68
US Aggregated after 2008 16.69 ¡3.13
ECB monetary policy (-1) until the GFC 9.39 52.96
ECB monetary policy (-2) until the GFC 114.43** 114.87**
ECB Aggregated before 2008 ¡105.04 ¡61.91
ECB monetary policy (-1) after the GFC 7.44 6.63
ECB monetary policy (-2) after the GFC 9.76 9.58
ECB Aggregated after 2008 ¡17.19 ¡16.22
Banks^ Excess Tier(-1) 0.15 0.05 0.38 0.24
D (Liabilities (-1)/Assets (-1)) 3.96** 4.29** 5.29*** 5.09***
Macro D (real wage (-1)) 0.66*** 0.63** 1.13*** 0.99***
D (unemployed 25–64 (-2)) 0.93 0.31 1.06 0.19
Seasonal Dummies Yes Yes Yes Yes
Data Break Dummies Yes Yes Yes Yes
C 0.03** 0.03* 0.020 0.03*
CRISIS2008 0.01 0.003 0.01 0.000
Observations 441 441 441 441
Adjusted R2 0.04 0.04 0.06 0.06
Notes: *p-value < 10%, **p-value < 5%, ***p-value < 1%. ^ The estimation also includes the change from liabilities to assets, the change in the banks’ real assets, the
change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were insignificant.
17
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Table 6a
Total Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Including ECB Monetary Policy
1 2 3 4
Basic – excl. interactions and Including Including global Including interactions and
global policy interactions policy global monetary policy
MaP Cumulative MaP_housing (-1), since 2004 0.0002 0.0002 0.0005 0.0001
MaP_general (-1) to (8) 0.0002 0.002 0.0004 0.002
MaP_concentration (-1) to (4) 0.006** 0.008** 0.008** 0.009**
MaP_concentration (-4) to (7) 0.008*** 0.009*** 0.007*** 0.007***
Monetary Policy Monetary policy (-1) until the GFC 0.10 0.68 1.66** 1.17
Monetary policy (-2) until the GFC 4.60*** 4.74*** 6.20*** 6.53***
Aggregated before 2008 ¡4.70*** ¡5.42*** ¡4.54*** ¡5.36***
Monetary policy (-1) after the GFC 0.59 0.17 1.23 0.13
Monetary policy (-2) after the GFC 1.76*** 2.31*** 1.40* 3.19*
Aggregated after 2008 2.35*** 2.48*** 2.62* 3.32
Interactions Cumulative MaP_housing (-1) Mon. policy (-1) 0.12 0.08
Cumulative MaP_housing (-1) Mon. policy (-2) 0.62*** 0.72**
Aggregated MaP_housing £ Mon. policy ¡0.74** ¡0.79*
MaP_general (-1) to (8) Mon. policy (-1) 2.69** 2.82*
MaP_general (-1) to (8) Mon. policy (-2) 4.31*** 4.01***
Aggregated MaP_general £ Mon. policy 6.99*** 6.83***
MaP_concentr.(-1) to (4) Mon. policy (-1) 7.83*** 6.25***
MaP_concentr.(-1) to (4) Mon. policy (-2) 0.17 0.03
Aggregated MaP_concentr. £ Mon. policy 7.66** 6.29**
MaP_concentr.(-4) to (7) Mon. policy (-1) 2.04 3.14
MaP_concentr.(-4) to (7) Mon. policy (-2) 1.49 3.34
Aggregated MaP_concentr. £ Mon. policy ¡0.55 0.20
Foreign US monetary policy (-1) until the GFC 2.44** 1.96*
US monetary policy (-2) until the GFC 2.04* 2.03*
US Aggregated before 2008 ¡0.41 0.07
US monetary policy (-1) after the GFC 0.81 1.59
US monetary policy (-2) after the GFC 2.34 0.72
US Aggregated after 2008 1.52 0.87
ECB monetary policy (-1) until the GFC 7.80 1.13
ECB monetary policy (-2) until the GFC 21.86** 28.59***
ECB Aggregated before 2008 ¡29.66** ¡2746*
ECB monetary policy (-1) after the GFC 0.93 1.75
ECB monetary policy (-2) after the GFC 1.58 0.21
ECB Aggregated after 2008 ¡2.50 ¡1.96
Banks^ Excess Tier(-1) 0.27*** 0.32*** 0.26** 0.29**
D (log (Real Assets (-1)) 0.08 0.08* 0.08 0.09*
Macro YoY(Real Wage (-2)) 0.01 0.04 0.08 0.02
D (Unemployed 25–64 (-2)) 1.15*** 1.01*** 1.31*** 0.95***
YoY(House Prices (-1)) 0.01 0.001 0.007 0.007
Seasonal Dummies Yes Yes Yes Yes
Data Break Dummies Yes Yes Yes Yes
C 0.002 0.002 0.003 0.001
CRISIS2008 0.007 0.009* 0.007 0.01*
Observations 441 441 441 441
Adjusted R2 0.28 0.32 0.29 0.32
Notes: *p-value < 10%, **p-value < 5%, ***p-value < 1%. ^ The estimation also includes the change from liabilities to assets, the change in the banks’ real assets, the
change in deposits as a share of assets, the change in liquid assets, and the change from excess reserves to assets, with varying lags. All results were insignificant.
References Caspi, I., Friedman, A., Ribon, S., 2021. The immediate impact and persistent effect of FX
purchases on the exchange rate. Forthcoming in the International Journal of Central
Banking.
Aiyar, S., Calomiris, C., Wieladek, T., 2014. Does macro-prudential regulation leak?
Cerutti, E., Claessens, S., Laeven, L., 2017. The use of macroprudential policies: new
Evidence from the UK policy experiment. J. Money Credit Bank. 46 (s1), 181–214.
evidence. J. Financ. Stabil. 28, 203–224.
Altavilla, C., Laeven, L., Peydr
o, J.L., 2020. Monetary and Macroprudential Policy
Cesa-Bianchi, A., Thwaites, G., Vicondoa, A., 2020. Monetary policy transmission in the
Complementarities: Evidence from European Credit Registers. CEPR Discussion
United Kingdom: a high-frequency identification approach. Eur. Econ. Rev. 123,
Papers 15539.
103375.
Andrade, P., Ferroni, F., 2021. Delphic and odyssean monetary policy shocks: evidence
Claessens, S., Ghosh, S., Mihet, R., 2013. Macro-prudential policies to mitigate financial
from the euro area. J. Monetary Econ. 117, 816–832.
system vulnerabilities. J. Int. Money Finance 39, 153–185.
Arena, M., Chen, T., Choi, S.M., Geng, N., Gueye, C.A., Lybek, T., Papageorgiou, E.,
Davis, E.P., Liadze, I., Piggott, R., 2019. Assessing the macroeconomic impact of
Zhang, Y.S., 2020. Macroprudential Policies and House Prices in Europe (IMF
alternative macroprudential policies. Econ. Modell. 80 (C), 407–428.
Departmental Paper).
De Jonghe, O., Dewachter, H., Ongena, S., 2020. Bank capital (requirements) and credit
Baudot-Trajtenberg, N., Tzur-Ilan, N., Frayberg, R., 2017. Assessing the impact of
supply: evidence from pillar 2 decisions. J. Corp. Finance 60, 101518.
macroprudential tools: the case of Israel. BIS Papers chapters. In: Bank for
De Marco, F., Kneer, C., Wieladek, T., 2021. The real effects of capital requirements and
International Settlements (Ed.), Macroprudential Policy Frameworks,
monetary policy: evidence from the United Kingdom. J. Bank. Finance 133, 106237.
Implementation and Relationships with Other Policies, vol. 94. Bank for International
Dell’Ariccia, G., Igan, D., Laeven, L., 2012. Credit booms and lending standards: evidence
Settlements, pp. 107–218.
from the subprime mortgage market. J. Money Credit Bank. 44 (2–3), 367–384 ().
Bussiere, M., Cao, J., de Haan, J., Hills, R., Lloyd, S., Meunier, B., Pedrono, J.,
Everett, M., de Haan, J., Jansen, D.-J., McQuade, P., Samarina, A., 2021. Mortgage
Reinhardt, D., Sinha, S., Sowerbutts, R., Styrin, K., 2021. The interaction between
lending, monetary policy, and prudential measures in small euro-area economies:
macroprudential policy and monetary policy: overview. Rev. Int. Econ. 29 (1), 1–19.
evidence from Ireland and The Netherlands. Rev. Int. Econ. 29 (1), 117–143.
Cao, J., Dinger, V., Grodecka-Messi, A., Juelsrud, R., Zhang, X., 2021. The interaction
Forbes, K., Reinhardt, D., Wieladek, T., 2017. The spillovers, interactions, and (Un)
between macroprudential and monetary policies: the cases of Norway and Sweden.
intended consequences of monetary and regulatory policies. J. Monetary Econ. 85,
Rev. Int. Econ. 29 (1), 87–116.
1–22.
18
J. Benchimol et al. Economic Modelling 112 (2022) 105872
Gambacorta, L., Murcia, A., 2020. The impact of macroprudential policies in Latin Kannan, P., Rabanal, P., Scott, A., 2012. Monetary and macroprudential policy rules in a
America: an empirical analysis using credit registry data. J. Financ. Intermediation model with house price booms. B E J. Macroecon. 12 (1), 1–44.
42, 100828. Kutai, A., 2020. Measuring the Effect of Forward Guidance in Small Open Economies: the
Gertler, M., Karadi, P., 2015. Monetary policy surprises, credit costs and economic Case of Israel. Bank of Israel Research Department. Discussion Paper 2020.03.
activity. Am. Econ. J. Macroecon. 7, 44–76. Niepmann, F., Schmidt-Eisenlohr, T., Liu, E., 2020. The effect of US stress tests on
Greenwood-Nimmo, M., Tarassow, A., 2016. Monetary shocks, macroprudential shocks monetary policy spillovers to emerging markets. Rev. Int. Econ. 29 (1), 165–194.
and financial stability. Econ. Modell. 56 (C), 11–24. Repullo, R., Suarez, J., 2013. The procyclical effects of bank capital regulation. Rev.
Gürkaynak, R.S., Sack, B., Swanson, E., 2005. Do actions speak louder than words? The Financ. Stud. 26 (2), 452–490.
response of asset prices to monetary policy actions and statements. Int. J. Centr. Bank. Yellen, J., 2014. Monetary Policy and Financial Stability. The 2014 Michel Camdessus
1 (1), 55–93. Central Banking Lecture.
International Monetary Fund, 2013. The Interaction of Monetary and Macroprudential
Policies.
19