Content uploaded by Jonathan Benchimol
Author content
All content in this area was uploaded by Jonathan Benchimol on Jul 30, 2023
Content may be subject to copyright.
1
The Interaction Between Domestic Monetary Policy
and Macroprudential Policy in Israel*
Jonathan Benchimol,‡ Inon Gamrasni,‡ Michael Kahn,‡ Sigal Ribon,‡ Yossi Saadon,‡
Noam Ben-Ze’ev,‡ Asaf Segal,‡ and Yitzchak Shizgal‡
July 2022
Abstract
The global financial crisis (GFC) triggered the use of
macroprudential policies imposed on the banking sector. Using bank-
level panel data for Israel for the period 2004–2019, we find that
domestic macroprudential measures changed the composition of bank
credit growth but did not affect the total credit growth rate. Specifically,
we show that macroprudential measures targeted at the housing sector
moderated housing credit growth but tended to increase business credit
growth. We also find that accommodative monetary policy surprises
tended to increase bank credit growth before the GFC. We show that
accommodative monetary policy surprises increased consumer credit
when interacting with macroprudential policies targeting the housing
market. Accommodative monetary policy interacted with nonhousing
macroprudential measures to increase total credit.
Keywords: Financial stability, Policy evaluation, Banking sector, Credit
markets, Regulation, Global Financial Crisis.
JEL Classification: E51, E52, E58, G01, G21, G28.
* 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.
‡ Bank of Israel, Jerusalem, Israel. Corresponding authors: inon.gamrasni@boi.org.il; sigal.ribon@boi.org.il.
3
1 Introduction
House prices in Israel have more than doubled since 2008, alongside a substantial
increase in housing credit (mortgages). This development, which may pose a
significant risk to the financial sector and particularly to banks, together with an
understanding of the importance of the financial sector stability as a whole, triggered
the imposition of macroprudential (MaP) measures on the banks. This paper analyzes
the effects of MaP policy and domestic monetary policy on domestic bank credit in
Israel between 2004 and 2019. We chose to investigate the effects of MaP policy on
housing, consumer (nonhousing), and business sector credit separately in order to
understand the direct effect of the MaP measures on the market at which they were
targeted and possible effects on the other bank credit sectors and on total credit.
We find that alongside the desired direct, moderating effect of the MaP measures
on the credit markets at which they were targeted, the effect of these measures on total
bank credit was weakened due to substitution effects. We show that the MaP measures
targeted at housing credit reduced its growth rate, but tended to increase the growth
rate of business credit. MaP measures targeted at reducing banks’ credit concentration
did not significantly affect consumer or business credit, but contributed to the growth
of housing credit. We find that other general MaP measures (capital requirements
from the banks) that were expected to slow the growth of business credit influenced
business credit with a significant lag, apparently due to the time frame given to the
banks to complete the adjustment of capital and risk assets. While the interaction
between accommodative monetary policy and MaP measures targeting the housing
market did not have any additional effect on housing credit, we find weak support for
its expansionary effect on consumer credit. Overall, MaP measures did not change the
development of total credit significantly, but they did change its composition.
Monetary policy was generally found to be effective, with less accommodative policy
slowing housing and consumer credit growth only before the global financial crisis
(GFC). In addition, we find that higher foreign policy rates tended to reduce housing
and consumer credit growth, similar to the impact of domestic policy, suggesting their
ability to indicate future domestic monetary policy.
MaP policy is a regulatory tool that seeks to ensure the financial system’s stability,
which is necessary for stable economic growth. The overall objective of MaP policy is
to limit the buildup of financial vulnerabilities in order to reduce the financial system’s
sensitivity to shocks. MaP policy targets a specific credit sector, such as the consumer,
business, or housing credit sector, to achieve a particular objective. Specifically, MaP
measures targeting housing credit are intended to limit banks’ exposure to mortgage
defaults. Consequently, most of these measures are expected to increase the mortgage
cost, moderating housing credit growth. A similar mechanism applies to capital
adequacy or banking concentration MaP measures. While MaP measures tend to
decrease the growth rate of the targeted credit, substitution effects may be reflected in
the growth of other credit sectors. For instance, MaP measures targeting housing
credit may increase consumer credit as the latter may serve as an alternative to the
former, whether for banks as suppliers of credit or for borrowers as demanders of
credit. Alternatively, consumer credit may decline if MaP measures decrease the
demand for housing, thus reducing the demand for housing-related consumption and
consumer credit. Another example relates to MaP measures that limit credit to large
4
companies. The supply of this credit may move to smaller businesses but can also be
routed to housing or consumer credit.
The use of MaP measures has expanded since the GFC (Cerutti et al., 2017). The
theoretical literature based on dynamic stochastic general equilibrium (DSGE) models
supports the view that MaP measures can dampen credit cycles. Using a DSGE model
with a housing sector, Kannan et al. (2012) demonstrated that monetary and
macroeconomic policies can help stabilize the economy, given the shocks that hit the
economy. When financial or housing demand shocks drive a credit and housing boom,
MaP measures can temper them and improve welfare. Nevertheless, DSGE models
offer limited empirical guidance on the influence of the interaction between MaP
measures and monetary policy on credit cycles. We aim to assess these empirical
linkages. Repullo and Suarez (2013) showed that the credit (financial) and business
cycles are not necessarily synchronized. Therefore, monetary policy alone cannot
smooth both cycles. Consequently, MaP policy plays a more significant role in
moderating the financial cycle according to their analysis, thus, promoting financial
stability, while monetary policy remains responsible for price and output stability.
The effects of the interaction between MaP and monetary policies is still unsettled
(Dell’Ariccia et al., 2012; IMF, 2013; Aiyar et al., 2014). De Marco et al. (2021) found
that stricter monetary policy and MaP measures reinforced each other for the UK, but
only for small banks. Similarly, Forbes et al. (2017) found that monetary policy can
amplify the impact of regulatory measures. By contrast, De Jonghe et al. (2020) found
a trade-off between monetary policy and supervisory capital requirements in the case
of Belgium: the higher the banks’ capital requirements are, the weaker the effect of
monetary accommodation on credit supply is. Cerutti et al. (2017) found that MaP
measures have less of an effect on credit growth in advanced economies that tend to
have alternative sources of nonbank credit and enable borrowers to obtain funds from
abroad in more open economies. We seek to understand how MaP policy affects each
component of bank credit, and whether the interaction with monetary policy has an
additional effect on these credit sectors.
Gambacorta et al. (2020) examined the effect of domestic MaP measures on
domestic credit through an empirical exercise similar to ours. Unlike in our
estimation, they used granular information on bank 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 the impact of MaP measures on credit. Finally, they found
that MaP measures that reinforce monetary policy (i.e., in the same direction of
influence) are more effective.
Everett et al. (2021) contributed to the literature with Dutch and Irish data.
1
They
employed a similar methodology to the one we adopt, examining the effect of
monetary policy and MaP policy and their interaction on domestic mortgage lending.
As both Ireland and the Netherlands are small economies in the euro system,
monetary policy shocks may be regarded as exogenous to these two countries. Like
Israel, the banking sector in Ireland and the Netherlands is relatively concentrated and
both have employed various MaP measures in the past decade against the background
1
Everett et al. (2021), Cao et al. (2021), and Bussière et al. (2021) were all part of the 2018 International
Banking Research Network (IBRN) project.
5
of rapid increases in house prices. Everett et al. (2021) complemented their analysis by
using confidential bank-level data for domestic lending. They found that positive (less
accommodative) domestic monetary policy shocks reduced mortgage lending growth
in both countries, while prudential regulations weakened this effect in Ireland but not
in the Netherlands. They found only weak evidence for an international lending
channel: a foreign (US, UK) monetary policy shock, with or without interaction with
the MaP measures, was found to have no significant effect on mortgage lending for
Ireland, while for the Netherlands, weak evidence was found that tightening
monetary policy in the UK reduces mortgage lending.
Cao et al. (2021) studied the interaction between MaP policy and monetary policy
in Norway and Sweden and found evidence that aggregate MaP policy depresses
aggregate lending, although different types of MaP measures do not lead to
significantly different effects. Bussière et al. (2021) summarized the main findings of
seven papers that related mainly to MaP policies targeted at capital flows and the
possible spillovers from monetary policies in core economies to recipient economies.
The findings show that MaP policies in recipient countries can partly offset monetary
policy spillover effects from the core (large) economies. Bussière et al. (2021) also
remarked that the impact differs considerably across MaP policy instruments,
suggesting the importance of a detailed analysis.
In the absence of any interest rate adjustment, a credit-restricting MaP shock could
reduce the debt to GDP ratio in the short term, but it is unlikely to reduce the ratio of
firms’ debt to their internal funds. Greenwood-Nimmo and Tarassow (2016)
suggested that MaP policy alone is likely to have ambiguous effects on financial
stability. Nevertheless, MaP policy is rarely used alone. When monetary policy adjusts
in response to an MaP shock, both ratios decrease substantially, indicating an increase
in financial stability. According to this finding, a combination of monetary and MaP
measures for achieving financial stability may be desirable, as Yellen (2014)
advocated. Arena et al. (2020) highlighted the possibility that MaP policies can limit
riskier mortgages and contribute to stronger bank balance sheets, countering the
accommodative monetary policy stance. Following this line of thought, we analyze
the interaction between MaP and monetary policies and their effect on credit growth.
Following Altavilla et al. (2020), we show that monetary policy and MaP measures
are mutually reinforcing in relation to consumer loans more than to mortgage loans.
This result may be related to the risk-weighted ratio of capital to assets. Part of the
banks’ income can be used to rebuild capital and increase headroom if the latter is
below its normal level, and margins on consumer loans can be raised to speed up the
process (Davis et al., 2019).
As a small open economy, more than half of Israel’s trades are related to the US
and the Eurozone. Imported inflation and capital flows from these places directly
influence domestic monetary policy and thus credit aggregates. Our findings that
higher foreign policy rates moderate housing and consumer credit growth while
tending to increase business sector credit growth complement the findings of
Niepmann et al. (2020). Although foreign credit is not significant in Israel, our results
indicate the importance of foreign monetary policy surprises for local credit
aggregates.
6
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 Appendix presents additional results.
2 Economic and Policy Background
2.1 Macroeconomic Background
Since the beginning of the new millennium, particularly after 2003 (and until 2019),
the Israeli economy has enjoyed a relatively stable macroeconomic environment
characterized by high growth rates and price stability. The economy was hit during
these years by two major crises. The first, at the end of 2001, was due to the dot-com
crisis coupled with local geopolitical hostilities. The second was the GFC, which the
Israeli economy weathered quite well, experiencing only a short period of slowdown
at the end of 2008 and the beginning of 2009 before returning to growth rates close to
its potential in the second half of 2009 (Figure 1).
Figure 1: Annual GDP Growth in Israel and Advanced and Emerging Economies
Notes: Annual GDP growth in percent. Sources: Bloomberg and Central Bureau of Statistics.
Compared with the crisis’ adverse effect on other advanced economies overall, the
impact in Israel was relatively mild, more akin to the experience of developing
economies. In particular, the crisis had a limited and moderate effect on Israel’s
financial system, and the financial institutions remained stable. The factors
contributing to the GFC’s relatively mild effect on Israel include a conservative
financial system (and in particular a conservative and closely supervised banking
system), a relatively balanced housing market, and a successful economic policy (both
fiscal and monetary). Since the beginning of the 2000s, Israel has enjoyed low inflation
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
%
Advanced Emerging Israel
7
rates within the framework of an inflation-targeting regime, with the target set
between 1 and 3 percent, and actual inflation fluctuated around this target range. In
recent years, inflation has been lower than the target, and monetary policy is generally
characterized as accommodative (see details in Section 2.4).
2.2 Housing Sector
Israel enjoyed a vast influx of immigrants during the 1990s, increasing its population
by about 20 percent (1 million people) within five years. The housing market—the
private sector, accompanied by government intervention—reacted, and housing starts
surged at the beginning of the 1990s, with prices increasing at double-digit rates. The
cycle ended in the early 2000s, with building activity and house prices declining by
about 12 percent from peak to trough. Between 2007 and 2018, house prices rose
steadily, with the price level more than doubling within a decade (Figure 2).
Figure 2: House Starts and House Prices in Israel
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.
On the supply side, building starts reacted sluggishly to the increase in demand
arising from both demographic needs and the rise in income and wealth, contributing
to the rapid price increases.
2
The housing supply side in Israel is generally inelastic. The planning process in
Israel is very long and takes about 13 years, on average. This slow supply response
results in large price effects in response to positive demand shocks.
A significant government program that was launched in 2015, which essentially
subsidizes new apartments for buyers who meet specific characteristics (young first-
2
See the Bank of Israel Annual Reports for 2017 and 2018.
-10
-5
0
5
10
15
20
25
0
10000
20000
30000
40000
50000
60000
70000
80000
Percent
House Starts House Prices
8
time buyers), has had a significant impact on the market. This is because it removes
both supply and demand from the free market, while in contrast, it may encourage
households that did not plan to purchase a home to enter the market. Even before that,
taxation had been changed to reduce investor demand.
In the last few years of our sample, signs of a market slowdown have emerged,
with declines in both housing starts and transactions and prices moderating and even
declining for several quarters. The annual price increase in 2019 was about 3.5 percent
(Figure 2).
As noted above, since 2008, house prices in Israel have increased nominally by
about 130 percent. This rate of increase was exceptional compared to other countries
(Figure 3). Looking at a different starting point partially reduces Israel’s deviations
from other countries. Comparing recent price-to-income ratios with past ones strongly
warns of relatively high house prices in Israel as of the end of 2019.
Figure 3: House Prices in Israel and Various Countries
Notes: The house price index is 100 in 2007. Source: OECD
2.3 Banking Sector
The Israeli banking sector is relatively concentrated and consists of a small number of
banks. Seven banking groups control most of the domestic banking system,
representing 99.6 percent of the banking sector, with two dominant banking groups
(Bank Leumi and Bank Hapoalim) accounting for more than 50 percent of the market.
In our analysis, we focus on the credit portfolio of these seven banks.
3
The
concentrated structure of the Israeli banking sector may be attributed to the small size
of the Israeli economy, together with the existence of economies of scale and scope.
3
Leumi, Hapoalim, Discount, Mizrahi-Tefahot, First International, Union Bank, and Bank of Jerusalem.
40
60
80
100
120
140
160
180
200
Canada Finland Germany
Greece Ireland Israel
New Zealand United Kingdom United States
9
Due to conservative management and very close supervision, the banks in Israel
exhibited substantial resilience to the developments of the GFC in the global financial
system.
In the past, not only the banking system suffered from a high level of
concentration, but the entire credit market did as well. Yet, the past two decades have
seen changes due to the government reducing its primary borrower role in the
economy. This change shifted private capital from investments in government bonds
to private sector investments, thus easing big companies’ access to credit from the
capital markets, creating a movement toward nonbank credit. In addition, the number
of loans given by institutional investors has grown, thanks to changes in the array of
pension plans in Israel. These loans also contribute to the reduction of the banks’ share
of business and consumer credit (Figure 4).
Figure 4: Banks’ Share of Business and Consumer Credit
Notes: Percentage of consumer and business credit issued by private banks in Israel. Source: Bank of
Israel, Banking Supervision Department.
The domestic credit portfolio among the seven big banks, which includes business
(nonfinancial) credit and credit to households, has been growing since 2001. The
largest component is the household sector, which includes rising levels of housing
loans. In the fourth quarter of 2019, the total amount of credit stood at NIS 970 billion,
the equivalent of approximately 69 percent of Israel’s annual GDP (Figure 5a).
The share of housing credit in GDP has grown in view of increasing house prices.
Consumer credit has remained relatively stable relative to GDP, and credit to large
businesses has declined (Figure 5b).
0.45
0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
0.95
Consumer Credit Business Credit
10
Figure 5a: Bank Credit Balance
Notes: Credit balance of the seven major banks in Israel in NIS billion. Source: Bank of Israel, Banking
Supervision Department.
Figure 5b: Bank Credit to GDP Ratio
Notes: Ratio of business, housing, and consumer credit to real GDP of the seven major banks in Israel.
Source: Bank of Israel, Banking Supervision Department.
0
50
100
150
200
250
300
350
400
450
500
NIS Billion
Business Credit Housing Credit Consumer Credit
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
%
Business Credit Housing Credit Consumer Credit
11
On the side of business credit, the reduction in credit to GDP ratio can be linked to
factors such as (1) requiring banks to reach regulatory capital targets by decreasing
big business credit’s share of total credit, (2) setting limits that should minimize
concentration among borrowers, and (3) increasing access to institutional and market
lending.
2.4 Monetary Policy
Since 2003, monetary policy in Israel has been conducted within the framework of a
“flexible inflation target” regime, with the target set within a 1–3 percent band, and
allowing for temporary deviations from the target range. The exchange rate regime
may be characterized as a “managed float,” as the Bank of Israel has been intervening
in the foreign exchange market since March 2008—at first with pre-announced daily
fixed amounts and since August 2009 in a discretionary manner, according to market
conditions (Caspi et al., 2022).
4
Figure 6: Bank of Israel, Federal Reserve, and ECB Interest Rates
Notes: Nominal interest rates. Sources: Bank of Israel, European Central Bank, and Board of Governors
of the Federal Reserve System.
Since the GFC, domestic monetary policy has been very accommodative. The Bank
of Israel did not set negative interest rates but, as noted above, it did use intervention
in the foreign exchange market and government bond purchases for a short period as
additional tools, and forward guidance to communicate the future monetary policy
path better. The interest rate remained the principal monetary policy instrument. It
4
In addition, the Bank of Israel purchased a pre-announced amount annually as part of a program intended
to offset the effects of natural gas production on the exchange rate. This program ceased at the beginning of
2019.
-1
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Bank of Israel European Central Bank Federal Reserve
12
was set to 0.1 percent in 2015, and increased to 0.25 percent in November 2018 (Figure
6).
Looking farther into the past, the decline in the Bank of Israel’s nominal interest
rate reflected both the adaptation to a lower inflation environment and a decline in
real interest rates, from about 6 percent at the beginning of the 2000s to about 2 percent
a decade into the millennium, and to around zero in recent years, in line with the very
accommodative global monetary policy.
2.5 Regulation and Macroprudential Measures
The debt to GDP ratio in Israel is low relative to other countries. Total debt to GDP is
about 180 percent, with household debt (housing and consumer) to GDP particularly
low at around 40 percent in 2019, with no significant change since 2000. Total business
sector debt (excluding financial institutions) to GDP was about 70 percent in 2019,
with bank credit constituting about half of this credit. Public debt to GDP in 2019 was
around 60 percent.
In the past decade, the risk stemming from the housing market was the main
concern for Israel’s financial stability. Of particular note was the banks’ large exposure
to the housing market, comprising housing credit that has grown considerably over
the past decade (Figure 5a), and credit exposure to the construction and real estate
sectors, which together account for 47 percent of total bank credit. Against the
background of the low interest rates since the GFC, households took a large portion
of housing loans in a floating interest rate, which was cheaper than a fixed interest
rate. However, the floating interest rate exposed households to risks stemming from
a possible rise in market interest rates.
The Bank of Israel began to implement MaP measures at a relatively early stage of
the financial cycle, about two years after the house price cycle started in 2008.
Although the MaP measures targeting the housing sector were substantial, other MaP
classes were also used in Israel during the past two decades. Some of them aimed to
reduce the risks arising from the concentration of the banks’ credit portfolio. Others
aimed to reduce general risks of the banks and were associated with the Basel
regulations. We divide these MaP measures into three major groups.
5
In the following
sections, we refer to the measures taken in each group. A summary of all MaP
measures taken by the Bank of Israel and included in these groups is presented in
Table 1.
2.5.1 Housing Market Measures
MaP measures in the housing sector contain LTV-oriented measures aimed at
influencing the quality of mortgages and borrowers’ risk, and thus the level of risk to
the banks from the mortgage market, and other measures aimed at strengthening the
banks’ resilience to a crisis, such as increased provisions and additional capital
requirements (Claessens et al., 2013; Aiyar et al., 2014).
5
See Section 3.4 for an explanation of the classification.
13
The first measures that were implemented in 2010 were capital requirements for
risky loans, and provisions and restrictions on the characteristics of the loans that can
be offered by the banks were set.
6
In addition, a requirement for a supplementary capital allowance regarding
housing loans was introduced. The risk-weighted assets (RWA) capital allocation
criterion is 100 percent for loans of groups of borrowers who buy properties
collectively. Later, in two consecutive measures (in July and October 2010), the RWA
was set to 100 percent for loans with an LTV greater than 60 percent when the variable-
rate credit component comprises more than 25 percent and the level of the loan is over
NIS 800,000.
Regarding the MaP measure taken in October 2010—restrictions on the level of
LTV—on house prices and borrowers, Baudot-Trajtenberg et al. (2017) found that the
impact on house prices was limited, since only 15 percent of mortgage borrowers were
affected by this move. The measure incentivized riskier borrowers to reduce leverage
with an LTV of over 60 percent. In response to these restrictions, borrowers bought
less expensive homes, farther from the center of the country, and in poorer
neighborhoods.
Between 2011 and 2014, several MaP measures were adopted to strengthen banks’
and households’ stability. Although influencing house prices was not a declared goal
of the Bank of Israel, moderated house prices would have been a welcome outcome.
In 2011, a limitation stating that only up to one-third of the total loan can bear
interest at a variable rate that is adjusted at a frequency of up to 5 years was
established. This measure was introduced to cope with the high demand for variable-
rate loans in view of low interest rates, which led to new loans with variable interest
rates reaching 80 percent of total new loans at that time.
Due to the continuing rise in housing prices and mortgage volumes, additional
measures were implemented to limit exposure to mortgage loans. In October 2012, the
LTV ratio was limited to 70 percent, except for first-time homebuyers (75 percent) and
investors, including nonresidents (50 percent).
In March 2013, risk weights on some housing loans increased, and at the end of
August 2013, four new measures were introduced: the PTI (payment-to-income) ratio
was limited to 50 percent of income; housing loans where the monthly repayment is
over 40 percent of income were weighted at 100 percent of RWA to calculate the capital
adequacy ratio; the portion of the loan at a variable rate was limited to two-thirds of
the total loan, and the portion of the loan at a variable rate that can change within five
years was limited to one-third of the total loan; the loan period was limited to 30 years
(inclusive).
In September 2014, an additional capital allowance requirement was introduced,
amounting to 1 percent of outstanding housing loans, to address vulnerabilities and
boost the banks’ loss absorption capacity. Until 2017, house prices continued to rise
despite the measures taken. The housing market remained very active, mortgage
interest rates continued to decline, and the volume of mortgage lending rose, as did
the volume of housing market transactions.
6
In 2009 banks were warned to inform their customers of risks related to loans with variable interest
rates. This warning is not included in the list of formal MaP measures that were implemented.
14
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 amended.
20/08/2003
31/03/2004
20/08/2003
CON
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 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.
11/07/2010
01/07/2010
11/07/2010
HSE
4
Risk weights increased on housing loans with LTVs above 60%, a floating component of
more than 25%, 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.
28/10/2010
26/10/2010
26/10/2010
HSE
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 up to 70% for those upgrading their homes.
01/11/2012
01/11/2012
01/11/2012
HSE
8
Risk weights on some housing loans are increased, depending on their LTV. Loans with
LTVs between 45% and 60% have a higher risk weight of 50%, and those with LTVs
between 60% and 75% are weighted at 75%.
21/03/2013
01/01/2013
01/01/2013
HSE
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 a variable rate that can change within 5 years from the date of
approval cannot exceed 1/3.
29/08/2013
01/09/2013
01/09/2013
HSE
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
15
Stability Measures
Decision Date
Application Date
Influence Date
Type
13
A. Banking corporations should increase their Common Equity Tier I Capital target
by a rate that 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.
28/09/2014
01/01/2015
28/09/2014
HSE
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 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.
09/06/2015
01/01/2016
09/06/2015
CON
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.
16
In addition to the MaP measures that the Bank of Israel implemented, other factors that
were no less important affected both the supply side, such as government measures to
increase the supply of land available for construction and to increase the number of housing
starts, and the demand side, such as government fiscal measures (taxation).
2.5.2 General Capital Requirement Measures
Another group of MaP measures, based on capital tools, aims to increase the banking
system’s stability and reduce systemic risks. Until 2010 capital requirements were only a
recommendation to the banks. In mid-2010, the Bank of Israel’s Banking Supervision
Department started implementing the Basel II framework and set the core capital ratio target
to a minimum of 7.5 percent. The banks had to achieve the Banking Supervision requirement
by the end of 2010. Naturally, achieving the minimum capital ratio target on time forced the
banks to adjust their core capital ratio right after the requirement was published.
The implementation of Basel III in Israel led the Banking Supervision in 2012 to
announce an update of the core capital ratio requirement to no lower than 9 percent by the
beginning of 2015. Two banks (out of the seven in our sample), classified as large banks,
were required to achieve a core capital ratio of no lower than 10 percent by the beginning of
2017. As before, the adjustment processes started right after the publication of the
requirements. Concerning the excess requirement of the large banks, we assume that the
adjustment processes to achieve the extra 1 percent core capital ratio started right after the
general requirement of 9 percent for all seven banks had to be fulfilled, i.e., the beginning
of 2015.
As part of the Basel III framework, the Banking Supervision Department implemented a
minimum liquidity coverage ratio of 100 percent. Liquidity requirements limit the credit
supply so that banking corporations have to increase the liquidity assets in their balance
sheets. Moreover, an additional leverage ratio surcharge also helps to control the banking
system’s leverage. It may influence the banks’ capital and, as a result, credit supply. In our
analysis, we refer to liquidity and capital tools together as bank stability tools.
2.5.3 Concentration Measures
Our analysis distinguishes between MaP measures targeting lending concentration, i.e.,
measures limiting credit to big borrowers, and general capital requirement measures, since
each of the two groups may affect banks’ credit differently. As a result of the big borrowers’
dominance in the business sector, a MaP directive limiting the indebtedness of borrowers
or groups of borrowers was announced in 2003. This MaP measure limited the banks’ capital
exposures to big borrowers, thus changing the banking sector supply for credit distribution
by increasing the proportions of small and medium businesses and households. In 2015,
more restrictive measures for big borrowers and groups of borrowers were imposed.
3 The Data
3.1 Bank Credit
The bank credit data are taken from the domestic banks’ confidential reports to the Bank of
Israel, submitted to the Banking Supervision Department on a quarterly basis in accordance
with Banking Supervision Directive 831.
17
The data include credit to the public,
7
deposits, and other assets and liabilities, excluding
bonds and securities that were borrowed or purchased under repurchase agreements. The
credit data is disaggregated to commercial credit, which we refer to as business credit,
8
housing credit, and other credit to households (consumer credit).
Other than the credit data, the source for the rest of the banking system’s data is the
publicly available quarterly reports of the banks, describing the level of liquid assets,
liabilities to assets, and deposits to assets. We include cash, deposits, and securities in the
liquid assets category, provided that they are not encumbered to the lenders.
Our panel includes the seven largest banks in the domestic banking sector, which
account for about 99 percent of the sector (see Section 2.3).
3.2 Capital Adequacy
Capital adequacy measures the amount of capital available to a bank to face unexpected
losses that can be incurred due to a realization of risks to which the bank is exposed. The
supervision policy is expressed in two ways: through a minimum capital requirement and
by ensuring the quality of the capital. Besides applying the various Basel rules in Israel, the
Bank of Israel’s Banking Supervision Department establishes further requirements.
Since 2010, the Banking Supervision Department’s requirements have dealt separately
with two sets of capital: total capital (including tier one, tier two, and tier three capital) and
tier one capital, or core capital, on its own—a distinction that preceded the final
recommendation of Basel III.
In our empirical exercise, we use the gap between the ratio of actual capital to risk
components of the banks and the Banking Supervision’s most recent requirement, even if it
had not yet been implemented. For example, in March 2012, the Banking Supervision
published minimum core capital targets, which the banks had to reach by the beginning of
2015. In the paper, the gap for the years 2012–15 is calculated as the difference between the
actual ratio for each period and the target ratio for 2015, with the understanding that the
banks’ policy changed during these years to reach the target.
The source of the capital adequacy data is the financial statements published by the
banks each quarter, yet the requirement for “tier one” capital was declared only in 2010.
Therefore, before 2010, the gap between the ratio of actual capital to risk components of the
banks and the most recent Banking Supervision requirement refers to the “total capital” data
that have been available since the 1990s.
3.3 Monetary Policy Surprises
As is customary in the literature (Gürkaynak et al., 2005; Gertler and Karadi, 2015; Cesa-
Bianchi et al., 2020; Andrade and Ferroni, 2021), we use the unexpected component of the
interest rates as an (exogenous) indicator of monetary policy to avoid endogeneity issues.
There are several possible empirical measurements of monetary policy surprises. The
most common measurement, based on the financial markets’ expectations, was first
analyzed by Gürkaynak et al. (2005) for the US. Although Kutai (2020) suggests a similar
7
During the first quarter of 2011, changes were made in the Banking Supervision 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 dummy variable in the estimation to represent the reporting change.
8
Including agriculture, industry, electricity, water, construction and real estate, commerce, transportation,
information and telecommunications, and other business credit, excluding credit for financial services.
18
methodology for Israel in line with the literature, due to data limitations, the period the
Israeli data covers, starting from 2007, is too short for our estimations. Consequently, the
monetary policy surprises we use in our estimations are derived from the difference
between the interest rate set by the Bank of Israel and the average of all the available
forecasts made by professional forecasters before the interest rate decision, which we
extracted from Bloomberg, starting from 2003.
Following the GFC, we can observe several periods of monetary surprises captured by
both methodologies. Between 2012 and 2014, some monetary policy decisions surprised
both market forecasts and expert forecasts.
Figure 7 shows that the two measures have similar patterns, with a few exceptions,
leading to a correlation coefficient of around 0.84. Our data for monetary policy surprises
derived using expert forecasts span a more extended period than the data constructed using
market-based forecasts. Thus, the correlation justifies our use of monetary policy surprises
based on expert forecasts in our estimations.
We aggregate the surprises to quarterly data by summing surprises related to each
decision during the corresponding quarter.
9
As shown in Figure 7, surprises are almost
absent after mid-2015, in view of the stable interest rate of 0.1 percent since that time.
Figure 7: Market- and Expert-Based Monetary Policy Surprises
Sources: Kutai (2020) and Bloomberg.
Our estimation also includes monetary surprises in the US, and in an alternative
specification, we add surprises in the eurozone (see the Appendix), both presented in Figure
8. As noted above, the Israeli banking system is entirely local, with practically no active
9
Until April 2017, interest rate decisions were made by the Bank of Israel 12 times a year, toward the end of
each month. Since April 2017, the Committee has decided on the interest rate eight times a year.
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
Forecasters Markets
19
foreign-owned banks. Moreover, the share of loans denominated in foreign currency is
relatively small, especially in the consumer and mortgage sectors. Still, as the ability to
substitute local currency credit with credit denominated in (or linked to) foreign currency
exists, we opt to test the importance of this factor on the bank credit growth rate.
Figure 8: Monetary Policy Surprises in the US, Eurozone, and Israel
Source: Bloomberg.
We derive monetary policy surprises in the US and eurozone economies using the same
methodology as for the Israeli economy (Figure 8). As seen in Figure 8, the magnitude of the
ECB’s monetary policy surprises is much smaller than that of the Fed. The correlation
between these measured surprises is about 0.5 between any two measures.
3.4 Prudential Measures
We refer in our estimation to the list of MaP measures summarized in Table 1. We classify
the MaP measures into three groups:
10
MaP measures on the housing market; measures
related to limitations on core capital level, which refer to the general bank credit market;
and measures related to credit concentration. As shown in Table 1, most of the MaP
measures target the housing sector. We cannot identify the importance of each of these
measures ex ante. Therefore, as is well accepted in this literature (e.g., Cerutti et al., 2017),
we assume that each of the measures has the same weight, gather all the measures into one
group, and construct a cumulative index of these measures.
10
Our classification of all MaP measures is based on the IBRN international dataset of MaP measures.
-0.9
-0.7
-0.5
-0.3
-0.1
0.1
0.3 Bank of Israel
-0.9
-0.7
-0.5
-0.3
-0.1
0.1
0.3 Federal Reserve
-0.9
-0.7
-0.5
-0.3
-0.1
0.1
0.3 European Central Bank
20
The first group includes measures aimed at the housing market, and can be characterized
as measures limiting borrowers, such as LTV (loan-to-value) and PTI (payment-to-income)
limits, or as measures that set limitations on the banks, such as differential or excess risk
weights.
The second group includes general capital and liquidity requirement measures, i.e.,
measures mainly referring to the implementation of Basel capital agreements (hereinafter:
“general”); the third group includes measures aimed at lessening banks’ credit
concentration by forcing limitations on credit to big borrowers. Even though the latter two
may be related to the bank’s business credit sector, these groups are separated in the
benchmark estimation since we want to test whether each of these two groups affected
housing and consumer credit differently.
11
We exclude two types of measures related to
reserve requirements on foreign currency-denominated accounts and interbank exposure
limits. These types include only three MaP measures that aimed to address specific issues
and were not supposed to affect the credit sector we investigated.
12
General MaP measures were aimed at increasing the bank’s soundness, which can affect
all bank credit sectors or a specific sector. Reducing the growth rate of business credit may
be the preferred reaction to the general MaP measures, due to their high weight as risk
assets. The concentration MaP measures are aimed at reducing the share of big borrowers
in each bank’s assets, i.e., reducing the bank’s business credit volume.
In the estimation of each of the types of credit—housing, consumer, business, and total
credit—we include the three groups of MaP measures. We expect measures that refer to the
studied credit sector to moderate its growth rate, while MaP measures that refer to other
types of credit, 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.
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 (according to the column “Influence Date”
in Table 1). We allow different time spans for MaP measures to affect the market. For
housing MaP measures, we include the accumulation of measures since the beginning of
the studied period as the indicator variable, lagged one period, as we assume that the
implemented measures continue to affect the market as long as they are in place and reflect
the MaP policy stance. General MaP measures are included in the estimation accumulated
over eight quarters, lagged one-quarter relative to the date they became effective. We use
this accumulated period since after the threshold for capital adequacy is achieved (plus
some buffer), the effect of the restriction on the credit growth rate ceases. Eight quarters is
a period that covers the maximum number of periods between the time a general MaP
measure began influencing banks’ behavior and the time it had to be applied (See Table 1).
All the thresholds were achieved by the banks before the due date, and so accumulating
over a longer period is unnecessary. Once banks achieve the recommended capital
adequacy, the MaP policy stance toward tightening banking risks is fulfilled.
The concentration measures were accumulated over four quarters, lagged one period.
We accumulate four quarters backward since the volume of business credit that a bank
extends, on which these measures are intended to operate, is approved and validated once
a year. Implementation of the concentration restrictions may therefore be fulfilled over one
year after the credit portfolio of all businesses come up for reapproval. For the estimation of
11
Our robustness checks aggregate the general and concentration groups discussed in Section 5.
12
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).
21
the business credit sector, we lag the accumulated measures by four quarters, rather than by
one quarter as in the other sectors. This is because while shifting credit from the business sector
to the other sectors can lead this process, the process of reducing the credit of the business sector
can be done only after the reapproval of a credit portfolio under the new concentration
restrictions.
Table 2 reveals that, with the exception of one MaP measure, all of the MaP measures
were implemented after the GFC. Therefore, there is no need to distinguish in the
estimations between pre- and post-GFC when relating to the MaP measures or their
interaction with monetary policy. It is also evident that the use of housing MaP measures is
more intensive than other measures due to the rapid increases in house prices during these
years (Figures 3 and 4).
Table 2: Macroprudential Measures by Category and Quarter Imposed
Concentration
General
Housing
Sep-03
1
Mar-10
1
Jun-10
1
Sep-10
1
Dec-10
1
Jun-11
1
Jun-12
1
Dec-12
1
Mar-13
1
Sep-13
1
Sep-14
1
Mar-15
1*
Jun-15
1
Notes: *Large banks only.
3.5 Other Macroeconomic Variables
In order to take into account the effect of the business cycle and housing market activity on
credit aggregates, we add control variables such as unemployment rate, change in real
wages, business sector GDP growth, and real change in house prices. A detailed description
of the macroeconomic variables in each specification is presented below in Section 4.
3.6 Sample Period
The estimation is carried out as a weighted cross-section panel estimation, correcting for
cross-section heteroscedasticity and contemporaneous correlation. The estimation is
performed for seven banks over 63 periods (2004:Q2–2019:Q4) for a total of 441
observations. The dependent variable is the log difference of the real quarterly volume of
credit.
Due to the substantial impact of the GFC on policy and the market, we distinguish
monetary policy surprises before and after the GFC, starting from the third quarter of 2008.
4 Estimation and Results
As described in Section 2, the banking sector in Israel is characterized by a small number of
domestic banks for which lending activity is concentrated in the domestic market, and its
22
sources are domestic as well. That is, the lending activity abroad of the domestic banks and
the activities of foreign banks in the domestic market are marginal. Naturally, those qualities
lead us to focus on estimates of the influence of the domestic monetary policy and domestic
MaP measures, and the interaction between them, on the domestic banks’ lending. Because
Israel is a small open economy, we complement the specification with monetary policy
surprises from the US,
13
as described in Section 3.2, in order to allow us to test the effect of
the interest rate differential between domestic and foreign financing on the growth of
domestic credit.
Our first objective is to assess the effect of domestic MaP policies—including their
interaction with monetary policy—on the housing credit (mortgage) market, which is one of
the components of the domestic banks’ total lending activity.
The second component of bank credit is lending to the business sector (excluding the
financial sector). We assess the effect of domestic MaP policies and of the interactions of
these policies with monetary policy on business sector lending.
Nonhousing consumer credit is the third component of bank credit. This credit is not
backed by the borrower’s housing, and is usually used for consumption, but may also
serve—for the banks as credit suppliers, and for households as credit demanders—as a
substitute for mortgages. As a final step, we assess the effect of monetary and MaP policies
on total bank credit, which is comprised of the above three components.
The general formulation of our estimation is
ΔCrediti,b,t = αi + β1iMaPt + β2iMonPolt + β3iMaPt×MonPolt + β4iForeignMonPolt + β5iMacroi,t
+ β6iBankChari,b,t +,
where t spans from 2004:Q2 to 2019:Q4, and b stands for the bank. is the log
change of domestic bank credit of type i (housing, consumer, business, or total credit) by
bank b in quarter t. MaPt are indices of MaP measures and MonPolt and ForeignMonPolt are
the unexpected change in the Bank of Israel and foreign interest rates. The estimated
parameters α and β, macroeconomic control variables (Macroi,t), and bank-specific control
variables (BankChari,b,t) vary by type of credit i. We will elaborate on the specific variables
and lags included in the estimations in the sections describing the results. are cross-
section-weighted and allow for conditional cross-section correlation between the
contemporaneous residuals but are restricted to be uncorrelated between different periods.
The estimation results are presented in Tables 3 to 6.
We choose to separately analyze the components of local bank credit—housing,
business, and consumer credit—in order to study the effect of each of the three classes of
MaP measures on each credit type. While measures targeting housing credit were meant to
moderate its growth in order to minimize the risk to the banking sector’s stability, they may
have induced the substitution of housing credit with other types of credit (such as business
credit), either due to a shift in the demand for this credit due to limitations on borrowers, or
due to supply-side considerations on the 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 and concentration MaP measures: they may have induced a possible indirect
substitution effect alongside the direct moderating effect on business, and possibly
consumer and housing, credit. In order to analyze the direct and indirect effects of the three
classes of MaP measures, we choose to include all three in the estimation of each of the
13
In an alternative specification, we also include monetary policy surprises from the eurozone. The qualitative
results remain unchanged (see the Appendix).
23
separate credit components. In addition, we estimate their effect on the growth rate of total
domestic bank credit, which comprises housing, consumer, and business credit.
As mentioned above, we include in the estimations each of the three MaP housing,
concentration, and general measures accumulated over different lengths of periods. While
we find that the effect of housing MaP 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 dynamics when we
consider the accumulation of the MaP measures over a shorter period into account (Figure
9).
Figure 9: Cumulative MaP Measures
Notes: MaP dummy variables were used in the estimations. Source: Bank of Israel Research Department.
The interaction between MaP measures and monetary policy is specified as the product
of the MaP measures index of dummies and the lagged one- and two-period monetary
policy. Foreign monetary policy is included in the estimations with one and two lags.
For each of the credit types, we present four specifications. Column 1 of each of the tables
includes the basic specification. Column 3 presents the specification including global
monetary policy surprises, and Columns 2 and 4 add the interaction terms between MaP
measures and monetary policy to these specifications. In the specifications shown in
Column 4 of Tables 3 to 6, we represent global monetary policy by including monetary
policy shocks in the US, as most of the foreign currency credit is denominated in dollars.
We present an alternative specification including eurozone monetary policy surprises in the
Appendix. Our qualitative results remain robust to this alternative specification.
In the following estimation results, the significance of the sum of the monetary policy
lags coefficients and the interaction lags coefficients is determined using the Wald test.
0
1
2
3
4
5
6
7
8
9
Cumulative Housing MaP
8-period Cumulative General MaP
4-period Cumulative Concentration MaP
24
4.1 Housing Credit Estimation Results
Table 3 shows that MaP measures relating to the housing market succeeded in reducing the
growth rate of housing credit according to all specifications. General MaP measures
contributed to reducing the growth rate of housing credit, although the results were
significant only in the specifications excluding the interaction terms.
Table 3: Housing Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global
policy
Including
interactions and
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.
We also find that MaP measures targeting the concentration of the banks’ activity
increased housing credit growth (the results are significant only in the full specification,
Column 4). This demonstrates a substitution effect on business credit from the supply side
25
of the banks, increasing mortgage lending as a substitute for business lending, particularly
to large borrowers. We find that positive monetary policy surprises (unexpected tightening)
significantly decreased the growth rate of housing credit, but turned insignificant in the
period after the financial crisis, starting in the third quarter of 2008 (Table 3). For some of
the specifications, the interaction between the general or concentration MaP measures and
monetary policy amplifies the substitution effect on mortgage credit.
Figure 10: Housing Credit Decomposition
Notes: Housing credit is expressed in log difference (dlog). “Other” includes the intercept, bank fixed effects,
dummies to control for seasonality, data break dummies, and other financial ratios. ***, **, and * indicate
significance at 1%, 5%, and 10% level, respectively. Source: Estimation, Table 3, Column 4.
As Israel is a small open economy, we look for the effect of global interest rates,
represented by US monetary policy. We find that higher rates abroad tended to moderate
local housing credit growth before the 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 of local mortgages
may indicate that US surprises are a proxy for expected domestic monetary policy, which
slowed the growth rate of mortgages before the GFC. Another possible interpretation links
the higher foreign interest rate to the actual and expected local currency appreciation. Since
house prices were generally denominated in US dollars before the GFC, an appreciation
would lead, at least in the short term, to a moderation of the local currency value of house
prices and, therefore, of mortgages.
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
0.25
Housing MaP*** Non-housing MaP**
Monetary Policy (MP)*** Housing MaP x MP
Non-housing MaP x MP* Macro controls**
Fed MP** Excess Tier 1***
Post-GFC dummy*** Other
Residuals Housing credit
26
The macroeconomic indicators—real wages, unemployment, and house prices—all have
the expected effect on the mortgage market. Higher prices indicate excess demand in the
market, but also increase the nominal value of a given number of transactions, inducing
higher demand for mortgages.
We also control for bank characteristics that may affect the banks’ tendency to offer
credit. An increase in the banks’ capital relative to the supervisory requirements (lagged
two periods), i.e., a positive gap, tends to increase the growth of housing credit.
Figure 10 presents the average annual contributions of the various factors that affect the
dynamics of housing credit. Housing MaP measures made a substantial downward
contribution to the development of housing credit after 2011. Foreign monetary policy had
some impact on local housing credit in the years before the GFC, and did not contribute to
credit dynamics in the years after the GFC. Macroeconomic variables and bank
characteristics contributed to the housing credit increase after 2009.
The post-GFC dummy reflects post-2008 factors, such as unconventional monetary
policies, that sustained the demand for housing credit. Figure 10 also shows that excess tier
1 capital increased housing credit.
4.2 Business Credit Estimation Results
Although the banks’ share of total business credit declined in the past decade from about 75
percent to 50 percent (Figure 5b), business credit is still an important sector, accounting for
about a 40 percent share of total bank credit.
We find that general MaP measures tended to reduce the growth rate of business credit,
as expected, although the results were significant only in some of the specifications (Table
4). In addition, our results show that MaP measures aimed at housing credit tended to
increase the supply of credit to the business sector, as those are substitutes on the asset side
of the banks’ balance sheets.
We do not find evidence that the concentration MaP measures, which were implemented
only once during our sample period, had any significant effect on business credit.
14
Similar
to our results for housing credit, higher policy rates tended to reduce business credit growth
before the GFC, but this influence diminishes and we find that higher policy rates tend to
increase business credit after the GFC, contrary to our basic intuition. We find that the
interactions between monetary policy and the MaP measures under the full specification do
not significantly affect business credit.
There is partial evidence that less accommodative foreign monetary policy reduced the
growth rate of domestic business credit after 2008. This suggests that the signaling effect of
a surprise in foreign interest rates, assuming one exists, is stronger than the possible
substitution effect on foreign credit. Similar to our results for housing credit,
macroeconomic indicators and bank-specific indicators have the expected effect on the
evolution of the business credit aggregates (Table 4).
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.
27
Table 4: Business Credit, Sample 2004:Q2—2019:Q4, 7 Banks
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global
policy
Including
interactions and
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.
4.3 Consumer Credit Estimation Results
As stated above, consumer credit comprises a relatively small share of the banks’ total
credit, and this share has been relatively constant during the past two decades. Therefore,
our ability to account for changes in consumer credit due to changes in policy or
macroeconomic factors is limited.
28
Table 5: Consumer Credit, Sample 2004:Q2—2019:Q4, 7 Banks
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global
policy
Including
interactions and
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.
The share of variance explained in our estimation is close to zero. Therefore, any
conclusions referring to the drivers of consumer credit are 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 accommodative monetary policy reduced the growth rate of consumer credit
only before the GFC, similar to the effect found for business and housing credit.
The expected effect of MaP measures aimed at housing credit on consumer credit is
ambiguous. MaP measures limiting housing credit or making it more expensive are
expected to increase consumer credit due to substitution effects on both the demand and
29
supply sides. By contrast, if housing (mortgage) credit and consumer credit are
complements, limiting mortgages will also tend to reduce demand for consumer credit.
Looking at the interaction between monetary policy and housing market MaP measures,
we find evidence for housing and consumer credit being substitutes. The interaction term
of these measures with monetary policy is negative, indicating, as for business credit, that
when stricter housing MaP measures accompany accommodative monetary policy,
consumer (nonhousing) credit will tend to increase. On the other hand, coupling a
surprising interest rate hike with accumulated housing market MaP measures intensifies
the effect of monetary surprises on consumer credit.
We find that until 2008 a positive surprise to the global interest rate tended to decrease
the growth rate of consumer credit. After 2008, this effect is not significant. As for housing
credit, interest rate surprises abroad may be an indication of the expected local monetary
policy and therefore affect the growth rate of consumer credit.
4.4 Total Credit Estimation Results
The previous sections analyzed the effect of MaP measures and monetary policy on each
component of bank credit: housing, business, and consumer credit. Although housing MaP
measures are naturally targeted at the housing market, the policy aims to influence the
macroeconomic conditions and financial stability.
Therefore, it is important to examine the effect of these measures on total bank credit.
We repeat our estimation for aggregate bank credit, which is the sum of the three
components mentioned above. Table 6 presents the detailed results.
We find that MaP measures targeted at specific credit sectors, which influence these
sectors, do not effect the dynamics of total credit according to our estimations. The effect of
housing and general MaP measures on total credit is insignificant, and the accumulated
effect of concentration-related measures is also null. As we found in the full specification
(Column 4) for some of the specific credit sectors, monetary policy affected credit growth
only before the GFC. A negative sign on the interaction between monetary policy and
housing MaP measures in the restricted specification (Column 2) indicates that
accommodative monetary policy tended to support total credit growth when coupled with
restrictive housing MaP measures.
Figure 11 shows that monetary policy is not central in explaining total bank credit
dynamics. The interaction between monetary policy and nonhousing MaP measures
decreased total credit growth between 2012 and 2014.
The housing MaP measures made a significant contribution to changes in housing and
business credit. However, the substitution between the types of credit offsets these effects,
resulting in nearly no contribution to changes in total credit. As Figure 11 shows, this
influence was augmented by the interaction of housing MaP measures and monetary policy,
contributing to total credit mainly through the effect of the interaction on consumer credit.
30
Table 6: Total Credit, Sample 2004:Q2—2019:Q4, 7 Banks
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global policy
Including
interactions and
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.
31
Figure 11: Total Credit Decomposition
Notes: Total credit is expressed in log difference (dlog). “Other” includes the intercept, bank fixed effects,
seasonal dummies, data break dummies, and other financial ratios. ***, **, and * indicate significance at 1%,
5%, and 10% level, respectively. Source: Estimations, Table 6, Column 4.
Table 7 presents the main qualitative effect of MaP policy on each of the different credit
components, based on the estimation including both the interaction terms and global
monetary policy, presented in the fourth 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 significantly by the
measures taken to curb housing credit or by the general MaP measures. The effect of
concentration MaP measures is also weaker due to their influence in opposite directions on
credit components.
We may conclude that the effect of MaP measures on total bank credit was less evident
than on each of the credit components due to the substitution effects, probably originating
from both the supply and demand sides. Housing MaP measures were aimed specifically at
the housing market, and housing credit intended to minimize the risks to financial stability
due to distress in this sector did succeed according to our estimation in moderating housing
credit growth.
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
Housing MaP Non-housing MaP***
Monetary Policy (MP)*** Housing MaP x MP
Non-housing MaP x MP*** Macro controls***
Fed MP Excess Tier 1***
Post-GFC dummy* Other
Residuals Total credit
32
Table 7: Summary of MaP Policy Qualitative Effect – Baseline Estimation
Housing
Business
Consumer
Total
Housing MaP
-***
+**
-
+
General MaP
-
-
+
+
Concentration MaP
+**
-
+
+***/-***
Housing MaP × MP
+
-
-*
-
General MaP × MP
+*
+
+
+***
Concentration MaP × MP
+
+
+
+**/-
Notes: ***, **, and * indicate significance at 1%, 5%, and 10% level, respectively. MP stands for monetary policy.
Concentration MaP appears with 1 to 4 lags (+) and 4 to 7 lags (-) according to the specification for housing,
business, and consumer credit. We refer to the estimation including the interaction term and the effect of US
monetary policy, shown in Tables 4 to 6, Column 4.
Other MaP measures targeted at banks’ capital structure may have achieved their goal
of minimizing financial risks by inducing changes in the banks’ balance sheets, while
affecting credit growth only marginally.
5 Robustness Checks
We perform several robustness checks in order to verify our results.
15
First, we check
whether aggregating general and concentration measures into a single index of dummies
matters and find that qualitative results did not change. The substitution effect is still
evident, as housing MaP measures tend to restrict housing credit growth but contribute to
business credit growth and, on the whole, do not affect the evolution of total credit. The
negative effect of housing MaP measures on consumer credit is significant in this alternative
specification, representing possible complementarity between housing credit and consumer
credit.
However, some of the effects of MaP and monetary policy surprises and their
interactions were less significant for the housing and total credit estimations than for the
business and consumer credit estimations. This result is expected since the effects in the
benchmark specification of the general and concentration MaP measures are in opposite
directions in the housing and total credit estimations. However, they are the same in the
business and consumer credit estimations, although both insignificant. Table 8 presents the
robustness checks aggregating both types of nonhousing MaP measures, including
interactions and global monetary policy.
15
We refer in this section to the results for the full specification (Column 4 of each table).
33
Table 8: Robustness Checks, Sample 2004:Q2–2019:Q4, 7 Banks
1
2
3
4
Housing
credit
Business
credit
Consumer
credit
Total credit
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**
D(BusinessGDP(-2))
0.17**
D(log(house_prices(-1))
0.09***
-0.00
Seasonal Dummies
Yes
Yes
Yes
Yes
Data Break Dummies
Yes
Yes
Yes
Yes
C
-0.001
-0.02***
0.01
-0.003
CRISIS2008
0.014**
0.01**
0.01
0.01*
Observations
441
441
441
441
Adjusted R2
0.17
0.31
0.07
0.28
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.
In an additional robustness check, we omit from the estimation the insignificant MaP
measures and find that our qualitative results do not change. We also experiment with
competing models by including each of the MaP measures—housing, general, and
concentration—as a cumulative index, compared to accumulating the general and
concentration measures only over a fixed number of past periods in the benchmark
specification. In this specification, the housing and business credit growth rates are not
(significantly) affected by MaP measures or their interaction with monetary policy, except
for the interaction of monetary policy with the housing MaP measures in the business credit
sector, which becomes significantly negative.
According to this specification, the interaction between accommodative monetary policy
and housing MaP measures tends to increase business credit. In other words, supply-side
34
substitution effects lead banks to increase the supply of business credit more when housing
market MaP measures are accompanied by accommodative monetary policy. For consumer
credit, we find that housing MaP measures become significantly positive and increase
consumer growth. By contrast, general MaP measures accumulated from the beginning of
the estimated period become significantly negative and tend to reduce it. Interactions do
not significantly affect consumer credit growth in this specification.
The last robustness check includes the relative surprise to the interest rate (surprise
divided by the level of the interest rate) in order to take into account the absolute level of
the interest rate relative to the size of the surprise. The results remain qualitatively
unchanged,
16
although the interaction terms with MaP policy become insignificant.
6 Concluding Remarks
This paper analyzes the effect of monetary and MaP policies on bank credit. We present the
results of an analysis of the effect of domestic monetary policy and MaP policy implemented
in Israel during the period 2004–19 on domestic bank credit. We estimate the effect of
monetary policy, MaP measures, and the interaction between them on housing credit,
consumer credit, credit to the business sector, and total credit.
We find that the MaP policy affected the development of bank credit. Our results show
that MaP measures targeted at housing credit did reduce the growth rate of this credit while
contributing to the increased 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 components.
Monetary policy was found to have a clear and significant effect on housing and total
credit growth rates before the GFC, but this effect did not persist after the crisis, at least
according to our estimation. We find some evidence for the effect of monetary policy when
it interacts with MaP measures that were taken after 2008. As the share of foreign currency
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 be a signal for possible future local
policy rather than a substitute.
We find that the interaction between monetary policy and MaP policy is important. We
show that accommodative monetary policy tended to increase consumer credit and total
credit in some specifications, when interacting with macroprudential policies targeting the
housing market.
We also find that macroeconomic conditions, i.e., output growth and employment
growth, are associated with credit growth rates, and that bank-specific characteristics,
particularly the bank’s capital relative to the supervisory requirements, affect the supply of
credit.
References
Aiyar, S., C. Calomiris, and T. Wieladek (2014). “Does Macro-Prudential Regulation Leak? Evidence
from the UK Policy Experiment.” Journal of Money, Credit and Banking, 46(s1): 181–214.
16
In this specification, there is no evidence for the substitution effect of housing MaP measures between
housing credit growth and business credit growth.
35
Altavilla, C., L. Laeven, and J.L. Peydró (2020). “Monetary and Macroprudential Policy
Complementarities: Evidence from European Credit Registers.” CEPR Discussion Papers 15539.
Andrade, P. and F. Ferroni (2021). “Delphic and Odyssean Monetary Policy Shocks: Evidence from
the Euro Area.” Journal of Monetary Economics, 117: 816–832.
Arena, M., T. Chen, S. M. Choi, N. Geng, C.A. Gueye, T. Lybek, E. Papageorgiou, and YS. Zhang
(2020). “Macroprudential Policies and House Prices in Europe.” IMF Departmental Paper.
Baudot-Trajtenberg, N., N. Tzur-Ilan, and R. Frayberg (2017). “Assessing the Impact of
Macroprudential Tools: The Case of Israel.”BIS Papers chapters, in: Bank for International
Settlements (ed.), Macroprudential policy frameworks, implementation and relationships with
other policies, volume 94, pages 107-218, Bank for International Settlements.
Bussière, M., J. Cao, J. de Haan, R. Hills, S. Lloyd, B. Meunier, J. Pedrono, D. Reinhardt, S. Sinha, R.
Sowerbutts, and K. Styrin (2021). “The Interaction between Macroprudential Policy and
Monetary Policy: Overview.” Review of International Economics, 29(1): 1–19.
Cao, J., V. Dinger, A. Grodecka-Messi, R. Juelsrud, and X. Zhang (2021). “The Interaction between
Macroprudential and Monetary Policies: The Cases of Norway and Sweden.” Review of
International Economics, 29(1): 87–116.
Caspi, I., A. Friedman, and S. Ribon (2021). “The Immediate Impact and Persistent Effect of FX
Purchases on the Exchange Rate.” International Journal of Central Banking, 18(5): 165–195.
Cerutti, E., S. Claessens, and L. Laeven (2017). “The Use of Macroprudential Policies: New
Evidence.” Journal of Financial Stability, 28: 203–224.
Cesa-Bianchi, A., G. Thwaites, and A. Vicondoa (2020). “Monetary Policy Transmission in the United
Kingdom: A High-Frequency Identification Approach.” European Economic Review, 123: 103375.
Claessens, S., S. Ghosh, and R. Mihet (2013). “Macro-prudential Policies to Mitigate Financial System
Vulnerabilities.” Journal of International Money and Finance, 39: 153–185.
Davis, E.P., I. Liadze, and R. Piggott (2019). “Assessing the Macroeconomic Impact of Alternative
Macroprudential Policies.” Economic Modelling, 80(C): 407–428.
De Jonghe, O., H. Dewachter, and S. Ongena (2020). “Bank Capital (Requirements) and Credit
Supply: Evidence from Pillar 2 Decisions.” Journal of Corporate Finance, 60: 101518.
De Marco, F., C. Kneer, and T. Wieladek (2021). “The Real Effects of Capital Requirements and
Monetary Policy: Evidence from the United Kingdom.” Journal of Banking & Finance, 133: 106237.
Dell’Ariccia, G., D. Igan, and L. Laeven (2012). “Credit Booms and Lending Standards: Evidence
from the Subprime Mortgage Market.” Journal of Money, Credit and Banking, 44(2–3): 367–384.
Everett, M., J. de Haan, D.-J. Jansen, P. McQuade, and A. Samarina, (2021). “Mortgage Lending,
Monetary Policy, and Prudential Measures in Small Euro-area Economies: Evidence from
Ireland and the Netherlands.” Review of International Economics, 29(1): 117–143.
Forbes, K., D. Reinhardt, and T. Wieladek (2017). “The Spillovers, Interactions, and (Un)intended
Consequences of Monetary and Regulatory Policies.” Journal of Monetary Economics, 85: 1–22.
Gambacorta, L. and A. Murcia (2020). “The Impact of Macroprudential Policies in Latin America:
An Empirical Analysis Using Credit Registry Data.” Journal of Financial Intermediation, 42: 100828.
Gertler, M. and P. Karadi (2015). “Monetary Policy Surprises, Credit Costs and Economic Activity.”
American Economic Journal: Macroeconomics, 7: 44–76.
Greenwood-Nimmo, M. and A. Tarassow (2016). “Monetary Shocks, Macroprudential Shocks and
Financial Stability.” Economic Modelling, 56(C): 11-24.
36
Gürkaynak, R. S., B. Sack, and E. Swanson (2005). “Do Actions Speak Louder Than Words? The
Response of Asset Prices to Monetary Policy Actions and Statements.” International Journal of
Central Banking, 1(1), 55-93.
International Monetary Fund (2013): “The Interaction of Monetary and Macroprudential Policies.”
Kannan, P., P. Rabanal, and A. Scott (2012). “Monetary and Macroprudential Policy Rules in a Model
with House Price Booms.” BE Journal of Macroeconomics, 12(1): 1–44.
Kutai, A. (2020). “Measuring the Effect of Forward Guidance in Small Open Economies: The Case of
Israel.” Discussion Paper 2020.03, Bank of Israel Research Department.
Repullo, R., and J. Suarez (2013). “The Procyclical Effects of Bank Capital Regulation.” Review of
Financial Studies, 26(2), 452-490.
Niepmann, F., T. Schmidt‐Eisenlohr, and E. Liu (2020). “The Effect of US Stress Tests on Monetary
Policy Spillovers to Emerging Markets.” Review of International Economics, 29(1): 165–194.
Yellen, J. (2014). “Monetary Policy and Financial Stability.” The 2014 Michel Camdessus Central
Banking Lecture.
37
Appendix – Estimation Including ECB Monetary Policy Surprises
Table 3a: Housing Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Incl. ECB Monetary Policy
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global
policy
Including
interactions and
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
Aggregated MaP_general × Mon. policy
3.80
5.65*
MaP_concentr.(-1) to (-4) × Mon. policy(-1)
4.08
2.59
MaP_concentr.(-1) to (-4) × Mon. policy(-2)
3.09
5.07*
Aggregated MaP_concentr. × Mon. policy
7.18*
7.66*
Foreign
US monetary policy(-1) until the GFC
-2.58**
-2.53*
US monetary policy(-2) until the GFC
-2.44*
-2.08
US Aggregated before 2008
-5.02***
-4.61**
US monetary policy(-1) after the GFC
-0.27
4.97
US monetary policy(-2) after the GFC
0.69
0.96
US Aggregated after 2008
0.41
5.93
ECB monetary policy(-1) until the GFC
24.93**
32.51**
ECB monetary policy(-2) until the GFC
-15.00
-15.14
ECB Aggregated before 2008
9.93
17.37
ECB monetary policy(-1) after the GFC
2.96
2.22
ECB monetary policy(-2) after the GFC
-1.02
--0.94
ECB Aggregated after 2008
1.94
1.27
Banks^
Excess Tier(-1)
0.38***
0.40***
0.59***
0.61***
Macro
D(real wage(-2))
0.12*
0.14**
0.18**
0.16*
D(unemployed 25-64(-5))
-0.77***
-0.61**
-0.64**
-0.57*
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
-0.003
-0.001
CRISIS2008
0.01**
0.02***
0.02***
0.02***
Observations
441
441
441
441
Adjusted R2
0.17
0.17
0.19
0.20
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.
38
Table 4a: Business Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Incl. ECB Monetary Policy
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global
policy
Including
interactions and
global monetary
policy
Cumulative MaP_housing(-1), since 2004
0.001**
0.001**
0.001**
0.001**
MaP
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**
D(Liabilities(-1)/Assets(-1))
-0.73
-1.06*
-1.02*
-1.21**
Macro
D(BusinessGDP(-2))
0.22***
0.20***
0.17*
0.17*
D(unemployed 25-64(-2))
-1.55***
-1.77***
-1.79***
-1.66***
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***
0.02***
Observations
441
441
441
441
Adjusted R2
0.30
0.30
0.30
0.30
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.
39
Table 5a: Consumer Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Incl. ECB Monetary Policy
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global
policy
Including
interactions and
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.
40
Table 6a: Total Credit, Sample 2004:Q2 – 2019:Q4, 7 Banks, Incl. ECB Monetary Policy
1
2
3
4
Basic – excl.
interactions
and global
policy
Including
interactions
Including
global policy
Including
interactions and
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.
