Recent experience has made it clear that housing markets are a major source of financial instability. This recognition has spawned a great deal of debate among policymakers and academics regarding the appropriate policy response. Monetary policy is one option. However, econometric estimates of interest rates' effects on house prices are rather small.1 Some, including Bernanke (2010), have therefore argued that an interest rate hike of sufficient magnitude to meaningfully restrain house-price growth would adversely affect other sectors of the economy and increase the likelihood of a recession.
The recognition of the limitations of interest rate policy has prompted a search for other policies to complement monetary policy in the stabilisation of housing markets. In June 2014, for example, the Bank of England imposed limits on banks’ exposure to loans exceeding 4.5 times household income. And in September 2013, the Swiss National Bank increased capital requirements for mortgages.
Many of these policies fall under the macroprudential rubric. Other tools, such as taxes and reserve requirements, have also been used. To distinguish them from conventional monetary policy and emphasise that they are a superset of macroprudential tools, we refer to these instruments collectively as non-interest rate policies.2
Drawing primarily on the dataset compiled by Shim et al. (2013), this article describes the usage of these non-interest rate tools in 60 economies, for time spans of up to four decades. In addition, it summarises some of the recent evidence on the policies’ efficacy in achieving their second objective, the dampening of housing cycles.
The two faces of housing cycles
Prices and credit are intimately intertwined. In the dataset used in Kuttner and Shim (2013), the contemporaneous correlation between price and credit growth is approximately 0.5. In a predictive sense, the causality between price and credit goes both ways. Vector autoregression analysis shows that rapid credit growth portends future appreciation. Symmetrically, rapid house price growth tends to precede credit expansions.
Figure 1 shows time series plots of real house prices and credit for Switzerland and the US. Although their boom-bust cycles occurred at different times, they exhibit similar patterns. In both countries, surges in credit issuance coincided with rapid price appreciation.
Figure 1. Real house price and credit growth in Switzerland and the US
One explanation for the strong co-movement is simply that both are driven by the same housing market fundamentals, such as household incomes and demographics. If this were always the case, a laissez-faire policy would be appropriate. But as recent experience has shown, the apparent positive feedback between prices and quantities can be highly destabilising, providing a more compelling rationale for policy intervention.
The use of non-interest rate policies across countries and over time
Research on the effects of non-interest rate policies has until recently been constrained by a lack of data. Although some countries have been using them for decades, the systematic documentation of their usage dates back only a few years.
The IMF has conducted two surveys of macroprudential policy usage. The 2010 IMF survey covered 49 countries for the 2000-2010 period. The second survey was conducted in 2013-14 and included information on twelve types of measures taken by 107 economies from 2000 to 2013.3
The database compiled by Shim et al. (2013), based primarily on official documents from central banks, regulatory authorities and ministries of finance, contains a large number of non-interest rate policy actions that were not taken for explicitly macroprudential purposes. The publicly available Shim et al. database encompasses 60 economies, in some cases extending as far back as 1990.
The dataset contains information on five different types of prudential policies: limits on loan-to-value and debt-service-to-income ratios, risk weighting and loan-loss provisioning requirements and exposure limits.4 Table 1 gives a breakdown of the 246 policy actions by type. Loan-to-value limits were used more frequently than any of the other policies.5 Risk weights on housing loans and maximum debt-service-to-income ratios were also used frequently. The 13 economies in the Asia-Pacific region were the most active users of prudential measures among all six regions. Table 2 shows that these prudential policy tools have been used with increasing frequency over the years. Over the entire 1990 through June 2012 period, 178 of the 246 actions can be classified as tightening (introduced to decrease credit) and 78 as loosening.
Table 1. Use of prudential policies by region
1 The values in brackets show the average number of policy actions per country per decade. 2 The figures in square brackets indicate the number of economies in each region.
Table 2. Use of prudential policies over time
1 When we calculate the number of policy actions, we first divide the total number of policy actions taken by all economies in a decade by the sum of the number of coverage years for each economy in the decade, and then multiply the average number of actions per country per year by 10 to rescale it to the number of actions taken in a decade.
Source: Reproduced from Shim et al. (2013).
The dataset also contains a wealth of detailed information on policy actions. For example, Table 3 shows that 55% of the 94 loan-to-value-related policy actions were to change the maximum ratio, while the other 45% were to introduce or abolish a maximum loan-to-value ratio or to prohibit certain types of housing loans. The average size of decreases in the maximum loan-to-value ratio was 11.2 percentage points, while that of increases was 14.1 percentage points. Similarly, Table 4 shows that 17 economies have taken 45 debt-service-to-income -related measures over the same period. The majority of debt-service-to-income -related measures were to introduce or abolish a maximum debt-service-to-income ratio.
Table 3. Distribution of policy actions related to LTV ratio rules
Source: Shim et al. (2013); authors’ calculation.
Table 4. Distribution of policy actions related to DSTI ratio rules
Source: Shim et al. (2013); authors’ calculation.
Evidence on the effectiveness of non-interest rate policies
As more data on non-interest rate policies have become available, a number of studies have attempted to gauge the policies’ effects on economy-wide leverage and credit growth. Lim et al. (2011) found that reserve requirements, dynamic provisioning, maximum loan-to-value and debt-service-to-income ratios and limits on foreign currency lending had measurable effects on the growth rate or cyclicality of private sector credit. Using a sample of approximately 2,800 banks in 48 economies from 2000 to 2010, Claessens et al. (2014) showed that maximum loan-to-value and debt-service-to-income ratios as well as limits on credit growth and foreign currency lending reduce bank leverage and asset growth during booms, but that few policies help stop declines in bank leverage and assets during downturns.
A few papers have also looked specifically at non-interest rate policies’ effects on housing markets. Crowe et al. (2011) and Cerutti et al. (2015) both concluded that policies such as loan-to-value limits are most likely to restrain real estate booms.
Kuttner and Shim (2013) systematically investigated the impact of non-interest rate policy measures on housing credit and house prices, using data from 57 economies over three decades. Of the five prudential policies considered, they concluded that changes in the maximum debt-service-to-income had the largest and most robust effects on housing credit growth. Their estimates in Table 5 indicate that a typical tightening action reduces real credit growth by 4-7 percentage points over the subsequent four quarters. Other policies, including changes in the maximum loan-to-value ratio and exposure limits also seem to affect credit growth.6 However, because of co-linearity, the loan-to-value variable becomes insignificant in the presence of the debt-service-to-income variable. None of the prudential policies had effects in housing prices that were both economically and statistically significant, however.
Table 5. Prudential policies’ estimated effects on real housing credit growth
Notes: The table shows the average response over four quarters of a typical policy action, defined as a dummy variable that takes on the value of 1 for a tightening and –1 for a loosening. Asterisks denote statistical significance: *** for 1%, ** for 5% and * for 10%.
Source: Reproduced from Kuttner and Shim (2013).
One important finding of Kuttner-Shim (2013) is that restrictions targeting specific characteristics of loans have larger effects on credit growth than those oriented towards bank capital. This may be because reducing the maximum loan-to-value or debt-service-to-income ratio directly affects households’ ability to borrow. Increasing capital requirements, on the other hand, works by increasing banks’ cost of funds, which is likely to be a secondary consideration in hot housing markets.
Another interesting finding of the Kuttner-Shim research is that changes in the maximum debt-service-to-income ratio appear to be more effective than adjustments to the maximum loan-to-value ratio. One possible explanation is that even with an unchanged loan-to-value ratio, house price appreciation enables a larger volume of borrowing. Thus, loan-to-value requirements are least effective when they are most needed. Debt-service-to-income limits, on the other hand, are unaffected by house price appreciation (although falling interest rates make them less binding). For this reason, capping the debt-service-to-income ratio may be the tool of choice for slowing credit growth during housing booms.
Conclusions and lessons for policy
With price stability and the mitigation of output volatility already in their mandates, central banks are eager to develop additional tools to further their financial-stability objectives. Because they can be deployed independent of monetary policy and targeted narrowly at housing markets, macroprudential and other non-interest rate measures are ideal for this purpose. Consequently, policymakers are pinning their hopes on these tools as a solution to their financial stability problems.
The research surveyed in this article indicates that certain types of macroprudential measures may be useful additions to the policy toolbox. The evidence is far from definitive, however, given the relatively small number of observations for each of the different types of policy. Our conclusions echo those of Tarullo (2015), who observed that “[e]xperience with macro-prudential policy measures in various countries is not extensive and may, in any case, have only limited applicability elsewhere because of differences in economic conditions, the relative importance of capital market and traditional bank intermediation, and many other factors”. We therefore concur with Borio (2014) that it would be unwise to rely solely on macroprodudential policies for taming financial booms and busts.
Disclaimer: The views expressed herein are those of the authors and are not necessarily those of the Bank for International Settlements.
Bernanke, B (2010), “Monetary Policy and the Housing Bubble”, speech delivered to the Annual Meeting of the American Economic Association, 3 January.
Borio, C (2014), “Macroprudential Frameworks: (Too) Great Expectations?”, Central Banking Journal, August.
Cerutti, E. S Claessens, and L Laeven (2015), “The Use and Effectiveness of Macroprudential Policies: New Evidence”, IMF Working paper 15/61, March.
Claessens, S, S R Ghosh and R Mihet (2014), “Macro-prudential Policies to Mitigate Financial System Vulnerabilities”, IMF Working Paper 14/155, August.
Crowe, C W, D Igan, G Dell’Ariccia, and P Rabanal (2011), “How to Deal with Real Estate Booms”, IMF Staff Discussion Note 11/02.
Kuttner, K N (2011), “Monetary Policy and Asset Price Volatility: Should We Refill the Bernanke-Gertler Prescription?” in E Kaufman and Malliaris (eds.) New Perspectives on Asset Price Bubbles: Theory, Evidence, and Policy, Oxford: Oxford University Press.
Kuttner, K N and I Shim (2013), “Can Non-interest Rate Policies Stabilise Housing Markets? Evidence from a Panel of 57 Economies”, BIS Working Paper no 433.
Lim, C H, F Columba, A Costa, P Kongsamut, A Otani, M Saiyid, T Wezel, and X Wu (2011), “Macroprudential Policy: What Instruments and How to Use Them? Lessons from Country Experiences”, IMF Working Paper 11/238.
Shim, I, B Bogdanova, J Shek and A Subelyte (2013), “Database for Policy Actions on Housing Markets”, BIS Quarterly Review: 83−95, September.
Tarullo, D (2015), “Advancing Macroprudential Policy Objectives”, Remarks at Office of Financial Research and Financial Stability Oversight Council’s 4th Annual Conference on Evaluating Macroprudential Tools: Complementarities and Conflicts, Arlington, Virginia, 30 January.
Vandenbussche, J, U Vogel and E Detragiache (2012), “Macroprudential Policies and Housing Prices. A New Database and Empirical Evidence for Central, Eastern and Southeastern Europe”, IMF Working Paper 12/303.
1 See Kuttner (2011) for a survey of the relevant evidence.
2 These polices have two complementary goals. The first is to enhance the resilience of the financial system to the inevitable housing downturn. Some of the tools, such as provisioning requirements, are intended to promote the build-up of buffers that would cushion financial institutions from declining asset values. The second goal is to attenuate price and credit fluctuations, thus reducing the likelihood of damaging booms and busts.
3 Detailed information has also been compiled for specific regions or sets of countries. Vandenbussche et al. (2012) for example, compiled a database of various types of actions taken by central and eastern European countries.
4 The dataset also includes information on changes in reserve requirements, limits on credit growth, and changes in housing-related taxes and subsidies.
5 These actions also included outright prohibitions on loans of certain types (such as mortgages to purchase third homes) or those exceeding a certain loan-to-value limit.
6 The results for exposure limits should be treated with caution, however, since there are only 12 such actions in the dataset.