In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.

Abstract

In recent years, the IMF has released a growing number of reports and other documents covering economic and financial developments and trends in member countries. Each report, prepared by a staff team after discussions with government officials, is published at the option of the member country.

Macroprudential Policies in Sweden—An Overview1

Sweden’s banking system meets most standard measures of financial soundness. But with its very large and wholesale-dependent banking sector, high and increasing household debt, and resurgent house price growth, additional measures are needed to contain mounting financial stability risks. On the supply side, this means continuing to strengthen capital and liquidity requirements. However, theoretical and empirical evidence points to a need to also limit credit demand, including through effective steps to increase the rate of mortgage amortization.

A. Financial Stability Assessment—An Update

1. After a brief period of uncertainty at the beginning of the crisis, the Swedish banking system performed well. The May 2013 Riksbank stress tests suggest that major banks can absorb direct loan losses triggered by large macroeconomic shocks. This resilience stems from banks’ strong earnings as well as a history of low default and loss-given-default rates. While the return on equity is still lower than during the pre-crisis period—reflecting, in part, the low interest rate environment—the gross margin on new mortgages now exceeds historical values and Swedish banks remain profitable compared to many European peers.

A01ufig1

Swedish Banks’ Return on Equity

(Percent, rolling 4Q)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.
A01ufig2

Gross Margin on the Major Swedish Banks’ New 3-month, Fixed-rate Mortgages

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.

2. However, Swedish banks are very large and wholesale-dependent, and high and increasing household debt and resurgent house prices add to financial stability risks.

  • In the context of household debt at around 175 percent of disposable income and standard indicators placing house prices 20 percent above fundamentals or higher, an unexpected drop in housing values would affect banks not only through rising non-performing loans (NPLs) and uncertainty in financial conditions, but also through household deleveraging and its knock-on effects on investment and corporate balance sheets. Staff calculations suggest a housing price correction in Sweden by 20 percent would reduce GDP by 2.6 percent, consumption by 3.7, and residential investment by 60 percent.2

  • The Swedish banks increasingly finance their liabilities from wholesale funding sources.3 The share of wholesale funding has increased from around 20 percent in 1996 to above 40 percent in 2014, with over 60 percent of the wholesale funding obtained in foreign currency. This practice introduces significant market and foreign liquidity risks that could quickly become systemic. Moreover, it is difficult to exclude scenarios where these risks could add to potential problems related to household credit (see above). Such a combination would put Sweden’s banks under significant stress, with potentially drastic consequences throughout the Nordic region (see IMF 2013).

A01ufig3

Sweden: Estimated Range of Impact

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Igan and Loungani (2012), IMF (2013) and Fund staff calculations.Note: This chart reports the estimated range of impact of a house price drop. The average impact is computed as a product of the point estimate from the impulse response function (IRF) and a decline in house prices implied by staff’s average estimate of house price gaps. The range is computed using one standard deviation from the IRF and staff’s end point estimates of house price valuation gaps.
A01ufig4

Swedish Banks’ Funding Structure

(SEK, trillions)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Statistics Sweden, Sveriges Riksbank and Fund staff calculations.

3. In principle, a range of potential macroprudential policy tools can be applied to address these and related risks. In particular, the European Systemic Risk Board (ESRB) emphasizes measures that operate on the banking (supply) side—such as strengthening capital buffers and liquid funding structures—and tools to influence credit demand.

Table 1.

ESRB Recommendations: Objectives and Macroprudential Instruments

article image
Sources: ESRB and Fund staff calculations.

B. Recent Reforms

4. Recent policy reforms in Sweden have focused on the supply side. The authorities have strengthened CET1 capital levels exceeding Basel III requirements ahead of many peers. At the same time, leverage ratios have remained at levels similar to other European banks (see text figure). Similarly, Swedish banks have been asked to fully implement the Liquidity Coverage Ratio (LCR) while the Riksbank’s measure of the banks’ structural liquidity position is still below the average of European peers.4 Moreover, banks’ reporting standard needs to be further improved, as transparency requirements are part of the Basel III regulations (Ingves, 2014).5

5. The track record regarding demand-side measures is weaker. Compared to many other European countries, Swedish households enjoy a low property tax rate along with income tax-deductable mortgage interest rate payments. For example, the Finnish government recently announced plans to reduce the share of mortgage interest expenses that can be deducted to 50 percent by lowering it 5 percentage points per year until 2018. At 85 percent, the LTV requirement introduced in 2010 is relatively high, and much higher compared with countries outside the Europe. In addition, some European countries are planning to reduce LTV ratios; for example, in the Netherlands, the Wijffels commission recently recommended a reduction of the LTV to 80 percent. Moreover, the average Swedish mortgage has a very long amortization period. To the extent that mortgages are amortized at all, a typical loan with a LTV ratio below 75 percent matures within 43 years. However, many mortgages are not amortized at all. Introducing minimum amortization requirements was a key policy recommendation in the 2013 Article IV consultation (see Table 2). Other countries have taken measures to limit the maximum amortization periods. For instance, the Monetary Authority of Singapore introduced a cap on maximum tenure of new mortgage loans at 35 years in 2012.

Table 2.

Enhancing Financial Stability: Progress Compared to the 2013 Article IV Recommendations

article image
Sources: IMF Sweden Article IV Consultation 2013 and Fund staff calculations.
A01ufig5

Demand and Supply Side Measures

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: ECB, OECD, Sveriges Riksbank and Fund staff calculations.Note: For each measure, countries that are further away from the origin, the better positioned they are. LCR, CET 1, structural liquidity and leverage ratio are as of 2012. Amortization rate is calculated as 1 over the average mortgage loan maturity as of 2007. Property tax revenue is 2012 data, and expressed as share of GDP. Share of fixed rate new mortgagesdata are as of 2007 for Europrean countries, and Sept. 2013 for Sweden . European countries include Belgium, Germany, Ireland, Grece, Spain, France, Italy, Cyprus, Malta, Netherlands, Austria, Portugal and Finland, however, data can vary depending on availability.
A01ufig6

Limits on LTV Ratios

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: National authorities and Fund staff calculations.Note: When calculating the sample average LTV, the mid-point of LTV ranges was used.

Supply Side: Progress But Not Complete

6. At around 4.1 percent, the leverage ratio for Swedish banks has improved little in recent years, despite significant improvements in risk weighted capital ratios.6 Banks’ risk weighted capital ratios have doubled since 2008, while the core Tier 1 capital as share of total assets has improved by 27 percent. Although the leverage ratio is above the minimum requirement listed by the Basel III committee, it remains low compared to the U.S., where the largest banks are asked to have leverage ratios no less than 5 percent.7

A01ufig7

Swedish Major Banks’ Core Tier 1 Capital Ratios and Tier 1 Capital in Relation to Total Assets

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.

7. Core Tier 1 capital ratios are relatively high and have been improving further—but primarily driven by changes in risk weights. While Swedish banks’ core Tier I capital buffers improved from 11.8 to 16.8 percent between December 2010 and December 2013—significantly above major EU bank’s average of 12.3 percent—only 0.9 percentage points came from higher capital while changing average risk weights contributed 4 percentage points.8

A01ufig8

Development of Major Swedish Banks’ Core Tier 1 Capital Ratios, Basel III

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.Note: Weighted average of the major banks’ core tier 1 capital ratios for December 2010 as a blue column and for December 2013 as an orange column. The intermediate columns show how different factors have contributed the change.
A01ufig9

Risk Weights

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: EBA, FSA, Statistics Sweden and Fund staff calculations.Note: The assumed risk weight for mortgages prior 2007 is estimated as the weighted average of the 50 percent risk weight applied to detached house and the 100 percent risk weight applied to tenant-owned apartments. The weights are calculated as the share of mortgages to single family dwellings for detached house and all other types of dwellings as tenant-owned dwellings.

8. Reductions in corporate loan risk weights played a key role in this regard.9 In anticipation of the current minimum requirement imposed by the FSA in May 2013, major Swedish banks increased the risk weights on mortgages to above 15 percent during 2012. However, simultaneously, banks lowered the average risk weights on corporate lending from 45 percent to 32 percent. A simple calculation suggests that this shift in risk weights alone—without any change in equity buffers—improved the average risk weighted capital ratio by about 3½ percentage points, similar to what was observed in the Swedish banking system between 2010 and 2013.10

9. This points to the fact that risk weight adjustments and leverage are related. Increasing the risk weight on mortgages has two implications: for any given lending portfolio, it reduces the total amount of lending banks can make with a given capital base, and it tilts banks’ lending incentives toward other lending (see the shift from point A to point B in the text chart). To attain their original level of lending, banks can either decrease risk weights on other lending categories (e.g., corporate risk weights—see point C), or raise additional equity (see point D). The latter would improve banks’ leverage ratio and involve a lending portfolio with a smaller proportion of the relatively riskier assets (mortgage lending). By contrast, unless reductions in corporate risk weights reflect changes in the underlying riskiness, the former would represent a distortion in lending.11

10. With little evidence that the Swedish corporate risk profile improved recently, strengthening the leverage ratio would have advantages. Adaption of advanced internal rating-based (IRB) models contributed to the reduction in corporate risk weights; while there is little evidence suggesting the underlying riskiness of corporate lending has changed. In fact, corporate bankruptcy rates have been relatively stable after the crisis and the Baltic exposure of Swedish banks has declined well before 2012. This would suggest limited scope to preserve overall lending levels given higher mortgage risk weights through decreasing corporate risk weights. Instead, an improvement in the leverage ratio would allow banks to preserve lending levels, in addition to increasing bank’s shock-absorbing equity buffer more generally, which increases the resilience of the banking system as a whole.

A01ufig10

Loan Risk Weights and Lending Decisions

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Source: Fund staff calculations.Note: The figure depicts a bank’s optimal lending allocation between corporate and mortgage loans, subject to the constraint of a risk-weighted capital ratio. With an initial lending portfolio A, an increase in mortgage risk weights implies an inward-tilt of the capital constraint and a new optimal portfolio B. where both mortgage and corporate lending are lower. To attain its original lending volume, the bank can either decrease risk weights on corporate lending, implying portfolio C, or raise its equity (i.e., improve its leverage ratio), with portfolio D as a result.
A01ufig11

Default Rate for Swedish Companies

(Percent of companies)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.
A01ufig12

Swedish MFIs’ Consolidated Claims in Baltic States

(SEK, billions)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Statistics Sweden, Sveriges Riksbank and Fund staff calculations.

11. Reflecting the size of their balance sheet, Swedish banks are highly dependent on wholesale funding. With a loan-to-deposit ratio of about 200 percent, around 50 percent of bank funding comes from wholesale sources. Bond loans and certificates provided roughly 20 percent and 16 percent of the outstanding wholesale funding in March 2014, respectively, out of which two-thirds of bonds and almost all of certificates are denominated in foreign currency.

A01ufig13

Outstanding Bonds Issued by Banks

(SEK, trillions)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

sources: Statistics Sweden, Sveriges Riksbank and Fund staff calculations.
A01ufig14

Outstanding Certificates Issued by Banks

(SEK, trilllions)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Statistics Sweden, Sveriges Riksbank and Fund staff calculations.

12. Structural liquidity has improved, but only gradually. Swedish banks already fulfill the liquidity coverage ratio requirement under Basel III. However, while lending largely consists of mortgages with long maturities, between 30 and 50 years, funding is predominantly through market instruments with an average maturity of four years.12 This implies that fulfilling the NSFR requirement is more difficult, and correspondingly measures of structural liquidity point to the presence of risks in the system. Based on the Riksbank’s measure of structural liquidity (which is conceptually akin to the NSFR prior to the revision in early 2014), the major Swedish banks improved their position in the aftermath of the global financial crisis—largely on the back of easy access to global funding markets. However, the speed of improvement has slowed in recent years and the levels are low compare to other European banks.

A01ufig15

Major Swedish Banks’ Structural Liquidity Measure

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.
A01ufig16

The Riksbank’s Structural Liquidity Measure

(December 2013, percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank.Note: The blue bars show a sample of European banks.

13. On May 8, 2014, the FSA announced plans for additional supply side measures, including the introduction of the Basel III requirements. The FSA plans to introduce a countercyclical buffer and to implement, for the four largest banks, a 3 percent CET1 systemic risk buffers as well as a 2 percent CET1 buffer within the Pillar 2 framework. According to the FSA, assuming a counter-cyclical buffer of 1.5 percent, most banks already fulfill the CET1 requirement as of 2014:Q1. As a result, the FSA announcement had little market impact, and the four major banks’ stock prices ended up higher at the end of the day. That said, if the counter-cyclical buffer was set at or close to 2½ percent, additional capital would be required. In addition, the FSA confirmed that risk weights for mortgages will be increased to 25 percent which market analysts expect could lead to a 20 basis points increase in mortgage rates.13

14. BIS methodology suggests the level of the countercyclical buffer should be set at or close to 2½ percent, the maximum value subject to international reciprocity. The buffer will be decided in the Fall 2014 after the FSA has consulted with the Riksbank, Ministry of Finance and the National Debt Office in the Financial Stability Council. Based on BIS methodology, credit growth in Sweden is much higher than its historical trend, indicating a positive credit gap above 6 percent of GDP at end 2013 (see text figure), with much of it driven by mortgage lending. Given recent upward trends in mortgage lending and house prices plus a persistent positive credit gap since 2005, these suggest a buffer at or close to the upper end of the maximum value subject to international reciprocity.14

A01ufig17

Estimated Credit Gap

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.

Demand Side: The Factors Driving Swedish Household Debt

15. Swedish household debt is high and increasing. Household debt as share of disposable income approached 175 percent at end of 2013, and over 190 percent if debt from tenant-owned housing associations is included. The continued willingness of bank to lend notwithstanding, strong household demand fostered by low interest rates and expectations of further house price increases in the future is a key factor driving the continued growth of the household debt stock.

A01ufig18

Swedish Household Indebtedness and Housing Price Expectations

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: SEB, Sveriges Riksbank and Fund staff calculations.Note: The debt ratios are expressed as shares of household disposable income. House price expectations are defined as the share of households believing housing prices will rise less the share of households believing they will fall
  • Due to very low rates of property taxation and the income tax deductibility of mortgage interest rate payments, there are strong tax incentives to save in the form of housing rather than other assets. The link between property taxation and house prices is well established. For example, Crowe and others (2011), using US state-level data, show that a one standard deviation ($5 per $1000 of assessed value) increase in property tax rate is found to be associated with a 0.9 percentage point decline in average annual house price from an annual growth of around 5.6 percent per year. (See Appendix 2 for a summary of selected papers in the existing literature).

  • After a long period of very low interest rates, households have taken on more debt with variable interest rates. Between 2012 and 2013, the share of variable interest rate mortgages has increased from around 55 to 67.2 percent and Swedish households expect mortgage rate to decline further.

  • Related to these developments, mortgage amortization periods have declined. Between 2007 and 2012, the share of unamortized mortgage loans has increased from 63 to 71 percent of new mortgages issued.15 This runs contrary to trends elsewhere—for example, the share of unamortized mortage loans has declined in the UK, where interest-only advances to residential mortgages fell from about 50 percent in 2007:Q3 to around 45 percent in 2008:Q3.

A01ufig19

Fixed-Rate Periods in Sweden for New Mortgages

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Statistics Sweden and Fund staff calculations.
A01ufig20

Households’ Mortgage Rate Expectations

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: National Institute of Economic Research (NIER)Note: The households’ mortgage-rate expectations refer to expectations regarding the variable mortgage rate.
A01ufig21

Interest-only Mortgages in Sweden

(Percent of new loans)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Sveriges Riksbank and Fund staff calculations.Note: The figure refers to mortages with loan-to-value ratio below 75 percent.

16. Rising household credit demand reflects rising house prices, in part resulting from very limited construction activity. Housing construction decreased substantially after the crisis in the early 1990s and remained subdued since then. Several factors are constraining the supply, including the monopoly enjoyed by local municipalities over planning and zoning, a highly regulated rental market,16 lengthy permit processes, and strict zoning and environmental regulations. The limited expansion of housing supply contributed to rising house and apartment prices, especially in metropolitan areas such as Stockholm, and in turn continued to push up credit demand by households.

A01ufig22

Number of Dwellings Completed

(Thousands)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Statistics Sweden and Fund staff calculations.
Table 3.

Demand-Side Issues

article image
Source: Fund staff calculations.

17. If progress under the current voluntary approach remains limited, a binding maximum amortization period for new mortgages would be helpful. Swedish banks are working to raise amortization through voluntary means. However, the picture remains largely unchanged. The share of new mortgages granted in 2013 that has an amortization plan has remained broadly constant from 2012. And, of the existing mortgage stock, a significant 40 percent of households either increased or did not decrease their debt in 2013. For the remaining 60 percent of households, it would take 100 years to fully pay down their debt if they continued to amortize at current speeds.

18. Lower LTV caps can contribute to address still high loan-to-value (LTV) ratios. The share of mortgages with LTV ratios exceeding 85 percent has significantly decreased after the introduction of the binding LTV cap in 2010, while the share of mortgages with a LTV ratio between 76 and 85 percent increased, indicating that there are strong incentives to keep the down payment to the minimum required by the LTV cap. The still relatively high LTV ratios, in turn, tend to lengthen amortization periods (as households seek to spread out higher mortgage payments), which leads to rising household debt. This suggests that lower, more binding LTV caps can play an important role in limiting debt accumulation. However, they do not necessary contain debt in relation to incomes if house values are rising fast. Also, evidence suggests that Swedish households are turning to additional (unsecured) loans to comply with the LTV maximum requirement.

A01ufig23

Loan-To-Values of Outstanding Mortgages

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Finansinspektionen and Fund staff calculations.Note: The loan-to-value ratio sare weighted by volume.
A01ufig24

Sweden: MFIs’ Unsecured and Consumption Lending to Households

(SEK, billions)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Statistics Sweden, Sveriges Riksbank and Fund staff calculations.

19. Debt service-to-income (DSTI) caps and binding amortization requirements are important complementary elements in the macroprudential toolkit. DSTI requirements tie households’ debt levels to their income level, independent of house values, and can work as an automatic stabilizer that helps contain excessive increases in credit (and asset prices) relative to income. They can also increase households’ resilience to income and interest rate shocks, complementing the LTV cap that affords resilience to asset price shocks. In addition, DSTI limits can help address arbitrage of LTV limits through secondary loans. And minimum mortgage amortization requirements, in addition to moderating loan demand, would improve the effectiveness of DSTIs, by helping to avoid that binding DSTIs lead to lower amortization payments rather than a lower debt burden.

A01ufig25

Limits On DSTI Ratios

(Percent)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: National authorities and Fund staff calculations.

C. The Road Ahead: Which Tools Are Most Effective?

20. Empirical evidence suggests that demand-side measures are effective in curbin household borrowing. A survey of the growing literature on the effects of macroprudential policy instruments shows that LTV limits are particularly helpful in slowing credit growth (see Appendix 2). For example, Crowe and others (2011), using a panel of US state level data, find that a 10 percent tightening of the LTV ratio leads to a decline in the rate of house prices appreciation of between 8 and 13 percentage points. Similar results have been reported for other countries including by: Ahuja and Nabar (2011) and Craig and Hua (2011) for HongKong; Igan and Kang (2011) for Korea, Jimenez and others (2012) for Spain, Krznar and Morsink (2014) for Canada and Elliot and others (2014) for the US housing market. In addition, other demand side measures, such as lower DSTI caps, limits on amortization periods and stamp duties have been found to be effective in slowing mortgage credit growth. In contrast, the effectiveness of supply-side measures such as higher sectoral capital requirements is less clear.

A01ufig26

Illustration of Demand Side Measures

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Source: Fund staff calculations.
Table 4.

Incidence of Macroprudential Policy Tightening

article image
Sources: Shim, Bogdanova, Shek and Subelyte (2013) and Fund staff calculations.Note: The table shows number of (tightening) macro-prudential policy actions for 60 countries until June 2012.

21. There is less evidence on the simultaneous use of these tools—a scenario particularly relevant for Sweden. To shed some light on the interactions between mortgage supply (mortgage risk weights) and mortgage demand (amortization rate and LTV) side measures, consider a simple partial equilibrium model (see Appendix 1). In the model, a representative household decides between spending a fixed budget on consumption or housing over a certain period, the latter with the help of a mortgage credit drawn from a bank that draws funding from deposits and equity capital. Despite its highly stylized nature, the model suggests that a combination of demand and supply side measures is more effective in containing credit growth than supply side measures alone. In particular:

  • Mortgage risk weights, limits on maximum amortization periods and LTV ratios are each effective in reducing mortgage lending, but operate through different channels. Higher risk weights on mortgages increase the bank’s funding costs (the additional equity to fulfill capital requirements is more expensive than deposit-based funding), driving up lending costs. More binding LTV caps and shorter maximum amortization period, instead, directly affect household mortgage demand.

  • An efficient policy mix will combine both supply and demand measures. The reason is that both types of measures are subject to diminishing returns, reflecting the (plausible) assumption that households value an additional unit of housing more when they have less of it.

  • When the level of both deposit rates and equity returns are low—along with overall mortgage rates—demand side measures tend to be more effective. On the one hand, at lower mortgage rates, household can afford larger mortgage loans, which tends to make a given reduction in the LTV cap more effective in reducing loan demand. On the other hand, the impact of risk weights is smaller as the spread between equity and deposit rates is low, making it less costly for the bank to switch from deposit to equity funding. A simple calibration of the model suggests that a 10 percentage points reduction in LTV maxima leads to a 18 percent decrease in mortgage credit, while a 10 percentage points increase in mortgage risk weights only reduces mortgage credit by 7.5 percent. (see Appendix 1 for the details of the calibration.)

22. The model also suggests that higher policy rates will impact both mortgage supply and demand.17 A higher policy rate reduces household wealth which, in turn, leads to lower mortgage demand. Simultaneously, a higher central bank rate will directly increase banks’ funding costs and lower supply. That said, the model does not include alternative funding options such as covered bonds, which could dampen the impact on overall funding costs.

A01ufig27

Results from the Calibrated Model

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

23. A simple structural VAR (SVAR) can shed light on the role of interest rates in the mortgage market. The VAR is estimated using monthly observations between September 2005 and March 2014 of inflation, industrial production growth, fixed-rate and variable-rate mortgage credit growth rates, the mortgage rate, the real effective exchange rate (REER), and the Stockholm Interbank Offered Rate (stibor).18 In addition, German bond yields and a time trend are included as exogenous variables in the system. Shocks are identified using sign restrictions in line with economic theory. Specifically, a positive monetary policy shock, as captured by changes in stibor, is assumed to have a negative impact on inflation, output, and variable-rate mortgage credit growth, but a positive impact on the fixed rate mortgage credit growth while the impact on the REER is left unconstrained.

24. The results suggest monetary policy affects both the level and the composition of mortgage credit. The SVAR finds a negative relationship between the monetary policy rate and variable-rate mortgage credit growth but a positive relationship with fixed-rate mortgage credit growth, and these results remain significant for 12 months after the initial shock. In particular, an exogenous positive increase in stibor rate by 25 basis points leads to a peak-increase of 30 basis-points of the average mortgage interest rate and reduces variable-rate mortgage credit growth by 2.7 percentage points after two months. At the same time, it increases fixed-rate mortgage credit growth, as households shift to mortgage contracts which promise protection from further interest rate increases. Overall, the monetary policy shock decreases the total mortgage credit stock by 0.5 percent after two months, and household debt-to-income ratio by 0.5 percentage points.19 These results would suggest that a more restrictive monetary policy will not only reduce the household debt ratio but also reduce the vulnerability of households to interest rate shocks.

25. The main results are robust to a number of alternative specifications. The main findings are qualitatively unchanged across different sample periods and alternative sign restrictions—for example, with regard to the contemporaneous correlation between the monetary policy shock and output and inflation. Similarly, introducing additional control variables (e.g., cover bond rates or long term German bond yields) or varying lag lengths does not significantly alter results. That said, any such exercise comes with significant uncertainty—related to the simplicity of the model and the available sample size—and should be interpreted with caution.

A01ufig28

Impulse Responses to a Monetary Policy Shock

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Sources: Fund staff calculations.Note: Impulse responses of variable-rate mortgage credit growth, fixed-rate mortgage credit growth and mortgage rates following a 25 basis point positive shock to the stibor rate, using a sign restriction identification scheme. Total mortgage credit growth is calculated taking the March 2014 stock of the variable and fixed rate mortgages as starting point, and applying the credit growth rates separately to the corresponding stock of mortgages.

References

  • Admati, Anat, and Martin Hellwig, 2013, “The Bankers’ New Clothes: What’s Wrong with Banking and What to Do About It,Princeton University Press.

    • Search Google Scholar
    • Export Citation
  • Ahuja, Ashvin, and Malhar Nabar, 2011, “Safeguarding banks and Containing Property Booms, Cross-Country Evidence on Macroprudential Policies and Lessons from Hong Kong SAR,IMF Working Paper 11/284 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Almeida, Heitor, Murillo Campello, and Crocker Liu, 2005, “The Financial Accelerator: Evidence from the International Housing Markets” (unpublished, New York University).

    • Search Google Scholar
    • Export Citation
  • Anderson et al, 2014, “Household Debt and Consumption During the Financial Crisis: Evidence From Danish Micro Data,Danmarks Nationalbank Working Papers 2014.89.

    • Search Google Scholar
    • Export Citation
  • Arregui, Nicolas, Jaromir Benes, Ivo Krznar, Srobona Mitra and Andre Oliveira Santos, 2013, “Evaluating the Net Benefits of Macrorprudential Policy: A Cookbook,IMF Working Paper 13/167 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Chen, Jiaqian, and Giuseppe Vera, forthcoming, “What do Monitors Monitor: Bank Balance Sheets and Borrower Performance.IMF Working Paper (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Craig, R. Sean, and Changchun, Hua,Determinants of Property Prices in Hond Kong SAR: Implications for Policy,IMF Working Paper WP/11/277

    • Search Google Scholar
    • Export Citation
  • Crowe, Christopher, Giovanni Dell’Ariccia, Deniz Igan, Pau Rabanal, 2011, “Policies for Macrofinancial Stability: Options to Deal with Real Estate Booms,IMF Staff Discussion Note 11/01 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Dell’Ariccia, G., D. Igan, and L. Leaven, 2012a, “Credit Booms and Lending Standards: Evidence from the Subprime Mortgage Market,Journal of Money, Credit and Banking, Vol. 44, pages 367384.

    • Search Google Scholar
    • Export Citation
  • Elliott J. Douglas, Greg Feldberg and Andreas Lehnert, 2014, “Cyclical American Macroprudential Policy,Federal Reserve Board mimeo.

    • Search Google Scholar
    • Export Citation
  • Finansinspektionen, 2013, “The Swedish Mortgage Market 2013”.

  • Igan, Deniz and Prakash Loungani, 2011, “Do Loan-to-Value and Debt-to-Income Limits Work? Evidence from Korea,IMF Working Paper 11/297 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Igan, Deniz, and Prakash Loungani, 2012, “Global Housing Cycles,IMF Working Paper 12/217 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2013, “The Interaction of Monetary and Macroprudential Policies,January (Washington).

  • International Monetary Fund, 2013, Nordic Regional Report, Country Report 13/274 (Washington: August).

  • Jiménez Gabriel, Steven Ongena, José-Luis Peydró and Jesús Saurina, 2012. “Macroprudential policy, countercyclical bank capital buffers and credit supply: Evidence from the Spanish dynamic provisioning experiments,Economics Working Papers 1315, Department of Economics and Business, Universitat Pompeu Fabra, revised Feb 2013.

    • Search Google Scholar
    • Export Citation
  • Kenneth Kuttner, and Ilhyock Shim, 2012, ”Taming the Real Estate Beast: The Effects of Monetary and Macroprudential Policies on Housing Prices and Credit,RBA Annual Conference Volume in, Property Markets and Financial Stability Reserve Bank of Australia, Alexandra Heath & Frank Packer and Callan Windsor (ed)

    • Search Google Scholar
    • Export Citation
  • Krznar, Ivo and James Morsink, 2014, “With Great Power Comes Great Responsibility: Macroprudential Tools at Work in Canada,IMF Working Paper 14/83 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Laseen, Stefan, and Ingvar Strid, 2013, “Debt Dynamics and Monetary Policy: A Note,Sveriges Riksbank Working Paper Series No. 283.

    • Search Google Scholar
    • Export Citation
  • Lim, C. H., F. Columba, A. Costa, P. Kongsamut, A. Otani, M. Saiyid, T. Wezel, X. Wu, 2011, “Macroprudential Policy: What Instruments and How Are They Used? Lessons from Country Experiences,IMF Working Paper 11/238 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Miles, David, Jing Yang and Gilberto Marcheggiano, 2013, “Optimal Bank Capital,The Economic Journal, Volume 123, Issue 567, pages 137.

    • Search Google Scholar
    • Export Citation
  • Shim, Ilhyock, Bilyana Bogdanova, Jimmy Shek and Agne Subelyte, 2013, “Database for Policy Actions on Housing Markets,BIS Quarterly Review, September 2013.

    • Search Google Scholar
    • Export Citation
  • Tovar, Camilo E, Mercedes Garcia-Escribano, and Mercedes Vera Martin, 2012, Credit Growth and the Effectiveness of Reserves Requirements and Other Macroprudential Instruments in Latin America,IMF Working Paper 12/142 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • 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 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Wong, Eric, Tom Fong, Ka-fai Li and Henry Choi, 2011, “Loan-to-Value Ratio as a Macroprudential Tools—Hong-Kong’s Experience and Cross-Country Evidence,HKMA Working Paper, 01/2011 (Hong Kong: Hong Kong Monetary Authority).

    • Search Google Scholar
    • Export Citation

Appendix I. A Simple Model of Demand and Supply of Mortgages

In a simple model framework, we illustrate how mortgage supply measures (e.g., mortgage risk weights) and mortgage demand measures (e.g., LTV) interact. The main implication of the model is that both measures exhibit diminishing marginal returns in terms of their effectiveness, suggesting that an efficient policy mix will generally involve both supply and demand side measures. In addition, a calibrated version of the model shows that demand-side measures tend to be more effective when both the equity returns and deposit rates are low, implying that at the current juncture, particular emphasis should be placed on demand-side measures.

Set-Up

Household—Assume a representative household receives an income w each period, and chooses consumption c and housing stock H to maximize utility subject to a borrowing constraint. That is:

maxH,cU(H,c)=H1γ11γ+c

s.t.

t0n(11+rD)t(wc)=t=0n(11+rD)trm(1δ)H+τδH+(11+rD)n+1(1δ)H

The household must borrow to finance the house over the period n, with the financing consisting of two parts: (1) a mortgage loan (1 δ)H which the principal needs to be paid at the end of the loan contract, and (2) a lump sum down paymentδH, where δ denotes the down payment (i.e. LTV = 1—δ), i.e., the amount that must be paid immediately.1 Banks providing the finance charge rate rm for the mortgage and a fixed fee τ for providing the down payment.2 The household borrows against his future discounted savings (income excludes consumption), which justifies the budget constraint above. For the LTV constraint to bind, we assume that τ>11rm1+rD, which ensures that the household strictly prefers to borrow from the mortgage loan. In addition, we denote rD as the per-period discount rate (or riskless rate) and rm as the mortgage rate. Finally, to simplify the notation, we define R=11+rD

and R¯=1R1Rn+1

Solving the household maximization problem, we obtain the demand for mortgages:

(1δ)H=LD=(1δ)(rm(1δ)+R¯τδ+R¯Rn+1(1δ))1γ

Bank—We assume that a bank specializes in mortgage lending, financed by either equity or bank deposits. The bank maximizes its profit subject to a common equity tier 1 (CET1) capital requirement, that is:

maxrm,k(rmrEkrD(1k))Lm

s.t.

kLmαLm=K¯

where rE denotes the return on equity, Lm the mortgage supply, k the share of equity in the bank’s balance sheet, a the risk weight, and K¯ the CET1 capital requirement.

We assume the bank enjoys monopoly power so that it takes the downward sloping demand curve as given from the household, and sets the mortgage rate (price) such that marginal cost equals marginal revenue.

Analytically—One can easily solve the model and show that the LTV, limits on amortization period and mortgage risk weights are effective in reducing mortgage credit growth. In particular:

Lα=(1δ)[(τδ)(1R)(1+γ)]1/γ[1R(rD+αK¯(rErD))]1γ1(RK¯(rErD))<0

Moreover, it can be shown that Lδ<0,Ln<0,2α2>0and 2Lδ2>0.

Calibration—The table below shows the model’s baseline calibration. Specifically, we choose the equity returns to match the observed return on equity in 2013, and the deposit rate is set at 2 percent, roughly the return on fixed maturity (less than two years) deposits in 2013. Moreover, we calibrate the CET1 capital requirement to be 15.5 percent consistent with the Riksbank’s recommendation, LTV maxima are set at 85 percent according to the FSA requirement, mortgage risk weights are set at 15 percent and average amortization period to be 100 years as indicated in the recent Riksbank mortgage report. Owing to the simplicity of the model, household preference parameter y and the service cost of down payment z are chosen to some reasonable values, but without matching particular empirical moments. However, the key results of the model do not change qualitatively by varying these two variables.

Table A1.

An Illustrative Calibration

article image
Source: Fund staff calculations.

Baseline result—The mortgage rate in the model is pinned down by the bank’s marginal cost, and the equilibrium mortgage credit volume is determined by the intersection between the marginal revenue and the marginal cost curves (see text figure). In what follows, we use variants of this figure to study the interaction between mortgage risk weights and LTV maxima.

Result 1—An efficient policy mix consists of both supply and demand side measures. As illustrated by the text figures below, both measures are subject to diminishing returns. For example, 5 percentage point reduction in LTV maxima starting from 85 percent leads to a larger decline in mortgage credit than when it takes place at the 80 percent level. Similarly, a 5 percentage point increase in mortgage risk weights reduces mortgage credit more when it takes place at a lower starting level of risk weights. These results follow from households’ concave preferences over housing (reflected in the convex marginal revenue curve above), that is, they value housing more when they have less of it.

A01ufig29

Model Equilibrium: Baseline

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Source: Fund staff calculations.

Result 2—When the level of both deposit rates and equity returns are low—along with overall mortgage rates—demand side measures tend to be more effective. Figure A1 illustrates two cases with low deposit rates supporting this claim. Households tend to demand larger loans when deposit rates, hence the mortgage rates, are low as illustrated in the top panel. This makes the demand side measures particularly powerful, as 1 percentage point reduction in LTV maxima leads to a decline in mortgage credit that is proportional to the house price. The effectiveness of credit risk weights hinges on the spread between equity and deposit returns, which, in turn, relates to the level of equity and deposit rates. When both are at low levels, it becomes less costly for the banks to switch from deposit to equity funding. At the same time we observe that the effectiveness of risk weights is also a decreasing function of the deposit rate because when the deposit rate is high, it costs less for bank to switch to equity funding (assuming that equity returns are constant). However, this diminishing effect is much smaller (relative to a same unit reduction in LTV maxima), because the deposit and equity returns, hence the spread, are at very low levels.

Scenario 1:
Scenario 1:

Change in LTV Maxima

(Further reduction in LTV maxima is diminishing in its effectiveness)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Source: Fund staff calculations.
Scenario 2:
Scenario 2:

Change in Mortgage Risk Weights

(Increase mortgage risk weights is diminishing in effectiveness)

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Source: Fund staff calculations.
Figure A1.
Figure A1.

Effectiveness of Demand and Supply-Side Measures

Citation: IMF Staff Country Reports 2014, 262; 10.5089/9781498339438.002.A001

Appendix II. Selected Literature Survey

article image
1

Note, as n→∞, mortgage loan is converging to an unamortized loan.

2

The down payment is assumed to be financed by a second loan.

1

Prepared by Jiaqian Chen.

2

See IMF (2013) for details of the calculations. Similarly, a recent panel study of Danish households by Andersen, Duus, and Jensen (2014) finds a positive and significant correlation between pre-crisis household leverage and the reduction in consumption during the crisis.

3

Note: wholesale funding consists of all sources of bank funding except deposits and equity.

4

The Net Stable Funding Ratios (NSFRs) of some Swedish banks have been revised upwards following the new Basel III guidance released in early 2014. However currently only two banks report the NSFR.

6

The leverage ratio is defined according to Basel III.

7

Recent studies, including, Admati and Hellwig (2013), Chen and Vera (forthcoming), Miles and others (2013), suggest that the “optimal” bank leverage could be higher than what is proposed in Basel III.

8

Core Tier 1 capital, in fact, increased by 2.3 percentage points, but 1.4 percentage points were offset by the change in total assets.

9

In this paper, risk weights are calculated as the ratio between risk weighted assets and exposure.

10

Consider a representative bank with a leverage ratio of 3 percent and a lending book consisting of only corporate and mortgage loans, with a 60 and 40 percent share, respectively. If this bank adjusts the risk weights for corporate and mortgage lending from about 45 and 8 percent to about 32 and 16 percent—an adjustment roughly representative of the change in risk weights observed for Swedish banks between 2010 and 2012—this implies a 1.7 percentage points improvement in the risk weighted capital ratio. Moreover, if the bank’s lending book shifted to a 40 and 60 split between corporate and households, its risk-weighted capital ratio would improve by about 3½ percentage points without any change in the leverage ratio.

11

Conceptually, one can think of a bank as a portfolio investor who chooses its investment allocation subject to a capital requirement constraint. Any changes in the risk weights, effectively makes it more costly to lend to that particular sector which in turn will affect bank’s investment allocation (see the text chart above). In a simple example, with risk weight shifting from, say, 45 and 8 percent for corporate and mortgage lending, respectively—an adjustment roughly representative of the change in risk weights observed for Swedish banks between 2010 and 2012—to 32 and 16 percent would imply an increase in bank lending to the corporate sector by 50 percent while mortgage lending would drop by 75 percent

12

This refers to the average maturity of Swedish covered bonds issued in the autumn of 2013, according to the Association of Swedish Covered Bond issuers. The average maturity for the stock of covered bonds was, at the same point in time, just below three years. Source: The Swedish mortgage market, Finansinspektionen, 2013.

13

A simple model simulation implies an effect of similar magnitude (see below).

14

See http://www.bis.org/publ/bcbs172.pdf for the BIS recommendations.

15

Focusing on households, which can hold multiple mortgages, rather than loans gives a broadly similar picture. Among the households with new mortgages and LTVs below 75 percent, about 60 percent did not amortize any of their mortgages in 2013.

16

In addition, rent controls could also have a distortionary impact on allocation of rental homes.

17

Empirically, Crowe and other (2011) have also found policy rate is a powerful tool in reducing mortgage credits, despite other potential costs.

18

Variable rate credit accounted for over 60 percent of the stock of total mortgage credit in March 2014.

19

In calculating the impact of monetary policy on households’ debt-to-disposable-income ratio, it is assumed that disposable income decreases by 0.5 percent following a 25 bp increase in the repo rate as estimated by Laseen and Strid (2013). In addition, it is assumed that a 25bp increase in the Stibor rate is equivalent to a 20bp increase in the repo rate estimated based on the historical relationship between the two rates. The impulse responses are then converted by taking the differences between the implied ratio and the assumed initial position (174.5 percent).

Sweden: Selected Issues
Author: International Monetary Fund. European Dept.