Switzerland: Selected Issues

International Monetary Fund. European Dept.

doi:10.5089/9798400212994.002.A002

Author:

International Monetary Fund. European Dept.

International Monetary Fund. European Dept.
Search for other papers by International Monetary Fund. European Dept. in
Current site
Google Scholar

Publication Date:: 20 Jun 2022

eISBN:: 9798400212994

Language:: English

Keywords:: loan risk characteristic; loss rate; mortgage portfolio; affordability risk; deep house-price correction; Housing prices; Loans; Real estate prices; Amortization; Global

Download PDF (750.6 KB)

Housing matters for economic activity and financial stability in Switzerland. The mortgage market is large relative to the size of the economy and banks are heavily exposed. House prices have significantly outpaced income growth, and this trend has accentuated during the pandemic. The Swiss authorities have taken decisive action to address unsustainable developments, but vulnerabilities have increased. This paper shows that a fuller set of macroprudential tools can be more effective to reduce systemic risk. Adequate calibration and a forward-looking approach are key given lags between policy announcements and policy effects. The paper quantifies a suite of LTV/DSTI caps, amortization requirements, and ‘speed limits’ calibrated at the vintage level to guard against the build-up of vulnerabilities and strengthen resilience.

Abstract

A Forward-Looking Approach to Calibrate Macroprudential Tools in Switzerland^¹

A. The Build-Up of Systemic Risk

1. Strong growth of credit in Switzerland over the past several years has resulted in a “too-big-to-fail” mortgage market. Mortgage volumes relative to GDP have experienced significant growth since 2008, while the ratio of other types of credit to GDP has decreased. The mortgage market reached 150 percent of GDP in 2021, a 40 percentage-point increase since the global financial crisis. Robust mortgage-credit growth has been echoed by rising household indebtedness, which topped 130 percent of GDP at end-2021. The two ratios—mortgages-to-GDP and household-debt-to-GDP—are high by European and international standards).

Credit Aggregates and Household Debt to GDP

(Percent)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: SNB; Haver Analytics; IMF staff calculations.

Price-to-Income Ratio

(Index; Long-term average=100; transaction prices)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: IAIZ; OECD; and IMF staff calculations.

2. The Swiss financial system is heavily exposed to mortgages. The Swiss banking sector is the largest in Europe relative to the size of the economy, reaching 5 times GDP. Mortgages are 85 percent of total bank loans and half of bank assets. Most household (95 percent) and corporate (65 percent) loans are mortgages. Credit markets are thus vulnerable to real-estate price corrections. Household mortgage credit, at 110 percent of GDP, is 3 times euro-area levels (Figure 1). The wider financial system is also heavily exposed to the real estate market: pension fund and insurers asset allocation is increasingly tilted to investment property and mortgages—23 and 12 percent of assets, respectively.^²

3. The residential real estate market shows clear signs of overheating. With negative policy interest rates of the Swiss National Bank (SNB) since December 2014, price growth has resulted in imbalances in residential real estate and mortgage markets.^³ Investors have increasingly turned to residential real estate in search-for-yield.^⁴ Prices accelerated further during the pandemic with households building up savings and increasingly working from home, and mortgage rates hovering at around historically low levels of 1.1 percent. Gains have been particularly strong in the Lake Geneva and Zurich regions. In the self-owner-occupied segment (SORE), strong demand and surging preference for ownership has outpaced supply, as construction remains below pre-Covid levels. Vacancy rates of dwellings are tight and declined in 2021 to 1.54 percent from 1.72 percent in 2020. Residential price-to-income and price-to-rent ratios have increased for privately-owned apartments, single-family homes, and apartment buildings and are 30 percent above long-term averages.

Price-to-Rent Ratio

(Index, Long-term average=100; transaction prices)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: IAIZ; OECD; and IMF staff calculations.

^{Figure 1.}

Macro-financial Relevance of the Mortgage Market

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

B. The Policy Response

4. The build-up of imbalances has triggered a series of policy responses by the Swiss authorities. The persistent increase in real estate prices and the build-up of vulnerabilities have prompted a series of responses by the Federal Council, SNB, FINMA, and the banking sector. Switzerland became the first country to activate a residential real estate countercyclical capital buffer (CCyB) in February 2013 at 1 percent. At that time, the financial and the business cycles were out of sync. The policy rate was zero as the inflation rate was subdued at -0.3 percent. With easy monetary conditions, house prices had risen sharply, at an average annual rate of 7.5 percent over the previous two years, supported by strong mortgage growth exceeding 5 percent. The sectoral CCyB was increased to 2 percent in January 2014, as real estate imbalances persisted (Figure 2).

^{Figure 2.}

Policy Response, Mortgage Credit, House Prices, and Interest Rates

(+ Tighten; – Relaxed; Year-on-Year; Percent)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: SNB; FINMA; Wuest Partner; IMF staff calculations.Note: The vertical lines represent policy response to address imbalances in the mortgage and residential real estate sectors. The red lines are macroprudential policy interventions (introduction; adjustments to CCyB buffer); the blue lines represent self-regulation rules (introduction; amendments). House prices by segment (apartments; single family) are transaction prices.

5. The authorities have directly involved banks in the design of macroprudential instruments under “self-regulation” guidelines.^⁵ A new self-regulatory regime entered into force from July 2012. For the first time, it laid out minimum requirements concerning down-payments by borrowers (at least 10 percent of the property value) and introduced compulsory amortization to two thirds of the value of the collateral within a maximum of 20 years. A revision to self-regulation rules was made in September 2014 involving a shorter period of 15 years for mandatory amortization. In January 2020, requirements of mortgage loans for investment properties were tightened: the maximum period for mandatory amortization was revised down from 15 to 10 years; and borrowers now need to provide a minimum down-payment of 75 percent of the property value. While the revised guidelines do not explicitly include the buy-to-let segment, most banks seem to apply the adjusted rules to this segment following FINMA’s recommendation.

6. FINMA has helped build bank resilience by tightening underwriting standards. As the Swiss supervisory authority, FINMA has deployed a range of supervisory tools to enhance resilience in residential real estate. It has introduced Pillar 2 multipliers on Internal Ratings-based (IRB) models, and applied the adjustments to the capital adequacy ordinance adopted by the Federal Council such that mortgages exceeding 80 percent of the property value have a risk-weight of 100 percent for the part of the loan over the threshold. Full Basel III implementation, expected in 2024, will introduce an output floor (a limit on the capital requirements that banks calculate using their internal models) and higher risk-weights for riskier loans.

7. Yet, despite these actions, affordability risks remain elevated (Figure 3). Despite recent tightening of self-regulation, affordability risks as measured by LTI ratios have increased. In the owner-occupied segment, the 75th value of the LTI distribution widened by 40 bps from 2018 to reach 7.2 at end-2021. Income-producing residential real estate (IPRRE) loan-to-rent ratios for both household and corporate investors have risen to reach 22 and 19 at the 75th percentile, respectively, high by international standards. As of Q4:2021, the LTV of rented-out property at the 90th percentile decreased to 75 percent for both households and companies from the 80 and 82 percent ratios reached in 2020 but remains elevated.^⁶

^{Figure 3.}

Leverage in Residential Real Estate Loans

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

8. High leverage exposes borrowers to rising interest rates (Figure 4). We estimate that a quarter (half) of new mortgage production could become unaffordable if rates were to increase to 3 percent (5 percent) across all sub-segments.^⁷ The 5-year fixed mortgage rate was 1.7 percent as of March 2022 (a 60-bps increase since December 2021), and long-term rates are 3.7 percent, so prospects of an increase to 3 percent are non-negligible. An important mitigating factor is that Swiss households with mortgage loans tend to have significant buffers in the form of financial wealth that they could use to fill potential liquidity shortfalls.

^{Figure 4.}

Impact of Interest Rate Shocks on Affordability Risk

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

1/ This simulation assumes that income generated by the property is the only source of income for assessing affordability risk. As households typically have other sources of income, and can draw on their financial wealth, these figures tend to overestimate the level of affordability risk. Note: The top left chart shows the policy rate (mid of the LIBOR 3m target range until 2019; policy rate after 2019), and the weighted mortgage rate for new business weighted by the relative share of variable mortgages, fixed-rate mortgages, and LIBOR mortgages. SORE stands for self-occupied real estate. IPRRE stands for income producing residential real estate. Affordability risk is considered high when imputed costs from mortgage servicing (including amortization, interest, and maintenance) exceed one-third of income (owner-occupied segment) or rents (investment-led segment). To compute affordability risk, we assume that average LTV is 80 percent and maintenance costs represent 1 percent of loan value. The chart ‘Debt Service Ratio Simulations in IPRRE-Companies’takes into account the tightening of the amortization requirement established by self-regulation in August 2019, effective in January 2020.

C. Transmission Channels of Macroprudential Tools

9. Macroprudential authorities can enhance banking system resilience by using capital-based tools. Tools include capital buffers on real state exposures to absorb unexpected losses in times of stress.^⁸ For instance, in the European Union, a 202 capital requirements directive (CRD V) increased the flexibility of the systemic risk buffer to target specific systemic risks at the sectoral level, i.e., real estate.^⁹ Capital buffers can also reduce the build-up of vulnerability through the credit channel. Banks can pass through capital charges to mortgage rates, therefore dampening credit demand (price effect), while capital-constrained banks may be forced to reduce credit supply (quantity effect). Lower credit is likely to feed into more sustainable house prices, hence stabilizing the financial cycle.

10. To contain the build-up of vulnerabilities, macroprudential authorities (and lenders) can rely on borrower-based tools that strengthen borrower balance sheets.^¹⁰ To address financial stability risks arising from rapid house price inflation and increasing household debt, authorities can use borrower-based tools including LTV, DTI and DSTI regulatory limits on mortgage lending. These instruments also build bank resilience indirectly by reducing the loss rate of the mortgage portfolio. While LTV restrictions reduce bank losses given default, debt serviceability requirements (DTI, DSTI) lessen borrower probabilities of default.

D. A Modeling Approach to Quantify Banking System Resilience

11. The calibration of a wide range of macroprudential tools warrants a modeling approach that assesses the performance of vintage pools. A vintage approach is needed as borrower-based tools are applied to the production of new mortgages whereas sectoral capital buffers are applied to the outstanding portfolio. A vintage analysis also allows for a deep understanding of the effects of loan maturation and external factors such as changes in real estate prices, interest rates, and disposable income. Mortgage vintage analysis tend to show the dispersion of delinquency between more recent vintages originated during the run-up of housing prices and older, more established vintages that were originated before the expansionary phase of the cycle.

12. A granular risk model allows macroprudential authorities to understand the credit risk of different real estate market sub-segments. A critical segmentation is by type of property. The model needs to quantify credit risk in the owner-occupier segment and income producing real estate segment separately, given differences in risk drivers and sensitivity to shifts in interest rates and other macroeconomic factors. The model also needs to produce credit risk projections according to the regulatory buckets used by macroprudential authorities to monitor vulnerabilities and activate regulatory limits.

13. Structural changes in the Swiss real estate market call for a structural approach. The limited data from past real estate crises in Switzerland, the presence of structural changes in housing finance (e.g., the structure of the market changed markedly in 1995, when Swiss residents were allowed to draw on their second and third pillar pension assets to fund part of the mortgage down payment), and adjustments to the regulatory framework (Section B) mean that structural models are more reliable than statistical approaches based on past loan performance. Statistical approaches are also hindered by the recent benign cycle of default risk in Switzerland, with average loss rates in the mortgage portfolio at just 5 basis points over the last 20 years.

The Model

14. We use a granular structural model by vintage to project mortgage risk.^¹¹ A loss event is defined by the ‘double trigger’ of default. A borrower defaults if they can no longer afford to service the loan and if the value of the house is lower than the value of the loan.^¹² This assumption implies that a borrower in financial distress could avoid default, as long as they have positive home equity that can be extracted to refinance the loan or repay the debt.^¹³ The model is calibrated by segment, i.e., owner-occupied mortgages and investment loans and risk buckets.

15. The probability that a borrower i gets into financial distress is driven by:

\begin{matrix} \Pr (F D_{i, t}) = Φ (D S T I_{i, t}) \cdot D_{i} + β_{1} \cdot Δ D S T I_{i, t}^{γ} + Φ (D S T I_{i, t}) \cdot (β_{2} \cdot U_{t} + β_{3} \cdot Δ U_{t}^{α}) & (1) \end{matrix}

which is a function of: (1) affordability risk, measured by the debt service-to-income ratio (DSTI); (2) the change in the debt servicing capacity since the last period (ADSTI); (3) the likelihood of being unemployed (U) and the change in the unemployment rate (AU); and (4) a demographic factor (D). The impact of idiosyncratic events (demographic shocks, migration to unemployment) is non-linear across affordability buckets.^¹⁴ The exponential effect of affordability shocks on financial distress captures an ‘expenditure-based’ approach whereby stretched borrowers struggle to maintain consumption, increasing default risk. We assume that idiosyncratic shocks have no impact on financial distress below a DSTI threshold level, then it raises gradually until an upper threshold where the effect has full impact.

16. Economic default is triggered if a financially distressed borrower cannot repay the loan by selling the house or drawing down on his wealth:

\begin{matrix} \Pr (E D_{i, t}) = (\begin{matrix} 1 & i f {\tilde{P}}_{i, t} - C + F i n W e a l t h_{i, t} < N P V (L_{i, t}, r_{t}^{t y p e, M}, r_{t}^{f}, T_{t, s}) | F D_{i, t} \\ 0 & O t h e r w i s e \end{matrix}) & (2) \end{matrix}

where ${\tilde{P}}_{i, t}$ is the market property value, C is the transaction cost of selling the property, and FinWealth_i,t is the borrower’s liquid financial assets. The net present value of the loan (NPV) consists of two elements: (1) the outstanding loan amount L_i,t and (2) the penalty for early prepayment, i.e., the discounted value of foregone interest payments, which increase with the mortgage rate locked-in at the time of default $r_{t}^{t y p e, M}$ (which depends on the type of mortgage and the resetting price schedule of the loan), and its remaining maturity (T_t,s – (t – s)) for a loan issued at time s.^¹⁵ The 5-year fixed mortgage rate $r_{t}^{f}$ is used to discount the amount of future interest payments.

17. To generate the probability of default (PD), we use a Monte Carlo algorithm over more than 250 vintage-LTV buckets:

\begin{matrix} P D_{i, t} = \Pr (F D_{i, t}) \cdot \Pr (F D_{i, t}) & (3) \end{matrix}

We divide the portfolio into risk buckets and simulations are carried out for each bucket and vintage, separately. For a given ‘vintage-LTV bucket’ of mortgages, a number N of borrowers already in financial distress is considered. For each of the N borrowers a house price draw is generated from a distribution with a mean equal to the average house price level in the tail risk scenario. For each of the house price draws, the model determines whether condition (2) is satisfied (i.e., if the borrower defaults). In the next step, the bucket-specific PD is calculated as the total number of defaults divided by the number of draws, 10,000, and multiplied by the bucket-specific probability of financial distress from equation (1). To reduce simulation noise, this simulation process is then repeated 2,000 times for each bucket. In the final step, the ultimate outputs, i.e., portfolio-wide PD, LGD, and the loss rate are calculated by combining separately estimated outputs for 252 vintage-LTV buckets (12 LTV buckets; 21 vintages) and weighting them by their outstanding share in bank portfolio at end 2021.

18. Upon default, a bank’s loss given default (LGD) is driven by the discounted sale price of the repossessed collateral net of the foreclosure discount:

\begin{matrix} L G D_{i, t} = N P V (L_{i, t}, r_{t}^{t y p e, M}, r_{t}^{f}, T_{t, s}) - (1 - δ) \cdot \frac{{\tilde{\tilde{P}}}_{i, t + n}}{{(1 + r_{t}^{f} + s p r e a d)}^{n}} & (4) \end{matrix}

where the first term denotes the outstanding debt and the second term the recovery value. δ denotes the foreclosure discount at which the bank sells the property at time t+n, where n reflects the time needed to sell off the collateral, and spread is the risk-adjusted spread used to discount the value of the risky asset.

19. The highly non-linear interaction between PD and LGD will have implications for the design of macroprudential tools. A large correction of house prices increases the chance of borrowers going into negative equity and therefore default (PD). It also increases the LGD for banks if borrowers fail to maintain their mortgage payments. Similarly, an upward shift of interest rates increases affordability risk (PD) and lowers the recovery rate of a defaulted loan (LGD). This suggests that LTV restrictions or interest-rate affordability tests can help prepare both households and banks for the potential fallout from a sharp reversal in the housing market or rising interest rates.

The Dataset

20. We use SNB survey data on new mortgages, which provide a rich characterization of loans approved since 2017. The quarterly dataset covers all loans granted by Swiss banks with a domestic mortgage lending volume of at least CHF 6 billion. It includes new mortgages that finance the purchase of real estate, as well as commutations (refinancing of a loan with another lender) but excludes rollover loans with the same lender. The dataset used to calibrate the model draws on SNB survey data but eliminates construction loans and removes outliers. The survey records a loan’s general characteristics (e.g., borrower, type of loan, credit limit, income, value of collateral, rent, interest rate, down payment). This allows a very granular segmentation. The data are segmented by type of business transaction (i.e., SORE and IPRRE segments) and vintage (20 vintages). In the IPRRE segment, income is defined as yearly rental income on the property. Within each segment/vintage, loan risk characteristics are grouped into 12 LTV buckets, 8 DTI buckets (LTI basis) and 11 DSTI buckets. To account for the correlation structure of risk factors, we construct matrices for 96 LTV/DSTI buckets, and 132 LTV/DTI buckets.

21. The data point to some recent improvement in LTV ratios, but with unequal effects across segments. High-LTV lending has declined in the IPRRE segment but remains elevated for self-occupiers (SORE). The recent tightening of self-regulation rules for investment properties is reflected in a decrease of high-risk LTV loans (>70 percent) from 41 in Q1:2017 to 36 percent in Q42021.^¹⁶ However, the share of high LTV loans in the SORE segment increased during the intervening period, with a bunching of loans below the 80 percent threshold.

LTV distribution of new mortgages by segment, 2021Q4

(Percent)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: SNB; and IMF staff calculations.Note: The chart shows the LTV distribution for newly approved mortgage loans in the last quarter of 2021. The LTV bucket segmentation used for the analysis is in 10pps increments and thus we cannot show the impact of adjustments to self-regulation for IPRRE in the [75%, 80%] bucket.

22. Borrower leverage, as captured by the DTI, has increased over time across segments.^¹⁷ Affordability risk is particularly elevated for highly leveraged borrowers with low equity buffers. The share of loans in the [70%, 80%] bucket with DTI exceeding 5 reached around 30 percent in both the SORE and IPRRE segments. This value seems high by international standards, where the most common maximum DTI cap in peer countries is 4.5.^¹⁸ On the other hand, debt servicing ratios remain contained due to the presence of non-amortizing mortgages for loan amounts below two thirds of the value of collateral at issuance.

23. To reconstruct vintages of mortgage flows we use mortgage stock data, aggregate issuances, self-regulation rules on amortization, and interest rate repricing data. Credit risk depends on Point-in-Time (PiT) risk parameters (LTV, DTI, DSTI) However, SNB survey data are at origination. Also, the dataset only starts from Q12017. To assess the risk of bank mortgage portfolios we need to estimate the volume of mortgages by the vintage outstanding as of end 2021. We also need to project PiT credit risk parameters by taking into account loan repayments, the housing cycle, and the macrofinancial cycle (e.g., interest rates, household income, rents). Mortgage stock and aggregate inflows data are available starting in 1990.^¹⁹ We use interest rate repricing data at issuance starting in 2017 and extend it back to 1990 assuming a pattern similar to the first vintage. For the average vintage), it is worth noting that around two-fifths of new loans reprice within a year.^²⁰ Starting in 1990, we use 6 repricing buckets and assume a uniform distribution of quarterly repricing within each bucket over the life of the mortgage. For each vintage, we estimate the repriced interest rate prevailing at end 2021 and apply interest rate shocks with pass-through effects driven by repricing schedules.

LTV and DTI of new mortgages by segment, 2021Q4

(Percent)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: SNB; and IMF staff calculations.Note: The chart shows the LTV/DTI distribution for newly approved mortgage loans in the last quarter of 2021.

24. Accounting for financial wealth of Swiss households is critical to assess mortgage risk. Wealthy borrowers may draw on their financial wealth to fill liquidity gaps in debt service repayments. Higher wealth is associated with lower default risk (equation (2)). The financial wealth of households in Switzerland has more than doubled over the past 20 years. Per capita net worth in Switzerland is around CHF 480,000, the highest level among peer countries (Annaheim, Heim, 2021). In standard credit risk models, borrower wealth is not typically considered as an input. Yet, Swiss banks use wealth indicators of borrowers in their underwriting affordability criteria. To account for borrower wealth, we use data from Swiss Wealth Statistics by individual. We compute wealth by household, merge this with SNB statistics on the breakdown of wealth, and apply a haircut linked to market liquidity by instrument (0 percent for cash, 10 percent for debt, and 20 percent for equity and collective investment schemes; insurance, pension schemes, and real estate are excluded). For the SORE segment, we consider household wealth at the 50th percentile, while for the IPRRE segment we take the 75th percentile to reflect the higher wealth of investors.

Share of Mortgages by Remaining Time to Repricing at Issuance

(Percent)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Sources: SNB.

Model Calibration

25. The model is calibrated using the loan loss experience in Switzerland in the early 1990s and in real estate crises in peer countries.^²¹ The 1990s Swiss recession, characterized by a sharp decline of real estate prices in combination with an economic slowdown, is used to estimate the default rate elasticity to macroeconomic risk factors, the allocation of stressed sales between those caused by unemployment shocks and interest rate shocks, and transaction costs. The UK real estate crisis of the 1990s is used to calibrate the relationship between default rates and financially distressed borrowers, the sensitivity of financial distress to initial unemployment, and the elasticity of financial distress to shifts in debt serviceability ratios by DSTI buckets. The Annex shows the main parameters of the model.

26. We compute Point-in-Time risk parameters for each LTV-vintage bucket. The procedure takes into account self-regulation amortization requirements, the housing cycle, macrofinancial fluctuations, and the repricing schedule from issuance to the current period (December 2021). We estimate the implied amortization rate, which is consistent with aggregate SNB statistics of mortgage stocks and flows. Then we apply self-regulation minimum amortization requirements, which depend on LTV at origination and the mortgage vintage (given adjustments to self-regulation in 2012, 2014, and 2020). See the Annex for more details on the procedure.

Stress Test Scenario

27. To assess the resilience of the banking sector to a sharp housing price correction, we consider a 3-year inflationary scenario combined with a domestic recession (Figure 5). The stress scenario tests the resilience of the Swiss banking sector to a deep recession, sharp falls in real estate prices, and higher global interest rates, which trigger large shifts in benchmark rates. In Switzerland, household disposable income in the scenario falls by 3.6 percent and house prices decline by an average of 25 percent peak-to-trough – at the upper-end of SNB’s estimated overvaluation (SNB, 2021). Price changes follow a stochastic process with fluctuations of up to 40 percent. The unemployment rate increases by 60 percent, and benchmark rates widen by 300 basis points. Tenant rent payments are less affected as under Swiss regulation, initial rents can be adjusted if interest rates increase.^²² Baseline projections follow IMF’s April 2022 WEO forecast over 2022–24.

^{Figure 5.}

Stress Test Scenario: Key Variables

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Note: The scenario generates quarterly projections of key macrofinancial variables. The table shows cumulative changes over 2022–24. Baseline projections are based on the IMF’s WEO forecast. Adverse projections are based on an inflationary recession scenario using econometric techniques. The size of the shocks are in line with SNB’s combined ‘protracted recession in the euro area’ scenario and ‘global interest rate shock’ scenario (FSR, 2021).

Model Performance

28. The model matches long-term averages of default risk under baseline conditions. A key component to the implementation of a model-based credit risk assessment is model validation. The aim is to ensure that the model structure and parameters are calibrated accurately, and the model performs consistently under baseline scenario assumptions. We conduct a validation exercise by assessing whether the model can replicate long-term default rates during the moderate but sustained expansion of the housing cycle that has taken place over the last 20 years. The results show that the weighted loss rate of the mortgage portfolio predicted by the model (weighted by the shares of the SORE and IPRRE segments) of 5 basis points, replicates the long-term average of loss rates in the Swiss banking system.

Cost of Risk of Mortgage Portfolio

(Basis Points; Annualized)

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Note: The cost of risk is defined as the expected loan loss rate of the mortgage portfolio. The chart shows the annualized 3-year ahead cost of risk of Swiss banks’ mortgage portfolio as of end 2021 under IMF’s WEO baseline conditions (left bar) against the long-term average cost of risk calculated over the 20-years (right bar).

E. Stress Test Results under Current Macroprudential Tools

29. Banks are, on aggregate, resilient to stress, but a real estate crisis could have a significant impact on macrofinancial stability. Stress tests results suggest that the annualized loss rate of the mortgage portfolio could rise from 5 basis points under the baseline to 90 basis points under stressed conditions. The impact of a sharp real estate correction is more severe in the IPRRE segment where default rates could jump to 4.5 percent (Figure 6). Over the 3-year stress test horizon, losses could reach CHF 27.6 billion (14 percent of CET1 capital), with the aggregate CET1 ratio declining by 230 basis points to 14 percent. The sectoral CCyB would absorb 25 percent of losses. Although banks could withstand the shock on aggregate, as they could use their managerial buffers to absorb losses, some banks could breach their capital conservation buffers. Banks with weakened balance sheets could curtail credit to financially-distressed borrowers amplifying the initial shock. Consumption would be affected by rising living costs, which would reduce real household disposable income, higher debt-servicing costs, and lower household wealth, creating negative feedback loops and deepening the recession.

^{Figure 6.}

Stress Test Results

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Note: Credit risk projections are Point-in-Time. Therefore, they reflect cyclical conditions and affect bank capital through loan loss provisions. On the other hand, risk weighted assets are calculated using through-the-cycle parameters in line with Basel standards.

F. Calibration of Borrower-Based Macroprudential Tools

30. The aggregate picture for banks and mortgages looks strong, but there are pockets of vulnerability in recent vintages. Recent borrowers are more exposed to rising interest rates and declining house prices than earlier borrowers as they have repaid less principal, experienced smaller home equity gains, and benefited less from income growth. Also, their serviceability capacity has been tested at lower interest rates.^²³ Our stress test results show that the loan-loss rate of recent vintages is between one third (IPRRE segment) and three quarters (SORE segment)—more than that of the portfolio on average.

31. Further tightening of self-regulation rules could contain the build-up of vulnerabilities in the mortgage and real estate markets. The Swiss authorities could encourage the banking sector to reduce the maximum period of amortization from 15 to 10 years in the owner-occupied or buy-to-let segment. Revised guidelines on mortgage lending could also increase the amortization rate by requiring that borrowers pay down the mortgage to less than two thirds of the lending value of the property.

32. The Swiss authorities could also introduce a range of borrower-based instruments to keep borrowing at sustainable levels. The set of possible instruments include LTV, DTI, DSTI or a combination of instruments. Design considerations include whether regulatory caps are hard limits applied to all new issuances or whether a speed limit might be used by required banks to reduce the volume of high-risk lending to below a specific share of new commitments. Another consideration is whether restrictions would apply temporarily when financial risk is elevated or whether the policy setting would be set to address structural risk (more) permanently.^²⁴

33. To account for lags between policy announcements and policy effects, we assume that new macroprudential tools are introduced before the adverse scenario materializes.^²⁵ First, we assume that in the absence of new macroprudential measures mortgage flows in the next 8 quarters share the average risk characteristic of issuance observed in the last four quarters of data (Figure 7). Meanwhile, some outstanding loans mature. Macroprudential limits reduce the relative mass of high-LTV or DSTI above the limits to zero if ‘hard’ limits are introduced, or to a specific percentage if ‘speed limits’ are applied. We assume that there is a bunching of new loans just below the regulatory limits as observed empirically. During the 8 quarters, house prices and macrofinancial factors follow April 2022 WEO assumptions. For each choice of macroprudential tools, we compare the losses on the aggregate portfolio generated in the adverse scenario with the losses observed on new vintages under ‘No limits’ and under each policy intervention.

^{Figure 7.}

Timeline of Macroprudential Policy Interventions

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Note: Regulatory limits on new vintage flows are introduced 8 quarters before the 3-year stress test horizon materializes. The model delivers tress test results for the outstanding portfolio (under the current macroprudential framework) and new vintages with and without regulatory policy interventions. Regulatory limits include LTV caps, DSTI caps, a combination of LTV and DSTI caps, with and without ‘speed limits’. We also consider changes to the maximum period of mandatory amortization and to the rate of mandatory amortization.

34. Introduction of an LTV-DSTI cap with a ‘speed limit’ or an increase in amortization requirements could guard against the rise in credit risk for new vintages. Figure 8 shows that different combinations of macroprudential limits can have a similar impact on expected losses of new mortgages. A simple rule that can guide the selection of the preferred tool/limit is to target the risk of new vintages to the average risk of older vintages. For instance, an LTV cap of 70 percent with a 20 percent ‘speed limit’,^²⁶ or an increase of amortization requirements to half of the value of the collateral (from two thirds under current self-regulation) would reduce by 40 percent the risk of new lending in the SORE segment. Likewise, a combination of LTV-DSTI limits of 75–25 percent with a 20 percent ‘speed limit’ or an increase in the amortization rate to 50 percent of the lending value of the property would cut the risk of new vintages by 25 percent in the IPRRE segment. In both cases, the risk of new issuance would be anchored to the risk of the outstanding portfolio (Figure 8).

^{Figure 8.}

Impact of Borrower-Based Tools Under Adverse Scenario by Vintage

Citation: IMF Staff Country Reports 2022, 172; 10.5089/9798400212994.002.A002

Note: The red bars represent the cumulative loss rate under the adverse scenario over the 3-year stress test horizon, and current macroprudential policy settings. The blue bars denote the stressed loss rate under alternative policy tools and calibrations (new instruments). The grey bars show the stressed loss rates under adjustments of self-regulation rules (current instruments).

35. Anchoring the risk of new vintages to the average portfolio using borrower-based limits would save about CHF 1 billion of regulatory CET1 by year. Lowering the risk of new vintages would also ease capital requirements for IRB banks. To estimate the average impact, we follow the next steps. First, using the Basel III IRB supervisory formula, we back out the effective maturity to match Swiss banks risk-weights on the mortgage portfolio to the reported PDs and LGDs. Second, we use the structural model to project PDs and LGDs of new vintages (no limits) and plug them into the IRB formula– adjusted to create Through-the-Cycle (TTC) projections – and calculate capital requirements. Third, using the same procedure we compute the capital required on new vintages (limits targeting the risk of the portfolio), assuming that 10 percent of the portfolio is being added yearly. The difference between the two amounts equals the savings in banks’ regulatory capital from applying the borrower-based tools calibrated in the paper.

G. Policy Implications

36. A real estate crisis could have a substantial impact on macrofinancial stability in Switzerland. The Swiss financial sector has proved resilient despite negative interest rates and challenges related to the pandemic, and more recently, to the war in Ukraine. Default rates have remained at historically low levels. However, the Swiss mortgage market is very large relative to the size of the economy. Swiss banks are highly exposed to mortgages with geographical concentration at the canton level. While Switzerland has not experienced a sharp housing boom in recent years, residential real estate prices have increased persistently, and a wide range of indicators point at rising vulnerabilities in the mortgage and real estate market across sub-segments. The pick-up in housing demand from ultra-low interest rates, search-for-yield, and shifting preferences during the pandemic may prove transitory rendering house prices unsustainable. A sharp price correction would be a drag on economic activity and lower household wealth affecting investment and consumption.^²⁷ Banks could be severely impacted and accentuate downturns through credit rationing, fueling adverse feedback loops to the real economy.

37. The Swiss authorities have rightly re-activated the CCyB buffer to build resilience. After the de-activation of the CCyB buffer in March 2020 to encourage banks to support the real economy during the Covid-19 pandemic, the Federal Council approved SNB’s proposal to reactivate the buffer at its maximum 2.5 percent level in January 2022 (effective from September 2022). This move was intended to increase financial system resilience given clear signs of overvaluation (and of sustained recovery from the pandemic shock). The reactivation complements FINMA’s close work with the banking sector to tighten self-regulation guidelines for investment property mortgages in January 2020.

38. A fuller set of macroprudential instruments could help contain the build-up of vulnerabilities, strengthen resilience, and preserve regulatory capital. While the banking sector is, on average, resilient to a real estate crisis, a deep house-price correction accompanied by a protracted recession could erode banks’ capital buffers. Under the adverse scenario presented in this paper, the 2.5 percent CCyB buffer could absorb 25 percent of cumulative losses over a 3-year stress period. Risks are accumulated in recent vintages as earlier borrowers have repaid more principal, seen larger gains in equity, and were tested at higher lending rates. An introduction of an LTV/DSTI cap with a ‘speed limit,’ or an increase in amortization requirements could guard against the buildup of vulnerabilities in new lending flows by reducing the share of homebuyers that could become financially stretched. This could also help strengthen banking sector resilience by decreasing losses under severe stress and save regulatory capital under baseline conditions. Our calculations suggest that anchoring the risk of new vintages to the average risk of the portfolio would save about CHF 1 billion of regulatory CET1 capital by year. More sustainable mortgage lending would also dampen house price fluctuations, thus supporting financial stability.

39. The Swiss authorities could start considering adding borrower-based tools to the macroprudential toolkit, making use of ‘speed limits’ to minimize impacts on mortgage-market access. One concern is that borrower-based measures could have a disproportionate impact on first-time home buyers or lower-income households. The use of ‘speed limits’ allows flexibility in implementation and can be easier to enforce, minimizing distortions to allocative efficiency. The Swiss authorities could start public consultation with the banking sector and relevant stakeholders to complete the macroprudential framework in case affordability risks keep increasing and new tools need to be activated. Beyond macroprudential measures, adjustments to taxation (e.g., abolition of imputed-rent taxation or phase-out of mortgage interest relief) and actions to support the rental market (e.g., targeted subsidies, social housing) could help decrease affordability risk and support sustainable house prices.

References

Annaheim, M., and Heim, L, (2021), “Household wealth in Switzerland: concepts and trends in an international comparison,” Swiss National Bank.
- Search Google Scholar
- Export Citation
Bank of International Settlements (2016), “Macroprudential Policy,” BIS Papers No. 86, September.
- Search Google Scholar
- Export Citation
Bank of International Settlements (2005), “An Explanatory Note on the Basel II IRB Risk Weight Functions,” July.
- Search Google Scholar
- Export Citation
Basten, C., and Koch, C. (2015), “Higher Bank Capital Requirements and Mortgage Pricing: Evidence from the Countercyclical Capital Buffer (CCyB),” BIS Working Paper No 511.
- Search Google Scholar
- Export Citation
Bloor, C., and Lu, B. (2019), “Have the LVR restrictions improved the resilience of the banking system?,” Reserve Bank of New Zealand Analytical Note Series, May.
- Search Google Scholar
- Export Citation
EBA (2020), “Final guidelines on the appropriate subsets of sectoral exposures to which competent or designated authorities may apply a systemic risk buffer in accordance with Article 133 (5) (f) of Directive 2013/36/EU,” September.
- Search Google Scholar
- Export Citation
EBA (2022), “Overview of national macroprudential measures,” March. Available at https://www.esrb.europa.eu/national_policy/html/index.en.html.
- Search Google Scholar
- Export Citation
ESRB (2014), “Recommendation of the European Systemic Risk Board on guidance for setting countercyclical buffer rates.”
- Search Google Scholar
- Export Citation
FINMA (2021), “FINMA Risk Monitor 2021,” November.
- Search Google Scholar
- Export Citation
Gornicka, L., and Valderrama, L. (2020), “Stress Testing and Calibration of Macroprudential Policy Tools,” IMF Working Paper 20/165, August.
- Search Google Scholar
- Export Citation
Harrison, I., and Mathew, C., (2008), “Project TUI: A Structural Approach to the Understanding and Measurement of Residential Mortgage Lending Risk,” Reserve Bank of New Zealand.
- Search Google Scholar
- Export Citation
Reserve Bank of New Zealand (2022), “Financial Stability Report,” May.
- Search Google Scholar
- Export Citation
Schneider, M. and Wagner, K. (2015), “Housing Markets in Austria, Germany and Switzerland,” Oesterreichische Nationalbank, Monetary Policy & The Economy, Q1/15.
- Search Google Scholar
- Export Citation
Swiss National Bank (2021), “Financial Stability Report,” June.
- Search Google Scholar
- Export Citation
Valderrama, L, (2021), “Macroprudential Policy Calibration in Slovakia,” Selected Issues, Chapter 2, IMF Country Report No. 21/134.
- Search Google Scholar
- Export Citation
Zurbrügg, F., (2022), “Macroprudential policy beyond the pandemic: Taking stock and looking ahead,” International Center of Monetary and Banking Studies, Swiss National Bank.
- Search Google Scholar
- Export Citation

Annex I. Technical Aspects of the Model

Model Calibration

^{Table A.1.}

Key Parameters of the Mortgage Model

Note: The calibration and parameterization of the model uses the loan loss experience of the burst of the real estate bubble in Switzerland and in the UK in the early 1990s. While the Swiss crisis was related to commercial real estate, the UK crisis also affected the owner-occupied residential segment. The calibration procedure using the UK crisis is adjusted to reflect the mortgage characteristics and regulatory framework in Switzerland.

^{Table A.1.}

Key Parameters of the Mortgage Model

Parameter	Value
Calibrate sensitivity of FD to shifts in DSTI	SORE β₁ = 0.0003 γ = 2.5 IPRRE β₁ = 0.00004 γ = 2.5
Calibrate sensitivity of FD to initial DSTI by bucket	$ϕ (D S T I) = {\begin{matrix} (\frac{D S T I - 10 %}{25 % - 10 %}) & 10 % < D S T I < 25 % \\ 1 & o t h e r w i s e \end{matrix}}$
Calibrate sensitivity of FD to initial unemployment	β₂ = 0.06
Allocate financial distress to unemployment shock	$\begin{matrix} (\begin{matrix} U n e m p & 10 % \\ I n t e r e s t & 90 % \end{matrix}) & α = 1 & β_{3} = 0.150 \end{matrix}$
Estimate sensitivity to demographic factors	D = 0.2
Estimate transaction costs	C = 4%
Estimate liquid financial assets (SORE)	$1.15 % \cdot \tilde{P}$
Estimate liquid financial assets (IPRRE)	$10.0 % \cdot \tilde{P}$

^{Table A.1.}

Key Parameters of the Mortgage Model

Parameter	Value
Calibrate sensitivity of FD to shifts in DSTI	SORE β₁ = 0.0003 γ = 2.5 IPRRE β₁ = 0.00004 γ = 2.5
Calibrate sensitivity of FD to initial DSTI by bucket	$ϕ (D S T I) = {\begin{matrix} (\frac{D S T I - 10 %}{25 % - 10 %}) & 10 % < D S T I < 25 % \\ 1 & o t h e r w i s e \end{matrix}}$
Calibrate sensitivity of FD to initial unemployment	β₂ = 0.06
Allocate financial distress to unemployment shock	$\begin{matrix} (\begin{matrix} U n e m p & 10 % \\ I n t e r e s t & 90 % \end{matrix}) & α = 1 & β_{3} = 0.150 \end{matrix}$
Estimate sensitivity to demographic factors	D = 0.2
Estimate transaction costs	C = 4%
Estimate liquid financial assets (SORE)	$1.15 % \cdot \tilde{P}$
Estimate liquid financial assets (IPRRE)	$10.0 % \cdot \tilde{P}$

^{Table A.2.}

Other Parameters of Mortgage Model

Note: Parameters provided by SNB based on expert judgment. The discount rate is the 5-year benchmark mortgage rate.

^{Table A.2.}

Other Parameters of Mortgage Model

Parameter	Value
Foreclosure discount	10%
Time to sell collateral after foreclosure	1 year
Discount rate	1.1%
Spread on discount rate on foreclosed collateral	1.0%

Note: Parameters provided by SNB based on expert judgment. The discount rate is the 5-year benchmark mortgage rate.

^{Table A.2.}

Other Parameters of Mortgage Model

Parameter	Value
Foreclosure discount	10%
Time to sell collateral after foreclosure	1 year
Discount rate	1.1%
Spread on discount rate on foreclosed collateral	1.0%

Note: Parameters provided by SNB based on expert judgment. The discount rate is the 5-year benchmark mortgage rate.

Credit Risk Parameters: From Origination to Point-in-Time (PiT)

We calculate the Point-in-Time (PiT) risk parameters of loans issued from 1990Q1 through 2021Q4 using information on the risk characteristics of the loans, and data provided by SNB on household disposable income, interest rates at origination, and real estate prices. We denote the time of issuance as s, the current period as t, and the maturity of the loan as T.

First, we compute the PiT LTV ratio by backing out the outstanding principal of the loan net of repayments at time t taking into account self-regulation rules on amortization at the time of the issuance and repricing the collateral at market value at time t:

\begin{matrix} L T V_{t, s} = \frac{L T V_{s} \cdot P_{s} - \max ((L T V_{s} - \min_{-} a m o r t_{s}) \cdot P \cdot (1 - \frac{T_{t, s} - (t - s)}{\min_{-} a m o r t_{s}}), 0)}{P_{t}} & (A .1) \end{matrix}

Then, we compute the income PiT using information extracting income at origination from DSTI and mortgage rates at origination, and quarterly income growth g:

\begin{matrix} I n c o m e_{s} = \frac{\max ((\frac{(L T V_{s} - \min_{-} a m o r t_{s}) \cdot P_{s}}{m a t u r i t y}), 0) + L T V_{s} \cdot P_{s} \cdot I n t e r e s t_{s | s}}{D S T I_{s, s}} & (A .2) \end{matrix}

\begin{matrix} I n c o m e_{t} = I n c o m e_{s} \cdot {(1 + g)}^{t - s} & (A .3) \end{matrix}

This allows us to compute DSTI PiT ratio as:

\begin{matrix} D S T I_{t, s} = \frac{\max ((\frac{(L T V_{s} - \min_{-} a m o r t_{s}) \cdot P_{s}}{m a t u r i t y}), 0) + L T V_{s} \cdot P_{s} \cdot I n t e r e s t_{t | s}}{I n c o m e_{t}} & (A .4) \end{matrix}

Similarly for DTI

\begin{matrix} D T I_{t, s} = \frac{\max ((\frac{(L T V_{s} - \min_{-} a m o r t_{s}) \cdot P_{s}}{m a t u r i t y}), 0)}{I n c o m e_{t}} & (A .5) \end{matrix}

where Interest_ts is the lending rate t t of a mortgage issued in s. During the stress test horizon at time t+j, we compute the shock to DSTI as:

\begin{matrix} Δ D S T I_{t + j, s} = \frac{(L T V_{s} \cdot P_{s}) \cdot I n t e r e s t_{t + j | s}^{s + λ}}{I n c o m e_{t} \cdot (1 + s h o c k_{j})} & (A .6) \end{matrix}

where $I n t e r e s t_{t + j | s}^{s + λ}$ is the lending rate as of t+j of a mortgage issued in s and with the last re-setting period of interest rate in s+λ.

Prepared by Laura Valderrama (EUR). The author thanks Mark Horton for helpful comments, Stéphane Riederer and Roland Goetschmann for their suggestions and productive discussions, and colleagues at the Swiss National Bank and FINMAfor providing detailed feedback.

Investment property accounts for 20 and 7 percent of assets for pension funds and insurers, while mortgages represent 3 and 5 percent of assets. These lenders represent 2 and 4 percent of the overall mortgage market, respectively.

The pass-through of the policy rate to benchmark mortgage rates is shown in Figure 2.

The gap between real house price and real household income growth rates rose from 60 bps pre-Covid to 130 bps.

These guidelines are professional rules of the Swiss Banking Association (SBA) for lending, credit monitoring, and reporting. These guidelines have been recognized by FINMA as a minimum regulatory standard in accordance with Circular 2008/10 “Self-regulation as a minimum standard”. The Federal Council changed the Capital Adequacy Ordinance (CAO) for banks to apply a risk weight of 100 percent for the entire loan amount which does not comply with the minimum standard.

Some EU countries have set limits on the investment-led segment, including Ireland and Latvia with 70 percent LTV caps, the Czech Republic with a pre-Covid limit of 60 percent (relaxed to 90 percent during the pandemic), and Belgium with an 80 percent threshold (first-time buyers benefit from a 10 percent exemption).

Affordability risk is considered high when imputed costs from mortgage servicing (including amortization, interest, and maintenance) exceed one-third of income (owner-occupied segment) or rents (investment-led segment). To compute affordability risk, we assume that average LTV is 80 percent and maintenance costs represent 1 percent of loan value.

Other instruments including regulatory provisions that increase buffers for expected losses and minimum risk weights that impose a capital floor on unexpected losses.

To date, only Germany has announced a sectoral SyRB on residential exposures—2 percent from February 2023, in addition to a 0.75 percent CCyB. Others with positive current CCyB (broad-based) rates: Bulgaria, Czech Republic (0.5 percent, to be increased to 1.5 percent and 2 percent, respectively in January 2023); Luxembourg (0.5 percent); Slovakia (1 percent); and Norway (1 percent to be increased to 2 percent in December 2022). Others with announced CCyB increases include Estonia (1 percent in December 2022); Iceland (2 percent in September 2022); Sweden (1 percent in September 2022 and UK (1 percent in December 2022 and 2 percent in June 2023).

While a tightening of lender’s underwriting standards will help contain mortgage growth, self-regulation guidelines are not designed to reduce systemic risk and thus help to prevent crisis, by contrast with macroprudential authorities’ goals.

The modeling approach follows Gornicka and Valderrama (2020), which builds on Harrison and Mathew (2008). A general application of this model is found in IMF (2021).

This definition is well established in the legal literature whereby a borrower becomes insolvent when he/she fails the ‘cash-flow test’ (cannot repay the debt as it comes due) and ‘the balance sheet test’ (the value of its assets fall below the value of its liabilities).

By contrast, the model does not consider “strategic defaults,” i.e., a situation where the borrower decides to stop repayments once the value of the underlying collateral falls below the value of the loan. Incentives to do so might exist in the case of non-recourse loans, which is not the case in Switzerland.

This is to capture the empirical observation that higher DSTI buckets are more sensitive to increases in servicing costs (UK crisis).

The penalty for early prepayment decreases with the current value of interest rates. The model uses the conservative assumption that the bank charges a positive penalty for early pre-payment even if market rates rise calculated as the net present value of foregone interest rate payments for a specified number of months.

The adjusted self-regulation rules now require borrowers to provide a minimum down payment of at least 25 percent of the LTV. As The LTV bucket segmentation used for the analysis is in 10pps increments we show the statistics at the 70 percent cap which lies near the 75 percent threshold.

The share of loans with a DTI higher than 6 has increased from about 35 to 50 percent over 2017–2021 in both, the SORE and the IPRRE segments.

E.g., Denmark (exception for wealthy borrowers); UK (15-percent exemption), Sweden (where higher LTIs trigger additional amortization). A cap of 3.5 is set in Ireland (20-percent first-time borrower exemption). Others have implemented a DTI limit with a broader debt concept (Norway—5; Latvia—6, 10-percent exemption; Slovakia—8 of net disposable income or 6.4 gross). A caveat to the comparison is that income is defined on a net basis in the SNB survey on new mortgages whereas some countries may use a gross definition of income.

A challenge of using aggregate mortgage stock data is that its coverage is broader than the granular dataset used to calibrate the model drawing on the SNB’s HypoB survey data which excludes refinancing loans with the same lender. For the calibration of the model, we also excluded commercial real estate (CRE) from the HypoB survey.

This is likely driven by mortgage loans for investment properties. But the repricing data is not broken down by sub-segment.

Switzerland experienced a house price bubble in the 1980s that burst in the early 1990s (Schneider and Wagner, 2015). During 1989–91, the mortgage variable ratejumped from 5.0 to 7.8 percent (peaking at 8.0 in Feb 1991), real estate prices dropped by 16 percent peak-to-trough, unemployment increased by over 50 basis points, and the estimated loss rate of the portfolio increased by over 50 basis points to 1.03 percent.

We use an econometric model to forecast rental payments over the stress test horizon as a function of change in real estate prices, household income, and mortgage rates.

While the Swiss Banking Association guidance stipulates that mortgage borrowers should meet an imputed affordability criteria at an assumed 5 percent interest rate, in practice there are numerous exemptions to this rule as Figure 3 illustrates.

The Reserve Bank of New Zealand (RBNZ) was relatively unique in using temporary LTV restrictions with speed limits in mid-2013 (BIS, 2016).

See Zurbrügg (2022) for a discussion of macroprudential policy in Switzerland and implementation lags.

A ‘speed limit’ allows for a certain proportion of the volume of new loans to be exempt from a particular measure. These limits could be unconditional or targeted to specific types of loans (e.g., first-time home buyers, green mortgage loans, etc.).

In Switzerland, real estate assets represent over half of total household net wealth.

Other Publishers

Save
Cite
Email this content

Share Link

Copy this link, or click below to email it to a friend
Email this content
or copy the link directly:

https://www.elibrary.imf.org/view/journals/002/2022/172/article-A002-en.xml
The link was not copied. Your current browser may not support copying via this button.

Link copied successfully