## 1. Uptrend in PTI and DTI Ratios in Selected Advanced Economies

## Source details

**Canonical URL:** [1. Uptrend in PTI and DTI Ratios in Selected Advanced Economies](https://www.imf.org/-/media/files/publications/wp/2018/wp18164.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18164.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18164.pdf.json)

---

### Introduction: recent patterns and macro-financial risks
- Over the past two decades, house prices have risen faster than income in many advanced economies (AEs), producing a strong uptrend in price-to-disposable income (PTI) ratios and similar increases in household debt-to-disposable income (DTI) ratios.
- Sizable reversals around the global financial crisis (GFC) occurred in Denmark, Ireland, and Spain; prices have recovered to some extent but remained below their pre-crisis peaks.
- Other AEs experienced almost uninterrupted booms with double-digit house price inflation and record-high price levels and household debt ratios, e.g., Sweden and Norway.
- Macro-financial risks:
  - Price corrections can cause household deleveraging, reduced consumption, and weakened financial intermediation.
  - High indebtedness amplifies shocks via the collateral channel.

### Fundamental valuation benchmarks and modeling approach
- Benchmarks and limitations:
  - PTI (price-to-disposable income) ratio: useful for affordability but omits other fundamentals (e.g., interest rates).
  - PTR (price-to-rent) ratio: compares house prices to user cost of housing; informative where rental data are reliable but less useful where rent controls distort rents (e.g., Sweden) or rental markets are thin (e.g., Norway).
  - Multivariate econometric modeling including disposable income, interest rates, demographics, and supply factors offers an alternative to estimate disequilibrium in house prices.
- Scope:
  - Aim: assess housing valuation risks by modelling sustainable levels of house prices for 20 AEs in the OECD.
  - Countries: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Israel, Italy, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, United States.
  - Novel focus: policy, institutional, and structural factors (tax incentives for home ownership, rent controls, long-run housing supply responsiveness) using a cross-country panel methodology.

### II. Fundamental determinants of house prices — demand, supply, and institutional factors
- Demand factors (long-run):
  - Positive: real per-capita disposable income (RPDI), household net financial wealth, demographic needs.
  - Negative: user costs and the housing stock.
  - Key empirical notes:
    - Average annual growth in RPDI is positively correlated with house price growth; bivariate slope exceeds unity.
    - For Sweden since the mid-1990s: about 60 percent of the rise in house prices explained by increased real disposable income; the rise in household real financial wealth accounts for slightly under 10 percent (Claussen (2013)).
    - Semi-elasticity of real housing prices w.r.t. interest rates in AEs ranges between close to zero and 6 percent.
    - Demographics (population growth and share at household-formation ages) boosted housing demand in Australia, Ireland, Israel, New Zealand, Norway.
- Supply factors:
  - Undersupply and lags in supply response can lead to price gains outpacing incomes.
  - Examples: Norway and Israel—lags in supply response increased population-to-dwelling ratios and prices; Sweden—prolonged underinvestment a key driver.
  - Estimated long-run price elasticities of new housing supply (quarterly 1989–2016, Caldera Sanchez et al. (2011) methodology):
    - Variation across 20 countries from about 0.2 in Switzerland to 1.6 in the US.
    - Long-run supply elasticity differences materially affect price responses to demand shifts.
- Structural and institutional factors:
  - Supply elasticity differences reflect topographical and man-made constraints (stringent land-use regulations, building permit procedures, construction capacity).
  - Tax treatment:
    - Capital gains taxes often exempted/deferred/reduced for principal residences.
    - Recurrent property taxes often of low economic importance due to low rates or outdated cadastral values.
  - Mortgage interest deductibility (MID):
    - Tends to incentivize borrowing and is capitalized into house prices.
    - MID often capped, but unbounded in some countries (Netherlands, Sweden, Norway).
    - Reforms: Portugal removed MID since 2012 (loans before end-2011 retain a tax credit at 15 percent up to a ceiling); Spain removed MID since 2013 (loans before end-2012 retain 15 percent credit up to a ceiling); Ireland removed MID from 2018. Denmark, Finland, Netherlands adopted gradual/moderate reductions.
    - OECD tax relief index (updated to 2016 reforms): as of 2016, tax relief most generous in the Netherlands and effectively zero in countries where mortgage loans are not tax favored.
  - Rent control:
    - Can reduce efficient use of housing, create lock-in effects, reduce effective supply, and raise house prices.
    - Sweden: most stringent rent controls in OECD sample, average waiting times for rental apartments of 10 years; combined with MID, this incentivized purchases and reduced rental supply.
    - Rent control index rescaled to 0–1 in the model (original scale 0–6 used in figures).

### III. Cross-country model of long-run house prices — theory and specification
- Theoretical setup:
  - Long-run equilibrium price p* where demand equals stock.
  - Long-run determinants: real per capita household disposable income (y), user cost (proxied by tax relief index and real mortgage rates), real per capita household net financial wealth (w), housing stock per capita (s).
  - Observed prices deviate from equilibrium by a stationary error term (cointegrating relationship).
- Practical model features:
  - User cost proxied by updated OECD tax relief index (ti), interacted with income to capture rising value of tax incentives with income.
  - Squared real mortgage rate included to allow non-linearities following present-value logic.
  - Rent control index (rescaled to 0–1) interacted with housing supply per capita.
  - Demeaned long-run supply elasticities (es) interacted with demand variables for country-specific variation.
- Reduced-form long-run relationship:
  - Real house prices modelled as function of demand variables (income and interactions), wealth, supply per capita and interactions with rent control, tax relief interacted with income, squared real mortgage rate, plus country fixed effects; residuals measure valuation gaps.
- Credit excluded due to endogeneity concerns (higher house prices increase borrowing via collateral).

### Empirical findings and estimation diagnostics
- Estimation sample: 2042 observations for 20 advanced countries in the OECD over 1990: Q3–2016: Q4.
- Robust standard errors clustered at the country level.
- Residuals confirmed stationary; equation (4) is a cointegrating relationship.
- Panel cointegration and unit-root test results for Model (5):
  - Kao (Engle-Granger based) t-Statistics: -3.806; Prob. 0.0001
  - Levin, Lin & chu t-Statistics: -2.705; Prob. 0.003
- Model fit and structure:
  - Observations: 2042
  - Number of countries: 20
  - Country fixed-effect: Y
  - Corrected for heteroskedasticity: Y (robust clustered SEs)
  - Adj. R-squared across models: 0.853, 0.853, 0.856, 0.857, 0.867 (models 1–5 respectively)

### Key estimated long-run coefficients (dependent variable: log of real house prices)
- y, log (real per capita disposable income):
  - Coefficients: 1.652; 1.638; 1.538; 1.544; 1.533
  - Significance: *** at 1 percent for all columns shown.
- morr, percent (real mortgage rate):
  - Coefficients: -1.922; -2.759; -2.234; -2.116; -1.776
  - Significance: *** for all reported.
- morr^2, percent:
  - Coefficients: 0.079; 0.066; 0.051; 0.058
  - Significance: **, **, (no asterisks), * respectively per columns shown.
- w, log (real per capita household net financial wealth):
  - Coefficients: 0.031; 0.033; 0.023; 0.020; 0.056
  - Significance: ***, ***, **, **, ** respectively per columns shown.
- s, percent (housing stock per capita):
  - Coefficients: -1.070; -1.080; -0.943; -1.267; -1.322
  - Significance: *** for all reported.
- tr * y (log) (tax relief interaction with income):
  - Coefficients: 0.362; 0.351; 0.487
  - Significance: ***
- rc * s (percent) (rent control interaction with housing stock):
  - Coefficients: 1.156; 0.436
  - Significance: *** and * respectively.
- sr * y (log): -0.007 (standard error [0.141], not significant)
- sr * morr: 1.133 (standard error [0.154], ***)
- sr * w (log): -0.060 (standard error [0.033], *)

### Interpretation of elasticities and cross-country variation
- Income elasticity:
  - A 1 percent rise in per capita disposable income raises long-run equilibrium house prices by a cross-country average of 1.5–1.7 percent, suggesting housing is a luxury good.
  - Country examples: impacts up to 2.1 percent in the Netherlands and as low as 1.6 percent in countries where housing finance is not tax favored (example: Canada), depending on tax relief and long-run supply elasticities.
- Interest-rate elasticity:
  - A 1 percentage point increase in the real mortgage rate reduces real house prices by a cross-country average of about 1.8–2.8 percent.
  - With varying long-run supply elasticities, the same increase in mortgage rates can have significantly different long-run impacts (amplified in less elastic markets like Switzerland; moderated in more elastic markets like the US).
  - A 2 percentage point increase in the real mortgage rate would reduce house prices by about 4–6 percent in equilibrium.
- Wealth and supply effects:
  - Real per capita household net financial wealth has a small positive impact on real house prices.
  - A 1 percent increase in the housing stock per capita is associated with a reduction in house prices by about 1.3 percent in countries with no rent control.
  - Rent controls offset part of the dampening effects of supply increases:
    - Sweden example: with stringent rent control, a 1 percent increase in housing stock per capita reduces long-run house prices by only 0.9 percent compared with 1.3 percent in countries without rent control.

### Estimated valuation gaps, scenarios, and country groupings (as of 2016: Q4)
- Average estimated overvaluation on current fundamentals (including policy and structural factors): modest at 6 percent on average across sample.
- For Model (5), estimated housing valuation gaps as of 2016: Q4:
  - Range from 12 percent undervaluation in Finland to 35 percent overvaluation in Canada.
- Interest-rate normalization scenario:
  - Because real mortgage rates are well below their average since 2000, normalization could lower equilibrium housing prices and increase measured overvaluation.
  - Example implication: a 2 percentage point increase in the real mortgage rate would reduce equilibrium house prices by about 4–6 percent, implying house prices could be up to 12 percent overvalued on average in the sample countries under such a normalization.
- Country groupings as of end-2016 (model-based):
  - Estimated to remain undervalued: Denmark, Finland, Germany, and the US.
  - Broadly fairly-valued: Ireland, Israel, Portugal, and Spain.
  - Back into overvaluation territory: Australia, Austria, the Netherlands, and New Zealand.
  - Estimated overvalued as of end-2016: Canada, Norway, Sweden, and the United Kingdom.
  - Experiencing persistent overvaluation: Belgium, France, Italy, Switzerland.
- Cautions:
  - Estimated equilibrium price levels subject to uncertainty.
  - For some countries (Belgium, France, Germany), estimated gaps are prolonged and actual prices cross equilibrium only once, suggesting model limitations or omitted characteristics.
  - Country-level valuations can conceal heterogeneous regional developments (example: Canada where Toronto and Vancouver have much lower supply elasticities than the rest of Canada).

### Policy implications and recommendations
- Policies, institutional, and structural factors (tax relief, rent controls, long-run supply responsiveness) play an important role in shaping long-run house prices.
- Recommended structural reforms alongside macroprudential instruments:
  - Reforms to raise the long-run elasticity of housing supply.
  - Phasing out rent control.
  - Reducing tax incentives for home ownership and debt financing.
- Rationale:
  - Structural reforms can improve housing affordability over time, reduce debt accumulation, and enhance financial stability.
  - Such reforms may complement macroprudential tools (e.g., limits on loan-to-value ratios) by shaping longer-term expectations in the housing market.

### Annex I — selected variable definitions and data sources
- General notes:
  - All variables are in log terms except for mortgage rates, the housing stock to population ratio, the tax relief and rent control indices, and long-run supply elasticities.
  - Robust standard errors clustered at the country level. Estimation sample: 2042 observations for 20 advanced countries over 1990: Q3–2016: Q4.
- Selected entries:
  - y: Real house price index, in log terms, seasonally adjusted. Source: OECD.
  - y (disposable income): Real per capita household personal disposable income, in log terms, seasonally adjusted. Source: Haver Analytics, national statistics websites.
  - morr: Real mortgage rate, in percent. Calculated as weighted nominal mortgage rate (all maturities) minus the HICP inflation rate. Source: Haver Analytics, national statistics websites.
  - w: Real per capita household net financial wealth, in log terms. Source: OECD, Haver Analytics, and national statistics websites. Quarterly data generated by interpolating annual data using a cubic spline.
  - s: Housing stock per capita, ratio. Source: OECD, Haver Analytics, national statistics websites, and country authorities.
  - tr: Index of tax relief on housing finance. Accounts for deductibility of mortgage interest, limits on period/amount of deduction, tax credits for loans, and taxation of imputed rent. Source: OECD, ESRB (2015), IMF country staff reports, country authorities, Fund staff calculations.
  - rc: Index of the strictness of rent controls. Composite indicator of extent of controls, determination of rent increases, and permitted pass-through of cost into rents. Source: OECD, Cuerpo et. al. (2014), IMF country staff reports, Fund staff calculations.
  - sr: Long-run elasticity of real dwelling investment w.r.t. real housing prices. Estimated using 1989: Q1–2016: Q4 data based on Caldera Sanchez et. al. (2011) methodology.

*Source: IMF working paper (wp18164), chapter "1. Uptrend in PTI and DTI Ratios in Selected Advanced Economies"; canonical source URL: https://www.imf.org/-/media/files/publications/wp/2018/wp18164.pdf*

### 1. Uptrend in PTI and DTI Ratios in Selected Advanced Economies..............................4

### 1. Uptrend in PTI and DTI Ratios in Selected Advanced Economies

### Introduction: recent patterns and risks
- Over the past two decades, house prices have risen faster than income in many advanced economies (AEs), producing a strong uptrend in price-to-disposable income (PTI) ratios and similar increases in household debt-to-disposable income (DTI) ratios.
- Sizable reversals around the global financial crisis (GFC) occurred in Denmark, Ireland, and Spain; prices have recovered to some extent but remained below their pre-crisis peaks.
- Other AEs experienced almost uninterrupted booms with double-digit house price inflation and record-high price levels and household debt ratios, e.g., Sweden and Norway.
- Macro-financial risks:
  - Price corrections can cause household deleveraging, reduced consumption, and weakened financial intermediation.
  - High indebtedness amplifies shocks via the collateral channel.

### Fundamental valuation benchmarks and limitations
- Common benchmarks:
  - PTI (price-to-disposable income) ratio: useful for affordability but not ideal for sustainability because it omits other fundamentals (e.g., interest rates).
  - PTR (price-to-rent) ratio: compares house prices to user cost of housing and can summarize a range of fundamentals; informative for markets like the U.S., but less useful where rental data are distorted by rent controls (e.g., Sweden) or where rental markets are thin (e.g., Norway).
- Alternative: multivariate econometric modeling including disposable income, interest rates, demographics, and supply factors to estimate disequilibrium in house prices.

### Scope of the paper
- Aim: assess housing valuation risks by modelling sustainable levels of house prices for 20 AEs in the OECD.
- The 20 countries in the sample are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, Israel, Italy, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and United States.
- Novel contribution: focus on policy, institutional, and structural factors (tax incentives for home ownership, rent controls, long-run housing supply responsiveness) using a cross-country panel methodology.

### Structure
- Section II: driving forces (demand, supply, institutional/structural).
- Section III: cross-country model of long-run equilibrium housing prices including policy and structural factors.
- Section IV: estimation results for 20 OECD countries.
- Section V: conclusion.

---

### II. Fundamental Determinants of House Prices

#### Demand factors
- Long-run positive determinants: real per-capita disposable income (RPDI), household net financial wealth, demographic needs.
- Long-run negative determinants: user costs and the housing stock.
- Key empirical relationships:
  - Average annual growth in RPDI is positively correlated with house price growth; the bivariate relationship slope exceeds unity.
  - Household net financial wealth accumulation has exerted upward pressure on housing demand.
    - Example: Claussen (2013) — for Sweden since the mid-1990s, about 60 percent of the rise in house prices explained by increased real disposable income; the rise in household real financial wealth accounts for slightly under 10 percent.
  - Interest rates have fallen substantially since 2000; real rates have been close to or below zero in many countries, reducing the user cost of housing.
    - From literature on AEs, the semi-elasticity of real housing prices with respect to interest rates ranges between close to zero and 6 percent.
  - Demographics: population growth and increases in the share at household-formation ages boosted housing demand (noted in Australia, Ireland, Israel, New Zealand, Norway).

#### Supply factors
- Undersupply and lags in supply response can lead to price gains outpacing incomes.
  - Example: Norway and Israel experienced lags in supply response, increasing the ratio of population to stock of dwellings and associate price rises.
  - Sweden: prolonged underinvestment identified as a key driver of house price inflation.
- Estimated long-run price elasticities of new housing supply (quarterly data 1989–2016, Caldera Sanchez et al. (2011) methodology):
  - Variation across the 20 countries from about 0.2 in Switzerland to 1.6 in the US.
  - Implication: long-run supply elasticity differences materially affect price responses to demand shifts.

#### Structural and institutional factors
- Supply elasticity differences reflect topographical constraints and man-made constraints (stringent land-use regulations, building permit procedures, construction capacity).
- Tax incentives and user cost:
  - Tax treatment of housing often favorable relative to other investment and can reduce user cost of housing, contributing to higher house prices and leverage.
  - Typical features:
    - Capital gains taxes often exempted/ deferred/ reduced for principal residences.
    - Recurrent property taxes may be of low economic importance due to low rates or outdated cadastral values.
  - Mortgage interest deductibility (MID) tends to incentivize borrowing and is capitalized into house prices.
    - MID often capped, but unbounded in some countries (Netherlands, Sweden, Norway).
    - Reforms: Portugal removed MID since 2012 (loans before end-2011 retain a tax credit at 15 percent up to a ceiling); Spain removed MID since 2013 (loans before end-2012 retain 15 percent credit up to a ceiling); Ireland removed MID from 2018. Denmark, Finland, Netherlands adopted gradual/moderate reductions.
  - OECD tax relief index (updated to 2016 reforms):
    - As of 2016, tax relief most generous in the Netherlands and effectively zero in countries where mortgage loans are not tax favored.
- Rent control:
  - Rent controls can reduce efficient use of housing, create lock-in effects, reduce effective supply, and raise house prices.
  - Sweden cited as having the most stringent rent controls among OECD countries, with average waiting times for rental apartments of 10 years; combined with mortgage interest deductibility, this has incentivized purchases and reduced rental supply.
  - Rent control index updated from OECD (rescaled to 0–1 in later model; original scale shown in figures uses 0–6, increasing in degree of control).

---

### III. A Cross-Country Model of Long-Run House Prices

- Theoretical setup:
  - Long-run equilibrium price of housing p* is that at which demand for housing equals the stock of housing.
  - Long-run determinants include real per capita household disposable income (y), user cost of housing (proxied by tax relief index and real mortgage rates), real per capita household net financial wealth (w), and housing stock per capita (s).
  - Observed prices deviate from long-run equilibrium by an error term; the error term is expected to be stationary (cointegrating relationship).
- Practical model features and variables:
  - User cost ideally captured by real after-tax mortgage rate; proxy used is the updated OECD tax relief index (ti), interacting tax relief with income to capture rising value of tax incentives with income.
  - A squared term of the real mortgage rate included to allow non-linearities following present-value logic.
  - Updated OECD rent control index (rescaled to 0–1) included and interacted with housing supply per capita, since rent control is expected to hinder efficient use of existing stock.
  - Demeaned long-run supply elasticities (es) are interacted with demand variables to allow country-specific variation in long-run impact of demand shocks.
- Estimated cross-country panel specification (reduced-form long-run relationship):
  - Real house prices are modelled as a function of:
    - Demand variables (income, interactions with supply elasticity and tax relief),
    - Wealth,
    - Supply per capita and interactions with rent control,
    - Tax relief interacted with income,
    - Squared real mortgage rate,
    - Country fixed effects and an error term used to measure valuation gaps.
- Note on credit: credit is not included in the model because it can be endogenous to house prices (higher house prices increase borrowing need and collateral).

---

*Source: IMF working paper (wp18164), chapter "1. Uptrend in PTI and DTI Ratios in Selected Advanced Economies"; canonical source URL: https://www.imf.org/-/media/files/publications/wp/2018/wp18164.pdf*

### Annex I summarizes the definition of variables and data sources. All variables are in log

### Annex I summarizes the definition of variables and data sources. All variables are in log terms except for mortgage rates, the housing stock to population ratio, the tax relief and rent control indices, and long-run supply elasticities. In addition, robust standard errors are clustered at the country level to allow for an arbitrary variance-covariance matrix within each country. The estimation sample includes 2042 observations for 20 advanced countries in the OECD over the period of 1990: Q3–2016: Q4.

### Empirical findings: long-run determinants and estimation diagnostics
- Estimation sample: 2042 observations for 20 advanced countries in the OECD over the period of 1990: Q3–2016: Q4.
- Robust standard errors clustered at the country level.
- Residuals confirmed to be stationary; equation (4) is a cointegrating relationship.
- Panel cointegration and unit-root test results for Model (5):
  - Kao (Engle-Granger based) t-Statistics: -3.806; Prob. 0.0001
  - Levin, Lin & chu t-Statistics: -2.705; Prob. 0.003
- Model fit and structure:
  - Observations: 2042 (for each reported model)
  - Number of countries: 20
  - Country fixed-effect: Y (included)
  - Corrected for heteroskedasticity: Y (robust clustered SEs)
  - Adj. R-squared across models: 0.853, 0.853, 0.856, 0.857, 0.867 (models 1–5 respectively)

### Key estimated long-run coefficients (selected from Table 2; dependent variable: log of real house prices)
- y, log (real per capita disposable income): 1.652; 1.638; 1.538; 1.544; 1.533
  - Standard errors reported in brackets; significance: *** at 1 percent for all columns shown.
- morr, percent (real mortgage rate): -1.922; -2.759; -2.234; -2.116; -1.776
  - All reported as statistically significant at 1 percent (***).
- morr^2, percent: 0.079; 0.066; 0.051; 0.058
  - Significance: **, **, (no asterisks), * respectively per columns shown.
- w, log (real per capita household net financial wealth): 0.031; 0.033; 0.023; 0.020; 0.056
  - Significance: ***, ***, **, **, ** respectively per columns shown.
- s, percent (housing stock per capita): -1.070; -1.080; -0.943; -1.267; -1.322
  - All reported as statistically significant at 1 percent (***).
- tr * y (log) (tax relief interaction with income): 0.362; 0.351; 0.487
  - Significance: *** across reported columns.
- rc * s (percent) (rent control interaction with housing stock): 1.156; 0.436
  - Significance: *** and * respectively.
- sr * y (log): -0.007 (standard error [0.141], not significant)
- sr * morr: 1.133 (standard error [0.154], ***)
- sr * w (log): -0.060 (standard error [0.033], *)

### Interpretation of main elasticities and cross-country variation
- Income elasticity:
  - A 1 percent rise in per capita disposable income raises long-run equilibrium house prices by a cross-country average of 1.5–1.7 percent, suggesting housing is a luxury good.
  - Cross-country example: Depending on tax relief and long-run supply elasticities, a 1 percent increase in per capita disposable income results in country-specific impacts ranging up to 2.1 percent in the Netherlands and as low as 1.6 percent in countries where housing finance is not tax favored (example: Canada).
- Interest-rate elasticity:
  - A 1 percentage point increase in the real mortgage rate reduces real house prices by a cross-country average of about 1.8–2.8 percent.
  - With varying long-run supply elasticities, the same increase in mortgage rates can have significantly different long-run impacts; amplified in less elastic markets (example: Switzerland) and moderated in more elastic markets (example: the US).
  - A 2 percentage point increase in the real mortgage rate would reduce house prices by about 4–6 percent in equilibrium.
- Wealth and supply effects:
  - Real per capita household net financial wealth has a small positive impact on real house prices.
  - A 1 percent increase in the housing stock per capita is associated with a reduction in house prices by about 1.3 percent in countries with no rent control.
  - Rent controls offset part of the dampening effects of supply increases on real house prices:
    - Example: Sweden (most stringent rent control in sample): a 1 percent increase in the housing stock per capita reduces long-run house prices by only 0.9 percent compared with 1.3 percent in countries without rent control.

### Estimated valuation gaps and scenarios
- Average estimated overvaluation on current fundamentals (including policy and structural factors): modest at 6 percent on average across sample.
- For Model (5), estimated housing valuation gaps as of 2016: Q4:
  - Range from 12 percent undervaluation in Finland to 35 percent overvaluation in Canada.
- Interest-rate normalization scenario:
  - Because real mortgage rates are well below their average since 2000, normalization could lower equilibrium housing prices and increase measured overvaluation.
  - Example implication: a 2 percentage point increase in the real mortgage rate would reduce equilibrium house prices by about 4–6 percent, implying house prices could be up to 12 percent overvalued on average in the sample countries under such a normalization.
- Country groupings as of end-2016 (model-based, grouping not definitive):
  - Estimated to remain undervalued: Denmark, Finland, Germany, and the US.
  - Broadly fairly-valued: Ireland, Israel, Portugal, and Spain.
  - Back into overvaluation territory: Australia, Austria, the Netherlands, and New Zealand.
  - Estimated overvalued as of end-2016: Canada, Norway, Sweden, and the United Kingdom.
  - Experiencing persistent overvaluation: Belgium, France, Italy, Switzerland.
- Cautions:
  - Estimated equilibrium price levels are subject to uncertainty.
  - For some countries (Belgium, France, Germany), estimated gaps are prolonged and actual prices cross equilibrium only once, suggesting model limitations or omitted characteristics.
  - Country-level valuations can conceal heterogeneous regional developments (example: Canada where Toronto and Vancouver have much lower supply elasticities than the rest of Canada).

### Policy implications and recommendations
- Policies, institutional, and structural factors (tax relief, rent controls, long-run supply responsiveness) play an important role in shaping long-run house prices.
- Structural reforms recommended alongside macroprudential instruments:
  - Reforms to raise the long-run elasticity of housing supply.
  - Phasing out rent control.
  - Reducing tax incentives for home ownership and debt financing.
- Rationale:
  - Structural reforms can improve housing affordability over time, reduce debt accumulation, and enhance financial stability.
  - Such reforms may complement macroprudential tools (e.g., limits on loan-to-value ratios) by shaping longer-term expectations in the housing market.

### Annex I — Variable definitions and data sources (selected entries)
- y: Real house price index, in log terms, seasonally adjusted. Source: OECD.
- y (disposable income): Real per capita household personal disposable income, in log terms, seasonally adjusted. Source: Haver Analytics, national statistics websites.
- morr: Real mortgage rate, in percent. Calculated as weighted nominal mortgage rate (all maturities) minus the HICP inflation rate. Source: Haver Analytics, national statistics websites.
- w: Real per capita household net financial wealth, in log terms. Source: OECD, Haver Analytics, and national statistics websites. Quarterly data generated by interpolating annual data using a cubic spline.
- s: Housing stock per capita, ratio. Source: OECD, Haver Analytics, national statistics websites, and country authorities.
- tr: Index of tax relief on housing finance. Indicator accounts for deductibility of mortgage interest, limits on period/amount of deduction, tax credits for loans, and taxation of imputed rent. Source: OECD, ESRB (2015), IMF country staff reports, country authorities, and Fund staff calculations.
- rc: Index of the strictness of rent controls. Composite indicator of extent of controls, determination of rent increases, and permitted pass-through of cost into rents. Source: OECD, Cuerpo et. al. (2014), IMF country staff reports, and Fund staff calculations.
- sr: Long-run elasticity of real dwelling investment with respect to real housing prices. Estimated using 1989: Q1–2016: Q4 data based on methodology in Caldera Sanchez et. al. (2011).

*Source: IMF Working Paper (content unit: wp18164 - Annex I summarizes the definition of variables and data sources).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18164.pdf_