## Don't Look Up: House Prices in Emerging Europe (WP/22/236) — Sections 1–3

## Source details

**Canonical URL:** [Don't Look Up: House Prices in Emerging Europe (WP/22/236) — Sections 1–3](https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022236-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2022/english/wpiea2022236-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2022/english/wpiea2022236-print-pdf.pdf.json)

---

### Abstract and primary findings
- Study scope: quarterly data covering 10 countries over the period 1998–2022 using a panel quantile regression approach.
- Key empirical conclusions:
  - Economic, financial and demographic factors explain changes in real house prices in emerging Europe, with income growth having the most significant impact.
  - Income growth matters more for higher housing prices than for lower quantiles of the property market (e.g., the coefficient on real GDP growth for the 25th percentile is smaller than for the 75th percentile; the difference is larger when comparing the 5th and 95th percentiles).
  - Increases in short-term or long-term interest rates have a price-dampening impact: a higher cost of borrowing is associated with lower real house prices.
  - Interest rate effects are not statistically significant at all points of the distribution; the interest rate elasticity becomes more significant across the distribution when using long-term interest rates.
- Robustness checks (including use of real household disposable income and household debt-to-income ratio, and IV quantile regression) confirm the baseline results.
- Policy-relevant implication: the downturn in house prices could deepen with the looming economic recession and soaring interest rates, and lower quantiles of the market are significantly more vulnerable to the end of cheap borrowing and slower income growth.

### Data overview and sample
- Sample: 10 emerging European markets (the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Russia, Slovenia, the Slovak Republic, and Türkiye) over 1998–2022.
- Housing price series: BIS nationwide residential property price index in real terms (harmonized series).
- Baseline explanatory variables (quarterly): Real GDP growth; Consumer price inflation; Unemployment rate; Short-term or long-term interest rates; Stock market returns in real terms; Real effective exchange rate (REER); Population growth.
- Granular variables used in robustness checks: Real household gross disposable income growth; Household debt-to-income ratio.
- Stationarity: Im-Pesaran-Shin (2003) panel unit root tests indicate variables are stationary in first-differences or after logarithmic transformation (test results available upon request).

### Summary statistics (selected)
- Real House Price Growth: Observations 620; Mean 2.7; Std. Dev. 12.3; Skewness -0.9; Kurtosis 8.1; Minimum -59.5; Maximum 41.5
- Real GDP Growth: Observations 620; Mean 2.6; Std. Dev. 4.9; Skewness -1.3; Kurtosis 6.9; Minimum -20.7; Maximum 20.2
- Short-Term Interest Rates: Observations 648; Mean 5.2; Std. Dev. 7.7; Skewness 1.7; Kurtosis 4.9; Minimum -0.6; Maximum 33.8
- Long-Term Interest Rates: Observations 651; Mean 5.0; Std. Dev. 4.1; Skewness 1.3; Kurtosis 5.4; Minimum -0.4; Maximum 22.7
- Unemployment Rate: Observations 654; Mean 8.2; Std. Dev. 3.6; Skewness 0.8; Kurtosis 3.4; Minimum 1.9; Maximum 21.3
- Population Growth: Observations 609; Mean -0.1; Std. Dev. 0.8; Skewness 0.1; Kurtosis 4.0; Minimum -2.3; Maximum 1.8
- Consumer Price Inflation: Observations 620; Mean 4.0; Std. Dev. 4.8; Skewness 3.6; Kurtosis 30.0; Minimum -3.9; Maximum 55.4
- Stock Market Returns: Observations 651; Mean 5.5; Std. Dev. 28.1; Skewness 0.4; Kurtosis 5.1; Minimum -81.0; Maximum 113.2
- REER: Observations 660; Mean 101.2; Std. Dev. 11.2; Skewness -0.2; Kurtosis 6.7; Minimum 52.1; Maximum 156.2
- Real Household Gross Disposable Income Growth: Observations 536; Mean 4.8; Std. Dev. 5.6; Skewness 0.7; Kurtosis 4.9; Minimum -11.6; Maximum 30.0
- Household Debt-Income Ratio: Observations 613; Mean 47.0; Std. Dev. 20.3; Skewness -0.2; Kurtosis 2.3; Minimum 2.2; Maximum 92.8
- Data sources: BIS; European Commission; Eurostat; Haver Analytics; OECD; and authors' calculations.

### Econometric strategy
- Method: panel quantile regression with fixed effects (Machado and Santos Silva, 2019) applied to quarterly first-differences of the log housing price series.
- Baseline model specification:
  - ∆hp_{i,t,τ} = θ_τ + γ_τ X_{i,t} + η_i + μ_t + ε_{i,t,τ}
  - Where ∆hp_{i,t,τ} is the logarithmic change in housing prices at quantile τ; X_{i,t} contains the explanatory variables listed above; η_i are country fixed effects; μ_t are time effects; ε_{i,t,τ} is an idiosyncratic error term.
- Rationale for quantile approach:
  - Captures heterogeneous effects of explanatory variables across the distribution of house prices.
  - Robust to outliers and non-normality and allows identification of conditional heterogeneous covariance effects.
- Robustness checks include IV quantile regression (Chernozhukov and Hansen, 2008) instrumenting real GDP growth with the trade-weighted average real GDP growth of trading partners, and separate pre- and post-GFC estimations.

### Key empirical results — Income and earnings
- Real GDP growth is the most important factor determining real house prices; income convergence is a key channel.
- Heterogeneous impact across the conditional distribution:
  - Reported coefficients on Real GDP Growth (Table 2): 1.489*** (OLS); 1.266** (FE); 1.403*** (5th Percentile); 1.487*** (25th Percentile); 1.575*** (50th Percentile); 1.698*** (95th Percentile).
  - Text-highlighted values: 25th percentile coefficient on real GDP growth = 1.403; 75th percentile coefficient on real GDP growth = 1.575.
  - Income elasticity of real house prices increases in magnitude and statistical significance moving to the top 95th percentile relative to the bottom 5th percentile.
- Real household disposable income growth (granular analysis):
  - Table 4 (with short-term interest rates): Real Household Disposable Income Growth: 0.493*** (OLS); 0.491** (FE); 0.492*** (5th); 0.493*** (25th); 0.493* (50th); 0.494 (95th).
  - Table 5 (with long-term interest rates): Real Household Disposable Income Growth: 0.477*** (OLS); 0.567* (FE); 0.513*** (5th); 0.476*** (25th); 0.440*** (50th); 0.406* (95th).
  - Coefficients on household disposable income growth are positive across the conditional distribution but generally smaller in magnitude than those on real GDP growth; statistical significance varies by quantile and interest-rate specification.

### Key empirical results — Interest rates and other factors
- Interest rates are the second most important factor shaping housing markets in the sample.
- Short-term interest rates (Table 2 short-term row): -0.649** (OLS); -1.400* (FE); -0.937** (5th Percentile); -0.656** (25th Percentile); -0.357 (50th Percentile); 0.057 (95th Percentile).
  - Interpretation: increase in short-term interest rates leads to a larger decline in real house prices at lower quantiles but not at the high end.
- Long-term interest rates (Table 3 long-term row): -0.512 (OLS); 0.440 (FE); -0.165 (5th Percentile); -0.501 (25th Percentile); -0.851** (50th Percentile); -1.318* (95th Percentile).
  - Interpretation: when using long-term interest rates the interest rate elasticity of real house prices is significantly more negative at higher quantiles.
- Other factors:
  - Unemployment Rate: coefficients generally negative across quantiles but often not statistically significant.
  - Population Growth: coefficients not statistically significant at conventional levels; sometimes positive and larger for lowest percentiles, possibly indicating entry-level demand.
  - Inflation: generally negative effect; statistically significant only at certain quantiles (e.g., 75th percentile when using long-term rates).
  - Real Stock Market Return: generally positive correlation with real house prices; statistical significance varies by specification and quantile.
  - REER: generally no significant effect across the distribution, though magnitudes sometimes increase with quantiles.

### Household leverage and macroprudential indicators
- Household Debt-to-Income Ratio:
  - Table 4: -0.044, -0.047, -0.045, -0.044, -0.042, -0.041.
  - Table 5: -0.042, -0.106, -0.068, -0.042, -0.016, -0.008.
  - Interpretation: negative sign as expected but generally not statistically significant at conventional levels.
- Average loan-to-value ratio: introduced as additional control; not statistically significant at conventional levels (attributed to low household leverage on average in these countries).

### Pre- and post-GFC findings and IV robustness
- House price movements in the sample: increased by as much as 46 percent before the GFC and declined by 36 percent after the GFC.
- Pre/post-GFC quantile results:
  - Real GDP growth had a greater magnitude of impact on real house prices before the GFC and a weaker effect after the GFC.
  - Short-term and long-term interest rates gained more prominence after the GFC.
- IV quantile regression (instrument: trade-weighted average real GDP growth of trading partners):
  - Confirms baseline findings that income growth and interest rates are the most significant determinants of real house prices.
  - With IV quantile regression, the coefficient on income growth declines only slightly in magnitude for each quantile but remains statistically significant.

### Quantitative evidence (selected appendix results)
- IV quantile regressions (Short-Term Interest Rates, Appendix Table A5): Real GDP Growth coefficients by quantile: 3.766***, 1.568***, 1.216***, 1.034***, 0.780***, 0.263. Short-Term Interest Rates coefficients by quantile: -0.266, -0.661***, -0.287**, -0.093, 0.177**, 0.726***.
- IV quantile regressions (Long-Term Interest Rates, Appendix Table A6): Real GDP Growth coefficients by quantile: 3.914***, 1.499***, 1.231***, 1.097***, 0.896***, 0.516**. Long-Term Interest Rate coefficients by quantile: 0.663**, -1.003***, -0.657***, -0.484***, -0.225, 0.266.
- Appendix panel quantile and OLS/FE specifications show consistently positive and often statistically significant coefficients on Real GDP Growth (examples: pre-GFC OLS Real GDP Growth 1.764**, FE 1.873***; post-GFC OLS Real GDP Growth 1.156***, FE 0.946*).

### Synthesis, risks, and policy recommendations
- Principal drivers: income growth (real GDP and real household disposable income growth) and interest rates dominate real house price dynamics in emerging European markets.
- Distributional heterogeneity:
  - Income effects are stronger at higher quantiles of the house price distribution.
  - Interest rate effects vary by maturity and quantile: short-term rates exert larger downward pressure at lower quantiles, long-term rates show more negative elasticities at higher quantiles.
- Near-term risks:
  - Monetary policy tightening and slower income growth could lead to major corrections in real housing prices, especially in countries with a greater share of variable-rate mortgages and higher household debt-to-income ratios.
  - Lower quantiles of the market are more vulnerable to the end of cheap borrowing and a slowdown in economic activity.
- Macroprudential and fiscal recommendations:
  - Build additional buffers in the banking system and increase borrower resilience to asset price or income shocks.
  - Conduct stress tests for banks and non-bank financial institutions with high real estate exposures and consider higher provisioning for vulnerable mortgage loans identified in stress tests.
  - During a significant housing downturn, consider relaxing sectoral macroprudential measures (capital requirements, loan-to-value limits, debt service-to-income ratios) to contain the procyclical feedback loop between lower credit and house prices.
  - Use real estate taxes as a countercyclical tool to stabilize house prices over the economic cycle where recurrent property taxation is well-developed.

*Section 1 of "Don't Look Up: House Prices in Emerging Europe", WP/22/236, Serhan Cevik and Sadhna Naik, December 2022.*

### Section 1

### Don't Look Up: House Prices in Emerging Europe — Section 1

### Abstract and primary findings
- Study scope: quarterly data covering 10 countries over the period 1998–2022 using a panel quantile regression approach.
- Key empirical conclusions:
  - Economic, financial and demographic factors explain changes in real house prices in emerging Europe, with income growth having the most significant impact.
  - Income growth matters more for higher housing prices than for lower quantiles of the property market (e.g., the coefficient on real GDP growth for the 25th percentile is smaller than for the 75th percentile; the difference is larger when comparing the 5th and 95th percentiles).
  - Increases in short-term or long-term interest rates have a price-dampening impact: a higher cost of borrowing is associated with lower real house prices.
  - Interest rate effects are not statistically significant at all points of the distribution; the interest rate elasticity becomes more significant across the distribution when using long-term interest rates.
- Robustness checks (including use of real household disposable income and household debt-to-income ratio, and IV quantile regression) confirm the baseline results.
- Policy-relevant implication: the downturn in house prices could deepen with the looming economic recession and soaring interest rates, and lower quantiles of the market are significantly more vulnerable to the end of cheap borrowing and slower income growth.

### Introduction and motivation
- Context:
  - Global surge in consumer prices and subsequent monetary tightening have increased the cost of capital and put downward pressure on asset prices, including housing.
  - Since the global financial crisis in 2008, housing prices experienced an uninterrupted boom across the world, increasing as much as 50 percent in emerging market economies in Europe.
- Rationale:
  - Housing markets can generate macroeconomic and financial stability spillovers; understanding determinants of residential property prices and the impact of slower income growth and sustained interest rate increases is crucial.
- Research contribution:
  - Provides granular investigation across 10 developing European countries using panel quantile regression to estimate impacts across the conditional distribution of housing prices.

### Data overview and sample
- Sample: 10 emerging European markets (the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Russia, Slovenia, the Slovak Republic, and Türkiye) over 1998–2022.
- Housing price series: BIS nationwide residential property price index in real terms (harmonized series).
- Baseline explanatory variables (quarterly):
  - Real GDP growth
  - Consumer price inflation
  - Unemployment rate
  - Short-term or long-term interest rates
  - Stock market returns in real terms
  - Real effective exchange rate (REER)
  - Population growth
- Granular variables used in robustness checks:
  - Real household gross disposable income growth
  - Household debt-to-income ratio
- Stationarity: Im-Pesaran-Shin (2003) panel unit root tests indicate variables are stationary in first-differences or after logarithmic transformation (test results available upon request).

### Summary statistics (Table 1)
- Real House Price Growth: Observations 620; Mean 2.7; Std. Dev. 12.3; Skewness -0.9; Kurtosis 8.1; Minimum -59.5; Maximum 41.5
- Real GDP Growth: Observations 620; Mean 2.6; Std. Dev. 4.9; Skewness -1.3; Kurtosis 6.9; Minimum -20.7; Maximum 20.2
- Short-Term Interest Rates: Observations 648; Mean 5.2; Std. Dev. 7.7; Skewness 1.7; Kurtosis 4.9; Minimum -0.6; Maximum 33.8
- Long-Term Interest Rates: Observations 651; Mean 5.0; Std. Dev. 4.1; Skewness 1.3; Kurtosis 5.4; Minimum -0.4; Maximum 22.7
- Unemployment Rate: Observations 654; Mean 8.2; Std. Dev. 3.6; Skewness 0.8; Kurtosis 3.4; Minimum 1.9; Maximum 21.3
- Population Growth: Observations 609; Mean -0.1; Std. Dev. 0.8; Skewness 0.1; Kurtosis 4.0; Minimum -2.3; Maximum 1.8
- Consumer Price Inflation: Observations 620; Mean 4.0; Std. Dev. 4.8; Skewness 3.6; Kurtosis 30.0; Minimum -3.9; Maximum 55.4
- Stock Market Returns: Observations 651; Mean 5.5; Std. Dev. 28.1; Skewness 0.4; Kurtosis 5.1; Minimum -81.0; Maximum 113.2
- REER: Observations 660; Mean 101.2; Std. Dev. 11.2; Skewness -0.2; Kurtosis 6.7; Minimum 52.1; Maximum 156.2
- Real Household Gross Disposable Income Growth: Observations 536; Mean 4.8; Std. Dev. 5.6; Skewness 0.7; Kurtosis 4.9; Minimum -11.6; Maximum 30.0
- Household Debt-Income Ratio: Observations 613; Mean 47.0; Std. Dev. 20.3; Skewness -0.2; Kurtosis 2.3; Minimum 2.2; Maximum 92.8
- Data sources: BIS; European Commission; Eurostat; Haver Analytics; OECD; and authors' calculations.

### Econometric strategy
- Method: panel quantile regression with fixed effects (Machado and Santos Silva, 2019) applied to quarterly first-differences of the log housing price series.
- Baseline model specification:
  - ∆hp_{i,t,τ} = θ_τ + γ_τ X_{i,t} + η_i + μ_t + ε_{i,t,τ}
  - Where ∆hp_{i,t,τ} is the logarithmic change in housing prices at quantile τ; X_{i,t} contains the explanatory variables listed above; η_i are country fixed effects; μ_t are time effects; ε_{i,t,τ} is an idiosyncratic error term.
- Rationale for quantile approach:
  - Captures heterogeneous effects of explanatory variables across the distribution of house prices (different impacts at lower vs. higher quantiles).
  - Robust to outliers and non-normality and allows identification of conditional heterogeneous covariance effects.

### Main policy implications and recommendations
- Near-term risks:
  - Monetary policy tightening and slower income growth could lead to major corrections in real housing prices, especially in countries with a greater share of variable-rate mortgages and higher household debt-to-income ratios.
  - Lower quantiles of the market are more vulnerable to the end of cheap borrowing and a slowdown in economic activity.
- Macroprudential recommendations:
  - Build additional buffers in the banking system and increase borrower resilience to asset price or income shocks.
  - Conduct stress tests for banks and non-bank financial institutions with high real estate exposures and consider higher provisioning for vulnerable mortgage loans identified in stress tests.
  - During a significant housing downturn, consider relaxing sectoral macroprudential measures (capital requirements, loan-to-value limits, debt service-to-income ratios) to contain the procyclical feedback loop between lower credit and house prices.
  - Use real estate taxes as a countercyclical tool to stabilize house prices over the economic cycle where recurrent property taxation is well-developed.

*Section 1 of "Don't Look Up: House Prices in Emerging Europe", WP/22/236, Serhan Cevik and Sadhna Naik, December 2022.*

### Section 2

### Section 2

### Empirical approach and scope
- Estimation method: panel quantile regression with OLS fixed-effects reported as a point of reference.
- Dependent variable: year-on-year change in real house prices.
- Sample: emerging market economies in Europe (10 countries). Observations reported in tables: 599 (Table 2), 602 (Table 3), 526 (Table 4), 529 (Table 5).
- Robustness checks: sensitivity analyses, IV quantile regression (Chernozhukov and Hansen, 2008) instrumenting real GDP growth with the trade-weighted average real GDP growth of trading partners, and separate pre- and post-GFC estimations.

### Key findings — Income (real GDP and household disposable income)
- Real GDP growth is the most important factor determining real house prices; income convergence is a key channel.
- Heterogeneous impact across the conditional distribution of house prices:
  - Reported coefficients on Real GDP Growth (Table 2): 1.489*** (OLS), 1.266** (FE), 1.403*** (5th Percentile), 1.487*** (25th Percentile), 1.575*** (50th Percentile), 1.698*** (95th Percentile).
  - Prose summary values highlighted in the text: 25th percentile coefficient on real GDP growth = 1.403; 75th percentile coefficient on real GDP growth = 1.575.
  - Interpretation: income growth matters more for higher housing prices than for lower quantiles; income elasticity of real house prices increases in magnitude and statistical significance moving to the top 95th percentile relative to the bottom 5th percentile.
- Real household disposable income growth (granular analysis):
  - Table 4 (with short-term interest rates) — Real Household Disposable Income Growth: 0.493*** (OLS), 0.491** (FE), 0.492*** (5th), 0.493*** (25th), 0.493* (50th), 0.494 (95th).
  - Table 5 (with long-term interest rates) — Real Household Disposable Income Growth: 0.477*** (OLS), 0.567* (FE), 0.513*** (5th), 0.476*** (25th), 0.440*** (50th), 0.406* (95th).
  - Comparison: coefficients on household disposable income growth are positive across the conditional distribution but generally smaller in magnitude than those on real GDP growth; statistical significance varies by quantile and by whether short- or long-term interest rates are included.

### Key findings — Interest rates
- Interest rates are the second most important factor shaping housing markets in the sample.
- Short-term interest rates (Table 2 — Short-Term Interest Rate row): -0.649** (OLS), -1.400* (FE), -0.937** (5th Percentile), -0.656** (25th Percentile), -0.357 (50th Percentile), 0.057 (95th Percentile).
  - Interpretation from text: magnitude larger and statistically significant for the 25th quantile compared to other quantiles; negative and larger for the 5th percentile; positive, smaller and statistically insignificant for the 95th percentile. Thus, an increase in short-term interest rates leads to a larger decline in real house prices at lower quantiles but not at the high end.
- Long-term interest rates (Table 3 — Long-Term Interest Rate row): -0.512 (OLS), 0.440 (FE), -0.165 (5th Percentile), -0.501 (25th Percentile), -0.851** (50th Percentile), -1.318* (95th Percentile).
  - Interpretation from text: when using long-term interest rates the interest rate elasticity of real house prices is significantly more negative at higher quantiles, suggesting differing wealth or financing effects across the price distribution.
- Granular results with household controls (Table 5 — Long-Term Interest Rate row): -0.587 (OLS), 0.563 (FE), -0.122 (5th), -0.600 (25th), -1.061** (50th), -1.502** (95th).

### Other economic, financial, and demographic factors
- Unemployment Rate:
  - Table 2 unemployment coefficients: -0.463, -0.312, -0.405, -0.462, -0.522, -0.605 (columns as reported).
  - Table 3 unemployment coefficients: -0.431, -0.909, -0.605, -0.437, -0.261, -0.027.
  - Interpretation: negative as expected across quantiles but generally not statistically significant at conventional levels (likely because real GDP growth captures the economic cycle).
- Population Growth:
  - Table 2: -0.453, 2.545, 0.697, -0.425, -1.618, -3.269.
  - Table 3: -0.586, 2.908, 0.686, -0.547, -1.831, -3.546.
  - Interpretation: not statistically significant at conventional levels; coefficients are positive and larger for lowest percentiles in some specifications, possibly indicating entry-level demand for low-cost housing.
- Inflation (proxy for macroeconomic stability):
  - Table 2 Inflation row: -0.117, -0.159, -0.011, -0.115, -0.225, -0.377.
  - Table 3 Inflation row: -0.447, -0.430, -0.441, -0.447, -0.453*, -0.461.
  - Interpretation: consistently negative effect; statistically significant only at certain quantiles (e.g., 75th percentile when model estimated with long-term interest rates).
- Real Stock Market Return:
  - Table 2 row: 0.039, 0.044, 0.041, 0.039, 0.036, 0.033.
  - Table 4 row: 0.109** (OLS), 0.080, 0.095***, 0.109***, 0.122*, 0.139.
  - Table 5 row: 0.114** (OLS), 0.131*, 0.121***, 0.114***, 0.107***, 0.100*.
  - Interpretation: positive correlation with real house prices but often not statistically significant and weaker at higher quantiles when long-term interest rates are included.
- Real Effective Exchange Rate (REER):
  - Table 2: -0.041, -0.111, -0.068, -0.041, -0.013, 0.025.
  - Table 5 REER row: -0.141, 0.017, -0.077, -0.142*, -0.206**, -0.266*.
  - Interpretation: generally no significant effect across the conditional distribution, though coefficient magnitudes sometimes increase with quantiles, potentially indicating weak non-resident investment demand at higher price tiers.

### Household leverage, loan-to-value and debt effects
- Household Debt-to-Income Ratio:
  - Table 4: -0.044, -0.047, -0.045, -0.044, -0.042, -0.041.
  - Table 5: -0.042, -0.106, -0.068, -0.042, -0.016, -0.008.
  - Interpretation: expected negative sign (higher indebtedness associated with lower property prices) but generally not statistically significant at conventional levels.
- Average loan-to-value ratio (macroprudential strength): introduced as additional control; not statistically significant at conventional levels (attributed to low household leverage on average in these countries).

### Pre- and post-GFC structural checks
- House price movements in the sample: increased by as much as 46 percent before the GFC and declined by 36 percent after the GFC.
- Pre/post-GFC quantile results:
  - Real GDP growth had a greater magnitude of impact on real house prices before the GFC and a weaker effect after the GFC.
  - Short-term and long-term interest rates gained more prominence after the GFC, reflecting realignment of risk premiums in lending and possibly further development of mortgage markets in emerging European countries.

### Instrumental-variable quantile regression (robustness)
- IV approach: instrument country real GDP growth with trade-weighted average real GDP growth of its trading partners.
- Main IV result: confirms baseline findings that income growth and interest rates are the most significant determinants of real house prices.
  - With IV quantile regression, the coefficient on income growth declines only slightly in magnitude for each quantile but remains statistically significant, supporting the plausibility of the instrument.

### Synthesis and implications
- Principal drivers: income growth (real GDP and real household disposable income growth) and interest rates are the dominant determinants of real house price dynamics in emerging European markets.
- Distributional heterogeneity:
  - Income effects are stronger at higher quantiles of the house price distribution; interest rate effects vary by maturity and are more strongly negative at higher quantiles when using long-term rates.
- Policy-relevant observations:
  - Monetary policy (through short- and long-term rates) has heterogeneous effects across the housing market; higher interest rates dampen prices overall but may exert limited downward pressure on the high-end of the market when measured by short-term rates.
  - Macroprudential indicators (loan-to-value, household debt-to-income) show limited statistical significance in this sample, which aligns with generally low household leverage.

*Source: wpiea2022236-print-pdf - Section 2*

### Section 3

### Section 3

### Findings and interpretation
- The house price cycle has already turned down in emerging Europe, which could deepen with the looming economic recession and soaring interest rates.
- Housing prices in the European sample remained buoyant as of the first half of 2022, but high-frequency data already shows growth deceleration in some countries.
- Empirical analysis indicates that monetary policy shocks and slower income growth could lead to a major correction in real housing prices, especially in countries with greater share of variable-rate mortgages and higher household debt-to-income ratios.
- Quantile regression results indicate that the lower quantiles of the real estate market in emerging European countries are significantly more vulnerable to the end of cheap borrowing and a slowdown in economic activity.
- Large house price adjustments can have adverse effects on economic performance and financial stability, as experienced during the GFC and other historical episodes.

### Quantitative evidence (selected results from appendix tables)
- IV quantile regressions (Short-Term Interest Rates, Appendix Table A5): Real GDP Growth coefficients by quantile: 3.766***, 1.568***, 1.216***, 1.034***, 0.780***, 0.263. Short-Term Interest Rates coefficients by quantile: -0.266, -0.661***, -0.287**, -0.093, 0.177**, 0.726***.
- IV quantile regressions (Long-Term Interest Rates, Appendix Table A6): Real GDP Growth coefficients by quantile: 3.914***, 1.499***, 1.231***, 1.097***, 0.896***, 0.516**. Long-Term Interest Rate coefficients by quantile: 0.663**, -1.003***, -0.657***, -0.484***, -0.225, 0.266.
- Panel quantile and OLS/FE specifications reported in Appendix Tables A1–A4 show consistently positive and often statistically significant coefficients on Real GDP Growth (examples: pre-GFC OLS Real GDP Growth 1.764**, FE 1.873***; post-GFC OLS Real GDP Growth 1.156***, FE 0.946*), and mixed signs and significance for Short-Term and Long-Term Interest Rates across quantiles and periods.
- Sample sizes and panels reported: Appendix Tables report Observations and Countries (examples: Appendix Table A5 Observations 599, Countries 10; Appendix Table A6 Observations 602, Countries 10). Notes: dependent variable is the year-on-year change in real house prices; standard errors are reported in brackets; significance denoted by *, **, and *** for 10%, 5%, and 1% levels.

### Policy recommendations
- Use macroprudential measures to build additional buffers in the banking system and increase resilience of borrowers to asset price or income shocks to minimize downside risks to financial stability.
- Before a housing market correction:
  - Conduct stress tests for banks and other non-bank financial institutions with high exposures to real estate.
  - Consider requiring higher provisioning for mortgage loans for institutions found vulnerable based on stress tests.
- During a significant housing market downturn:
  - Relax sectoral macroprudential measures, such as capital requirements and limits on loan-to-value and debt service-to-income ratios, to contain the procyclical feedback loop between lower credit and house prices.
- Use real estate taxes as a countercyclical tool to stabilize house prices over the economic cycle, at least in countries with a well-developed system of recurrent property taxation.

*Source: wpiea2022236-print-pdf - Section 3*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2022/english/wpiea2022236-print-pdf.pdf_