## wpiea2023256-print-pdf

## Source details

**Canonical URL:** [wpiea2023256-print-pdf](https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023256-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2023/english/wpiea2023256-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2023/english/wpiea2023256-print-pdf.pdf.json)

---

### Key findings and quantitative results
- Using a panel of 20 European countries over the period 1980–2023, households consume more when house prices and income growth increase in real terms; effects on consumer spending are short-lived and diminish over time.
- Seasonally-adjusted quarter-on-quarter elasticities:
  - Private consumption falls by 0.13 percentage points on average for a one percent decrease in real house prices in the first quarter after the shock.
  - Private consumption falls by 0.02 percentage points on average for a one percent decrease in real gross disposable income in the first quarter after the shock.
- Cumulative impact example:
  - The observed quarter-on-quarter decline of -1.96 percent in real house prices in Q1 2023 could dampen consumer spending by about -0.51 percentage points in real terms on a cumulative basis over a horizon of eight quarters (in the sample of European countries).
- Housing accounts for, on average, about 55 percent of aggregate household wealth in Europe (with significant cross-country variation).
- Sample descriptive statistics (quarter-on-quarter growth, 1980–2023):
  - House Price Growth: Count 1750; Mean 0.42; SD 1.94; Kurtosis 5.09; Min -9.48; Max 7.74.
  - Gross Disposable Income Growth: Count 1710; Mean 0.30; SD 2.49; Kurtosis 7.01; Min -13.49; Max 13.48.
  - Consumption Growth: Count 1750; Mean 0.37; SD 2.57; Kurtosis 35.19; Min -26.79; Max 18.87.

### Context and economic interpretation
- European households face a "double crisis": the worst inflation shock since World War II and a sudden correction in house prices, driven by higher interest rates and increased uncertainty.
- Residential property is the dominant component of household wealth in Europe, so housing price cycles materially influence net worth and borrowing constraints, thereby affecting consumption smoothing.
- There is substantial heterogeneity across countries in both housing price cycles and consumption responses, and downside risks are larger where real wealth destruction extends beyond housing.

### Data and sample
- Panel dataset: quarterly observations for 20 countries in Europe during 1980–2023.
- Countries included: Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Netherlands, Norway, Poland, Portugal, Romania, Slovenia, Spain, Sweden, and the United Kingdom.
- Main series:
  - Real residential house price index from BIS (harmonized series per Eurostat, 2013), nationwide residential property, real terms adjusted for consumer price inflation.
  - Real household disposable income and real private consumption from BIS, Eurostat, and Haver Analytics.
- Time-series treatment:
  - Series are seasonally adjusted, transformed in logarithms, and expressed in quarter-on-quarter growth rates.
  - Cross-sectional dependence: Pesaran (2004) test rejects the null of cross-sectional independence; data are transformed by time demeaning to account for common factors.
  - Stationarity: Im-Pesaran-Shin (2003) and Pesaran (2007) tests indicate stationarity after logarithmic transformation and first differencing (results available upon request).

### Econometric strategy and model details
- Baseline model: three-variable Panel Vector Autoregression (PVAR) with variables: real house prices, real household disposable income, and real private consumption.
- Estimation approach:
  - Fixed effects to control for unobserved individual heterogeneity and common time fixed effects.
  - Generalized Method of Moments (GMM) with lagged regressors as instruments.
  - Lag selection: MMSC (Andrews and Lu, 2001) indicates lag 1 for quarter-on-quarter growth rates and lag 5 for year-on-year growth rates; alternative lags tested with similar results.
- Identification and inference:
  - Impulse Response Functions (IRFs) and variance decompositions computed using Love and Zicchino (2006) approach.
  - Residual covariance matrix decomposed via Cholesky (recursive VAR) ordering for shock identification.
- Model properties:
  - PVAR has all eigenvalues within the unit circle, indicating stability and invertibility; non-cumulative IRFs converge to zero.

### Methodology and identification details
- Identifying assumption: variables earlier in the ordering affect following variables contemporaneously and with a lag; variables later affect previous variables only with a lag (earlier = more exogenous; later = more endogenous).
- PVAR implementation specifics:
  - Restriction that underlying structure is the same for each cross-sectional unit; fixed effects 푥푖 introduced to allow for individual heterogeneity in levels.
  - Forward mean-differencing (the “Helmert procedure”) used to eliminate fixed effects to avoid bias from correlation between fixed effects and lagged dependent variables (Arellano and Bover, 1995).
  - Time fixed effects removed by subtracting the means of each variable calculated for each country–year, preserving orthogonality between transformed variables and lagged regressors.
  - Lagged regressors used as instruments and coefficients estimated by the GMM method; GMM helps deal with potential serial correlation in the error term, especially for year-on-year growth rates.
- Impulse response uncertainty:
  - Standard errors and confidence intervals for IRFs derived using Monte Carlo simulations.
  - Procedure: randomly generate a draw of coefficients β using the estimated coefficients and their variance–covariance matrix; re-calculate impulse responses; repeat 1,000 times and generate 5th and 95th percentiles as confidence intervals.
  - IRFs presented are non-cumulative (effect of the shock on other variables in a given period).
- Model diagnostics:
  - Appendix Table A1: All the eigenvalues lie inside the unit circle therefore passing the stability test.

### Main estimation results (impulse responses and magnitudes)
- Shock definition:
  - A positive one standard deviation shock is equivalent to a growth rate of 1.3 percentage points on average (housing prices shock).
- Housing price shock on consumer spending (quarter-on-quarter growth rates):
  - A one standard deviation positive shock to real house price growth (equivalent to an increase of 1.3 percentage points on average) raises consumer spending by about 0.2 percentage points in the first quarter.
  - Cumulative effect over the two-year period: 0.3 percentage points.
  - The impact is short-lived, grows smaller over time, and plateaus after six quarters.
- Gross disposable income shock on consumer spending:
  - A one standard deviation positive shock to real gross disposable income growth (equivalent to an increase of 2.2 percentage points on average) boosts private consumption in real terms by about 0.04 percentage points in the first quarter.
  - Cumulative effect over the two-year horizon: 0.03 percentage points.
- Two-year cumulative effects reported in conclusion:
  - Housing prices: cumulatively amounting to 0.34 percentage points over the two-year period.
  - Gross disposable income: cumulatively amounting to 0.03 percent over the two-year period.
- Example quantitative implication for 2023Q1:
  - Seasonally-adjusted quarter-on-quarter decline of 1.96 percent in real house prices in the first quarter of 2023 could dampen consumer spending by about -0.51 percentage points, on average, in real terms on a cumulative basis over the next eight-quarter horizon.
- Monte Carlo simulation draws for IRF confidence intervals: 1,000 repetitions; confidence intervals use the 5th and 95th percentiles.
- Cross-country heterogeneity and historical context:
  - Results broadly consistent with previous studies and the experience during the GFC in Spain and the Netherlands, when the collapse in housing prices constrained consumer spending in the subsequent two-year period by 5 percent and 3 percent, respectively.
  - Significant heterogeneity in housing price cycles and the impact on private consumption across Europe; some countries continue to experience positive growth rates, others face potential for larger slowdowns.

### Interpretation, mechanisms, and risks
- Consistent with economic theory: households take into account fluctuations in house prices and income growth and smooth spending over time.
- Mechanisms:
  - Increasing (decreasing) house prices and gross disposable income can stimulate (dampen) consumption by raising household income and wealth and easing (tightening) borrowing constraints, especially in the short run.
- Heterogeneity caveats:
  - Effects vary significantly at the household level depending on financial constraints, precautionary saving behavior, and changes in the ratio of mortgage payments to household income.
  - Findings are robust to a sample of only advanced economies in the sample (Appendix Figures A1 and A2).
- Policy-relevant implication:
  - Short-lived but economically meaningful housing wealth effects imply that recent real house price declines can depress consumption growth substantially over subsequent quarters.
  - Heterogeneity across countries implies uneven macro-financial vulnerabilities; larger real wealth losses beyond housing could amplify declines in private consumption and affect macro-financial stability.

*Source: IMF Working Paper "Feeling Rich, Feeling Poor: Housing Wealth Effects and Consumption in Europe", WP/23/256, December 2023.*

### Section 1

### Feeling Rich, Feeling Poor: Housing Wealth Effects and Consumption in Europe (Section 1)

### Key findings and quantitative results
- Using a panel of 20 European countries over the period 1980–2023, households consume more when house prices and income growth increase in real terms; effects on consumer spending are short-lived and diminish over time.
- Seasonally-adjusted quarter-on-quarter elasticities:
  - Private consumption falls by 0.13 percentage points on average for a one percent decrease in real house prices in the first quarter after the shock.
  - Private consumption falls by 0.02 percentage points on average for a one percent decrease in real gross disposable income in the first quarter after the shock.
- Cumulative impact example:
  - The observed quarter-on-quarter decline of -1.96 percent in real house prices in Q1 2023 could dampen consumer spending by about -0.51 percentage points in real terms on a cumulative basis over a horizon of eight quarters (in the sample of European countries).
- Housing accounts for, on average, about 55 percent of aggregate household wealth in Europe (with significant cross-country variation).
- Sample descriptive statistics (quarter-on-quarter growth, 1980–2023):
  - House Price Growth: Count 1750; Mean 0.42; SD 1.94; Kurtosis 5.09; Min -9.48; Max 7.74.
  - Gross Disposable Income Growth: Count 1710; Mean 0.30; SD 2.49; Kurtosis 7.01; Min -13.49; Max 13.48.
  - Consumption Growth: Count 1750; Mean 0.37; SD 2.57; Kurtosis 35.19; Min -26.79; Max 18.87.

### Context and economic interpretation
- European households face a "double crisis": the worst inflation shock since World War II and a sudden correction in house prices, driven by higher interest rates and increased uncertainty.
- Residential property is the dominant component of household wealth in Europe, so housing price cycles materially influence net worth and borrowing constraints, thereby affecting consumption smoothing.
- There is substantial heterogeneity across countries in both housing price cycles and consumption responses, and downside risks are larger where real wealth destruction extends beyond housing.

### Data and sample
- Panel dataset: quarterly observations for 20 countries in Europe during 1980–2023.
- Countries included (determined by data availability on the three variables): Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Netherlands, Norway, Poland, Portugal, Romania, Slovenia, Spain, Sweden, and the United Kingdom.
- Main series:
  - Real residential house price index from BIS (harmonized series per Eurostat, 2013), nationwide residential property, real terms adjusted for consumer price inflation.
  - Real household disposable income and real private consumption from BIS, Eurostat, and Haver Analytics.
- Time-series treatment:
  - Series are seasonally adjusted, transformed in logarithms, and expressed in quarter-on-quarter growth rates.
  - Cross-sectional dependence: Pesaran (2004) test rejects the null of cross-sectional independence; data are transformed by time demeaning to account for common factors.
  - Stationarity: Im-Pesaran-Shin (2003) and Pesaran (2007) tests indicate stationarity after logarithmic transformation and first differencing (results available upon request).

### Econometric strategy and model details
- Baseline model: three-variable Panel Vector Autoregression (PVAR) with variables: real house prices, real household disposable income, and real private consumption.
- Estimation approach:
  - Fixed effects to control for unobserved individual heterogeneity and common time fixed effects.
  - Generalized Method of Moments (GMM) with lagged regressors as instruments.
  - Lag selection: MMSC (Andrews and Lu, 2001) indicates lag 1 for quarter-on-quarter growth rates and lag 5 for year-on-year growth rates; alternative lags tested with similar results.
- Identification and inference:
  - Impulse Response Functions (IRFs) and variance decompositions computed using Love and Zicchino (2006) approach.
  - Residual covariance matrix decomposed via Cholesky (recursive VAR) ordering for shock identification.
- Model properties:
  - PVAR has all eigenvalues within the unit circle, indicating stability and invertibility; non-cumulative IRFs converge to zero.

### Implications and risks
- Short-lived but economically meaningful housing wealth effects imply that recent real house price declines can depress consumption growth substantially over subsequent quarters.
- Heterogeneity across countries implies uneven macro-financial vulnerabilities; larger real wealth losses beyond housing could amplify declines in private consumption and affect macro-financial stability.

*Source: IMF Working Paper "Feeling Rich, Feeling Poor: Housing Wealth Effects and Consumption in Europe" (Section I: Introduction; Sections II–IV overview), WP/23/256, December 2023.*

### Section 2

### wpiea2023256-print-pdf - Section 2

### Methodology and Identification
- Identifying assumption: variables earlier in the ordering affect following variables contemporaneously and with a lag; variables later affect previous variables only with a lag (earlier = more exogenous; later = more endogenous).
- Panel VAR (PVAR) implementation:
  - Restriction imposed that underlying structure is the same for each cross-sectional unit; to allow for individual heterogeneity in levels, fixed effects 푥푖 are introduced.
  - Forward mean-differencing (the “Helmert procedure”) is used to eliminate fixed effects to avoid bias from correlation between fixed effects and lagged dependent variables (Arellano and Bover, 1995).
  - Time fixed effects are removed by subtracting the means of each variable calculated for each country–year; this preserves orthogonality between transformed variables and lagged regressors.
  - Lagged regressors are used as instruments and coefficients are estimated by the GMM method.
  - The GMM approach also helps deal with potential serial correlation in the error term, especially when the model is estimated using year-on-year growth rates.
- Impulse response uncertainty:
  - Standard errors of IRFs and confidence intervals are derived using Monte Carlo simulations.
  - Procedure: randomly generate a draw of coefficients β using the estimated coefficients and their variance–covariance matrix; re-calculate impulse responses; repeat this procedure 1,000 times and generate 5th and 95th percentiles as confidence intervals.
  - IRFs presented in this paper are non-cumulative (interpretable as the effect of the shock on other variables in a given period).
- Model diagnostics:
  - Appendix Table A1: All the eigenvalues lie inside the unit circle therefore passing the stability test.

### Main Estimation Results (Impulse Responses and Magnitudes)
- Shock definition:
  - A positive one standard deviation shock is equivalent to a growth rate of 1.3 percentage points on average (housing prices shock).
- Housing price shock on consumer spending (quarter-on-quarter growth rates):
  - A one standard deviation positive shock to real house price growth (equivalent to an increase of 1.3 percentage points on average) raises consumer spending by about 0.2 percentage points in the first quarter.
  - Cumulative effect over the two-year period: 0.3 percentage points.
  - The impact is short-lived, grows smaller over time, and plateaus after six quarters.
- Gross disposable income shock on consumer spending:
  - A one standard deviation positive shock to real gross disposable income growth (equivalent to an increase of 2.2 percentage points on average) boosts private consumption in real terms by about 0.04 percentage points in the first quarter.
  - Cumulative effect over the two-year horizon: 0.03 percentage points.
- Two-year cumulative effects reported in conclusion:
  - Housing prices: cumulatively amounting to 0.34 percentage points over the two-year period.
  - Gross disposable income: cumulatively amounting to 0.03 percent over the two-year period.
- Example quantitative implication for 2023Q1:
  - Seasonally-adjusted quarter-on-quarter decline of 1.96 percent in real house prices in the first quarter of 2023 could dampen consumer spending by about -0.51 percentage points, on average, in real terms on a cumulative basis over the next eight-quarter horizon.
- Cross-country heterogeneity and historical context:
  - Results are broadly consistent with previous studies and the experience during the GFC in Spain and the Netherlands, when the collapse in housing prices constrained consumer spending in the subsequent two-year period by 5 percent and 3 percent, respectively.
  - There is significant heterogeneity in housing price cycles and the impact on private consumption across Europe; while some countries continue to experience positive growth rates, others face potential for larger slowdowns in consumer spending growth.

### Interpretation and Economic Mechanisms
- Aligns with economic theory: households take into account fluctuations in house prices and income growth and smooth spending over time.
- Mechanisms:
  - Increasing (decreasing) house prices and gross disposable income can stimulate (dampen) consumption by raising household income and wealth and easing (tightening) borrowing constraints, especially in the short run.
- Heterogeneity caveats:
  - Effects vary significantly at the household level depending on factors including financial constraints, precautionary saving behavior, and changes in the ratio of mortgage payments to household income.
  - Findings are robust to a sample of only advanced economies in the sample (Appendix Figures A1 and A2).

### Key Quantitative Summary (exact figures preserved)
- One standard deviation shock to real house price growth: 1.3 percentage points on average.
- One standard deviation shock to real gross disposable income growth: 2.2 percentage points on average.
- Immediate consumer spending response (first quarter):
  - Housing price shock → +0.2 percentage points.
  - Income shock → +0.04 percentage points.
- Cumulative over two years:
  - Housing prices → 0.3 percentage points (text) and 0.34 percentage points (conclusion).
  - Gross disposable income → 0.03 percentage points (quarter-on-quarter estimate) and 0.03 percent (conclusion wording preserved).
- 2023Q1 observed change:
  - Real house prices decline: 1.96 percent (seasonally-adjusted q-o-q).
  - Implied cumulative impact on consumer spending over eight quarters: -0.51 percentage points.
- Monte Carlo simulation draws for IRF confidence intervals: 1,000 repetitions; confidence intervals use the 5th and 95th percentiles.

_Italic: Source — Authors’ calculations and text from wpiea2023256-print-pdf - Section 2._

---


_Source: https://www.imf.org/-/media/files/publications/wp/2023/english/wpiea2023256-print-pdf.pdf_