## wp1876

## Source details

**Canonical URL:** [wp1876](https://www.imf.org/-/media/files/publications/wp/2018/wp1876.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp1876.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp1876.pdf.json)

---

### Introduction and research questions
- Research aims:
  - Assess generalizability of negative relationship between household debt and macro outcomes across a broad set of countries.
  - Identify channels and factors explaining this relationship.
  - Evaluate role of institutional factors.
- Complementary mechanisms highlighted:
  - Debt overhang effects depend on differences in marginal propensities to consume across households.
  - Household credit booms raise future probability of banking crises.
  - Sentiment-driven rapid increases in household debt are associated with neglect of crash risks.

### Empirical analysis — sample, model, and core magnitudes
- Sample:
  - 80 countries total: 39 advanced economies; 41 emerging market economies.
  - Sample span cited elsewhere: up to 65 years (1950–2016) for macro analysis.
- Forecasting specification:
  - Panel regression with country and year fixed effects; standard errors clustered by year and country.
  - Future real GDP growth related to past three-year changes in household debt to GDP and corporate debt to GDP; controls include lagged GDP growth and past three-year change in government debt to GDP.
  - Forecast horizon varied: k = 0 to k = 6 years.
- Key sample statistics:
  - Mean household debt to GDP ratio: 35 percent.
  - Mean annual increase in household debt: about 1 percentage point.
  - Mean firm (nonfinancial corporate) debt to GDP ratio: 60 percent, increasing by slightly less than 1 percentage point per year.
  - Mean public debt to GDP ratio: 52 percent, rising by 2 percentage points per year.
  - Household debt-GDP ratio at 10th and 90th percentiles: between 6 percent and 72 percent.
  - Standard deviation of three-year change in household debt to GDP ratio: 5.89 percentage points.
  - Standard deviation of three-year change in corporate debt to GDP ratio: 18.97 percentage points.
  - A one standard deviation increase in the household debt ratio → 1.2 percentage points lower GDP growth over a three-year horizon.
- Main empirical finding:
  - Regression coefficients on household debt to GDP ratio are strongly negative across horizons 0–6 years and statistically significant at the 1 percent level; effect strongest at the three- to four-year horizon and diminishes thereafter.
  - Coefficients on corporate debt are more than three-times smaller than for household debt; negative and significant effects of corporate debt on future income are absent at horizons beyond four to five years.

### Section III — Household debt and future income growth; heterogeneity
- Advanced vs emerging markets:
  - Negative correlation between past household debt growth and future income growth is negative and statistically significant at horizons of one to five years in advanced-economy sample; effect diminishes at seven-year horizon.
  - Negative correlation present for emerging markets but statistical significance is weak at three- and five-year horizons due to shorter data spans.
- Controls and robustness:
  - All regressions include lagged GDP growth for preceding two years.
  - Results robust to inclusion of past three-year change in government debt to GDP.
  - For 27 countries with quarterly data (1998–2015) innovations-based analysis yields similar results to OLS.

### Institutional factors and distributional characteristics — empirical framework
- Baseline interaction regression (illustrative h = 5 years):
  - Δy_{i,t+h} = α_i + β1 ΔDebt_{i,t-1} + β2 ⋅ ΔDebt_{i,t-1} ⋅ Z_i + γ X_{i,t-1} + ε_{i,t+h}
  - Z_i = indicator for institutional or distributional characteristic; coefficient of interest = β2.
  - Qualitatively similar results for horizons at three or seven years and using panel VAR.

### B.1 Exchange rate regime and capital account openness
- Concept:
  - Fixed exchange rate regimes constrain monetary policy; expected stronger negative effects of household debt growth on future income under fixed regimes.
- Indicators:
  - IMF classification: indicator for fixed exchange rate regime = 1 for fixed, 0 otherwise.
  - High capital account openness = index above the median.
- Empirical findings:
  - Interaction between past household debt growth and fixed exchange rate indicator is negative and highly significant while household debt growth by itself becomes insignificant (Column (1), Table 3).
  - High capital account openness amplifies negative effect (Columns (2)–(3), Table 3).
  - Results persist when both fixed regime and high openness included; confirmed with panel VAR (Figure 5).
  - Not driven solely by euro-area countries; interaction remains significant at the 10 percent level when including euro-area dummy.

### B.2 Financial development, transparency, and financial risk index
- Indicators:
  - Financial Development Index: indicator = 1 if index in top quartile as of 2014.
  - Transparency index and Financial Risk Index capture credit information transparency and financial risk.
- Findings (Columns (4)–(6), Table 3):
  - High Financial Development Index mitigates negative effect of household debt: interaction coefficient offsets household debt term, indicating negative effects concentrated in low-Financial-Development countries.
  - Better transparency and better financial risk also mitigate negative effect, though interaction coefficients smaller than for Financial Development Index.
- Interpretation:
  - Better and more efficient financial markets/institutions improve credit-risk assessment, allocation, and management of consequences, reducing medium-term macro-financial costs of rising household debt.

### B.3 Distributional characteristics: participation rate vs. DTI (micro-based indicators for 30 countries)
- Indicators constructed from micro-level data (latest waves for 30 countries):
  1. LowIncPart (mortgage participation rate of low-income households):
     - Defined as participation for bottom two quintiles (bottom 40 percent).
     - LowIncPart = 1 if mortgage participation rate for bottom 40 percent ranks within top quartile (roughly above 20 percent), 0 otherwise.
     - Finding: high mortgage participation rate for low-income households mitigates negative effects of household debt on future income growth (Column (7), Table 3); interpreted as financial inclusion/efficient screening.
  2. LowDTI (average DTI of low-income households):
     - Weighted average DTI for bottom 40 percent, weighted by share of debt held by these households.
     - LowDTI = 1 if weighted DTI for these households is low (within lower quartile), 0 otherwise.
     - Finding: low average DTI for low-income households appears to reduce negative impact of aggregate household debt, though statistical significance is weaker (Columns (8)–(9), Table 3); result robust to interaction with EM indicator.
- Interpretation:
  - Participation (access) and average DTI capture distinct aspects: participation reflects financial inclusion and development; DTI reflects danger of over-indebtedness.
- Confirmation:
  - Distributional results further confirmed in panel VAR (Figures 5–6).

### C. Systemic banking crises — model, thresholds, and main magnitudes
- Model:
  - Logit panel for systemic banking crisis event: log odds = Ψ0 + Ψ1 X_{it} + Ψ2 X_{it} × I(HiDebt_{it}) + country fixed effects + error.
  - X_{it} includes lagged changes and levels of household and corporate debt (scaled by GDP); I(HiDebt) = 1 if debt exceeds thresholds.
- Data and thresholds:
  - Logit panel covers 34 countries over 1970–2015 (sample-size variations in tables).
  - Example thresholds:
    - HI household debt threshold = 65 percent of GDP (top quintile).
    - HI sovereign debt threshold = 60 percent of GDP (top 1/3).
  - Average change in household credit-to-GDP ratio in regression sample: about 1 percentage point.
  - Standard deviation of that change: around 2.5 percentage points.
  - Up to four lags considered; baseline includes significant lags (third lag often used).
- Main findings:
  - Household debt is a good early warning indicator for banking crises; both household and corporate debt-to-GDP ratios positively associated with crisis probability (Table 4).
  - Changes in household debt are more important than levels in some specifications (Column 3).
  - Household debt effects tend to dominate corporate debt effects (Column 4).
  - Relationship between increasing household debt and crises stronger when household debt level exceeds 65 percent of GDP (Column 5).
  - High sovereign debt (above 60 percent of GDP) amplifies crisis probability (Column 6).
- Quantitative magnitudes:
  - Average marginal effect of changes in household debt ≈ 1 percentage point, almost double effect of firm debt increases.
  - When household debt level is high, crisis probability boosted by another 80 basis points.
  - Unconditional crisis probability in considered countries ≈ 3.5 percent.
- Model performance diagnostics (selected):
  - Coefficients on household debt change across specs: 40.05***; 35.01***; 35.60***; 31.25***; 30.86***; 26.47***.
  - Interaction terms example: ݄݄݀Δ x I(Hi Gov Debt) = 22.62*; 24.12*; ݄݄݀Δ x I(Hi hhd) = 24.41*; 25.93* (standard errors reported in table).
  - AUC by model: 0.700; 0.791; 0.806; 0.840; 0.845; 0.850; 0.856.
  - Pseudo R2 ranges from 0.0612 to 0.228 across reported models.

### D. Neglected Downside Risk — bank equity returns, crash probabilities, and abnormal returns
- D1. Bank stock returns (70 countries, 1973–2016; h = 1, 3, 5; k = 3):
  - Past three-year change in household debt ratio negatively associated with future bank equity returns.
  - One standard deviation increase in annual growth of household debt ratio → lower equity return of 2 percent to 2.7 percent one year later.
  - One standard deviation increase → lower equity returns of 12 percent to 15 percent at three- and five-year horizons.
  - Selected coefficients (Table 6): k = 1: -0.020* (0.011); k = 3: -0.027** (0.013); k = 5: -0.120** (0.032) and variants.
- D2. Probability of bank stock crashes (probit):
  - Dependent indicator: at least one annual return below -30 percent in next k years.
  - One standard deviation increase in growth rate of household debt ratio → 8 percent higher probability of a bank equity crash within next five years (main text).
  - Selected marginal effects (Table 6): positive and significant coefficients include 0.053** (0.011); 0.099** (0.017); 0.057** (0.020) in some specs.
- D3. Abnormal returns (two-stage CAPM residuals; 30-country sample):
  - Past three-year growth in household debt associated with negative future abnormal returns at two- to three-year horizons; corporate debt also negatively correlated in 30-country subsample.
  - Selected coefficients (Table 7):
    - k = 1: Δhousehold-debt -0.049 (0.063); Δcorporate-debt -0.198** (0.081).
    - k = 2: Δcorporate-debt -0.401*** (0.098); other lags show larger negative and significant coefficients (-0.503*** (0.134), -0.723*** (0.161)).
    - k = 3: Δhousehold-debt -0.228*** (0.079) and variants.

### E. Debt overhang: micro-level evidence (euro area household panel)
- Data:
  - Micro-level longitudinal data for five euro area countries (Belgium, Cyprus, Germany, Malta, the Netherlands) for two waves (2010 and 2014); about 3,000 households with borrowing and consumption information.
- Household-level regression (change in household food consumption, percent of income):
  - Key regressor: DTI_i,2010 (debt-to-income ratio).
  - Controls: household characteristics, country fixed effects; errors clustered at country level.
- Main findings and magnitudes:
  - Higher indebtedness (DTI or LTV) increases household vulnerability to income shocks.
  - Negative effects intensify when total debt exceeds 300 percent of household disposable income.
  - Magnitudes:
    - 100 percentage point increase in DTI ratio → 4 percentage points drop in consumption.
    - For households with total debt > 300 percent of disposable income → about 7 percentage points drop in consumption.
  - Selected coefficients (Table 8):
    - DTI (lag) coefficients: -0.0396*** (0.00235); -0.0404*** (0.00183); -0.0401*** (0.00226); -0.0152* (0.00619) across columns.
    - LTV (lag): -0.123*** (0.0218); -0.128** (0.0302); -0.131*** (0.0154).
    - DTI x I(DTI>300) (lag): -0.0537*** (0.00816).
    - I(DTI>300) (lag): 26.32*** (2.010).
  - Robustness: results hold controlling for age, size, education, employment, net wealth, and time-invariant country features.

### Robustness: Panel VAR evidence and heterogeneity
- Panel VAR specification:
  - Five-variable recursive panel VAR with 5 lags; quarterly data for 27 countries over 1998Q1–2015Q4; forecast horizon 32 quarters.
  - Variables: real GDP, corporate debt, household debt, house prices, short-term interest rates.
  - Identification: Cholesky order — log real GDP, log real corporate debt, log real household debt, log real house prices, short-term interest rates.
  - Estimation: least squares dummy variable estimator with robustness checks (different orderings, lags, first differences, GMM-style, mean group).
- Dynamic responses:
  - Household debt shock:
    - Short-term: higher real house prices and higher output.
    - Medium-term (about two to five years): declining output.
    - After six to seven years: cumulative economic impact ≈ 0.2 (quarterly) log points of real output, offsetting short-term positives.
  - House price shock:
    - Short-term first four to six quarters: positively associated with output.
    - Medium-term: negatively associated with output; induces household debt increases short- and medium-term then mean reversion.
  - Corporate debt shock:
    - No short-term positive significant effect on real output.
    - A standard deviation increase in corporate debt → output loss ≈ 0.25 quarterly log points after five to six years.
  - Interest rate shock:
    - Negative effects on real output, corporate and household debt, and real house prices.
    - Cumulative negative effects on real GDP reach almost 0.4 log (quarterly) points.
- Heterogeneity by exchange-rate flexibility (sensitivity):
  - Country groups (Hi FX vs Lo FX) listed in source.
  - Short-term positive effects similar across groups; medium-term negative effects four times larger in low FX flexibility countries than high FX group.
  - Negative effects in fixed (low FX flexibility) countries almost double in medium term compared to flexible regimes.
- Heterogeneity by distributional features:
  - Splits by participation rate and debt share of low-income borrowers show:
    - Financial inclusion (higher participation of low-income borrowers) dampens negative effects.
    - Higher debt share held by low-income borrowers amplifies negative effects.

### Conclusions — summarized quantitative and policy-relevant findings
- Main quantified result:
  - A one standard deviation increase in the household debt ratio is, on average, associated with a 1.2 percentage points lower output growth over the following three years (sample: 80 countries).
- Mechanisms and magnitudes:
  - Banking-crisis channel:
    - Household debt increases probability of banking crises; effect ≈ twice as large as for nonfinancial corporate debt.
    - Effect stronger when household debt level > 65 percent of GDP.
  - Sentiment and mispricing:
    - Household credit growth predicts lower bank stock returns and higher probability of bank stock crashes; one standard deviation increase → 8 percent higher probability of a bank equity crash within five years.
  - Distributional and consumption-sensitivity channel:
    - Micro evidence: 100-percentage point increase in DTI → 4 percentage points drop in household consumption (2010–2014); very high DTI (>300 percent) associated with ≈ 7 percentage points drop.
- Institutional and policy factors mitigating risk:
  - Flexible exchange rates help mitigate risks from rising household debt.
  - Higher financial development, better transparency, and lower financial risk mitigate negative medium-term macro-financial effects.
  - Greater mortgage participation by low-income households (financial inclusion) appears to reduce negative effects; low average DTI among low-income borrowers also reduces risk though with weaker significance.
- Data and methods notes:
  - Macro debt data sources include BIS, Jordà-Schularick-Taylor (JST), and IMF Monetary and Financial Statistics.
  - Micro-level data: euro area household finance and consumption survey (2010 and 2014 waves) for a panel of five countries.
  - Empirical approaches: panel OLS regressions with country and time fixed effects, panel VAR, logit panel for crisis probability, probit for crash probability, CAPM-based abnormal returns; standard errors clustered or dually clustered as appropriate.

*Source: wp1876 - Appendix I; Section III; Results and Conclusions (IMF staff computations).*

### Appendix I. Data Description and Methodology ...........................................................................

### Appendix I. Data Description and Methodology

### Introduction and research questions
- Recent studies link increases in household debt to lower output growth, higher unemployment, and greater probability of future banking crises (Mian, Sufi, and Verner, 2017; Jordà, Schularick, and Taylor, 2016).
- Paper addresses: (1) generalizability of the negative relationship between household debt and macro outcomes across a broad set of countries; (2) channels and factors explaining this relationship; (3) role of institutional factors.
- Main complementary mechanisms identified:
  - The macro effect of a debt overhang situation, magnitude depends on differences in marginal propensities to consume across households.
  - Household credit booms being reflected in a higher future probability of banking crises.
  - Sentiment driven by rapid increases in household debt being associated with a neglect of crash risks.

### The role of debt in macro-financial models
- Financial frictions and collateral constraints on nonfinancial corporates amplify supply-side shocks (Bernanke and Gertler, 1989; Kiyotaki and Moore, 1997; Caballero and Krishnamurthy, 2003; Brunnermeier and Sannikov, 2014).
- Debt-driven demand channel: when credit supply tightens after a credit boom, nominal rigidities and monetary policy constraints can amplify downward pressure on growth (Eggertsson and Krugman, 2012; Korinek and Simsek; Farhi and Werning, 2016).
- Behavioral biases (present-bias, extrapolative expectations, heterogeneous beliefs, underestimation of risk) can lead to excessive borrowing and amplified leverage cycles (Laibson, 1997; Cheng, Raina, and Xiong, 2014; Geanakoplos, 2010; Baron and Xiong, 2017).
- Demographics and distributional factors matter: younger households borrow against future income; income and wealth inequality linked to rapid household debt growth (Blundell, Browning, and Meghir 1994; Rajan, 2010; Kumhof, Rancière, and Winant, 2015).

### Empirical analysis — A. Household Debt and Output Growth in 80 Countries
- Sample and data description:
  - Sample size: 80 countries.
  - Advanced economies: 39.
  - Emerging market economies: 41.
  - Table 1 describes countries and data sources (Appendix I).
- Forecasting equation (panel regression with country and year fixed effects; standard errors clustered by year and country) relates future real GDP growth to past three-year changes in household debt to GDP and corporate debt to GDP, with controls including lagged GDP growth and past three-year change in government debt to GDP. Forecast horizon varied from k = 0 to k = 6 years.
- Key empirical findings:
  - Regression coefficients on the household debt to GDP ratio are strongly negative across forecasting horizons zero to six years and statistically significant at the 1 percent level.
  - The negative effect is strongest at the three to four-year horizon and diminishes as the horizon increases.
  - Robustness: effects remain significant when sample is split (before/after year 2000, before/after the global financial crisis).
- Economic magnitudes and comparative statistics:
  - Mean household debt to GDP ratio across the sample: 35 percent.
  - Mean annual increase in household debt: about 1 percentage point.
  - Mean firm (nonfinancial corporate) debt to GDP ratio: 60 percent, increasing by slightly less than 1 percentage point per year.
  - Mean public debt to GDP ratio: 52 percent, rising by 2 percentage points per year.
  - Heterogeneity: household debt-GDP ratios at the tenth and ninetieth percentile range between 6 percent and 72 percent.
  - Standard deviation of the three-year change in household debt to GDP ratio: 5.89 percentage points.
  - A one standard deviation increase in the household debt ratio is associated with 1.2 percentage points lower GDP growth over a three-year horizon.
  - Standard deviation of the three-year change in corporate debt to GDP ratio: 18.97 percentage points.
  - Coefficients on corporate debt are more than three-times smaller than those for household debt; negative and significant effects of corporate debt on future income are absent at horizons beyond four to five years.
- Supplementary methods:
  - Results are verified using a panel vector autoregression (VAR) approach (discussed elsewhere in paper).

*Source: wp1876 - Appendix I. Data Description and Methodology*

### Section III.

### Section III.

### Household debt and future income growth
- Sample split:
  - 39 advanced economies (Columns (1)–(4) of Table 2b).
  - Emerging markets sample results reported in Columns (5)–(8) of Table 2b.
- Main findings:
  - The correlation between past growth in household debt and future income growth is negative and statistically significant with a forecasting horizon of one to five years in the advanced-economy sample.
  - The negative effect diminishes at the seven-year horizon in advanced economies.
  - The negative correlation is present for emerging markets, although statistical significance is weak at the three- and five-year horizons due to shorter data span for many emerging market economies.
- Methodological notes:
  - All regressions include lagged GDP growth for the preceding two years as controls.
  - Results remain the same when including the past three-year change in the government debt to GDP ratio (not reported here).
  - For 27 countries with quarterly household debt and GDP data between 1998 and 2015, an analysis of innovations in household debt controlling for corporate debt, house prices, and short-term interest rates yields results very similar to the OLS regressions.

### What drives the negative relationship? (micro vs. cross-country)
- Micro-level studies (e.g., Mian and Sufi 2014a) analyze deleverage after large negative shocks but are typically limited to a few advanced economies due to data quality and face external validity concerns.
- Cross-country analysis provides broader variation that can help overcome external validity issues when assessing macrofinancial policies across countries.
- The paper studies institutional factors and distributional characteristics likely to matter most in the context of household debt.

### Institutional factors and distributional characteristics — empirical framework
- Baseline regression (forecasting horizon fixed at five years for illustration):
  - Δy_{i,t+h} = α_i + β1 ΔDebt_{i,t-1} + β2 ⋅ ΔDebt_{i,t-1} ⋅ Z_i + γ X_{i,t-1} + ε_{i,t+h}
  - Z_i is an indicator for an institutional factor or distributional characteristic of household debt.
  - Coefficient of interest: β2 (interaction term between past household debt growth and indicator Z_i).
  - Qualitatively similar results obtained for forecasting horizons at three or seven years.
  - Alternative panel VAR methodology (Section III) yields qualitatively similar results.

### B.1 Exchange rate regime and capital account openness
- Conceptual mechanism:
  - Increase in household debt transfers funds from savers to borrowers; a negative credit constraint shock can force deleveraging and reduce borrowers’ consumption.
  - Optimal monetary policy should lower interest rates to encourage savers’ consumption to offset borrowers’ cutbacks.
  - Fixed exchange rate regimes constrain monetary policy flexibility; thus, negative effects of household debt growth on future income growth are expected to be stronger under fixed regimes.
- Data and indicators:
  - IMF classification of exchange rate regimes for all 80 countries in the sample; indicator for fixed exchange rate regime equals 1 for fixed, 0 otherwise.
  - High capital account openness defined as having the capital account openness index above the median.
- Empirical findings:
  - Column (1) of Table 3 (controlling for firm debt and lagged GDP growth): the interaction between past household debt growth and the fixed exchange rate indicator is negative and highly significant; household debt growth by itself becomes insignificant.
  - Columns (2) and (3): high capital account openness amplifies the negative effect of household debt on future GDP growth.
  - When both fixed exchange rate regime and high capital account openness are included, results persist: significant interactions with household debt growth while household debt growth by itself is insignificant.
  - Results confirmed using a panel VAR approach (Figure 5; see Section III).
  - Robustness: result not driven solely by euro-area countries; regression remains statistically significant at the 10 percent level when interaction with euro area dummy is included.

### B.2 Financial development, transparency, and financial risk index
- Indicators:
  - Financial Development Index: indicator = 1 if index is within the top quartile of countries in the sample as of 2014, 0 otherwise.
  - Transparency index and Financial Risk Index capture credit information transparency and financial risk, respectively.
- Empirical findings (Columns (4)–(6) of Table 3):
  - High Financial Development Index mitigates the negative effect of household debt: coefficient on the interaction term equals the coefficient on household debt term, suggesting negative effects concentrated in countries with low Financial Development Index.
  - Better transparency and better financial risk likewise mitigate the negative effect, though interaction coefficients are smaller in magnitude than for Financial Development Index.
- Interpretation:
  - Better and more efficient financial markets and institutions help overcome negative medium-term macro-financial effects associated with rising household debt, likely because such systems are better able to assess credit risk, allocate credit, and deal with consequences.

### B.3 Distributional characteristics of household debt: participation rate vs. DTI
- Rationale:
  - Distributional characteristics (who holds debt) matter because differences in marginal propensities to consume (MPCs) between borrowers and savers can generate aggregate effects from debt overhang (Korinek and Simsek 2016).
  - Empirical literature (Mian and Sufi 2009, 2014a) shows concentration of debt among low-income areas or highly leveraged counties correlates with worse outcomes.
- Two micro-based indicators (constructed from latest micro-level data for 30 countries):
  1. Mortgage participation rate of low-income households:
     - Defined as participation for bottom two quintiles (bottom 40 percent) of income distribution.
     - Indicator LowIncPart = 1 if mortgage participation rate for bottom 40 percent ranks within the top quartile of countries (roughly above 20 percent), and 0 otherwise.
     - Column (7) of Table 3: high mortgage participation rate for low-income households mitigates the negative effects of household debt on future income growth.
     - Qualitatively and quantitatively similar to Financial Development Index results (the Financial Development Index uses a larger sample of 80 countries).
     - Interpretation: higher mortgage participation for low-income households signals more efficient banking sector screening and better financial intermediation (financial inclusion).
  2. Average debt-to-income (DTI) ratio of low-income households:
     - Weighted average DTI for bottom 40 percent of income distribution, weighted by share of debt held by these households as percent of total outstanding household debt.
     - Indicator LowDTI = 1 if weighted DTI for these households is low (within the quartile of all countries), and 0 otherwise.
     - Columns (8)–(9) of Table 3: low average DTI for low-income households appears to reduce the negative impact of aggregate household debt on the economy, though statistical significance is weaker.
     - Result remains when controlling for interaction with emerging market (EM) indicator (Column 9).
     - Interpretation: high indebtedness of low-income borrowers likely worsens negative impacts; debt participation (access) and average DTI capture distinct aspects — the former linked to financial inclusion and development, the latter to danger of over-indebtedness.
- Confirmation:
  - These distributional results are further confirmed in Figures 5 and 6 using an alternative panel VAR approach (see Section III).

### C. Systemic banking crises
- Research questions:
  - Do household and corporate credit growth matter for financial crises?
  - What role does level of debt play? Is there a threshold effect?
- Empirical model:
  - Probabilistic logit framework for systemic banking crisis event for country i and year t:
    - log odds = Ψ0 + Ψ1 X_{it} + Ψ2 X_{it} × I(HiDebt_{it}) + country fixed effects + error
    - X_{it} includes lagged changes and levels of household and corporate debt (scaled by GDP).
    - I(HiDebt) equals 1 if household or sovereign debt exceeds specified thresholds.
  - Country fixed effects control for time-invariant country-specific characteristics.
- Key data and parameters:
  - Logit panel estimation covers 34 countries over 1970–2015.
  - Thresholds used as examples:
    - HI household debt threshold = 65 percent of GDP (top quintile of country-time distribution).
    - HI sovereign debt threshold = 60 percent of GDP (top 1/3 of the distribution).
  - Average change in household credit-to-GDP ratio in regression sample: about 1 percentage point.
  - Standard deviation of that change: around 2.5 percentage points.
  - Up to four lags for household and corporate credit changes considered; only significant lags included in baseline specification.
- Main findings:
  - Household debt is a good early warning indicator for banking crises.
  - Both household and corporate debt-to-GDP ratios are positively associated with greater probability of systemic banking crises (Table 4).
  - Changes in household debt are found more important than levels (Column 3).
  - Effects of household debt tend to dominate those of corporate debt (Column 4).
  - The relationship between increasing household debt and financial crises is more pronounced when household debt level exceeds 65 percent of GDP (Column 5).
  - Probability of banking crisis is larger when sovereign debt levels are high (above 60 percent of GDP) suggesting constrained government capacity to support banks (Column 6).
- Quantitative magnitudes:
  - The average marginal effect of changes in household debt is about 1 percentage point, almost double the effect of firm debt increases.
  - When household debt level is high, the probability of systemic banking crises is boosted by another 80 basis points.
  - Unconditional crisis probability for the countries considered is about 3.5 percent.
- Model performance:
  - Crisis predictability compared using Area Under Curve (AUC) metric across four nested models:
    - xb0: country fixed effects only.
    - xb1: country fixed effects + level of household debt.
    - xb2: country fixed effects + changes in household debt.
    - xb3: country fixed effects + levels and changes of household and corporate debt + interactions of household debt with high household debt and high sovereign debt levels.
  - Performance comparison assesses incremental predictive power of debt levels, changes, and interactions.

*Source: wp1876 - Section III.*

### 0.87 dominates the second (xb1) and the third model (xb2) which in turn do a better job than

### wp1876 - 0.87 dominates the second (xb1) and the third model (xb2) which in turn do a better job than

### D. Neglected Downside Risk — overview
- Research question: whether investors systematically neglect downside risk to asset returns and economic outcomes during the buildup of household leverage, potentially causing overborrowing and reinforcing other mechanisms in the paper.
- Behavioral channels noted:
  - Extrapolative expectations about future house prices (example: professionals in mortgage securitization were aggressive buyers in the 2000s housing boom).
  - “This time is different” thinking and mean-reverting investor sentiment.
- Implication: mispricing of risk in banking sector implies banks may require larger capital buffers than market-implied measures (e.g., value-at-risk), supporting macroprudential policy rationale.

### D1. Predictability of Bank Stock Returns
- Empirical setup:
  - Dependent variable: h-year ahead excess return of country i’s banking sector index relative to the risk-free rate.
  - Key regressor: past k-year growth in the household debt to GDP ratio, normalized by the standard deviation of annual changes for each country.
  - Controls (X): include past k-year growth in corporate debt to GDP ratio and other controls.
  - Sample: 70 countries between 1973 and 2016.
  - Fixed effects: country- and year-fixed effects.
  - Main specification: h = 1, 3, 5 and k = 3. Standard errors clustered at country level.
- Main findings:
  - One-year horizon (k=1 / past three-year change in household debt):
    - Past three-year change in household debt ratio is negatively associated with one-year ahead bank equity returns.
    - Relationship remains significant controlling for corporate debt growth and past debt levels.
    - Magnitude: one standard deviation increase in the annual growth of the household debt ratio → lower equity return of 2 percent to 2.7 percent one year later.
  - Three- and five-year horizons:
    - Relationship becomes strongly significant.
    - Magnitude: one standard deviation increase in the annual growth rate of the household debt ratio → lower equity returns of 12 percent to 15 percent at the three- and five-year horizons.
  - Interpretation: results are deviations from country averages (country fixed effects included); among countries with similar average conditions, those with higher household credit growth have lower future equity returns.

### D2. Probability of Bank Stock Crashes
- Empirical setup:
  - Probit model where dependent indicator = at least one annual return below -30 percent in the next k years.
  - Independent variables normalized by annual standard deviation; coefficients represent marginal effects per one standard deviation increase.
- Main finding:
  - Columns (8) and (9): past increases in the household debt ratio are associated with significantly higher probability of bank equity crashes three to five years later.
  - Magnitude: one standard deviation increase in the growth rate of the household debt ratio → 8 percent higher probability of having a bank equity crash within the next five years.

### D3. Abnormal Returns of Bank Stocks
- Two-stage approach:
  - Stage 1: estimate banking sector market beta using CAPM; residuals = abnormal returns (bank return minus beta × market return). Quarterly regressions with rolling windows; require at least 20 past quarters. Sample restricted to 30 countries due to data history.
  - Stage 2: regress abnormal returns (one- to three-year ahead) on past three-year growth in household credit. Time fixed effects dropped since market factor controlled for.
- Main findings:
  - Past three-year growth in the household debt ratio is associated with negative future abnormal returns for the banking sector.
  - Relationship statistically significant at the two- to three-year horizons.
  - Including past three-year change in the corporate debt ratio as control does not alter results.
  - In the 30-country subsample, corporate debt ratio is also negatively (and slightly more strongly) correlated with future banking sector abnormal returns.

### E. Debt Overhang: Micro-level Evidence
- Aggregate and micro patterns:
  - Aggregate: private consumption fell more after the crisis in countries with steeper pre-GFC household debt increases; moderate household credit growth countries saw modest consumption increases.
  - Micro: similar pattern found in household-level data.
- Household-level regression (change in household food consumption, percent of income):
  - Key regressor: debt-to-income ratio (DTI_i,2010) as proxy for past household indebtedness.
  - Controls: household characteristics (size of household main residence, employment, education, age of reference person), country fixed effects; errors clustered at country level.
  - Data: micro-level longitudinal data for five euro area countries (Belgium, Cyprus, Germany, Malta, the Netherlands) for two waves (2010 and 2014); about 3000 households with borrowing and consumption information; population weights considered.
- Main findings:
  - Higher indebtedness (DTI or loan-to-value ratios) increases household vulnerability to income shocks.
  - Negative effects intensify when total debt exceeds 300 percent of household disposable income.
  - Magnitudes:
    - 100 percentage point increase in DTI ratio → 4 percentage points drop in consumption.
    - For households with total debt in excess of 300 percent of disposable income → about 7 percentage points drop in consumption.
  - Robustness: results hold controlling for age, size, education, employment, net wealth, and time-invariant country features.
- Conclusion: micro evidence confirms debt overhang channel consistent with macro results.

### Robustness: Panel VAR Analysis
- Model specification:
  - Five-variable recursive panel VAR with 5 lags; quarterly data for 27 countries over 1998Q1–2015Q4.
  - Variables: real GDP, corporate debt, household debt, house prices, short-term interest rates.
  - Country and time fixed effects included.
  - Identification: Cholesky decomposition ordering — log real GDP, log real corporate debt, log real household debt, log real house prices, short-term interest rates.
  - Estimation: least squares dummy variable estimator (LSDV) with robustness checks (different orderings, different lags, first differences, GMM-style estimator, mean group estimator).
  - Sample of 27 countries listed in source.
  - Note: small T not an issue (T = 68).
- Key dynamic results:
  - Household debt shock:
    - Short-term: leads to higher real house prices and higher output.
    - Medium-term (after about two to five years): results in declining output.
    - After six to seven years: cumulative economic impact ≈ 0.2 (quarterly) log points of real output, offsetting short-term positive effects.
  - House price shock:
    - Short-term (first four to six quarters): positively associated with output.
    - Medium-term: negatively associated with output.
    - House price shock causes household debt to increase steadily short- and medium-term, then revert to long-term mean.
  - Corporate debt shock:
    - No short-term positive and significant effect on real output.
    - A standard deviation increase in corporate debt → output loss of about 0.25 quarterly log points after five to six years.
  - Interest rate shock:
    - Negative effects on real output, corporate and household debt, and real house prices.
    - Cumulative negative effects on real GDP reach almost 0.4 log (quarterly) points.
- Robustness and interpretation caveats noted (differences in sample, frequency, methodology).

### A. Exchange Rate Regime (sensitivity)
- Countries split into high and low FX flexibility groups based on sample average de facto classification.
  - Hi FX group: AUS, CAN, COL, GBR, JPN, KOR, NOR, NZL, SGP, SWE, THA, USA.
  - Low FX group: BEL, CZE, DEU, DNK, ESP, FIN, FRA, GRC, HKG, HUN, ISR, ITA, LTU, NLD, PRT.
- Findings:
  - Short-term positive effects of household debt on output similar across groups.
  - Medium-term negative effects are four times larger in low FX flexibility countries than in high FX flexibility countries.
  - Negative effects in fixed exchange rate (low FX flexibility) countries are almost double in magnitude over the medium term compared to flexible regimes.

### B. Distributional Characteristics of Household Debt
- Role of distributional features:
  - Low-income households (bottom 40 percent) have less access to credit, higher debt-to-income ratios (proxy for riskiness), lower participation rates, and higher marginal propensities to consume (MPC).
  - Highly indebted low-income households reduce consumption sharply when facing income shocks, amplifying macro negative effects.
- Sensitivity analyses:
  - Split countries by credit participation rate of low-income households (financial inclusion):
    - Financial inclusion found to dampen negative effects associated with household debt on output.
  - Split countries by debt share of low-income households:
    - Countries with higher debt share held by low-income borrowers are impacted more negatively than countries with lower debt share.
- Conclusion: both financial inclusion (credit participation) and riskiness of low-income borrowers (debt share) matter for future output growth and exert opposite effects on the transmission of household debt to macro outcomes.

*Source: wp1876 (IMF working paper content as provided in the source text).*

### CONCLUSIONS

### CONCLUSIONS

### Main findings on household debt and growth
- Over the business cycle, high growth in household borrowing is negatively associated over the medium term with economic growth.
- Sample: 80 advanced and emerging market economies spanning 65 years (1950–2016).
- Magnitude: a one standard deviation increase in the household debt ratio is, on average, associated with a 1.2 percentage points lower output growth over the following three years.
- Effect heterogeneity: the negative effect appears stronger for advanced economies than for emerging markets.
- Results are in line with Mian, Sufi and Verner (2017), which first documented such relationship for 30 advanced economies.

### Role of country characteristics and household composition
- Institutional and policy factors that mitigate risks from rising household debt:
  - Flexible exchange rates help mitigate risks associated with increasing household debt.
  - Higher financial development helps mitigate risks associated with increasing household debt.
- Micro-household and country characteristics (smaller panel sample):
  - Higher participation by low-income households, suggestive of greater financial inclusion, appears to reduce the negative effect of household debt on medium-term GDP growth.
  - A higher debt share (potentially reflecting a debt overhang effect) is associated with a more negative effect.

### Quantified mechanisms linking household indebtedness to future growth declines
- Three complementary, related mechanisms (which can mutually reinforce each other and interact with contractionary effects of binding borrowing constraints):
  - Banking-crisis channel:
    - Household debt increases the probability of banking crises and the effect is twice as large as for non-financial corporate debt.
    - The effect is even stronger when the level of household debt is above 65 percent of GDP.
    - Links to prior evidence that increases in private debt raise the probability of systemic banking crises and that economic costs associated with financial crises are higher than in normal downturns.
  - Sentiment, mispricing, and bank equity returns:
    - Sentiment-driven behavior is associated with neglected crash risk and abnormal bank stock returns.
    - Analysis of bank stock performance in 70 countries over the past 40 years found that household credit growth systematically predicts lower bank stock returns (both excess stock returns and market-adjusted abnormal returns), as well as bank stock crashes, in the next two to three years.
    - This predictability suggests neglected crash risks associated with growth in household credit; resulting price corrections typically trigger sharp asset price declines, increases in risk premiums, and significant resource reallocations.
  - Distributional and consumption-sensitivity channel:
    - Distributional characteristics of household debt matter because marginal propensities to consume differ across indebted households and higher leverage increases sensitivity to negative income shocks or tighter credit constraints.
    - Micro-level results for Europe: a 100-percentage point increase in the debt-to-income ratio translates into a 4 percentage point drop in household consumption between 2010 and 2014.
    - More indebted households exhibit higher sensitivity to shocks.

### Data and methodological notes (summary)
- Macro debt data sources: BIS, Jordà-Schularick-Taylor (JST), and IMF’s Monetary and Financial Statistics; country coverage and time span vary, with most advanced economies starting in the 1950s and emerging markets in the 1990s.
- Micro-level household debt data: euro area household finance and consumption survey for two waves (2010 and 2014); panel available for Belgium, Cyprus, Germany, Malta, and the Netherlands where the same households report consumption and debt in both waves.
- Empirical approach: regressions control for country and time fixed effects and lagged GDP growth; standard errors dually clustered on country and year.

*Source: wp1876 - CONCLUSIONS*

### 2014. Transparency takes value 1 if consumer credit transparency index is 1. LowIncPart takes

### wp1876 - 2014. Transparency takes value 1 if consumer credit transparency index is 1. LowIncPart takes

### Key variable definitions and model setup
- Transparency: takes value 1 if consumer credit transparency index is 1.
- LowIncPart: takes value 1 if the mortgage participation rate for the bottom 4 percent of households in the income distribution is within the top 25 percent of countries in the most recent year where data are available.
- FINRISK: takes value 1 if financial risk rating is above the median (higher rating indicates less risk).
- LowDTI: takes value 1 if the weighted debt-to-income ratio for the bottom 40 percent of households (mortgage borrowers) in the income distribution is within the lower 25 percent of countries in the most recent year where data are available.
- High household debt I(Hi hhd): dummy = 1 if household debt level exceeds 65 percent of GDP.
- High government debt I(Hi Gov Debt): dummy = 1 if government debt level exceeds 60 percent of GDP.
- All independent variables are lagged in crisis probability regressions; the third lag of household debt change is used in core specifications.
- Regressions include country and time fixed effects; standard errors are dually clustered on country and year (or clustered at the country level in crisis models).

### Probability of Systemic Banking Crisis (Table 4 and robustness)
- Main logit panel estimates (selected coefficients and diagnostics):
  - Coefficient on household debt level (݄݄݀) reported across specifications: 4.037***; 2.501***; 1.270; 1.727; 2.091; 2.479 (standard errors reported in table).
  - Coefficient on household debt change (݄݄݀Δ) across specs: 40.05***; 35.01***; 35.60***; 31.25***; 30.86***; 26.47*** (standard errors reported in table).
  - Coefficients on corporate debt change (Δfd): 13.13***; 12.64***; 15.62***; 15.33*** (standard errors reported in table).
  - Interaction terms:
    - ݄݄݀Δ x I(Hi Gov Debt): 22.62*; 24.12* (standard errors reported).
    - ݄݄݀Δ x I(Hi hhd): 24.41*; 25.93* (standard errors reported).
  - Constant terms: around -5.949*** to -5.534*** across specifications.
  - Sample sizes and diagnostics:
    - N: 1,223; 1,033; 1,033; 1,020; 1,020; 1,020; 1,020 across columns.
    - Countries: 40; 34; 34; 34; 34; 34; 34.
    - Number of crises: 46; 37; 37; 37; 37; 37; 37.
    - AUC (area under curve) by model: 0.700; 0.791; 0.806; 0.840; 0.845; 0.850; 0.856.
    - Pseudo R2 ranges from 0.0612 to 0.228 across reported models.
- Robustness to lags (Table 5a):
  - Δhhd third lag (݄݄݀Δ (l3)) is strongly positive and significant: 32.93***; 29.89***; 26.99*** (standard errors reported).
  - Δ Corporate Debt (l1) significant in some specs: 9.528**; 11.48***; 13.94*** (standard errors reported).
  - Samples: N 938; 938; 972. Countries: 33; 33; 34.
- Robustness to estimation methods (Table 5b):
  - Δhhd strong and significant across methods: 34.19***; 30.73***; 24.61***; 29.57*** (logit FE, panel logit, Firth logit, Poisson).
  - Δfd remains positive and in some methods significant: 6.413**; 5.675; 3.136; 5.527*.
  - Observations: 786 in robustness-method panel; Number of crises: 35; Countries: 32.

### Bank equity returns and crash probabilities (Table 6)
- Relationship between past household debt growth and future bank stock returns / equity crash probability:
  - Future bank stock returns (selected coefficients, k horizons):
    - k = 1: Δhousehold-debt coefficient -0.020* (standard error (0.011)).
    - k = 3: -0.027** (0.013).
    - k = 5: -0.120** (0.032) and related specifications -0.113** (0.037), -0.159** (0.050), -0.123** (0.055).
  - Probability of future equity crashes (marginal effects, selected):
    - Some specifications show positive and significant effects on crash probability (e.g., coefficients 0.053** (0.011); 0.099** (0.017); 0.057** (0.020)).
  - Samples and fit:
    - N reported for return regressions: N 1,488; 1,319; 1,348; 1,319; 1,208; 1,179.
    - Country coverage: 70 countries for return regressions; R2 ranges 0.27 to 0.37 for return models and 0.11 to 0.15 for crash probability models.

### Abnormal returns for bank stocks (Table 7)
- Past household debt growth and future abnormal returns (CAPM residuals), horizons k = 1, 2, 3:
  - Selected coefficients:
    - k = 1: Δhousehold-debt coefficient -0.049 (0.063); Δcorporate-debt -0.198** (0.081).
    - k = 2: Δhousehold-debt 0.003 (0.069); Δcorporate-debt -0.401*** (0.098); other lags show larger negative and significant coefficients (e.g., -0.503*** (0.134), -0.723*** (0.161)).
    - k = 3: Δhousehold-debt -0.228*** (0.079); -0.145* (0.083); -0.289* (0.090); -0.289*** (0.097) in various specs.
  - Samples:
    - N: 723; 723; 722; 722; 693; 693 across columns.
    - Countries: 30 countries in these analyses.
    - R2 ranges from 0.00 to 0.11 across horizons.

### Euro area micro evidence: household debt overhang and consumption (Table 8; Figures 2–3)
- Cross-section regressions linking past indebtedness and changes in consumption-to-income ratio for euro area households:
  - DTI (lag) coefficients:
    - Column (1): DTI (lag) -0.0396*** (0.00235).
    - Column (2): -0.0404*** (0.00183).
    - Column (3): -0.0401*** (0.00226).
    - Column (4): -0.0152* (0.00619).
  - LTV (mortgages, lag) coefficients:
    - -0.123*** (0.0218); -0.128** (0.0302); -0.131*** (0.0154) across columns.
  - Interaction and threshold variables:
    - DTI x I(DTI>300) (lag): -0.0537*** (0.00816).
    - I(DTI>300) (lag): 26.32*** (2.010).
  - Household controls with examples:
    - Size of household main residence: 0.0294** (0.00694); -0.0506* (0.0173); 0.0200* (0.00779).
    - Education of reference person: 0.986*** (0.0584); 0.557 (2.923); 0.721*** (0.108).
    - Age of reference person: 0.110 (0.0469); 0.116 (0.0720); 0.132** (0.0264).
  - Sample sizes and fit:
    - N: 2,925; 699; 2,925; 699; 2,744; 656; 2,744 across columns.
    - R2: 0.102; 0.059; 0.103; 0.113; 0.109; 0.133; 0.142.
    - Country FE: Y in all reported regressions; Net Wealth FE: Y in all reported regressions.

### Panel VAR evidence and cross-country heterogeneity (Figures 4–8)
- Panel VAR baseline (Figure 4):
  - Countries included: AUS, BEL, CAN, COL, CZE, DEU, DNK, ESP, FIN, FRA, GBR, GRC, HKG, HUN, ISR, ITA, JPN, KOR, LTU, NLD, NOR, NZL, PRT, SGP, SWE, THA, USA.
  - Forecast horizon: 32 quarters. Estimation period: 1998:Q1–2015:Q4.
  - Identification: Cholesky decomposition order: log real GDP, log real corporate debt, log real household debt, log real house prices, short-term interest rates.
  - Dashed lines: 95 percent confidence intervals (1000 Monte Carlo simulations).
- Heterogeneity by exchange-rate flexibility (Figure 5):
  - “Hi_FX” group: AUS, CAN, COL, GBR, JPN, KOR, NOR, NZL, SGP, SWE, THA, USA.
  - “Lo_FX” group: BEL, CZE, DEU, DNK, ESP, FIN, FRA, GRC, HKG, HUN, ISR, ITA, LTU, NLD, PRT.
  - Forecast horizon: 32 quarters. Estimation period: 1998:Q1–2015:Q4.
- Participation rate and financial development (Figure 6):
  - Participation for low-income borrowers refers to mortgages in the bottom 40 percent of the income distribution, micro waves 2011–2013.
  - Positive cross-country relationship depicted between Financial Development Index and Participation Rate of Low Income Households (%).
- Heterogeneity by low-income borrower participation (Figure 7):
  - “Low Participation” group: PRT, HUN, BEL, FIN, FRA, GRC, DEU, ITA, ISR.
  - “High Participation” group: NOR, SWE, KOR, JPN, NLD, ESP, USA, GBR, AUS, CAN.
  - Forecast horizon: 32 quarters. Estimation period: 1998Q1–2015Q4.
- Heterogeneity by debt share of low-income borrowers (Figure 8):
  - “High Debt Share” group: AUS, ESP, GRC, HUN, ITA, JPN, KOR, NLD.
  - “Low Debt Share” group: BEL, CAN, DEU, FIN, FRA, GBR, PRT, USA.
  - Forecast horizon: 32 quarters. Estimation period: 1998:Q1–2015:Q4.

### Substantive takeaways
- Changes in household debt (particularly substantial positive changes and third lags of household debt change) are strongly associated with higher probability of systemic banking crises (coefficients on ݄݄݀Δ large and statistically significant across main and robustness specifications).
- Corporate debt change (Δfd) also contributes positively to banking crisis probability in multiple specifications.
- Interactions suggest that household-debt growth raises crisis risk more when sovereign debt or household debt levels are high (positive and significant interaction coefficients with I(Hi Gov Debt) and I(Hi hhd)).
- Past household debt growth is negatively associated with future bank stock returns and positively associated with the probability of large equity crashes over multiple horizons.
- Abnormal returns analyses indicate negative effects of debt dynamics on bank abnormal returns, with stronger effects at longer horizons in several specifications.
- Euro-area microdata: higher lagged DTI and higher lagged LTV are associated with declines in consumption-to-income ratios; DTI effects are statistically significant and economically meaningful (e.g., DTI (lag) -0.0396***).
- Panel VAR results demonstrate dynamic responses of GDP, corporate debt, household debt, house prices, and short-term interest rates to shocks in corporate and household debt, with heterogeneity across groups defined by exchange-rate flexibility, low-income borrower participation, and low-income borrower debt share.

*Source: IMF staff computations, wp1876.*

### REFERENCES

### REFERENCES

### References list

- Albanesi, Stefania, Giacomo De Giorgi, and Jaromir Nosal. 2017. “Credit Growth and the Financial Crisis: A New Narrative”, NBER Working Paper 23740, National Bureau of Economic Research, Cambridge, MA.
- Arcand, Jean Louis, Enrico Berkes, and Ugo Panizza. 2015. “Too Much Finance?” Journal of Economic Growth 20 (2): 105–48.
- Badarinza, Cristian, Vimal Balasubramaniam, and Tarun Ramadorai. 2016. “The Indian Household Savings Landscape.” NCAER Working Paper, National Council of Applied Economic Research, New Delhi, India.
- Badarinza, Cristian, John Y. Campbell, and Tarun Ramadorai. 2016. “International Comparative Household Finance.” Annual Review of Economics 8 (1): 111–44.
- Baron, Matthew, and Wei Xiong. 2017. “Credit Expansion and Neglected Crash Risk.” Quarterly Journal of Economics: 713–64.
- Beck, Thorsten, Berrak Büyükkarabacak, Felix K. Rioja, and Neven T. Valev. 2012. “Who Gets the Credit? And Does It Matter? Household vs. Firm Lending across Countries.” B.E. Journal of Macroeconomics 12 (1): 1–46.
- Beck, Thorsten, and Ross Levine. 2004. “Stock Markets, Banks, and Growth: Panel Evidence.” Journal of Banking and Finance 28 (3): 423–42.
- Beck, Thorsten, and Norman Loayza. 2000. “Finance and Sources of Growth.” Journal of Financial Economics 58 (1–2): 261–300.
- Bernanke, Ben and Mark Gertler. 1989. “Agency Costs, Net Worth, and Business Fluctuations.” American Economic Review, Vol. 79, No. 1, pp. 14-31.
- Blundell, Richard, Martin Browning, and Costas Meghir. 1994. “Consumer Demand and the Life-Cycle Allocation of Household Expenditures.” Review of Economic Studies 61 (1): 57–80.
- Brunnermeier, Markus K., and Yuliy Sannikov. 2014. “A Macroeconomic Model with a Financial Sector.” American Economic Review 104 (2): 379–421.
- Brunnermeier, Markus, Darius Palia, Karthik Sastry, and Christopher Sims. 2017. “Feedbacks: Financial Markets and Economic Activity.” Working Paper.
- Caballero, Ricardo J., and Arvind Krishnamurthy.2003. “Excessive Dollar Debt: Financial Development and Underinsurance,” Journal of Finance 58: 867–893.
- Calza, Alessandro, Tommaso Monacelli, and Livio Stracca. 2013. “Housing Finance and Monetary Policy.” Journal of the European Economic Association 11 (1): 101–22.
- Cheng, Ing-Haw, Sahíl Raina, and Wei Xiong. 2014. “Wall Street and the Housing Bubble.” American Economic Review 104 (9): 2797–829.
- Chinn, Menzie D. and Hiro Ito (2008). "A New Measure of Financial Openness." Journal of Comparative Policy Analysis 10 (3): 309–22 (September).
- Coibion, Olivier, Yuriy Gorodnichenko, Marianna Kudlyak, and John Mondragon. 2017. “Does Greater Inequality Lead to More Household Borrowing? New Evidence from Household Data.” Federal Reserve Bank of Minneapolis Working Paper 17-04, Minneapolis, MN.
- Crowe, Christopher, Giovanni Dell’Ariccia, Deniz Igan, and Pau Rabanal. 2013. “How to Deal with Real Estate Booms: Lessons from Country Experiences.” Journal of Financial Stability 9 (3): 300–19.
- De Gregorio, Jose. 1996. “Borrowing Constraints, Human Capital Accumulation, and Growth.” Journal of Monetary Economics 37 (1): 49–71.
- Deaton, Angus. 1991. “Saving and Liquidity Constraints.” Econometrica 59 (5): 1221–48.
- Dell’Ariccia, Giovanni, Luc Laeven, Deniz Igan, and Hui Tong, “Policies for Macro-financial Stability: How to Deal with Credit Booms,” IMF Staff Discussion Note, 2012.
- Demirgüç-Kunt, Asli, and Leora Klapper. 2012. “Measuring Financial Inclusion: The Global Findex Database.” Policy Research Working Paper 6025, World Bank, Washington, DC.
- De Nicolò, Gianni, Giovanni Favara, and Lev Ratnovski. 2012. “Externalities and Macroprudential Policy.” IMF Staff Discussion Note 12/05, International Monetary Fund, Washington, DC.
- Drehmann, Mathias, Mikael Juselius, and Anton Korinek. 2017. "Accounting for Debt service: The Painful Legacy of Credit Booms.", BIS Working Paper No. 645.
- Drehmann, Mathias, and Kostas Tsatsaronis. 2014. “The Credit-to-GDP Gap and Countercyclical Capital Buffers: Questions and Answers.” BIS Quarterly Review (March).
- Eggertsson, Gauti B., and Paul Krugman. 2012. “Debt, Deleveraging, and the Liquidity Trap: A Fisher-Minsky-Koo Approach.” Quarterly Journal of Economics 127 (3): 1469–513.
- Fisher, Irving. 1933. “The Debt-Deflation Theory of Great Depressions.” Econometrica 1 (4): 337–57.
- Friedman, Milton. 1957. A Theory of the Consumption Function. Princeton, NJ: Princeton University Press.
- Foote, Christopher L., Lara Loewenstein, and Paul S. Willen. 2016. “Cross-sectional patterns of mortgage debt during the housing boom: Evidence and implications.” National Bureau of Economic Research Working Paper no. w22985.
- Fuster, Andreas, David Laibson, and Brock Mendel. 2010. “Natural Expectations and Macroeconomic Fluctuations.” Journal of Economic Perspectives 24 (4): 67–84.
- Gan, Li, Zhichao Yin, Nan Jia, Shu Xu, Shuang Ma, and Lu Zheng. 2013. Data You Need to Know about China. Research Report of China Household Finance Survey. Heidelberg: Springer.
- Geanakoplos, J. 2010. “The Leverage Cycle.” NBER Macroeconomics Annual 2009, Vol. 24. Edited by Darron Acemoglu, Kenneth Rogoff, and Michael Woodford. Cambridge, MA: National Bureau of Economic Research.
- Gennaioli, Nicola, Andrei Shleifer, and Robert Vishny. 2012. “Neglected Risks, Financial Innovation, and Financial Fragility.” Journal of Financial Economics 104 (3): 452−68.
- Gourinchas, Pierre-Olivier, and Maurice Obstfeld. 2012. “Stories of The Twentieth Century for the Twenty-First.” American Economic Journal: Macroeconomics 4 (1): 226–65.
- Hall, Robert E. 1978. “Stochastic Implications of the Life Cycle–Permanent Income Hypothesis: Theory and Evidence.” Journal of Political Economy 86 (6): 971–87.
- Ilzetzki, Ethan, Carmen M. Reinhart, and Kenneth S. Rogoff. 2017. “Exchange Arrangements Entering the 21st Century: Which Anchor Will Hold?” NBER Working Paper 23134, National Bureau of Economic Research, Cambridge, MA
- Jappelli, Tullio, and Marco Pagano. 1994. “Saving, Growth, and Liquidity Constraints.” Quarterly Journal of Economics 109 (1): 83–109.
- Jordà, Òscar, Moritz Schularick, and Alan M. Taylor. 2016. “The Great Mortgaging: Housing Finance, Crises and Business Cycles.” Economic Policy 31 (85): 107–52.
- Kiyotaki, Nobuhiro, and John Moore. 1997. “Credit Cycles.” Journal of Political Economy 105 (2): 211–48.
- Korinek, Anton, and Alp Simsek. 2016. “Liquidity Trap and Excessive Leverage.” American Economic Review 106 (3): 699–738.
- Kumhof, Michael, Romain Rancière, and Pablo Winant. 2015. “Inequality, Leverage, and Crises.” American Economic Review 105 (3): 1217–45.
- Laeven, Luc, and Fabián Valencia. 2013. “Systemic Banking Crises Database.” IMF Economic Review 61 (2): 225–70.
- Laibson, David. 1997. “Golden Eggs and Hyperbolic Discounting.” Quarterly Journal of Economics 112 (2): 443–78.
- Levine, Ross. 1998. “The Legal Environment, Banks, and Long-Run Economic Growth.” Journal of Money, Credit and Banking 30 (3): 596–613.
- Lombardi, Marco, Madhusudan Mohanty, and Ilhyock Shim. 2017. “The Real Effects of Household Debt in the Short and Long Run.” BIS Working Paper 607, Bank for International Settlements, Basel.
- Mian, Atif, and Amir Sufi. 2009. “The Consequences of Mortgage Credit Expansion: Evidence from the U.S. Mortgage Default Crisis.” Quarterly Journal of Economics 124 (4): 1449–96.
- Mian, Atif, Kamalesh Rao, and Amir Sufi. 2013. “Household Balance Sheets, Consumption, and the Economic Slump.” Quarterly Journal of Economics 128 (4): 1687–726.
- Mian, Atif, Amir Sufi. 2011. “House Prices, Home Equity–Based Borrowing, and the U.S. Household Leverage Crisis.” American Economic Review 101 (5): 2132–56.
- Mian, Atif, and Amir Sufi. 2014a. “What Explains the 2007-2009 Drop in Employment?” Econometrica 82 (6): 2197–223.
- Mian, Atif, and Amir Sufi. 2014b. House of Debt. Chicago: University of Chicago Press.
- Mian, Atif, and Emil Verner. 2017. “Household Debt and Business Cycles Worldwide.” Quarterly Journal of Economics: 1–63.
- Mussa, Michael, Giovanni Dell’Ariccia, Barry J. Eichengreen, and Enrica Detragiache. 1998. “Capital Account Liberalization: Theoretical and Practical Aspects.” IMF Occasional Paper 172, International Monetary Fund, Washington, DC.
- Rajan, Raghuram G. 2010. Fault Lines: How Hidden Fractures Still Threaten the World Economy. Princeton, NJ: Princeton University Press.
- Sahay, Ratna, Martin Čihák, Papa N’Diaye, Adolfo Barajas, Ran Bi, Diana Ayala, Yuan Gao. 2015. “Rethinking Financial Deepening: Stability and Growth in Emerging Markets.” IMF Staff Discussion Note 15/08, International Monetary Fund, Washington DC.
- Sanders, Anthony. 2008. “The Subprime Crisis and its Role in the Financial Crisis.” Journal of Housing Economics 17(4): 254–261, December.
- Schmitt-Grohé, Stephanie, and Martín Uribe. 2016. “Downward Nominal Wage Rigidity, Currency Pegs, and Involuntary Unemployment.” Journal of Political Economy 124 (5): 1466–514.
- Sheedy, Kevin D. 2014. “Debt and Incomplete Financial Markets: A Case for Nominal GDP Targeting.” Brookings Papers on Economic Activity Spring: 301–72.
- Stulz, Rene M. 1999. “Globalization of Equity Markets and the Cost of Capital.” NBER Working Paper 7021, National Bureau of Economic Research, Cambridge, MA.
- Svirydzenka, Katsiaryna. 2016. “Introducing a New Broad-Based Index of Financial Development.” IMF Working Paper 16/5, International Monetary Fund, Washington, DC.
- Uribe, Martín, and Stephanie Schmitt-Grohé. 2017. Open Economy Macroeconomics. Princeton, NJ: Princeton University Press.

*Reference list from "REFERENCES" (wp1876).*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp1876.pdf_