## 5. Impact of Private Deleveraging on Growth and Public Debt (IPWRA)

## Source details

**Canonical URL:** [5. Impact of Private Deleveraging on Growth and Public Debt (IPWRA)](https://www.imf.org/-/media/files/publications/wp/2018/wp18141.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18141.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18141.pdf.json)

---

### I. Introduction and main findings
- Governments respond systematically to private-sector deleveraging with countercyclical rises in deficits and public debt, effectively transferring private debt burdens to public balance sheets via weaker growth and fiscal responses.
- Key quantitative observations:
  - Advanced economies experienced slow private deleveraging since the GFC while government borrowing rose from 2007 through 2014 before leveling off.
  - A decade after the onset of the GFC, advanced economies are "39 percent of GDP deeper in debt."
  - Spain: public debt has increased by "close to 59 percent of GDP since 2007" while only "4¾ percent of GDP stems from capital injections and called guarantees."
- Core mechanisms:
  - Primary channel: growth — private deleveraging weighs on economic activity, prompting cyclical deterioration in public finances and a countercyclical rise in public debt.
  - Debt substitution (private → public) occurs irrespective of whether deleveraging coincides with a financial crisis.
  - Outcome hinge: severity and duration of growth slowdown determine whether total debt falls or rises after deleveraging:
    - Benign slowdowns → overall debt burden ultimately goes down.
    - Protracted slowdowns → "growthless deleveraging" leaves countries with more debt than at the outset.
- Timing note: decreases in borrowing flows, growth slowdown, and rise in public debt often occur before a visible decline in private debt-to-GDP.

### II. Data, definitions, and measurement
- Data source and coverage:
  - Global Debt Database (GDD) provides private, public, and total debt for 190 countries from 1950.
  - Sample coverage expands from 45 countries (accounting for "63 percent of global output in 1950") to "99 percent of the world’s GDP in 2016."
  - Analysis sample restricted to countries with both public and private debt data: "158 countries."
- Preferred aggregates:
  - Private debt measure includes loans and securities (domestic and external).
  - Total debt = nonfinancial private debt + public debt.
- Private deleveraging metrics:
  1. Debt-ratio metric: distance from peak to ensuing trough in private debt-to-GDP ratio; peaks/troughs defined as local maxima/minima within five years (Harding and Pagan (2002)); minimum deleveraging duration = two years.
  2. Leverage-flows metric: experimental; negative "leverage flows" (net borrowing/lending excluding interest) indicate deleveraging.
- Decomposition identity (preserved notation in source):
  - 푑푑푡 − 푑푑푡−1 = �(푟푟 − 푔푔)/(1 + 푔푔)� 푑푑푡−1 − 푝푝푡 + 푆푆푆푆푆푆
    - 푔푔 = nominal GDP growth; 푟푟 = interest rate; 푝푝푡 = net borrowing/lending excluding interest payments; 푆푆푆푆푆푆 = stock-flow adjustments.
  - “Leverage flows” = excess growth in nominal private debt relative to debt service, as a share of GDP; negative when new issuances < what is needed to service existing stock.

### III. Stylized facts (1950–2016)
- Episodes and frequency:
  - "535 private deleveraging episodes, involving 159 countries over 1950–2016."
  - Episodes per decade: low of "17 episodes in the 1950s" to high of "114 in the 1990s."
  - Three regional/decade waves: "2010s for advanced economies (21 episodes over the 6 years covered)"; "2000s in emerging market economies"; "1990s in low-income countries."
  - Incidence: virtually every country undergoes deleveraging at some point; wide heterogeneity across countries.
- Mean episode characteristics (World / Advanced / Emerging / Low-income; mean (SD) where provided):
  - Length (years): World "3.6 (1.9)"; Advanced "3.9 (2.2)"; Emerging "3.4 (1.6)"; Low-income "3.5 (2.0)".
  - Size (peak→trough, % of GDP): World "-11.6 (22.3)"; Advanced "-18.7 (37.7)"; Emerging "-14.5 (20.3)"; Low-income "-6.1 (9.4)".
  - Rate (Size/Length): World "-3.2 (4.8)"; Advanced "-4.3 (5.6)"; Emerging "-4.4 (5.4)"; Low-income "-1.7 (2.5)".
  - Distance between episodes (years): World "8.5 (7.6)"; Advanced "10.9 (9.4)"; Emerging "7.9 (6.9)"; Low-income "8.0 (6.9)".
  - Peak private debt (% of GDP): World "54.1 (60.9)"; Advanced "131.2 (75.7)"; Emerging "51.8 (41.4)"; Low-income "20.4 (20.6)".
  - Pre-deleveraging 5-yr cumulative change in debt (% of GDP): World "9.2 (17.6)"; Advanced "19.0 (27.6)"; Emerging "9.4 (16.1)"; Low-income "3.9 (6.8)".
  - Pre-deleveraging 5-yr avg growth (percent): World "3.5 (3.9)"; Advanced "3.1 (2.5)"; Emerging "4.1 (3.7)"; Low-income "3.3 (4.5)".
  - Growth during deleveraging (avg, percent): World "3.8 (4.2)"; Advanced "3.5 (2.5)"; Emerging "4.0 (4.2)"; Low-income "3.8 (4.8)".
  - Number of episodes: World "535"; Advanced "111"; Emerging "184"; Low-income "240".
  - Probability of deleveraging (share of observations): World "27%"; Advanced "25%"; Emerging "29%"; Low-income "26%".
- Relationship with crises:
  - Less than "15 percent of deleveraging episodes occur within a +/- 3-year window of a financial crisis."
  - Crisis vs Non-crisis (selected comparisons; p-values reported):
    - Length (years): Crisis "4.03" vs Non-crisis "3.51"; p-value "0.03".
    - Size (% of GDP): Crisis "-19.21" vs Non-crisis "-9.34"; p-value "0.00".
    - Rate (size/length): Crisis "-5.26" vs Non-crisis "-2.67"; p-value "0.00".
    - Peak private debt (% of GDP): Crisis "89.67" vs Non-crisis "50.32"; p-value "0.00".
    - Pre-deleveraging 5-yr cum change in debt (% of GDP): Crisis "17.41" vs Non-crisis "7.66"; p-value "0.00".
    - Growth during deleveraging (avg, percent): Crisis "2.82" vs Non-crisis "4.03"; p-value "0.02".
    - Number of episodes: Crisis "77"; Non-crisis "456".
  - Interpretation: most deleveraging episodes do not coincide with financial crises; crisis-related episodes are longer, larger, and associated with higher peak private debt and lower growth.

### IV. Typical macro dynamics around deleveraging episodes (event-study preview)
- Timing (Period 0 = start of deleveraging by debt-to-GDP metric):
  - Private debt ratio: grows faster than normal before the episode and contracts afterwards.
  - Leverage flows: slow progressively before deleveraging and turn negative one year before deleveraging starts — indicating net debt repayment begins prior to visible ratio decline.
  - GDP growth: switches from slight expansion to a deep contraction relative to normal times, hitting a trough one year before start of deleveraging (coincident with negative leverage flows); growth returns to normal about one year after the start.
  - Public debt: increases faster than normal, with the surge front-loaded — starts one year before deleveraging, peaks when leverage flows turn negative and growth troughs, continues one year into deleveraging, then tapers off as growth recovers.
  - Drivers of public-debt increase (advanced-economy decomposition): largely cyclically adjusted primary deficits (CAPB) and macro-related changes; stock-flow adjustments are not major drivers in this sample.
  - Total debt: rises in run up, stabilizes early in deleveraging, and eventually declines as deleveraging proceeds.

### V. Heterogeneity: growthless vs high-growth deleveraging
- Classification thresholds (25th percentile benchmarks since 1950 by income group):
  - Advanced economies: 1.6 percent
  - Emerging market economies: 1.9 percent
  - Low-income countries: 1.1 percent
- Distinguishing outcomes:
  - Growthless deleveraging:
    - Longer (about two years longer than high-growth episodes),
    - Require larger private-sector efforts,
    - Entail larger and continuous increases in public borrowing,
    - Result in more total debt at the end of the process.
  - Magnitudes:
    - Private debt cumulative decline: "17 percent of GDP" in growthless vs "15 percent of GDP" in high-growth.
    - Cumulative leverage flows: "-24 percent of GDP" in growthless vs "-11% of GDP" in high-growth.
    - Public debt cumulative increase (faster than normal): "17 percent of GDP" in growthless vs "6 percent of GDP" in high-growth.
- Prevalence: growthless episodes ≈ one quarter of the sample; occur across income levels.

### VI. Fiscal policy role within growthless deleveraging
- Within growthless episodes, splitting by government support (top vs bottom half of cumulative public debt changes in two years pre-deleveraging) yields:
  - Episodes with the largest government support tend to have the weakest growth (lowest-growth of the growthless group).
  - Strongly supported episodes involve larger private-sector deleveraging and often result in higher total debt than before.
- Caution: correlation does not imply causation; endogeneity of fiscal policy to growth outcomes limits causal interpretation in descriptive splits.

### VII. Toward causality: IPWRA estimator and identification
- Objective: estimate causal impact of private deleveraging on GDP and public debt using Inverse Propensity Weighting Regression Adjustment (IPWRA) — a doubly robust estimator.
- Treatment and outcomes:
  - Treatment d_{i,t} = 1 if year t is the first year leverage flows turn negative for country i.
  - Outcome ΔY_{i,t}^S(d) = Y_{i,t+S}(d) − Y_{i,t}, where Y is GDP or public debt.
  - Causal parameter: β^S = E[ΔY(1)] − E[ΔY(0)] under selection-on-observables.
- Identification assumption:
  - Conditional on observable country characteristics X_{i,t}, private deleveraging is as good as random: ΔY_{i,t}^S(d) ⟂ d_{i,t} | X_{i,t}.
- Estimation strategy (summary):
  - IPW: estimate pr(d_{i,t}=1 | X_{i,t}); compute weights w_{i,t} = 1 / E[d_{i,t}] · [ p̂(d_{i,t}=1) / (1 - p̂(d_{i,t}=1)) ] and reweight control observations to emulate treated ones.
  - RA: regress outcomes ΔY on X separately in treated and control groups to model counterfactuals.
  - IPWRA: combine IPW and RA — reweight and then estimate outcome regressions; causal impact β̂_S computed from weighted predicted differences.
- Covariates used (model selection via iterative pruning with AIC & BIC):
  - Growth model pool: lagged GDP per capita, lagged GDP growth, Law and Order index, trade openness, inflation, government spending, stocks and 3-year cumulative growths in private and public debt.
  - Public debt covariates: lagged GDP growth, GDP per capita, elderly (65+) dependency ratio, effective interest rate on public debt, inflation, political fractionalization, institutional quality, crisis dummy (past 5 years), exchange rate depreciation, lagged private and public debt levels.
  - Private deleveraging covariates: lagged GDP growth, lagged private and public debts, lagged 3-year cumulative private-debt growth, private interest rates, GDP per capita, US Treasury 10-year yields, institutional quality, crisis dummy; housing and stock prices in restricted 16-AE sample.
- Model performance:
  - Predictive accuracy: AUROC consistently > 0.7 for most specifications.
  - Including country and time fixed effects improves predictive power but can cause IPWRA GMM instability and convergence issues in about half of specifications.
- Validation:
  - IPWRA reweighting removes correlation between private deleveraging and past growth/public debt (panel checks in Annex 4).
  - Visual overlap: distributions of estimated propensity scores show significant overlap between treatment and control in full-sample specification.

### VIII. Causal estimates and heterogeneity (IPWRA results)
- Full-sample 5-year horizon (cumulative impacts):
  - On average, onset of private deleveraging induces a "7.2 percent of GDP" rise in public debt over the following 3 years.
  - Effect declines from t+4 onward and becomes statistically insignificant subsequently.
  - IPWRA estimate slightly larger than event-study baseline: "7.2 vs 5.3 percent of GDP."
- Income-group heterogeneity (summary):
  - Advanced economies: peak public-debt impact up to "9.6 percent of GDP."
  - Emerging markets: impact peaks at "5.7 percent of GDP."
  - Low-income countries: estimated impact positive but not statistically significant.
  - Growth slowdown associated with deleveraging observed across income groups; deleveraging-induced rises in public borrowing more prevalent at higher income levels.
- Robustness:
  - Excluding episodes starting within a 5-year window around a financial crisis leaves results virtually unchanged — private deleveraging still causes growth slowdown and a rise in public debt.

### IX. Conclusions and policy implications
- Main empirical conclusions:
  - Excess private debt tends to spill over into the public balance sheet even absent a financial crisis.
  - When deleveraging coincides with a prolonged near-halt in growth, countries can end up with more total debt than before deleveraging ("growthless deleveraging").
- Interpretive perspective:
  - Government issuance of new public debt effectively provides an "implicit guarantee" to private liabilities — an indirect form of bailout at the macro level.
  - Moral hazard likely differs from conventional bank bailouts given the indirect, macro-scale nature of these interventions.
- Open questions for future work:
  - Which features of leverage/deleveraging cycles determine whether deleveraging is growthless or high-growth?
  - What is the optimal role of fiscal policy during deleveraging episodes?
  - Are indirect bailouts welfare-improving?
- Policy recommendation:
  - Adopt a more integrated approach to debt sustainability analysis: avoid siloed treatment of private and public debt and pay closer attention to the total stock of debt in the economy given the blurring line between private and public debt.

*Source: wp18141 — IMF Working Paper excerpt: "5. Impact of Private Deleveraging on Growth and Public Debt (IPWRA)".*

### References ________________________________________________________________38

### References ________________________________________________________________38

### ANNEXES
- 1. Data: Country Coverage ___________________________________________________27
- 2. Private Debt And Leverage Flows  ___________________________________________28
- 3. Deleveraging Episodes By Income Groups  ____________________________________32
- 4. Additional Figures And Tables ______________________________________________33

### FIGURES
- 1. Advanced Economies: Gross Debt ____________________________________________3
- 2. Spain: Gross Debt _________________________________________________________4
- 3. Private Debt in the Global Debt Database  ______________________________________6
- 4. Number of Deleveraging Episode _____________________________________________9
- 5. Probability of Deleveraging by Decades  _______________________________________9
- 6. Incidence of Private Deleveraging, 1950–2016 __________________________________9
- 7. Macroeconomic Dynamics During Private Deleveraging _________________________13
- 8. Sources of Change in Public Debt During Private Deleveraging in Advanced Countries _13
- 9. "Growthless" vs "High Growth" Deleveraging" _________________________________15
- 10. "Post-GFC" vs. "Pre-GFC" Deleveraging Episodes" ____________________________17
- 11. Deleveraging Episodes ___________________________________________________17
- 12. Cumulative Impact of Private Deleveraging on GDP and Public Debt ______________24

### TABLES
- 1. Average Features of Private Deleveraging Episodes _____________________________10
- 2. Private Deleveraging Episodes: Crisis vs. Non-crisis_____________________________11
- 3. Average Macro-Economic Conditions during Normal Times ______________________12
- 4. Determinants of Private Deleveraging ________________________________________23

*Source: wp18141 - References ________________________________________________________________38*

### 5. Impact of Private Deleveraging on Growth and Public Debt (IPWRA) _______________25

### 5. Impact of Private Deleveraging on Growth and Public Debt (IPWRA)

### I. Introduction and main findings
- Public bailouts of financial institutions provoke strong public backlash; e.g., U.S. Treasury plans to purchase up to $700 billion worth of mortgage-backed securities (September 2008) spurred broad protests.
- The paper documents a widespread, less visible form of “bailout”: when the private sector deleverages after a debt overhang, governments systematically respond with countercyclical rises in government deficits and debt, effectively transferring private debt burdens to public balance sheets via weaker growth and fiscal responses.
- Key quantitative observations:
  - Advanced economies have experienced a slow private deleveraging since the GFC while government borrowing rose as early as 2007 and continued through 2014 before leveling off.
  - A decade after the onset of the GFC, advanced economies are "39 percent of GDP deeper in debt."
  - Spain: public debt has increased by "close to 59 percent of GDP since 2007" while only "4¾ percent of GDP stems from capital injections and called guarantees."
- Core mechanisms and implications:
  - The channel is primarily growth: private deleveraging weighs on economic activity, prompting cyclical deterioration in public finances and a countercyclical rise in public debt as governments borrow to support demand and cushion the slowdown.
  - The debt substitution (private → public) occurs irrespective of whether private deleveraging is associated with a financial crisis; it is a recurrent phenomenon across countries and time.
  - Whether total debt rises or falls through a deleveraging episode critically depends on the severity and duration of the growth slowdown:
    - Benign slowdowns → overall debt burden ultimately goes down.
    - Protracted slowdowns → countries are left with more debt than at the outset ("growthless deleveraging" episodes).
  - Much of the macro-dynamics—decrease in borrowing flows, growth slowdown, and rise in public debt—occur before a decline in private debt-to-GDP is visible.

### II. Contributions and conceptual framing
- Three main contributions relative to the literature:
  - Establishes a causal link between private deleveraging and public debt.
  - Shows that private debt spillovers during crises are part of a broader phenomenon that occurs whenever the private sector deleverages.
  - Leverages the Global Debt Database (GDD) to examine private deleveraging dynamics across a far larger cross-section of countries and a long historical span.
- Characterization of this phenomenon as a “form of bailout”:
  - Negative spillovers: over-borrowing by some imposes burdens on all via future taxes or reduced public services.
  - Indirect relief to indebted households/firms: government support for growth (demand creation, tax reliefs, employment support) eases deleveraging.
  - Some moral hazard exists, though arguably less acute than in standard bank bailouts.

### III. Data: Global Debt Database (GDD) and sample coverage
- The GDD provides private, public, and total debt for 190 countries going as far back as 1950.
- Coverage expansion and descriptive metrics:
  - Sample coverage expands from 45 countries accounting for "63 percent of global output in 1950" to "99 percent of the world’s GDP in 2016."
  - The GDD nearly doubles the country coverage of the largest existing private-debt datasets, with particular gains for emerging market economies ("56") and low-income countries ("65"), and additional advanced economies ("8").
  - For analysis in the paper, the authors restrict to countries with both public and private debt data, resulting in a sample of "158 countries."
- Preferred debt aggregates:
  - Preferred private debt measure includes loans and securities (both domestic and external).
  - Total debt metric used is the sum of nonfinancial private and public debt.

### IV. Definitions and measurement of private deleveraging
- Two complementary metrics of private deleveraging:
  1. Debt-ratio metric (standard): distance from peak to ensuing trough in the private debt-to-GDP ratio.
     - Peaks/troughs defined as local maxima/minima within a five-year period using Harding and Pagan (2002) algorithm, with a minimum deleveraging duration of two years.
  2. Leverage-flows metric (experimental): based on debt decompositions (Escolano 2010) separating macro-related changes from net borrowing/lending (leverage flows). Private deleveraging defined as periods of negative "leverage flows".
- Decomposition equation (as presented):
  - 푑푑푡 − 푑푑푡−1 = �(푟푟 − 푔푔)/(1 + 푔푔)� 푑푑푡−1 − 푝푝푡 + 푆푆푆푆푆푆
    - where 푔푔 = nominal GDP growth, 푟푟 = interest rate, 푝푝푡 = net borrowing/lending excluding interest payments, and 푆푆푆푆푆푆 = stock-flow adjustments.
  - “Leverage flows” = excess growth in nominal private debt relative to the debt service, expressed as a share of GDP; they are negative when new debt issuances are lower than what is needed to service the existing stock of debt.

### V. Stylized facts from 1950–2016
- Episodes and frequency:
  - Identified "535 private deleveraging episodes, involving 159 countries over 1950–2016."
  - Number of episodes per decade ranges from a low of "17 episodes in the 1950s" to a high of "114 in the 1990s."
  - Three private deleveraging waves by region/decade: "2010s for advanced economies (21 episodes over the 6 years covered)"; "2000s in emerging market economies"; "1990s in low-income countries."
  - Overall incidence: virtually every country in the sample undergoes deleveraging at some point, with wide cross-country heterogeneity.
- Summary statistics and cross-country patterns (from Table 1 and accompanying text):
  - Private deleveraging episodes become less frequent, longer, and larger in size as countries develop.
    - Advanced economies: deleveraging occurs at lower frequency ("27 percent of the time") and more sparsely (on average every "11 years").
    - EMEs and LIDCs: deleveraging more common (every "8 years"), slightly shorter (about "3½ years"); private debt peaks at significantly lower levels.
  - Mean characteristics (from Table 1; World / Advanced / Emerging / Low-income groups):
    - Length (years): World "3.6 (1.9)"; Advanced "3.9 (2.2)"; Emerging "3.4 (1.6)"; Low-income "3.5 (2.0)".
    - Size (drop from peak to trough, % of GDP): World "-11.6 (22.3)"; Advanced "-18.7 (37.7)"; Emerging "-14.5 (20.3)"; Low-income "-6.1 (9.4)".
    - Rate (Size/Length): World "-3.2 (4.8)"; Advanced "-4.3 (5.6)"; Emerging "-4.4 (5.4)"; Low-income "-1.7 (2.5)".
    - Distance between episodes (years): World "8.5 (7.6)"; Advanced "10.9 (9.4)"; Emerging "7.9 (6.9)"; Low-income "8.0 (6.9)".
    - Peak private debt (average, % of GDP): World "54.1 (60.9)"; Advanced "131.2 (75.7)"; Emerging "51.8 (41.4)"; Low-income "20.4 (20.6)".
    - Pre-deleveraging change in debt (average of 5-yr cumulative change, % of GDP): World "9.2 (17.6)"; Advanced "19.0 (27.6)"; Emerging "9.4 (16.1)"; Low-income "3.9 (6.8)".
    - Pre-deleveraging growth (average of 5-yr, percent): World "3.5 (3.9)"; Advanced "3.1 (2.5)"; Emerging "4.1 (3.7)"; Low-income "3.3 (4.5)".
    - Growth during deleveraging episodes (average, percent): World "3.8 (4.2)"; Advanced "3.5 (2.5)"; Emerging "4.0 (4.2)"; Low-income "3.8 (4.8)".
    - Number of episodes: World "535"; Advanced "111"; Emerging "184"; Low-income "240".
    - Probability of deleveraging (share of deleveraging obs.): World "27%"; Advanced "25%"; Emerging "29%"; Low-income "26%".
- Relationship with crises (from Table 2 and text):
  - Less than "15 percent of deleveraging episodes occur within a +/- 3-year window of a financial crisis."
  - Crisis vs non-crisis episodes (Table 2):
    - Length (years): Crisis "4.03" vs Non-crisis "3.51"; p-value "0.03".
    - Size (drop, % of GDP): Crisis "-19.21" vs Non-crisis "-9.34"; p-value "0.00".
    - Rate (size/length): Crisis "-5.26" vs Non-crisis "-2.67"; p-value "0.00".
    - Peak private debt (average, % of GDP): Crisis "89.67" vs Non-crisis "50.32"; p-value "0.00".
    - Pre-deleveraging change in debt (5-yr cum., % of GDP): Crisis "17.41" vs Non-crisis "7.66"; p-value "0.00".
    - Pre-deleveraging growth (5-yr avg, percent): Crisis "3.18" vs Non-crisis "3.59"; p-value "0.41".
    - Growth during deleveraging (avg, percent): Crisis "2.82" vs Non-crisis "4.03"; p-value "0.02".
    - Number of episodes: Crisis "77"; Non-crisis "456".
  - Interpretation:
    - Most deleveraging episodes do not coincide with financial crises.
    - Deleveraging episodes during crises are longer and larger, with peak private debt substantially higher (almost "40 percent of GDP higher") and growth during deleveraging more than "1 percentage" point lower in crisis episodes.

### VI. Empirical methodology (event-study framework)
- Event-study framework adapted from Gourinchas and Obstfeld (2012):
  - Regression specification:
    - 푦푖푡 = 훼푖 + Σ푘 훽푡+푘 T푖푘 + 휀푖푡
    - Where T푖푘 is a dummy equal to 1 when country i is k years away from the start of a deleveraging episode in year t.
    - Most estimates use 9 dummies (four years before to four years after the start).
  - Interpretation:
    - Each parameter 훽푡+푘 captures the average change in 푦 at time t+ k relative to observations outside the 9-year window (interpreted as “normal times”).
    - Parameters are directly comparable across variables because they are estimated relative to the same benchmark.
    - Cumulative impacts over windows {t−k ... t+ k} are computed by summing corresponding 훽 coefficients.
    - Additivity: for any decomposition y1 = y2 + y3, the corresponding 훽 coefficients satisfy 훽1 = 훽2 + 훽3.
- “Normal times” correspond to periods of leveraging as well as private debt declines that last only a year, representing "71–75 percent of observations" depending on country group.
- Table 3 (not reproduced fully in this excerpt) provides average values of variables during “normal times” to serve as benchmarks.

### VII. Interim empirical insights (preview)
- Growth slowdown tends to occur before the private debt ratio visibly declines, implying that deleveraging dynamics (decline in borrowing flows, growth slowdown, rise in public debt) start prior to observable drops in private debt-to-GDP.
- The debt-ratio and leverage-flow measures typically identify the same deleveraging periods, except in “abnormal times” (crises, recessions, high growth), where the measures can diverge—these divergences are central to the subsequent analysis.

*Source: IMF Working Paper — "5. Impact of Private Deleveraging on Growth and Public Debt (IPWRA)" (excerpt from wp18141).*

### 1. The Typical Private Deleveraging

### 1. The Typical Private Deleveraging

### Baseline macroeconomic dynamics around private deleveraging episodes
- Period 0 denotes the start of the private deleveraging episode based on the standard debt-to-GDP metric.
- Figure 7 panels a–d summarize estimated 훽푡+k parameters for:
  - annual change in private debt (and subcomponents),
  - real growth,
  - annual change in public debt,
  - cumulative change in total debt (sum of annual-change 훽푡+k parameters), with a starting total debt stock of 100 in t −5.
- Dotted lines denote 90 percent confidence intervals.
- Key empirical patterns:
  - The private debt ratio grows faster than “normal times” in the run up to the deleveraging episode and contracts afterwards.
  - Leverage flows grow at a progressively slower pace before deleveraging and turn negative one year before deleveraging starts, indicating net debt repayment once debt service costs are accounted for.
  - Private agents begin net repayment efforts before the debt-to-GDP ratio starts to decline—i.e., deleveraging in leverage-flow terms begins earlier than visible in the private debt ratio.
  - GDP growth switches from a slight expansion to a deep contraction relative to normal times, hitting a trough one year before the start of deleveraging—coincident with leverage flows turning negative.
  - The recession is short-lived: growth returns to normal about a year after the start of the deleveraging episode.
- Sample note: due to restrictions on private interest rate data necessary to calculate leverage flows, these results are based on a smaller sample of 283 deleveraging episodes.

### Drivers and timing of public debt increases
- Public debt tends to increase faster than in normal times during private deleveraging episodes.
- The surge in public debt is front-loaded:
  - Public debt starts to increase one year before the start of the deleveraging episode.
  - The largest increase in public debt coincides with leverage flows turning negative and growth reaching its trough.
  - Public debt increases continue one year into the deleveraging episode, then taper off as growth recovers and deleveraging takes hold.
- Decomposition (Figure 8, advanced-economy sample):
  - Changes in the public debt ratio are broken into cyclically adjusted primary balance (CAPB), “macro-related changes”, and residual stock-flow adjustments.
  - Most of the increase in public debt during private deleveraging spells is driven by cyclically adjusted primary deficits along with macro-related flows, suggesting a true countercyclical policy component.
  - Stock-flow adjustments do not appear to be a major driver of the public debt surge in this sample.

### Total debt behavior
- Total debt:
  - Rises in the run up to the deleveraging episode,
  - Stabilizes early in the deleveraging process,
  - Eventually declines as deleveraging proceeds.
- Overall narrative: private deleveraging begins before the decline in the debt-to-GDP ratio is visible; as private deleveraging takes hold, activity recovers, public-debt surge subsides, and total debt progressively declines to a lower level.

### Heterogeneity: “Growthless” versus “High growth” deleveraging
- Definition:
  - “Low growth” benchmarks (25th percentiles since the 1950s by income groups): 1.6 percent for advanced economies, 1.9 percent for emerging market economies, and 1.1 percent for low-income countries.
  - Deleveraging episodes are classified as “growthless” if average growth throughout the episode is lower than the above benchmarks; otherwise “high growth”.
- Contrast in outcomes (Figure 9):
  - “High growth” deleveraging: similar to average episodes—moderate government support and a short-lived slowdown in activity.
  - “Growthless” deleveraging:
    - Last two years longer than high-growth episodes,
    - Require larger private-sector efforts,
    - Entail a larger and continuous increase in public borrowing,
    - Result in more total debt at the end of the process.
  - Magnitudes:
    - Private debt declines cumulatively by 17 percent of GDP in growthless cases versus 15 percent in high-growth cases.
    - Cumulative leverage flows: -24 percent of GDP in growthless cases versus -11% in high-growth cases.
    - Public debt increases cumulatively by 17 percent of GDP faster than normal in growthless episodes versus 6 percent of GDP in high-growth deleveraging cases.
- Prevalence: growthless episodes account for about a quarter of the sample and occur across income levels.

### Role of fiscal policy during growthless deleveraging
- Experiment: within the growthless sample, split episodes into “strong” and “limited government support” based on the top and bottom half of cumulative changes in public debt over the two years leading up to deleveraging.
- Findings:
  - Episodes with the largest government support tend to have the weakest growth (the lowest-growth of the growthless group).
  - These strongly supported episodes involve larger private-sector deleveraging efforts and often result in higher total debt than before.
- Interpretation caution:
  - The correlation between larger public-debt increases and lower growth does not imply causation from fiscal expansion to poor growth; endogeneity of public debt to growth outcomes is likely, and the descriptive framework cannot fully resolve this.

### Subsample checks: Post-GFC, pre-GFC, and crisis-related episodes
- Pre- vs post-2007 (Figure 10):
  - Baseline patterns hold generally for pre-GFC episodes.
  - Post-GFC episodes often begin like typical episodes but quickly morph into growthless ones.
- Crisis-related classification (Figure 11):
  - “Crisis-related” deleveraging = starts within a centered 5-year window around a financial crisis.
  - Patterns are broadly similar with and without accompanying financial crises: growth slows, public debt rises, and total debt eventually declines after a temporary increase.

### Toward causality: IPWRA framework and identification
- Concern: private debt, public debt, and growth are mutually endogenous; observed correlations could reflect multiple causal directions.
- Objective: establish whether causality runs from private deleveraging to public debt and growth (supporting the “indirect bailout” interpretation).
- Method: Inverse Propensity Weighting Regression Adjustment (IPWRA) estimator (doubly robust: combines IPW and Regression Adjustment).
- Treatment and outcomes:
  - Treatment d_{i,t} = 1 denotes the start of a private deleveraging episode in country i at time t; operationalized as the first year that leverage flows turn negative.
  - Outcome Y_{i,t} stands for GDP or public debt; potential outcomes Y_{i,t+s}(d).
  - Outcome of interest ΔY_{i,t}^s(d) = Y_{i,t+s}(d) − Y_{i,t}: potential growth in GDP and public debt s periods after the start of the deleveraging spell.
- Causal parameter:
  - Individual causal effect: β_i^S = ΔY_{i,t}^S(1) − ΔY_{i,t}^S(0).
  - Average treatment effect β^S = E[ΔY(1)] − E[ΔY(0)] if treatment were randomly assigned.
- Endogeneity and identification:
  - Private deleveraging is not randomly assigned; selection bias exists.
  - Identification rests on the selection-on-observables assumption: conditional on observable country characteristics X_{i,t}, private deleveraging is as good as random:
    - ΔY_{i,t}^S(d) ⟂ d_{i,t} | X_{i,t}.
  - IPWRA addresses endogeneity under this assumption by modeling selection and outcomes, providing protection from misspecification in either component.

*Source: IMF Working Paper — “1. The Typical Private Deleveraging”, wp18141*

### 1. Inverse Propensity Weighing (IPW)

### 1. Inverse Propensity Weighing (IPW)

### IPW method and implementation
- IPW addresses endogeneity of selection by re-weighting observations to replicate random selection.
- First step: estimate the probability P[d_{i,t}=1 | X_{i,t}] (Equation (5)) of going through a deleveraging spell as a function of X_{i,t}.
- Weights are constructed following Hirano, Imbens, and Ridder (2003) as:
  - w_{i,t} = 1 / E[d_{i,t}] · [ p̂(d_{i,t}=1) / (1 - p̂(d_{i,t}=1)) ] (Equation (6) — notation preserved from source).
- Intuition: give greater weight to non-deleveraging countries that resemble deleveraging countries.
- Average treatment effect on treated units computed as reweighted difference in means:
  - β̂_S = E[ w_{i,t} · ΔY_{i,t}^S ] (Equation (7)).
- Identification requires overlap/common support:
  - 0 < pr(d_{i,t}=1 | X_{i,t}) < 1 (Equation (8)).

### Key technical assumption
- For any value of X_{i,t}, there must be a non-zero probability to observe both deleveraging and non-deleveraging observations.

---

### 2. Regression Adjustment (RA)

### RA method and implementation
- RA models outcomes (growth and public debt) rather than the treatment.
- Structural form maintained from source:
  - ΔY_{i,t}^S(d) = π_d^S X_{i,t} + ε_{i,t}^d (Equation (9)).
  - E[ ε_{i,t}^d | X_{i,t}, d_{i,t} ] = 0 (Equation (10)) — follows from selection-on-observables.
- Counterfactual outcomes obtained by estimating π_d^S separately in treatment and control groups (Kline (2011) approach).
- Average treatment effect:
  - β̂_S = E[ X_{i,t} π̂_1^s − X_{i,t} π̂_0^s ] (Equation (11)).
- Imbens and Wooldridge (2009) representation (Equation (12)) shows RA subtracts difference in means of X across treatment status weighted by estimated marginal impacts:
  - β̂_S = ΔȲ_S(1) − ΔȲ_S(0) − ( N_0/(N_0+N_1) · π̂_1^S + N_1/(N_0+N_1) · π̂_0^S )′ ( X̄_1 − X̄_0 ).

### Interpretation
- RA accounts for differences in country characteristics across treatment status via outcome regression.

---

### 3. IPWRA: Combining IPW and RA

### Estimator construction
- IPWRA simultaneously implements IPW and RA:
  - Step 1: estimate probability of deleveraging and compute weights w_{i,t} per (6).
  - Step 2: reweight the sample and estimate linear regressions of ΔY_{i,t}^S on X_{i,t} for both reweighted treatment and control groups.
  - Causal impact computed as:
    - β̂_S = E[ ( X_{i,t} π̂_1^s − X_{i,t} π̂_0^s ) · w_{i,t} ] (Equation (12) in source).

---

### B. Estimating the Causal Impact of Private Deleveraging

### 1. Model selection (covariates and procedure)
- Procedure:
  - Start from a large pool of tested determinants.
  - Iteratively drop non-significant determinants one at a time starting from longest lag.
  - Select specification with best predictive power based on Akaike and Bayesian Information Criteria (AIC & BIC).
- Growth model starting pool includes:
  - (i) lagged GDP per capita and lagged GDP growth; (ii) ICRG’s Law and Order index; (iii) trade openness (imports+exports as share of GDP); (iv) inflation; (v) government spending; (vi) stocks and 3-year cumulative growths in private and public debt as share of GDP.
- Public debt covariates include:
  - (i) lagged GDP growth and GDP per capita; (ii) elderly (65+) dependency ratio; (iii) effective interest rate on public debt; (iv) inflation; (v) political fractionalization index; (vi) quality of institutions; (vii) dummy for crisis in last 5 years; (viii) annual exchange rate depreciation; (x) lagged levels in private and public debts.
- Private deleveraging model covariates include:
  - (i) lagged GDP growth; (ii) lagged private and public debts as share of GDP; (iii) lagged 3-year cumulative growth in private debt; (iv) private interest rates; (v) GDP per capita; (vi) US Treasury 10-year bond yields (global liquidity); (vii) quality of institutions; (viii) crisis dummy (past 5 years); plus housing and stock prices for restricted 16-AE sample.

### Model performance and stability
- Estimated parameters are broadly stable and bear expected signs across specifications.
- Predictive accuracy for private deleveraging assessed via AUROC:
  - AUROC interpretation: 1 = perfect, 0.5 = coin toss. AUROC close to 0.7 is generally considered adequate.
  - Models consistently achieve AUROCs above 0.7 for most specifications.
- Including country and time fixed effects improves predictive power (especially for LICs where ROCs rise well above 0.7), but makes IPWRA GMM function unstable and causes convergence issues in about half of specifications.

---

### 2. Validation tests
- Test of selection-on-observables: check that IPWRA reweights past growth and public debt to be orthogonal to private deleveraging.
- Panel estimates (Annex 4) show:
  - Without IPWRA: private deleveraging significantly correlated with past growth and public debt (selection bias present).
  - With IPWRA correction: correlation disappears, including for 4th and 5th lags — indicating successful correction of selection bias.
- Visual overlap check (Figure A.4.2): distributions of estimated probabilities of deleveraging in treatment and control groups show significant overlap for full-sample specification.

---

### 3. Results (IPWRA estimates and heterogeneity)
- Cumulative impact of private deleveraging started in t on GDP and public debt over a five-year horizon (full sample):
  - Private deleveraging causes a slowdown in growth and a rise in public debt.
  - On average, onset of a private deleveraging spell induces a 7.2 percent of GDP rise in public debt over the following 3 years.
  - Effect declines from t+4 onward and becomes statistically insignificant subsequently.
  - Size is slightly larger than baseline event-study estimate in Section IV: 7.2 vs 5.3 percent of GDP.
- Heterogeneity across income groups (Table 5 summary):
  - Advanced economies: private deleveraging induces on average up to 9.6 percent of GDP higher public debt.
  - Emerging markets: impact peaks at 5.7 percent of GDP.
  - Low-income countries: estimated impact positive but fails to meet common significance standards.
  - Growth slowdown associated with deleveraging observed at all income levels, but deleveraging-induced rises in public borrowing (“indirect bailouts”) more prevalent at higher income levels.
- Robustness:
  - Replication excluding deleveraging episodes that start within a 5-year window around a financial crisis (per crisis/non-crisis typology): results virtually unchanged (private deleveraging still causes growth slowdown and rise in public debt).

---

### VI. Concluding remarks — key takeaways and policy implications

- Main empirical findings:
  - Excess private debt tends to spill over into the public-sector balance sheet even in the absence of a financial crisis.
  - When growth grinds close to a halt during deleveraging spells, countries tend to end up with more total debt than at the beginning of private deleveraging.
- Interpretive points:
  - An interpretation: governments’ issuance of new public debt on behalf of taxpayers provides an “implicit guarantee” to household and corporate liabilities.
  - Moral hazard may be less prevalent due to the indirect and macro-level nature of these bailouts; individual borrowers may not internalize the macro-level bailout probability.
- Open questions highlighted:
  - What characteristics of leverage/deleverage cycles condition whether growth is strong or weak during deleveraging?
  - What should be the role of fiscal policy in deleveraging episodes?
  - Are “indirect bailouts” optimal from a welfare perspective? (Not answered in this paper.)
- Policy recommendation:
  - Move toward a more integrated approach to debt sustainability: avoid analyzing private and public debts in silos and pay closer attention to the total stock of debt in the economy, given the blurring line between private and public debt.

*Source: wp18141 - 1. Inverse Propensity Weighing (IPW), IMF working paper (source content provided).*

### ANNEX 1. DATA: COUNTRY COVERAGE

### ANNEX 1. DATA: COUNTRY COVERAGE

### Advanced Economies
- Australia
- Austria
- Belgium
- Canada
- Cyprus
- Czech Republic
- Denmark
- Estonia
- Finland
- France
- Germany
- Greece
- Hong Kong SAR
- Iceland
- Ireland
- Israel
- Italy
- Japan
- Latvia
- Lithuania
- Luxembourg
- Malta
- Netherlands
- New Zealand
- Norway
- Portugal
- Republic of Korea
- Singapore
- Slovak Republic
- Slovenia
- Spain
- Sweden
- Switzerland
- United Kingdom
- United States

### Emerging Market Economies
- Albania
- Algeria
- Argentina
- Azerbaijan
- Bahamas
- Bahrain
- Botswana
- Brazil
- Bulgaria
- China, Mainland
- Colombia
- Costa Rica
- Croatia
- Czech Republic
- Dominican Republic
- Ecuador
- Egypt
- El Salvador
- Georgia
- Guatemala
- Hungary
- India
- Indonesia
- Iran
- Iraq
- Jamaica
- Jordan
- Kazakhstan
- Kuwait
- Lebanon
- Macedonia, FYR
- Malaysia
- Mauritius
- Mexico
- Mongolia
- Morocco
- Oman
- Pakistan
- Paraguay
- Peru
- Philippines
- Poland
- Qatar
- Romania
- Russian Federation
- Saudi Arabia
- Serbia
- South Africa
- Sri Lanka
- Thailand
- Trinidad & Tobago
- Turkey
- Ukraine
- United Arab Emirates
- Uruguay
- Venezuela
- Vietnam
- Vanuatu

### Low-Income Developing Countries
- Afghanistan
- Bangladesh
- Benin
- Bhutan
- Burkina Faso
- Burundi
- Cambodia
- Cameroon
- Cape Verde
- Central African Republic
- Chad
- Comoros
- Cote D'Ivoire
- Dem. Rep. of the Congo
- Djibouti
- Dominica
- Eritrea
- Ethiopia
- Gambia
- Ghana
- Grenada
- Guinea
- Guinea-Bissau
- Haiti
- Kenya
- Kyrgyz Republic
- Laos
- Lesotho
- Liberia
- Madagascar
- Malawi
- Maldives
- Mali
- Mauritania
- Micronesia
- Moldova
- Mozambique
- Myanmar
- Nepal
- Nicaragua
- Niger
- Nigeria
- Papua New Guinea
- Republic of Congo
- Rwanda
- Samoa
- Sao Tome and Principe
- Senegal
- Sierra Leone
- Solomon Islands
- South Sudan
- St. Lucia
- St. Vincent and Grenadines
- Tajikistan
- Tanzania
- Timor Leste
- Togo
- Tonga
- Uganda
- Vanuatu
- Yemen
- Zambia
- Zimbabwe

*Source: wp18141 - ANNEX 1. DATA: COUNTRY COVERAGE*

### References

### References

### Complete list

- Acemoglu, D., S. Johnson, and J.A. Robinson. 2001. “The colonial origins of comparative development: An empirical investigation,” American Economic Review 91 (5), pp. 1369–1401.  
- Acemoglu, D., S. Naidu, P. Restrepo, and J. A. Robinson. Forthcoming. “Democracy Does Cause Growth,” Journal of Political Economy.  
- Alesina, A. and G. Tabellini, 1990. “A Positive Theory of Fiscal Deficits and Government Debt,” Review of Economic Studies 57, pp. 403-414.  
- Alter A., A . X. Feng, and N. Valckx. 2018. “Understanding the Macro-Financial Effects of Household Debt: A Global Perspective,” IMF Working Paper 18/76, International Monetary Fund, Washington, DC.  
- Angrist, J. D., O. Jordà, and Guido Kuersteiner. 2013. “Semiparametric Estimates of Monetary Policy Effects: String Theory Revisited,” NBER Working Paper No. 19355, National Bureau of Economic Research, Cambridge, MA.  
- Auerbach A. and Y. Gorodnichenko. 2012. “Measuring the Output Responses to Fiscal Policy,” American Economic Journal: Economic Policy, 4(2), pp. 1–27.  
- Arcand, J. L., E. Berkes and U. Panizza. 2015. “Too much finance?” Journal of Economic Growth, vol. 20, pp 105‒148.  
- Barro, R. J. 1990. “Government Spending in a Simple Model of Endogeneous Growth” Journal of Political Economy, vol. 98(5), pp. 103–25.  
- Barro, R. J., and X. Sala-i-Martin. 1992. ‘‘Convergence,’’ Journal of Political Economy, pp. 223–251.  
- Batini, N., G. Melina, and S. Villa. 2016. “Fiscal Buffers, Private Debt and Stagnation: The Good, the Bad, and the Ugly.” IMF Working Paper 16/104, International Monetary Fund, Washington, DC.  
- Beck, T., H. Degryse, and C. Kneer. 2014. “Is More Finance Better? Disentangling Intermediation and Size Effects of Financial Systems,” Journal of Financial Stability, vol. 10, pp. 50-64.  
- Bernardini, M. and L. Forni. 2017. “Private and Public Debt: Are Emerging Markets at Risk,” IMF Working Paper 17/61, International Monetary Fund, Washington, DC.  
- Blanchard, O., and D. Leigh. 2013. “Growth Forecast Errors and Fiscal Multipliers,” American Economic Review Papers and Proceedings, 103(3), pp. 117–120.  
- Blanchard, O. J. and R. Perotti. 2002. “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output,” The Quarterly Journal of Economics, 117, pp. 1329–1368.  
- Bova, E., M. Ruiz-Arranz, F. Toscani, and H. E. Ture. 2016. “The Fiscal Costs of Contingent Liabilities: A New Dataset,” IMF Working Paper 16/14, International Monetary Fund, Washington, DC.  
- Bouis, R. 2015. “Household Deleveraging and Saving Rates: A Cross-Country Analysis”, Available at SSRN: https://ssrn.com/abstract=2394121.  
- Buttiglione, L., P. R. Lane, L. Reichlin, and V. Reinhart. 2014. Deleveraging? What Deleveraging? Geneva Reports on the World Economy 16. Geneva: International Center for Monetary and Banking Studies and Centre for Economic Policy Research.  
- Cecchetti, S.G., and E., Kharroubi. 2012. “Reassessing the Impact of Finance on Growth,” BIS Working Paper Series No. 381. Basel: Bank for International Settlements.  
- Cecchetti, S.G., and E., Kharroubi. 2015. “Why does financial sector growth crowd out real economic growth?” BIS Working Paper Series No. 490. Basel: Bank for International Settlements.  
- Cespedes, L.F., Chang, R., Velasco, V. 2004 "Balance Sheets and Exchange Rate Policy," American Economic Review, 94, pp. 1183–1193.  
- Chudik, A., K. Mohaddes, M. H. Pesaran, and M. Raissi. 2017. “Is there a Debt-threshold Effect on Output Growth?” Review of Economics and Statistics, 99(1), 135-150.  
- Chen, S., M. Kim, M. Otte, K. Wiseman, and A. Zdzienicka. 2015. “Private Sector Deleveraging and Growth Following Busts,” IMF Working Paper 15/35, International Monetary Fund, Washington, DC.  
- Claessens, S., D. Klingebiel, and L. Laeven. 2001. “Financial Restructuring in Banking and Corporate Sector Crises: What Policies to Pursue?” NBER Working Paper 8386, National Bureau of Economic Research, Cambridge, MA.  
- Claessens, S., A. M. Kose, and M. Terrones. 2011. “Financial Cycles: What? How? When?” IMF Working Paper 10/245, International Monetary Fund, Washington, DC.  
- Claessens, S., A. Kose, L. Laeven, and F. Valencia. 2013. "Understanding Financial Crises: Causes, Consequences, and Policy Responses," CEPR Discussion Papers 9310.  
- Crivelli, E., Gupta, S., C. M. Granados, and C. Correa-Caro. 2016. “Fragmented Politics and Public Debt,” IMF Working Paper 16/190, International Monetary Fund, Washington, DC.  
- Dembiermont, C., M. Drehmann, and S. Muksakunratana. 2013. “How Much Does the Private Sector Really Borrow? A New Database for Total Credit to the Private Non-Financial Sector.” BIS Quarterly Review (March): pp. 65–81.  
- Eggertsson, G and P Krugman. 2012. “Debt, deleveraging, and the liquidity trap: A Fisher Minsky-Koo approach,” The Quarterly Journal of Economics, pp 1469–1513.  
- Englund, P. 1999. “The Swedish Banking Crisis: Roots and Consequences,” Oxford Review of Economic Policy, Vol.15. Iss 3, pp. 80–97.  
- Eschenbach, F. and Schuknecht, L. 2004. “Budgetary Risks from Real Estate and Stock Markets,” Economic Policy, vol. 39, pp. 313–46.  
- Escolano, J. 2010. A Practical Guide to Public Debt Dynamics, Fiscal Sustainability, and Cyclical Adjustment of Budgetary Aggregates. Technical Notes and Manuals 10/02. Washington, DC: International Monetary Fund.  
- Fischer, S. 1983. “Inflation and growth” Working Paper 1235. National Bureau of Economic Research  
- Frankel, J. A. and D. Romer. 1999. "Does Trade Cause Growth?" The American Economic Review, (June), pp. 379-399.  
- Gourinchas, P. and M. Obstfeld. 2012. “Stories of the Twentieth Century for the Twenty-First,” American Economic Journal: Macroeconomics, Vol. 4(1), pp. 226–265.  
- Harding, D. and A. Pagan. 2002. “Dissecting the Cycle: A Methodological Investigation,” Journal of Monetary Economics (49), pp. 365–381.  
- Hasan, I., R. Horvath, and J. Mares. 2016. "What type of finance matters for growth? Bayesian Model Averaging Evidence," Policy Research Working Paper Series 7645, The World Bank  
- Hirano, K., G. W. Imbens, and G. Ridder. 2003 “Efficient Estimation of Average Treatment Effects Using the Estimated Propensity Score,” Econometrica, 71(4), pp. 1161– 1189.  
- Hoggarth, G., R. Reis, and V. Saporta. 2002. “Costs of Banking System Instability: Some Empirical Evidence,” Journal of Banking and Finance 26 (5), pp. 825–55.  
- Honohan, P., and D. Klingebiel. 2003. “The Fiscal Cost Implications of an Accommodating Approach to Banking Crises,” Journal of Banking and Finance 27 (8), pp. 1539–60.  
- Ilzetzki, E., E. G. Mendoza, and C. A. Végh. 2013. “How Big (Small?) Are Fiscal Multipliers?” Journal of Monetary Economics 60 (2): 239–254.  
- Imbens, G. and J. M. Wooldridge. 2009. “Recent Developments in the Econometrics of Program Evaluation,” Journal of Economic Literature 47, pp. 5-86.  
- Institute of International Finance (IIF). 2017. Global Debt Monitor (July), Washington, DC.  
- International Monetary Fund (IMF). 2016. “Debt—Use It Wisely.” Fiscal Monitor (October), Washington, DC.  
- Jensen, S. and Nielsen, S. 1996. “Population Ageing, Public Debt and Sustainable Fiscal Policy,” Fiscal Studies, 16(2), pp. 1–20.  
- Jordà Ò., M. Schularick, and A. M. Taylor. 2011. "Financial Crises, Credit Booms, and External Imbalances: 140 Years of Lessons," IMF Economic Review, vol. 59(2), pp. 340-378.  
- ———. 2013. “When Credit Bites Back,” Journal of Money, Credit and Banking 45 (S2, December), pp. 3–28.  
- ———. 2015a. “Betting the House,” Journal of International Economics (96), pp. S2-S18.  
- ———. 2015b. "Leveraged bubbles," Journal of Monetary Economics, Elsevier, vol. 76(S), pages 1-20  
- ———. 2016a. “Sovereigns versus Banks: Credit, Crises, and Consequences,” Journal of the European Economic Association 14 (1), pp. 45–79.  
- ———. 2016b. “The Great Mortgaging: Housing Finance, Crises and Business Cycles,” Economic Policy, pp. 107–52.  
- ———. 2017. “Macrofinancial History and the New Business Cycle Facts.”    In NBER Macroeconomics Annual 2016, volume 31, edited by Martin Eichenbaum and Jonathan A. Parker. Chicago: University of Chicago Press.  
- Kaminsky G. L., Reinhart C. M., and C. Végh. 2005. "When It Rains, It Pours: Procyclical Capital Flows and Macroeconomic Policies," NBER Chapters in: NBER Macroeconomics Annual 2004, Volume 19, pp. 11-82 National Bureau of Economic Research.  
- Kneer, C. 2013. “Finance as a Magnet for the Best and Brightest: Implications for the Real Economy” DNB Working Paper 392  
- Koo, R. 2008. “The Holy Grail of Macroeconomics: Lessons from Japan’s Great Recession,” Wiley.  
- Laeven, L., and F. Valencia. 2013. "Systemic Banking Crises Database," IMF Economic Review, Vol. 61, pp. 225–270.  
- Mbaye, S., Moreno-Badia, M., and K. Chae. 2018a. “The Global Debt Database: Methodology and Sources.” IMF Working Paper (Forthcoming), International Monetary Fund, Washington, DC.  
- ———. 2018b. “Stories on Global Debt,” IMF Working Paper (Forthcoming), International Monetary Fund, Washington, DC.  
- Mian, A., K. Rao, and A. Sufi. 2009. “The Consequences of Mortgage Credit Expansion: Evidence from the U.S. Mortgage Default Crisis,” Quarterly Journal of Economics, 124(4), 1449-1496.  
- ———. 2013. "Household Balance Sheets, Consumption, and the Economic Slump," Quarterly Journal of Economics, 128, pp. 1687–1726.  
- Mian, A., and A. Sufi. 2010. “Household Leverage and the Recession of 2007 to 2009,” IMF Economic Review, 58(1), 74-117.  
- ———. “House Prices, Home Equity-Based Borrowing, and the U.S. Household Leverage Crisis,” American Economic Review, 101: 2132-2156.  
- ———. 2014a. House of Debt, University of Chicago Press  
- ———. 2014b. "What Explains the 2007-2009 Drop in Employment," Econometrica 82, pp. 2197–2223  
- Mian, A., Sufi, A., and E. Verner. 2017. “Household Debt and Business Cycles Worldwide,” Quarterly Journal of Economics, 132(4), pp. 1755–1817.  
- Pagano, M. 2012. "Finance: Economic Lifeblood or Toxin?" CSEF Working Papers 326, Centre for Studies in Economics and Finance (CSEF), University of Naples, Italy  
- Pescatori, A., D. Sandri, and J. Simon. 2014. “Debt and Growth: Is There a Magic Threshold?” IMF Working Paper 14/34, International Monetary Fund, Washington, DC.  
- Reinhart, C. M., and K. S. Rogoff. 2009a. “The Aftermath of Financial Crises,” American Economic Review, Vol. 99, pp. 466–72.  
- ———.2009b. This Time is Different: Eight Centuries of Financial Folly, Princeton Press.  
- ———. 2011. “From Financial Crash to Debt Crisis,” American Economic Review, Vol. 101(5), 1676-1706.  
- ———. 2013. “Banking Crises: An Equal Opportunity Menace,” Journal of Banking and Finance vol. 37, vol. 11, pp. 4557–73.  
- Reinhart, C., and B. Sbrancia. 2015. “Debt Liquidation,” Economic Policy (April). pp. 291–333.  
- Schularick M., and A. M. Taylor. 2012. “Credit Booms Gone Bust: Monetary Policy, Leverage Cycles, and Financial Crises, 1870–2008,” American Economic Review 102(2), pp. 1029–1061.  
- Suarez-Serrato, J.C., and P. Wingender. 2016. “Estimating Local Fiscal Multipliers.” NBER Working Paper No. 22425, National Bureau of Economic Research, Cambridge, MA.  
- Talvi, E. and C. Vegh. 2005. “Tax Base Variability and Procyclicality of Fiscal Policy,” Journal of Development Economics 78, no. 1, 156-190.  
- Wooldridge, J. 2007. “Inverse Probability Weighted M-Estimators for General Missing Data Problems,” Journal of Econometrics 141 (2), pp.1281-1301.  
- Woodward, D.B. 1947. "Public Debt and Institutions," American Economic Review (May), pp. 157–83.

*Content unit: wp18141 - References*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18141.pdf_