## 5. Sectoral Debt Cycles and Vulnerability to Dollar Cycles_2SLS Estimation

## Source details

**Canonical URL:** [5. Sectoral Debt Cycles and Vulnerability to Dollar Cycles_2SLS Estimation](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024030-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024030-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024030-print-pdf.pdf.json)

---

### Purpose and approach
- Objective: Assess how sectoral debt cycles (household and nonfinancial corporate debt) relate to medium-run GDP growth in FMEs and EMEs and how they shape vulnerability to US dollar (“dollar”) cycles.
- Key methods:
  - Regressions of 3-year GDP per capita growth on 3-year sectoral debt growth (equation (1)) and on debt levels (equation (2)).
  - Interaction regressions with US dollar log-differences over 3 years (Δ3 USD) (equation (3)).
  - Two-stage least squares–style procedure using debt levels to predict future (t→t+3) debt growth and then using projected debt growth in GDP regressions (equations (4)–(5)).
  - Local projections (Jordà, 2005) to trace dynamic impulse responses (equation (6)), and extensions to realized debt growth, consumption, and investment (equations (7)–(8)).

### Data and sample (highlights)
- Annual data: 2005 to 2021.
- Sample: 52 FMEs and EMEs divided into three groups by 2011 GDP per capita (2010 international dollars):
  - Group 1 (<7,000): 18 economies.
  - Group 2 (7,000–15,000): 17 economies.
  - Group 3 (15,000–40,000): 17 economies.
- Selected summary statistics (Full sample unless noted):
  - GDP per capita growth (3-year avg): Mean 2.11; Std.Dev. 2.75; Min -9.91; Max 12.17.
  - Household Debt (% of GDP): Mean 19.45; Std.Dev. 17.60; Min 0.00; Max 105.79.
  - Nonfinancial Corp. Debt (% of GDP): Mean 33.86; Std.Dev. 28.51; Min 0.00; Max 160.28.
  - Government Debt (% of GDP): Mean 50.05; Std.Dev. 28.21; Min 3.20; Max 319.09.
  - Household Debt Growth (3-year change): Mean 1.52; Std.Dev. 3.87; Min -14.14; Max 21.53.
  - Nonfin. Corp. Debt Growth (3-year): Mean 2.72; Std.Dev. 8.18; Min -23.72; Max 73.37.
  - Government Debt Growth (3-year): Mean 6.10; Std.Dev. 16.04; Min -56.56; Max 185.48.
- Group-specific household debt-to-GDP means:
  - Group 1: 8.00
  - Group 2: 23.96
  - Group 3: 28.26
- Countries in each group (Annex A):
  - Group 1: Benin; Cameroon; Congo; Cote d’Ivoire; Ethiopia; Ghana; India; Kenya; Lao P.D.R.; Morocco; Mozambique; Nigeria; Pakistan; Rwanda; Senegal; Tajikistan; Tanzania; Zambia.
  - Group 2: Angola; Brazil; China; Dominica; Ecuador; El Salvador; Indonesia; Jordan; Jamaica; Mongolia; Serbia; South Africa; Sri Lanka; Thailand; Tunisia; Ukraine.
  - Group 3: Argentina; Chile; Costa Rica; Croatia; Czech Republic; Hungary; Israel; Kazakhstan; Korea; Malaysia; Maldives; Mexico; Oman; Poland; Romania; Russia; Turkiye.

### Main findings — Sectoral debt cycles and GDP growth (Tables 2 and 3)
- Debt growth regressions (dependent = GDP per capita growth t→t+3):
  - Full sample: Household debt growth coefficient = -0.056*** (Driscoll-Kraay SE 0.011). Nonfin. corp. debt growth = -0.015*** (0.004).
  - Group 3 (highest income): Household debt growth = -0.165*** (0.021); Nonfin. corp. debt growth = -0.018* (0.010).
  - Group heterogeneity: Household debt growth’s negative association with future GDP strongest in Group 3; Group 1 and Group 2 show different and less clear household dominance.
- Debt level regressions (dependent = GDP per capita growth t→t+3):
  - Full sample: Household debt level = -0.013* (0.007); Nonfin. corp. debt level = -0.024*** (0.004).
  - Group 1: Household debt level = -0.139*** (0.042).
  - Group 2: Nonfin. corp. debt level = -0.053*** (0.014).
  - Group 3: Household debt level = -0.068*** (0.011); Nonfin. corp. debt level = -0.023*** (0.005).
- Interpretation: Household debt dominance in slowing future growth resembles Mian et al. (2017) but holds primarily for relatively rich EMEs (Group 3). For lower-income FMEs/EMEs, nonfinancial corporate debt can be more influential.

### Key finding — Debt and vulnerability to US dollar cycles (Table 4)
- Specification: Interaction of debt level × dollar shock (Δ3 USD between t and t+3).
- Notable coefficients:
  - Household debt level main effect (Group 1): -0.148*** (0.041).
  - Household debt level · dollar shock (Group 1): 0.127** (0.050).
  - Household debt level · dollar shock (Group 3): 0.245*** (0.056).
  - Nonfinancial corporate debt level · dollar shock (Group 2): 0.218*** (0.029).
- Empirical pattern: Debt level × dollar shock coefficients are positive for Groups 1 and 3 — higher debt levels are associated with weaker negative effects of US dollar appreciation on GDP growth, contrary to standard leverage-based vulnerability predictions.

### Two-stage (2SLS-style) estimation and projected debt growth (Table 5)
- Rationale: Debt levels may proxy slower future debt growth (deleveraging). Use debt level to predict future (t→t+3) debt growth and use fitted values (projected debt growth) in GDP regressions to distinguish mechanisms.
- Second-stage estimates (selected):
  - Projected household (HH) debt growth (Group 1): 0.183*** (0.050).
  - Projected HH debt growth (Group 3): 0.219*** (0.026).
  - Projected HH debt growth · dollar cycle (Group 1): -0.169** (0.074).
  - Projected HH debt growth · dollar cycle (Group 3): -0.656*** (0.079).
  - Projected NFC debt growth · dollar cycle (Group 2): -0.271*** (0.057).
- Interpretation: Faster projected household debt growth amplifies negative GDP responses to dollar appreciation (negative interaction). This reconciles the positive debt level × dollar coefficients in Table 4: high debt levels can appear insulating only because they predict slower projected debt growth; it is projected debt growth (leveraging) that increases fragility to dollar shocks.

### Dynamics from local projections (summarized results)
- Group 3 (relatively rich EMEs):
  - Given a dollar appreciation shock, each additional 1pp of projected household debt growth is associated with 1.5pp slower GDP growth in year t+1.
  - Cumulative effect peaks in year t+2 with 1.7pp slower GDP growth per 1pp of projected household debt growth.
- Nonfinancial corporate debt in Group 3:
  - Each 1pp higher expected NFC debt growth → 0.3pp slower GDP growth in year t+1 after a dollar appreciation shock; effect becomes insignificant by t+2.
- Group 2:
  - No significant role for expected household debt growth in transmission; expected NFC debt growth shows short-lived amplification in year t+1.

### Channels — realized debt growth, consumption, and investment (dynamics and magnitudes)
- Realized household debt growth:
  - Higher expected household debt growth significantly amplifies the negative effect of dollar appreciation on realized household debt growth.
  - Peak effect around h = 6 years: for every 1pp increase in expected debt growth, dollar shocks slowed future realized debt growth by an additional 3–4pp over 6 years (footnote 20).
- Consumption:
  - Higher expected household debt growth strengthens the negative transmission of dollar appreciation to consumption; statistically significant for the first 3 years following the shock; effect fades by year 4.
  - Pattern: households initially reduce consumption rather than immediately adjust debt.
- Investment:
  - Effects on investment are more persistent; amplification by expected household debt growth peaks later (4 to 6 years after the shock).
  - Interpretation: initial consumption cuts are followed by later deleveraging and reduced real-estate and related investments, with longer-lived investment declines.

### Robustness checks
- Controlling for foreign currency external debt (excluding direct investment debt and, in baseline, local-currency external debt) and adjusting Group 3 composition:
  - Local projection responses remain qualitatively and quantitatively similar to baseline (Figure 7). Household debt results slightly weaker; nonfinancial corporate debt results slightly stronger.
  - Two-stage estimations with foreign currency debt control (unreported) broadly confirm baseline findings.

### Main implications and policy takeaways
- Debt growth (projected leveraging/deleveraging) matters as much as, or more than, debt levels for economic vulnerability to external dollar cycles in FMEs and EMEs.
- Higher debt levels can appear less fragile if they are associated with expected deleveraging; cross-country comparisons focused only on debt levels without context on debt cycle phase may be misleading.
- For relatively rich EMEs (Group 3), projected household debt growth amplifies the negative impact of dollar appreciation on GDP, consumption, investment, and realized debt growth — implying targeted monitoring of household credit growth is critical for macro-financial stability.
- Policy relevance: macroprudential frameworks and financial stability assessments should incorporate debt growth dynamics and their interaction with global dollar cycles, not only debt levels.

*Source: IMF Working Paper — “Sectoral Debt and Global Dollar Cycles in Developing Economies” — Section 5 and accompanying tables and figures.*

### References .............................................................................................................

### wpiea2024030-print-pdf - References

### Document sections
- References ......................................................................................................................................................... 31
- Annex A. Data ...................................................................................................................................................... 33
- Annex B. Additional Tables .............................................................................................................................. 34

### Figures
- 1.  Further Investigation of the Role of Sectoral Debt in the Transmission of Dollar Shocks ..................... 7
- 2.  Sectoral Debt Cycles and Vulnerability to USD Appreciation Shocks ...................................................... 8
- 3.  US dollar Index and Private Sectoral Debts in FMEs and EMEs ............................................................. 18
- 4.  Impulse Response Functions of GDP to Expected Debt Growth × Dollar Appreciation Shocks ......... 24
- 5.  Impulse Response Functions of Debt Growth to Expected Debt Growth × Dollar Shocks .................. 26
- 6.  Impulse Response Functions of Consumption and Investment to Expected Debt Growth × Dollar 
  Shocks .......................................................................................................................................................... 27
- 7.  Impulse Response Functions of GDP to Expected Debt Growth × Dollar Appreciation Shocks with 
  Control of Foreign Currency External Debt .............................................................................................. 29

### Tables
- 1.  Summary Statistics ...................................................................................................................................... 12
- 2.  Sectoral Debt Cycles and GDP Growth ...................................................................................................... 15
- 3.  Sectoral Debt Levels and GDP Growth 4 ................................................................................................... 16
- 4.  Sectoral Debt Levels and Vulnerability to Dollar Cycles  ......................................................................... 19

*Source: wpiea2024030-print-pdf - References*

### 5.  Sectoral Debt Cycles and Vulnerability to Dollar Cycles_2SLS Estimation ........................................... 

### 5.  Sectoral Debt Cycles and Vulnerability to Dollar Cycles_2SLS Estimation

### Purpose and approach
- Objective: Assess how sectoral debt cycles (household and nonfinancial corporate debt) relate to medium-run GDP growth in FMEs and EMEs and how they shape vulnerability to US dollar (“dollar”) cycles.
- Key methods:
  - Regressions of 3-year GDP per capita growth on 3-year sectoral debt growth (equation (1)) and on debt levels (equation (2)).
  - Interaction regressions with US dollar log-differences over 3 years (Δ3 USD) (equation (3)).
  - Two-stage least squares–style procedure using debt levels to predict future (t→t+3) debt growth and then using projected debt growth in GDP regressions (equations (4)–(5)).
  - Local projections (Jordà, 2005) to trace dynamic impulse responses (equation (6)), and extensions to realized debt growth, consumption, and investment (equations (7)–(8)).

### Data and sample (highlights from Table 1 and grouping)
- Annual data from 2005 to 2021.
- Sample construction:
  - 52 FMEs and EMEs divided into three groups by 2011 GDP per capita (2010 international dollars): Group 1 (<7,000), Group 2 (7,000–15,000), Group 3 (15,000–40,000).
  - Group sizes: Group 1 = 18 economies, Group 2 = 17 economies, Group 3 = 17 economies.
- Selected summary statistics (Full sample unless otherwise noted):
  - GDP per capita growth (3-year avg): Mean 2.11, Std.Dev. 2.75, Min -9.91, Max 12.17.
  - Household Debt (% of GDP): Mean 19.45, Std.Dev. 17.60, Min 0.00, Max 105.79.
  - Nonfinancial Corp. Debt (% of GDP): Mean 33.86, Std.Dev. 28.51, Min 0.00, Max 160.28.
  - Government Debt (% of GDP): Mean 50.05, Std.Dev. 28.21, Min 3.20, Max 319.09.
  - Household Debt Growth (3-year change): Mean 1.52, Std.Dev. 3.87, Min -14.14, Max 21.53.
  - Nonfin. Corp. Debt Growth (3-year): Mean 2.72, Std.Dev. 8.18, Min -23.72, Max 73.37.
  - Government Debt Growth (3-year): Mean 6.10, Std.Dev. 16.04, Min -56.56, Max 185.48.
- Group-specific household debt-to-GDP means:
  - Group 1: 8.00
  - Group 2: 23.96
  - Group 3: 28.26

### Main findings — Sectoral debt cycles and GDP growth (Tables 2 and 3)
- Debt growth regressions (Table 2, dependent = GDP per capita growth from t to t+3):
  - Full sample: Household debt growth coefficient = -0.056*** (Driscoll-Kraay SE 0.011). Nonfin. corp. debt growth = -0.015*** (0.004).
  - Group 3 (highest income): Household debt growth = -0.165*** (0.021) (column (4)); nonfin. corp. debt growth = -0.018* (0.010).
  - Group heterogeneity: Dominance of household debt growth in predicting lower future GDP is strongest in Group 3; Group 1 and Group 2 show different patterns and less clear household dominance.
- Debt level regressions (Table 3, dependent = GDP per capita growth t→t+3):
  - Full sample: Household debt level = -0.013* (0.007); Nonfin. corp. debt level = -0.024*** (0.004).
  - Group 1: Household debt level = -0.139*** (0.042).
  - Group 2: Nonfin. corp. debt level = -0.053*** (0.014).
  - Group 3: Household debt level = -0.068*** (0.011); Nonfin. corp. debt level = -0.023*** (0.005).
- Interpretation: Household debt dominance in slowing future growth resembles Mian et al. (2017) but holds primarily for relatively rich EMEs (Group 3). For lower-income FMEs/EMEs, nonfinancial corporate debt can be more influential.

### Key finding — Debt and vulnerability to US dollar cycles (Table 4)
- Specification includes interaction: Debt level × dollar shock (Δ3 USD between t and t+3).
- Notable results (Table 4):
  - Household debt level main effect (Group 1): -0.148*** (0.041).
  - Household debt level · dollar shock (Group 1): 0.127** (0.050).
  - Household debt level · dollar shock (Group 3): 0.245*** (0.056).
  - Nonfinancial corporate debt level · dollar shock (Group 2): 0.218*** (0.029).
- Counterintuitive empirical pattern: Coefficients on debt level × dollar shock are positive for Groups 1 and 3—i.e., higher debt levels are associated with weaker negative effects of US dollar appreciation on GDP growth, contrary to standard leverage-based vulnerability predictions.

### Two-stage (2SLS-style) estimation and projected debt growth (Table 5)
- Rationale: Current debt levels may proxy slower future debt growth (deleveraging); use debt level to predict future (t→t+3) debt growth and use fitted values (projected debt growth) in GDP regressions to separate mechanisms.
- Second-stage estimates (Table 5):
  - Projected household (HH) debt growth (Group 1): 0.183*** (0.050).
  - Projected HH debt growth (Group 3): 0.219*** (0.026).
  - Projected HH debt growth · dollar cycle (Group 1): -0.169** (0.074).
  - Projected HH debt growth · dollar cycle (Group 3): -0.656*** (0.079).
  - Projected NFC debt growth · dollar cycle (Group 2): -0.271*** (0.057).
  - Interpretation: Faster projected household debt growth amplifies negative GDP responses to dollar appreciation (negative interaction). This reconciles Table 4: high debt levels can appear insulating only because they predict slower projected debt growth; it is projected debt growth (leveraging) that increases fragility to dollar shocks.

### Dynamics from local projections (summarized results)
- Group 3 (relatively rich EMEs) — expected household debt growth amplifies and prolongs negative GDP responses to dollar appreciation shocks:
  - Given a dollar appreciation shock, each additional 1pp of projected household debt growth is associated with 1.5pp slower GDP growth in year t+1; cumulative effect peaks in year t+2 with 1.7pp slower GDP growth per 1pp of projected household debt growth.
- Nonfinancial corporate debt in Group 3: each 1pp higher expected NFC debt growth → 0.3pp slower GDP growth in year t+1 after a dollar appreciation shock; effect becomes insignificant by t+2.
- Group 2: No significant role for expected household debt growth in transmission; expected NFC debt growth shows short-lived amplification in year t+1.

### Channels — realized debt growth, consumption, and investment (dynamics and magnitudes)
- Realized household debt growth:
  - Higher expected household debt growth significantly amplifies the negative effect of dollar appreciation on realized household debt growth.
  - Peak effect occurs around h = 6 years: for every 1pp increase in expected debt growth, dollar shocks slowed future realized debt growth by an additional 3–4pp over 6 years (footnote 20).
- Consumption:
  - Higher expected household debt growth strengthens the negative transmission of dollar appreciation to consumption; statistically significant for the first 3 years following the shock; effect fades by year 4.
  - Pattern: households initially reduce consumption rather than immediately adjust debt.
- Investment:
  - Effects on investment are more persistent; amplification by expected household debt growth peaks later (4 to 6 years after the shock).
  - Interpretation: initial consumption cuts are followed by later deleveraging and reduced real-estate and related investments, with longer-lived investment declines.

### Robustness checks
- Controlling for foreign currency external debt (excluding direct investment debt and, in baseline, local-currency external debt) and adjusting Group 3 composition:
  - Local projection responses remain qualitatively and quantitatively similar to baseline (Figure 7). Household debt results slightly weaker; nonfinancial corporate debt results slightly stronger.
  - Two-stage estimations with foreign currency debt control (unreported) broadly confirm baseline findings.

### Main implications and policy takeaways
- Debt growth (projected leveraging/deleveraging) matters as much as, or more than, debt levels for economic vulnerability to external dollar cycles in FMEs and EMEs.
- Higher debt levels can appear less fragile if they are associated with expected deleveraging; therefore, cross-country comparisons focused only on debt levels without context on debt cycle phase may be misleading.
- For relatively rich EMEs (Group 3), projected household debt growth amplifies the negative impact of dollar appreciation on GDP, consumption, investment, and realized debt growth — suggesting targeted monitoring of household credit growth is critical for macro-financial stability.
- Policy relevance: macroprudential frameworks and financial stability assessments should incorporate debt growth dynamics and their interaction with global dollar cycles, not only debt levels.

*Source: IMF Working Paper — “Sectoral Debt and Global Dollar Cycles in Developing Economies” — Section 5 and accompanying tables and figures.*

### References

### wpiea2024030-print-pdf - References

### References
- Arslanalp, S. and Tsuda, T. (2014). “Tracking Global Demand for Emerging Market Sovereign Debt,” IMF Working Paper, WP/14/39.
- Bahadir, B. and Gumus, I. (2016). Credit decomposition and business cycles in emerging market economies. Journal of International Economics, 103:250–262.
- Benetrix, A. S., D. Gautam, L. Juvenal, and M. Schmitz. (2020). “Cross-border currency exposures: new evidence based on an enhanced and updated dataset,” ECB Working Paper, No. 2417.
- Bergant, K., Grigoli, F., Hansen, N. J. H., & Sandri, D. (2020). Dampening global financial shocks: can macroprudential regulation help (more than capital controls)? IMF Working Papers, 2020 (106).
- Bertaut, C. C., Bruno, V., and Shin, H. S. (2021). Original sin redux. Available at SSRN 3820755.
- Bornhorst, F. and Arranz, M. R. (2014). Growth and the importance of sequencing debt reductions across sectors. Jobs and Growth: Supporting the European Recovery, page 13.
- Bruno, V. and Shin, H. S. (2015). Cross-border banking and global liquidity. The Review of Economic Studies, 82(2):535–564.
- Bruno, V. and Shin, H. S. (2020). Currency depreciation and emerging market corporate distress. Management Science, 66(5):1935–1961.
- Cavallino, P. and Hofmann, B. (2022). Capital flows and monetary policy trade-offs in emerging market economies. Available at SSRN.
- Cecchetti, S. G., Mohanty, M. S., and Zampolli, F. (2011). The real effects of debt. BIS Working Papers, No. 352
- Chen, P., Karabarbounis, L., and Neiman, B. (2017). The global rise of corporate saving. Journal of Monetary Economics, 89:1–19. Carnegie-Rochester-NYU Conference Series on the Macroeconomics of Liquidity in Capital Markets and the Corporate Sector.
- Devereux, M. B. and Wu, S. P. Y. (2022). Foreign reserves management and original sin. Working Paper 30418, National Bureau of Economic Research.
- Geanakoplos, J. (2010). The leverage cycle. NBER macroeconomics annual, 24(1):1–66.
- Han, B. (2022). Original sin dissipation and currency exposures in emerging markets. Available at SSRN 4066583.
- Han, B. (2023). Transmission of global financial shocks: Which capital flows matter? 77th issue (March 2023) of the International Journal of Central Banking, 19(1):55–109.
- Hofmann, B., Patel, N., and Wu, S. P. Y. (2022). Original sin redux: a model-based evaluation. Available at SSRN 4051502.
- Jiang, Z., Krishnamurthy, A., and Lustig, H. (2020). Dollar safety and the global financial cycle. (No. w27682). National Bureau of Economic Research.
- Jordà, Ò. (2005). Estimation and inference of impulse responses by local projections. American economic review, 95(1):161–182.
- Jordà, Ò., Kornejew, M., Schularick, M., and Taylor, A. M. (2022). Zombies at large? corporate debt overhang and the macroeconomy. The Review of Financial Studies, 35(10):4561–4586.
- Kalemli-Özcan, Ṣ. (2019). US monetary policy and international risk spillovers (No. w26297). National Bureau of Economic Research.
- Kekre, R. and Lenel, M. (2021). The flight to safety and international risk sharing. (No. w29238). National Bureau of Economic Research.  (No. w29238). National Bureau of Economic Research.
- Kiyotaki, N. and Moore, J. (1997). Credit cycles. Journal of Political Economy, 105(2):211–248.
- Lombardi, M. J., Mohanty, M. S., and Shim, I. (2017). The real effects of household debt in the short and long run. BIS Working Papers, No. 607.
- Ma, C. and Wei, S.-J. (2020). International equity and debt flows: Composition, crisis, and controls. (No. w27129). National Bureau of Economic Research.
- Mian, A., Sufi, A., and Verner, E. (2017). Household Debt and Business Cycles Worldwide. The Quarterly Journal of Economics, 132(4):1755–1817.
- Mian, A., Sufi, A., and Verner, E. (2020). How does credit supply expansion affect the real economy? the productive capacity and household demand channels. The Journal of Finance, 75(2):949–994.
- Miranda-Agrippino, S. and Rey, H. (2020). Us monetary policy and the global financial cycle. The Review of Economic Studies, 87(6):2754–2776.
- Obstfeld, M. and Zhou, H. (2023). The global dollar cycle. (No. w31004). National Bureau of Economic Research.
- Rey, H. (2013). Dilemma not trilemma: The global financial cycle and monetary policy independence. in global dimensions of unconventional monetary policy. Technical report, Federal Reserve Bank of Kansas City.

### Annex A. Data — List of the countries in each of the groups
- Group 1:
  - Benin
  - Cameroon
  - Congo
  - Cote d’Ivoire
  - Ethiopia
  - Ghana
  - India
  - Kenya
  - Lao P.D.R.
  - Morocco
  - Mozambique
  - Nigeria
  - Pakistan
  - Rwanda
  - Senegal
  - Tajikistan
  - Tanzania
  - Zambia
- Group 2:
  - Angola
  - Brazil
  - China
  - Dominica
  - Ecuador
  - El Salvador
  - Indonesia
  - Jordan
  - Jamaica
  - Mongolia
  - Serbia
  - South Africa
  - Sri Lanka
  - Thailand
  - Tunisia
  - Ukraine
- Group 3:
  - Argentina
  - Chile
  - Costa Rica
  - Croatia
  - Czech Republic
  - Hungary
  - Israel
  - Kazakhstan
  - Korea
  - Malaysia
  - Maldives
  - Mexico
  - Oman
  - Poland
  - Romania
  - Russia
  - Turkiye

*IMF Working Papers — Sectoral Debt and Global Dollar Cycles in Developing Economies (References and Annex A: Data).*

### Annex B. Additional Tables

### Annex B. Additional Tables

### Table A.1: 1st Stage Estimations in 2SLS Estimations
- Dependent variables: Household Debt Growth (columns (1)-(3)), Nonfin. Corp. Debt Growth (columns (4)-(6))
- Household debt level:
  - (1) -0.805*** (0.053)
  - (2) -0.338** (0.120)
  - (3) -0.364*** (0.092)
- Nonfin. corp. debt level:
  - (4) -0.688*** (0.075)
  - (5) -0.625*** (0.146)
  - (6) -0.630*** (0.083)
- GDP per-capita:
  - (1) 0.112 (0.353)
  - (2) 1.241*** (0.234)
  - (3) 0.410** (0.155)
  - (4) -4.240*** (0.798)
  - (5) 2.335 (1.606)
  - (6) 0.536** (0.241)
- Inflation:
  - (1) -0.058** (0.020)
  - (2) -0.020 (0.068)
  - (3) -0.114 (0.071)
  - (4) 0.021 (0.061)
  - (5) -0.367*** (0.100)
  - (6) 0.065 (0.132)
- Population growth:
  - (1) 28.487 (19.203)
  - (2) 19.965 (35.324)
  - (3) 23.856 (23.862)
  - (4) 38.278 (129.500)
  - (5) -12.918 (112.521)
  - (6) 38.136 (32.171)
- Institutional quality:
  - (1) 1.563*** (0.351)
  - (2) -1.840 (2.224)
  - (3) 1.066 (1.875)
  - (4) 7.738*** (1.444)
  - (5) 3.702* (1.923)
  - (6) -13.407** (4.685)
- Fiscal balance:
  - (1) 0.022 (0.022)
  - (2) 0.077** (0.029)
  - (3) -0.162 (0.125)
  - (4) -0.065 (0.063)
  - (5) 0.280 (0.187)
  - (6) -0.600** (0.258)
- Trade openness:
  - (1) 0.001 (0.007)
  - (2) -0.042* (0.023)
  - (3) -0.104*** (0.019)
  - (4) 0.032*** (0.007)
  - (5) -0.140* (0.071)
  - (6) -0.098 (0.056)
- Observations: 236, 235, 226, 236, 235, 226
- R-squared: 0.498, 0.188, 0.286, 0.417, 0.335, 0.465
- F-statistic: 302.1, 8.761, 47.198, 81.787, 97.892, 68.033
- Group: #1, #2, #3, #1, #2, #3
- Country FE: YES; Year FE: YES
- Notes: 1) The debt levels and all other variables are as in year t. The dependent variable is the debt growth between year t+3 and t. 2) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels. 3) Reported in brackets are Driscoll-Kraay standard errors.

### Table A.2: 1st Stage Estimations in Local Projection Estimations
- Dependent variables: Household Debt Growth (columns (1)-(2)), Nonfin. Corp. Debt Growth (columns (3)-(4))
- Household debt level:
  - (1) -0.353** (0.118)
  - (2) -0.443*** (0.105)
- Nonfin. corp. debt level:
  - (3) -0.651*** (0.131)
  - (4) -0.692*** (0.078)
- GDP per-capita:
  - (1) 0.889** (0.343)
  - (2) 0.145 (0.093)
  - (3) 2.229* (1.134)
  - (4) 0.158 (0.096)
- Inflation:
  - (1) -0.023 (0.063)
  - (2) -0.019 (0.069)
  - (3) -0.344*** (0.073)
  - (4) 0.109 (0.135)
- Population growth:
  - (1) 16.785 (35.980)
  - (2) 6.965 (25.295)
  - (3) -9.287 (104.521)
  - (4) 41.668 (38.293)
- Institutional quality:
  - (1) -1.250 (2.310)
  - (2) 1.148 (1.127)
  - (3) 3.975** (1.512)
  - (4) -13.306** (4.556)
- Fiscal balance:
  - (1) 0.124* (0.061)
  - (2) 0.002 (0.080)
  - (3) 0.242 (0.169)
  - (4) -0.374* (0.200)
- Trade openness:
  - (1) -0.032* (0.016)
  - (2) -0.114*** (0.018)
  - (3) -0.113* (0.056)
  - (4) -0.093* (0.048)
- Observations: 235, 226, 235, 226
- R-squared: 0.192, 0.327, 0.337, 0.531
- F-statistic: 15.231, 118.827, 90.800, 90.205
- Group: #2, #3, #2, #3
- Country FE: YES; Year FE: YES
- Notes: 1) The debt levels and all other variables are as in year t. The dependent variable is the debt growth between year t+3 and t. 2) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels. 3) Reported in brackets are Driscoll-Kraay standard errors.

### Table A.3: Cumulative Responses to Expected Debt Growth × Dollar Shocks — Household Debt — Group 3
- Dependent: Cumulative responses of GDP per-capita Year 1–Year 5
- Projected HH debt growth:
  - Year 1: 0.255*** (0.063)
  - Year 2: 0.370*** (0.099)
  - Year 3: 0.409*** (0.119)
  - Year 4: 0.110 (0.209)
  - Year 5: -0.138 (0.281)
- Projected HH debt growth · Dollar shock:
  - Year 1: -1.529*** (0.437)
  - Year 2: -1.769*** (0.486)
  - Year 3: -0.895 (0.603)
  - Year 4: -0.482 (0.694)
  - Year 5: -0.013 (0.560)
- GDP per-capita controls:
  - Year 1: -0.180 (0.114)
  - Year 2: -0.434** (0.167)
  - Year 3: -0.376 (0.255)
  - Year 4: -0.991*** (0.320)
  - Year 5: -1.562** (0.571)
- Inflation significant negatives across years:
  - Year 1: -0.197*** (0.052)
  - Year 2: -0.402*** (0.063)
  - Year 3: -0.218* (0.105)
  - Year 4: -0.469*** (0.115)
  - Year 5: -0.552*** (0.116)
- Fiscal balance positive coefficients in early years:
  - Year 1: 0.235*** (0.072)
  - Year 2: 0.263* (0.123)
  - Year 3: 0.401** (0.157)
  - Year 4: 0.115 (0.193)
  - Year 5: 0.462** (0.160)
- Trade openness:
  - Year 1: 0.034** (0.015)
  - Year 2: 0.062* (0.031)
  - Year 3: 0.097* (0.049)
  - Year 4: 0.089 (0.050)
  - Year 5: 0.069 (0.053)
- Dollar shock:
  - Year 1: 35.823* (16.759)
  - Year 2: 31.215 (45.632)
  - Year 3: 25.676 (27.029)
  - Year 4: -33.059 (34.019)
  - Year 5: -144.399*** (32.079)
- F.Dollar shock:
  - Year 1: -0.257 (5.589)
  - Year 2: -21.034 (18.136)
  - Year 3: -11.695** (4.539)
  - Year 4: 13.020* (6.457)
  - Year 5: 22.983*** (4.590)
- World GDP shock:
  - Year 3: 3.473*** (0.317)
  - Year 5: 3.804*** (0.671)
- Commodity price shock and interaction with exporter dummy:
  - Commodity price shock Year 1: 16.903*** (3.978)
  - Commodity price shock · Commodity exporter dummy Year 1: 19.149*** (2.954)
- International reserve and interaction with dollar shock:
  - International reserve Year 5: 0.381** (0.142)
  - International reserve · Dollar shock Year 5: 4.181*** (0.659)
- Observations: 207, 207, 207, 192, 177
- R-squared: 0.552, 0.479, 0.567, 0.399, 0.533
- Group: #3; Country FE: YES; Year FE: NO
- Notes: 1) Each column shows the cumulative response of GDP per capita in year t+h relative to year t for h = 1 - 5. 2) Projected HH debt growth is projected household debt growth between year t and year t+3 based on information in year t. 3) All other variables are as in year t. 4) Dollar shock is the log-difference between year t and t+1. 5) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels. 6) Reported in brackets are Driscoll-Kraay standard errors. 7) Fewer observation numbers than Table 5 are due to the drop of Maldives as the data of international reserves are missing for Maldives.

### Table A.4: Cumulative Responses to Expected Debt Growth × Dollar Shocks — Corporate Debt — Group 3
- Dependent: Cumulative responses of GDP per-capita Year 1–Year 5
- Projected NFC debt growth:
  - Year 1: 0.038 (0.028)
  - Year 2: 0.041 (0.037)
  - Year 3: 0.033 (0.040)
  - Year 4: -0.013 (0.044)
  - Year 5: -0.083** (0.029)
- Projected NFC debt growth · Dollar shock:
  - Year 1: -0.345*** (0.067)
  - Year 2: -0.225 (0.205)
  - Year 3: -0.070 (0.231)
  - Year 4: 0.009 (0.210)
  - Year 5: 0.029 (0.159)
- GDP per-capita:
  - Year 1: -0.230** (0.106)
  - Year 2: -0.518*** (0.141)
  - Year 3: -0.480* (0.254)
  - Year 4: -1.041*** (0.292)
  - Year 5: -1.575*** (0.489)
- Inflation:
  - Year 1: -0.216*** (0.056)
  - Year 2: -0.428*** (0.071)
  - Year 3: -0.244* (0.116)
  - Year 4: -0.477*** (0.126)
  - Year 5: -0.534*** (0.105)
- Trade openness:
  - Year 1: 0.009 (0.018)
  - Year 2: 0.027 (0.031)
  - Year 3: 0.061 (0.050)
  - Year 4: 0.075* (0.040)
  - Year 5: 0.071* (0.034)
- Fiscal balance:
  - Year 1: 0.237*** (0.072)
  - Year 2: 0.270* (0.138)
  - Year 3: 0.413** (0.171)
  - Year 4: 0.135 (0.195)
  - Year 5: 0.526** (0.188)
- Dollar shock:
  - Year 1: 39.071** (15.628)
  - Year 2: 38.959 (43.309)
  - Year 3: 34.227 (28.306)
  - Year 4: -30.030 (34.178)
  - Year 5: -148.023*** (30.312)
- F.Dollar shock:
  - Year 1: -1.866 (6.045)
  - Year 2: -23.523 (18.834)
  - Year 3: -14.338** (4.835)
  - Year 4: 11.941* (6.487)
  - Year 5: 23.171*** (5.341)
- World GDP shock:
  - Year 3: 3.291*** (0.259)
  - Year 5: 3.932*** (0.557)
- Commodity price shock and interaction:
  - Commodity price shock Year 1: 17.077*** (3.861)
  - Commodity price shock · Commodity exporter dummy Year 1: 19.748*** (2.980)
- International reserve and interaction with dollar shock:
  - International reserve Year 5: 0.366** (0.164)
  - International reserve · Dollar shock Year 5: 4.258*** (0.524)
- Observations: 207, 207, 207, 192, 177
- R-squared: 0.527, 0.458, 0.552, 0.397, 0.539
- Group: #3; Country FE: YES; Year FE: NO
- Notes: same note structure as Table A.3; fewer observations due to Maldives.

### Table A.5: Cumulative Responses to Expected Debt Growth × Dollar Shocks — Household Debt — Group 2
- Dependent: Cumulative responses of GDP per-capita Year 1–Year 5
- Projected HH debt growth:
  - Year 1: -0.030 (0.127)
  - Year 2: -0.362 (0.221)
  - Year 3: -0.644** (0.279)
  - Year 4: -1.069*** (0.249)
  - Year 5: -0.981*** (0.233)
- Projected HH debt growth · Dollar shock:
  - Year 1: -1.021 (0.744)
  - Year 2: -1.263 (1.478)
  - Year 3: -1.243 (1.728)
  - Year 4: -1.827 (1.589)
  - Year 5: -1.801 (2.104)
- GDP per-capita:
  - Year 1: -0.890*** (0.282)
  - Year 2: -1.797*** (0.411)
  - Year 3: -2.254*** (0.686)
  - Year 4: -3.503*** (0.726)
  - Year 5: -4.674*** (0.768)
- Trade openness strongly positive:
  - Year 1: 0.068*** (0.014)
  - Year 2: 0.105*** (0.020)
  - Year 3: 0.148*** (0.011)
  - Year 4: 0.181*** (0.043)
  - Year 5: 0.221*** (0.058)
- Fiscal balance positive and significant:
  - Year 1: 0.147** (0.067)
  - Year 2: 0.419*** (0.075)
  - Year 3: 0.579*** (0.088)
  - Year 4: 0.615*** (0.076)
  - Year 5: 0.510*** (0.140)
- Dollar shock:
  - Year 1: 13.595 (12.483)
  - Year 2: 14.172 (23.879)
  - Year 3: 15.937 (16.207)
  - Year 4: -4.362 (33.948)
  - Year 5: -84.936** (34.071)
- F.Dollar shock:
  - Year 1: 2.494 (6.072)
  - Year 2: -11.285 (11.897)
  - Year 3: -12.157** (4.507)
  - Year 4: 5.945 (5.906)
  - Year 5: 22.133** (7.385)
- World GDP shock:
  - Year 2: 1.267* (0.678)
  - Year 3: 3.127*** (0.213)
  - Year 5: 3.140*** (0.792)
- Commodity price shock:
  - Year 1: 7.158** (3.201)
  - Year 5: -20.055*** (5.707)
  - Commodity price shock · Commodity exporter dummy Year 1: 10.919*** (2.926)
- International reserve:
  - Year 1: 0.034 (0.037)
  - Year 4: -0.232* (0.120)
  - Year 5: -0.312** (0.112)
- Observations: 216, 216, 216, 199, 182
- R-squared: 0.445, 0.490, 0.617, 0.577, 0.642
- Group: #2; Country FE: YES; Year FE: NO
- Notes: fewer observations due to drop of Sri Lanka (missing international reserves).

### Table A.6: Cumulative Responses to Expected Debt Growth × Dollar Shocks — Corporate Debt — Group 2
- Dependent: Cumulative responses of GDP per-capita Year 1–Year 5
- Projected NFC debt growth:
  - Year 1: 0.087* (0.046)
  - Year 2: 0.116 (0.106)
  - Year 3: 0.211 (0.129)
  - Year 4: 0.136 (0.111)
  - Year 5: 0.227* (0.115)
- Projected NFC debt growth · Dollar shock:
  - Year 1: 0.371*** (0.081)
  - Year 2: 0.283 (0.266)
  - Year 3: 0.465* (0.226)
  - Year 4: 0.263 (0.161)
  - Year 5: -0.193 (0.298)
- GDP per-capita:
  - Year 1: -0.713** (0.261)
  - Year 2: -1.448*** (0.370)
  - Year 3: -1.645** (0.623)
  - Year 4: -2.701*** (0.770)
  - Year 5: -3.877*** (0.906)
- Trade openness and fiscal balance positive and significant across years
  - Trade openness Year 1: 0.072*** (0.017); Year 5: 0.262*** (0.047)
  - Fiscal balance Year 2: 0.366*** (0.108); Year 4: 0.501*** (0.086)
- Dollar shock:
  - Year 1: 6.953 (11.316)
  - Year 2: 3.856 (26.705)
  - Year 3: 2.287 (18.809)
  - Year 4: -24.056 (37.374)
  - Year 5: -99.185** (36.087)
- F.Dollar shock:
  - Year 1: 1.925 (6.141)
  - Year 2: -12.028 (11.871)
  - Year 3: -13.607** (5.093)
  - Year 4: 4.223 (5.847)
  - Year 5: 20.204*** (6.502)
- World GDP shock:
  - Year 2: 1.418* (0.689)
  - Year 3: 3.386*** (0.217)
  - Year 5: 3.424*** (0.745)
- Commodity price shock and interaction:
  - Commodity price shock Year 1: 6.934** (3.091)
  - Commodity price shock · Commodity exporter dummy Year 1: 11.060*** (2.685)
- International reserve and interaction with dollar shock:
  - International reserve Year 5: -0.308** (0.134)
  - International reserve · Dollar shock Year 1: 0.847*** (0.183)
- Observations: 216, 216, 216, 199, 182
- R-squared: 0.462, 0.490, 0.621, 0.547, 0.629
- Group: #2; Country FE: YES; Year FE: NO
- Notes: fewer observations due to drop of Sri Lanka.

### Table A.7: Cumulative Responses of Debt Growth to Expected Debt Growth × Dollar Shocks — Corporate Debt — Group 3
- Dependent: Cumulative responses Year 1–Year 8
- Projected HH debt growth:
  - Year 1: 0.217** (0.092)
  - Year 2: 0.583*** (0.156)
  - Year 3: 0.951*** (0.237)
  - Year 4: 1.337*** (0.349)
  - Year 5: 1.757*** (0.345)
  - Year 6: 2.171*** (0.164)
  - Year 7: 2.393*** (0.130)
  - Year 8: 2.430*** (0.181)
- Projected HH debt growth · Dollar shock:
  - Year 1: -0.602*** (0.182)
  - Year 2: -0.886** (0.315)
  - Year 3: -1.039* (0.489)
  - Year 4: -1.598** (0.725)
  - Year 5: -2.358*** (0.738)
  - Year 6: -3.602*** (1.128)
  - Year 7: -3.418** (1.108)
  - Year 8: -2.992 (1.630)
- GDP per-capita shows increasing positive coefficients in later years:
  - Year 5: 0.435* (0.208)
  - Year 6: 0.552*** (0.172)
  - Year 7: 0.651*** (0.143)
  - Year 8: 0.734*** (0.130)
- Trade openness increases over horizon, significant in later years:
  - Year 5: 0.146** (0.055)
  - Year 6: 0.234*** (0.024)
  - Year 7: 0.303*** (0.013)
  - Year 8: 0.343*** (0.019)
- Fiscal balance signs vary; notable negatives in years 5–6:
  - Year 5: -0.351*** (0.103)
  - Year 6: -0.331*** (0.085)
- Dollar shock coefficients vary over horizon; notable:
  - Year 1: -17.869** (7.045)
  - Year 5: 30.955 (18.372)
  - Year 8: -53.717 (43.452)
- F.Dollar shock series includes significant negatives and positives:
  - Year 5: -13.832** (4.551)
  - Year 6: -15.292*** (3.460)
  - Year 7: 8.954** (9.118)
- World GDP shock negative early, then mixed:
  - Year 1: -0.294** (0.123)
  - Year 3: -1.101*** (0.274)
- Commodity price shock and exporter interaction show mixed signs; some significant negatives in later years for exporter interaction (Years 4–7)
- International reserve generally negative and significant in later years:
  - Year 6: -0.179** (0.071)
  - Year 7: -0.168*** (0.040)
  - Year 8: -0.128** (0.049)
- Observations: 207, 207, 207, 192, 177, 163, 149, 135
- R-squared: 0.269, 0.330, 0.418, 0.487, 0.569, 0.643, 0.672, 0.689
- Group: #2; Country FE: YES; Year FE: NO
- Notes: 1) Each column shows the cumulative response of GDP per capita in year t+h relative to year t for h = 1 - 8. 2) Projected NFC debt growth is projected household debt growth between year t and year t+3 based on information in year t. 3) All other variables are as in year t. 4) Dollar shock is the log-difference between year t and t+1. 5) *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels. 6) Reported in brackets are Driscoll-Kraay standard errors. 7) Fewer observation numbers than Table 5 are due to the drop of Maldives as the data of international reserves are missing for Maldives.

### Table A.8: Cumulative Responses of Consumption Growth to Expected Household Growth × Dollar Shocks — Household Debt — Group 3
- Dependent: Cumulative responses of consumption (noted as investment growth in notes) Year 1–Year 5
- Projected HH debt growth:
  - Year 1: 0.323*** (0.065)
  - Year 2: 0.863*** (0.231)
  - Year 3: 1.132*** (0.334)
  - Year 4: 1.300*** (0.330)
  - Year 5: 1.417*** (0.297)
- Projected HH debt growth · Dollar shock:
  - Year 1: -1.736*** (0.436)
  - Year 2: -2.902*** (0.787)
  - Year 3: -2.413* (1.140)
  - Year 4: -1.692 (1.472)
  - Year 5: -1.942** (0.819)
- Inflation negative and significant across years:
  - Year 1: -0.259*** (0.047)
  - Year 2: -0.518*** (0.114)
  - Year 3: -0.385** (0.131)
  - Year 4: -0.592*** (0.158)
  - Year 5: -0.449*** (0.127)
- Trade openness positive and significant across years:
  - Year 1: 0.072*** (0.023)
  - Year 2: 0.213*** (0.051)
  - Year 3: 0.311*** (0.093)
  - Year 4: 0.427*** (0.105)
  - Year 5: 0.503*** (0.085)
- Investment growth lags and signs:
  - Inv. Growth Year 1: 0.133*** (0.012)
  - L.Inv. Growth Year 1: 0.049*** (0.014)
  - L2.Inv. Growth Year 1: 0.020** (0.009)
  - L.Inv. Growth Year 4: -0.145*** (0.044)
  - L2.Inv. Growth Year 4: -0.093*** (0.019)
- Dollar shock and F.Dollar shock:
  - Dollar shock Year 5: -95.301* (44.953)
  - F.Dollar shock Year 1: -7.704*** (1.960)
  - F.Dollar shock Year 2: -35.786** (13.772)
  - F.Dollar shock Year 3: -29.758*** (8.284)
  - F.Dollar shock Year 5: 16.721** (5.915)
- World GDP shock Year 2: 1.509* (0.802); Year 3: 3.886*** (0.654)
- Observations: 180, 180, 179, 164, 151
- R-squared: 0.674, 0.578, 0.609, 0.536, 0.623
- Group: #3; Country FE: YES; Year FE: NO
- Notes: 7) Fewer observation numbers than Table A_4 are due to the drop of Oman and Kazakhstan as the data of consumption are missing for EMEs.

### Table A.9: Cumulative Responses of Investment Growth to Expected Household Growth × Dollar Shocks — Household Debt — Group 3
- Dependent: Cumulative responses of investment growth Year 1–Year 5
- Projected HH debt growth:
  - Year 1: 0.776** (0.273)
  - Year 2: 2.000*** (0.463)
  - Year 3: 2.566*** (0.457)
  - Year 4: 2.396*** (0.486)
  - Year 5: 2.373*** (0.469)
- Projected HH debt growth · Dollar shock:
  - Year 1: -1.933* (1.046)
  - Year 2: -3.027 (2.218)
  - Year 3: -4.522** (2.049)
  - Year 4: -4.669** (1.706)
  - Year 5: -4.795** (1.664)
- Con. Growth` (consumption growth) strong positive in early years:
  - Year 1: 1.461*** (0.276)
  - Year 2: 1.562** (0.590)
  - Year 3: 0.535* (0.267)
- Lagged investment growth:
  - L.L.Inv. Growth Year 1: -0.593*** (0.146)
  - L.L.Inv. Growth Year 2: -0.892*** (0.279)
  - L.L.Inv. Growth Year 3: -0.713** (0.251)
  - L.L.Inv. Growth Year 4: -1.304*** (0.190)
  - L.L.Inv. Growth Year 5: -1.343*** (0.329)
- Dollar shock and F.Dollar shock:
  - Dollar shock Year 1: 125.535 (74.510)
  - Dollar shock Year 3: 130.282* (69.136)
  - Dollar shock Year 5: -360.840* (182.355)
  - F.Dollar shock Year 4: 23.316* (10.959)
  - F.Dollar shock Year 5: 40.756** (15.259)
- World GDP shock Year 3: 7.431*** (2.343); Year 5: 8.712*** (1.833)
- Commodity price shock notable positives in Year 1 and Year 3:
  - Year 1: 38.628*** (9.424)
  - Year 3: 42.349** (17.732)
- International reserve and interaction:
  - International reserve Year 5: 1.066** (0.459)
  - International reserve · Dollar shock Year 5: 9.109** (3.968)
- Observations: 180, 180, 179, 164, 151
- R-squared: 0.619, 0.488, 0.496, 0.401, 0.493
- Group: #2; Country FE: YES; Year FE: NO
- Notes: 7) Fewer observation numbers than Table A_4 are due to the drop of Oman and Kazakhstan as the data of investments are missing for EMEs.

*Annex B. Additional Tables, Working Paper No. WP/2024/30 — Sectoral Debt and Global Dollar Cycles in Developing Economies*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024030-print-pdf.pdf_