## wpiea2024093-print-pdf - Appendix Tables B.2 and B.3 present country-level estimates of the impact of monetary policy shocks on financial variables and subsequent sections on macroeconomic and firm-level transmission

## Source details

**Canonical URL:** [wpiea2024093-print-pdf - Appendix Tables B.2 and B.3 present country-level estimates of the impact of monetary policy shocks on financial variables and subsequent sections on macroeconomic and firm-level transmission](https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024093-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2024/english/wpiea2024093-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2024/english/wpiea2024093-print-pdf.pdf.json)

---

### Country-level financial impacts
- Estimates reported over 1- and 14-day windows.
- Whenever statistically significant, the impact of monetary policy shocks on government bond yields is positive.
- Considerable heterogeneity across countries in the strength of the impact, especially on the day of the monetary policy announcement.
- Country-level regressions confirm that monetary policy shocks tend to appreciate the exchange rate on impact.
- Examples from Appendix Tables B.2 and B.3 (contemporaneous and 2-week-ahead effects; coefficients, Obs., R2 reported per country-variable pair):
  - Brazil (contemporaneous): 1-year govt. yield coeff. 0.549***; Obs. 134; R2 0.12. 2-year govt. yield coeff. 0.427**; Obs. 134; R2 0.06. Exchange rate coeff. -1.014**; Obs. 207; R2 0.04.
  - India (contemporaneous): 1-year govt. yield coeff. 0.500**; Obs. 39; R2 0.16. 2-year coeff. 0.483***; Obs. 39; R2 0.19. 5-year coeff. 0.570***; Obs. 39; R2 0.21. Exchange rate coeff. -1.385**; Obs. 39; R2 0.10.
  - Poland (2 weeks after): 1-year govt. yield coeff. 1.056***; Obs. 206; R2 0.18. 2-year coeff. 1.065***; Obs. 206; R2 0.13. 5-year coeff. 0.780***; Obs. 206; R2 0.07.

### Macroeconomic empirical specification and identification
- Local-projection panel regression estimated:
  - Y_{c,t+h} − Y_{c,t−1} = α^h_c + β^h I_{c,t} + A^h(L) ΔY_{c,t−1} + B^h(L) P_{c,t−1} + τ^h_t + ε^h_{c,t}
- Outcome Y_{c,t} includes: log of industrial production, unemployment rate, log headline CPI or core CPI, log bilateral exchange rate vs US dollar, and one-year government bond yield in country c in month t.
- Horizons h ≥ 0 up to 3 years; local projections cover 36 months following shocks.
- ΔY_{c,t−1} uses yearly changes; A^h(L) and B^h(L) are matrix polynomials of degree 11 (allowing for 12 lags).
- Controls P_{c,t−1}: log of new COVID-19 cases, index of lockdown stringency, indicator for income support and debt relief measures (Oxford COVID-19 Government Response Tracker).
- Fixed effects: country α^h_c and time τ^h_t to purge global shocks.
- I_{c,t} captures exogenous variations in interest rates associated with monetary policy decisions; β^h measures causal effect.

### Data cleaning, transformations, and identification details
- Monetary policy shocks converted to monthly frequency by summing all shocks within each month.
- Financial variables (bond yields and exchange rate) use average values in the last three working days of each month.
- Light data cleaning: trim the 0.25 top and bottom percentiles.
- Smoothing: three-month moving averages applied to industrial production, unemployment, and CPI series.
- Autocorrelation addressed with rich lag structure and clustering standard errors by country.
- Nigeria excluded from regressions because it lacks monthly industrial production data.

### Main OLS local-projection results (shocks rescaled to a 1 percentage point bond-yield impact)
- Shocks rescaled to generate an increase in bond yields by 1 percentage point on impact.
- Key financial and macro responses:
  - Bond yields remain elevated for about 9 months.
  - Industrial production contracts by more than 2 percent; effect starts after a lag of a few months and peaks 8 months after the shock.
  - Unemployment rate increases by about 0.5 percentage points (gradual rise).
  - Producer prices decline starting about 6 months after the shock and cumulate to about a 6 percent reduction.
  - Core CPI follows producer prices but smaller: total cumulative decline of about 2 percent.
  - Headline CPI begins to decline about a year after the monetary tightening (longer lag).
  - Exchange rate tends to appreciate, building over time to a temporary appreciation of about 5 percent.
- No evidence of an exchange rate puzzle (i.e., puzzling depreciation) found in some prior studies.

### Instrumental-variable (IV) robustness and orthogonalization
- IV approach: 1-year sovereign bond yield instrumented with orthogonalized monetary policy shocks; first-stage F-statistics well above the reference critical value.
- IV impulse responses are remarkably similar to OLS results.
- Interpretation: in emerging markets monetary policy shocks tend to originate mostly around policy announcements; instrumentation is not essential to capture additional shocks.
- Orthogonalization improves precision:
  - Without orthogonalization, point estimates remain directionally consistent but are quantitatively smaller and confidence bands considerably wider.
  - Orthogonalization detects limited predictability of policy rate decisions beyond analysts’ forecasts; it is important to improve estimate precision.
- Sample restriction excluding non-inflation-targeting countries (Egypt and Malaysia) leaves point estimates largely unchanged; restricting to inflation-targeting countries reduces confidence bands around CPI impacts and shows faster and stronger unemployment effects.

### Comparison with recent advanced-economy (US) evidence
- Transmission magnitude and timing in emerging markets closely resemble recent US estimates (Bauer and Swanson (2023b)):
  - Peak contraction in industrial production: emerging markets 2.5 percent; US between 1.6 to 2.8 percent depending on specification.
  - Unemployment increase: about 0.5 percentage points in both the US and emerging markets.
  - CPI decline: US estimates show a maximum CPI decline between 0.8 and 4 percent; emerging-markets estimates are on the higher end of that range, suggesting steeper Phillips curves in emerging markets.

### Firm-level data and scope
- Firm-level analysis uses Refinitiv Datastream quarterly data for publicly listed firms.
- Sample includes 9,378 firms.
- Motivation: assess heterogeneity of monetary transmission across firms by financial characteristics, given likely more pervasive borrowing constraints in emerging markets.

### Firm-level local-projection methodology
- Firm-level local projections estimated for y_f,t+h − y_f,t−1 with:
  - Firm fixed effects α^h_f and sector-time fixed effects τ^h_s,t.
  - Controls: sum of monetary policy shocks I_c,t (quarter t in country c), 4-quarter lags of firm-level changes ∆y_f,t−1, and country-level controls X_c,t−1 (including ∆Y_c,t−1 and pandemic indicators P_c,t−1).
  - Sample: firms with at least 5 years of data; exclude public sector and finance, insurance and real estate sectors; drop observations where quarter-on-quarter growth rates of the dependent variable exceed 200 percent.
  - Standard errors clustered by firm and time.
  - Shock scale: contractionary monetary policy shock that generates a 1 percent increase in the one-year bond yield.
  - Responses reported over the 12 quarters following the shock; confidence intervals are 90 percent.

### Main firm-level findings (average responses to a 1pp monetary policy shock)
- Aggregate effects (average across firms):
  - Fixed-capital stock declines by about 3 percent.
  - Sales decline by about 3 percent.
  - Employment falls by about 2 percent.
- Timing and dynamics:
  - Investment exhibits a swift contraction following the shock.
  - Sales contract with a lag of about 3 quarters.
  - Employment contracts within the first year and again after three years.
- Interpretation:
  - Quick investment response consistent with higher borrowing costs and expected demand contraction.
  - Delayed sales response aligns with delayed effects on household consumption.
  - Employment dynamics broadly consistent with unemployment responses from macro estimates.

### Heterogeneity by firm financial characteristics
- Interaction specification includes firm characteristic F_f as leverage, liquidity, or dividend-paying dummy.
  - Leverage = ratio between firms’ assets and liabilities.
  - Liquidity = ratio of current assets to current liabilities.
  - Dividend dummy = 1 if firm ever pays dividends during sample, 0 otherwise.
  - For leverage and liquidity, comparisons use levels at the 10 and 90 percentiles of the sample distribution.
- Findings:
  - Monetary tightening has a stronger impact on firms facing tighter financial conditions.
  - Leverage is a particularly important amplifier:
    - Firms with higher leverage reduce investment more strongly and persistently than firms with lower leverage.
    - Firms with sufficiently low leverage do not reduce investment (possible market-share gains).
  - Point estimates indicate stronger investment responses for firms with lower liquidity and for firms that do not pay dividends; these differences are quantitatively more modest and less precisely estimated than leverage effects.
- Identification refinement:
  - Interaction coefficient γ^h estimated with country-sector-time fixed effects to control for aggregate domestic shocks; this tests whether financial conditions affect monetary policy impact but does not recover the average effect.

### Robustness of leverage amplification
- Leverage amplification of monetary policy on investment is robust:
  - Holds when sample restricted to manufacturing sector.
  - Holds when using firms’ time-varying (lagged) leverage rather than median levels.
  - Holds when rescaling firms’ leverage by the average country level.

### Additional descriptive statistics (selected entries from Appendix Tables)
- Table A.2: Monetary policy rates in emerging markets (Percent). Selected entries:
  - Brazil: Obs. 213; Mean 13.2; SD 15.4; Min 12.00; Max 26.50; Share≤0.50 0.0
  - Chile: Obs. 237; Mean 3.74; SD 2.04; Min 0.50; Max 11.25; Share≤0.50 4.6
  - India: Obs. 93; Mean 6.54; SD 1.38; Min 4.00; Max 9.00; Share≤0.50 0.0
  - Nigeria: Obs. 71; Mean 12.61; SD 1.39; Min 8.00; Max 16.50; Share≤0.50 0.0
- Table A.3: Descriptive statistics of firm-level data (selected entries):
  - Brazil: Start 2000q1; Firms 417; Leverage ratio 1.3; Liquidity ratio 1.87; Share of firms paying dividends 5.1
  - India: Start 2000q2; Firms 3,046; Leverage ratio 0.6; Liquidity ratio 3.1; Share of firms paying dividends 65.9
  - Indonesia: Start 2001q1; Firms 6,830; Leverage ratio 0.5; Liquidity ratio 3.1; Share of firms paying dividends 67.5
  - Mexico: Start 2000q1; Firms 1,270; Leverage ratio 0.5; Liquidity ratio 1.9; Share of firms paying dividends 59.2
  - Russia: Start 2004q1; Firms 7,800; Leverage ratio 0.6; Liquidity ratio 1.9; Share of firms paying dividends 62.0
  - Türkiye: Start 2001q1; Firms 3,820; Leverage ratio 0.5; Liquidity ratio 2.2; Share of firms paying dividends 64.3

### Broader conclusions and policy implications
- Dataset and identification: novel dataset of monetary policy shocks based on analysts’ forecasts from Bloomberg; survey forecast errors used to construct meeting-level monetary policy surprises; surprise measures orthogonalized with respect to variables known prior to each policy meeting to reduce endogeneity.
- Aggregate results for 18 emerging markets:
  - Monetary policy shocks have highly significant and persistent effects on sovereign bond yields at short- and medium-term maturities.
  - Monetary tightening generates a temporary appreciation of the exchange rate and a reduction in stock prices.
  - Real economy effects: temporary drop in industrial production and a gradual increase in the unemployment rate.
  - Inflation: rapid pass-through to producer prices and a more delayed effect on consumer prices.
  - Firm-level: investment of highly leveraged firms is particularly sensitive to monetary tightening.
- Policy implication:
  - Evidence suggests monetary policy in emerging markets can meaningfully influence domestic financial and macroeconomic conditions and might be better positioned than commonly assumed to mitigate financial and macroeconomic disruptions from global financial shocks, motivating further research on this channel.

*Source: wpiea2024093-print-pdf - Appendix Tables B.2 and B.3 present country-level estimates of the impact of monetary policy shocks on financial variables and subsequent sections on macroeconomic and firm-level transmission.*

### Appendix Tables B.2 and B.3 present country-level estimates of the impact of mone-

### wpiea2024093-print-pdf - Appendix Tables B.2 and B.3 present country-level estimates of the impact of mone-

### Country-level financial impacts (Appendix Tables B.2 and B.3)
- Country-level estimates of monetary policy shocks on financial variables are reported over 1- and 14-day windows.
- Whenever statistically significant, the impact of monetary policy shocks on government bond yields is positive.
- Considerable heterogeneity exists in the strength of the impact, especially on the day of the monetary policy announcement.
- Country-level regressions confirm that monetary policy shocks tend to appreciate the exchange rate on impact.

### Macroeconomic effects of monetary policy shocks — empirical specification
- Regression (panel local projections) estimated:
  Y_{c,t+h} − Y_{c,t−1} = α^h_c + β^h I_{c,t} + A^h(L) ΔY_{c,t−1} + B^h(L) P_{c,t−1} + τ^h_t + ε^h_{c,t}
- Y_{c,t} includes: log of industrial production, unemployment rate, log headline CPI or core CPI, log bilateral exchange rate vs US dollar, and one-year government bond yield in country c in month t.
- Horizon h ≥ 0 extends up to 3 years; local projections cover 36 months following shocks.
- ΔY_{c,t−1} = Y_{c,t−1} − Y_{c,t−1−12} includes yearly changes; A^h(L) and B^h(L) are matrix polynomials of degree 11 (allowing for 12 lags).
- Controls in P_{c,t−1}: log of new COVID-19 cases, an index of lockdown stringency, and an indicator for income support and debt relief measures from the Oxford COVID-19 Government Response Tracker.
- Regression includes country fixed effects α^h_c and time fixed effects τ^h_t to purge global shocks.
- I_{c,t} reflects exogenous variations in interest rates associated with monetary policy decisions; β^h measures the causal effect.

### Data cleaning, transformations, and identification
- Monetary policy shocks converted to monthly frequency by summing all shocks within each month.
- Financial variables (bond yields and exchange rate) use average values in the last three working days of each month.
- Light data cleaning: trim the 0.25 top and bottom percentiles.
- Smoothing: three-month moving averages applied to series for industrial production, unemployment, and CPI.
- Autocorrelation addressed with rich lag structure and clustering standard errors by country.
- Nigeria excluded from regressions because it lacks monthly industrial production data.

### Main OLS local-projection results (rescaled shocks)
- Monetary policy shocks rescaled to generate an increase in bond yields by 1 percentage point on impact.
- Bond yields remain elevated for about 9 months.
- Industrial production contracts by more than 2 percent; effect starts after a lag of a few months and peaks 8 months after the shock.
- Unemployment rate increases by about 0.5 percentage points (gradual rise).
- Producer prices decline starting about 6 months after the shock and cumulate to about a 6 percent reduction.
- Core CPI broadly follows producer prices but quantitatively smaller: total cumulative decline of about 2 percent.
- Headline CPI begins to decline about a year after the monetary tightening (longer lag).
- Exchange rate tends to appreciate, building over time to a temporary appreciation of about 5 percent.
- Results show no evidence of an exchange rate puzzle (i.e., puzzling depreciation) found in some prior studies.

### Instrumental-variable (IV) robustness and orthogonalization
- IV approach: 1-year sovereign bond yield instrumented with orthogonalized monetary policy shocks; first-stage F-statistics well above the reference critical value.
- IV impulse responses are remarkably similar to OLS results.
- Interpretation: in emerging markets monetary policy shocks tend to originate mostly around policy announcements; instrumentation is not essential to capture additional shocks.
- Orthogonalization of monetary policy surprises improves precision:
  - Without orthogonalization, point estimates remain directionally consistent but are quantitatively smaller and confidence bands are considerably wider.
  - Orthogonalization detects limited predictability of policy rate decisions beyond analysts’ forecasts; it is important to improve estimate precision.
- Sample restriction excluding non-inflation-targeting countries (Egypt and Malaysia) leaves point estimates largely unchanged; restricting to inflation-targeting countries reduces confidence bands around CPI impacts and shows faster and stronger unemployment effects.

### Comparison with advanced-economy evidence (US)
- Magnitude and timing of transmission in emerging markets closely resemble recent US estimates (Bauer and Swanson (2023b)):
  - Peak contraction in industrial production: emerging markets 2.5 percent; US between 1.6 to 2.8 percent depending on specification.
  - Unemployment increase: about 0.5 percentage points in both the US and emerging markets.
  - CPI decline: US estimates show a maximum CPI decline between 0.8 and 4 percent; emerging-markets estimates are on the higher end of that range, suggesting steeper Phillips curves in emerging markets.

### Monetary transmission across firms (data and scope)
- Firm-level analysis uses Refinitiv Datastream quarterly data for publicly listed firms.
- Sample includes 9,378 firms.
- Motivation: assess heterogeneity of monetary transmission across firms by financial characteristics, given likely more pervasive borrowing constraints in emerging markets.

*Source: wpiea2024093-print-pdf - Appendix Tables B.2 and B.3 present country-level estimates of the impact of monetary policy shocks on financial variables and subsequent sections on macroeconomic and firm-level transmission.*

### Appendix Table A.3 reports country-level summary statistics of the data.

### wpiea2024093-print-pdf - Appendix Table A.3 reports country-level summary statistics of the data.

### Methodology: firm-level local projections
- Estimated local projections (equation (7)) for firm-level outcomes y_f,t+h − y_f,t−1 with specification:
  - Includes firm fixed effects α^h_f and sector-time fixed effects τ^h_s,t.
  - Controls: sum of monetary policy shocks I_c,t (quarter t in country c), 4-quarter lags of firm-level changes ∆y_f,t−1, and country-level controls X_c,t−1 (including ∆Y_c,t−1 and pandemic indicators P_c,t−1).
  - Sample: all firms with at least 5 years of data; exclude public sector and finance, insurance and real estate sectors; drop observations with quarter-on-quarter growth rates of the dependent variable above 200 percent.
  - Standard errors clustered by firm and time.
  - Focus on contractionary monetary policy shock that generates a 1 percent increase in the one-year bond yield.
  - Responses reported over the 12 quarters following the shock; confidence intervals are 90 percent, constructed with standard errors clustered at the firm and time level.

### Main findings: average firm-level responses to a 1pp monetary policy shock
- Aggregate effects (average across firms):
  - Fixed-capital stock declines by about 3 percent.
  - Sales decline by about 3 percent.
  - Employment falls by about 2 percent.
- Timing and dynamics:
  - Investment exhibits a swift contraction following the shock.
  - Sales contract with a lag of about 3 quarters.
  - Employment contracts within the first year of the monetary policy tightening and then again after three years.
- Interpretation:
  - Investment responds quickly consistent with increased borrowing costs and expected demand contraction.
  - Delayed sales response aligns with delayed effects of monetary policy on household consumption.
  - Employment dynamics are broadly consistent with unemployment rate dynamics estimated from macro data.

### Heterogeneity: transmission by firms’ financial characteristics
- Extended specification with interaction between monetary policy shock and firm financial characteristic F_f (equation (8)):
  - F_f alternately defined as firm leverage, firm liquidity, or a dividend-paying dummy.
  - Leverage = ratio between firms’ assets and liabilities.
  - Liquidity = ratio of current assets to current liabilities.
  - Dividend dummy = 1 if firm ever pays dividends during sample, 0 otherwise.
  - For leverage and liquidity, comparisons use levels equal to the 10 and 90 percentiles of the sample distribution.
- Findings on fixed capital investment:
  - Monetary tightening has a stronger impact on firms facing tighter financial conditions.
  - Leverage is a particularly important amplifier:
    - Firms with higher leverage reduce investment more strongly and persistently than firms with lower leverage.
    - Firms with sufficiently low leverage do not reduce investment (possible market-share gains).
  - Point estimates suggest stronger investment responses for firms with lower liquidity and for firms that do not pay dividends; these differences are quantitatively more modest and less precisely estimated than leverage effects.
- Identification refinement:
  - Interaction coefficient γ^h estimated using country-sector-time fixed effects (equation (9)) to control for aggregate domestic shocks (e.g., fiscal policy, structural reforms).
  - This specification tests whether financial conditions affect monetary policy impact but does not recover the average monetary policy effect.

### Robustness checks on leverage effects
- Leverage amplification of monetary policy on investment is consistent across checks:
  - Holds when sample restricted to manufacturing sector (more homogeneous set of firms).
  - Holds when using firms’ time-varying (lagged) leverage rather than median levels.
  - Holds when rescaling firms’ leverage by the average country level, addressing cross-country leverage variation concerns.

### Broader conclusions and implications
- Dataset and identification:
  - Novel dataset of monetary policy shocks based on analysts’ forecasts from Bloomberg; survey forecast errors used to construct meeting-level monetary policy surprises.
  - Surprise measures orthogonalized with respect to variables known prior to each policy meeting to reduce endogeneity.
- Aggregate results for 18 emerging markets:
  - Monetary policy shocks have highly significant and persistent effects on sovereign bond yields at short- and medium-term maturities.
  - A monetary tightening also generates a temporary appreciation of the exchange rate and a reduction in stock prices.
  - Real economy effects: temporary drop in industrial production and a gradual increase in the unemployment rate.
  - Inflation: rapid pass-through to producer prices and a more delayed effect on consumer prices.
  - Firm-level: investment of highly leveraged firms is particularly sensitive to monetary tightening.
- Policy implication:
  - Evidence suggests monetary policy in emerging markets can meaningfully influence domestic financial and macroeconomic conditions.
  - Findings indicate monetary policy might be better positioned than commonly assumed to mitigate financial and macroeconomic disruptions from global financial shocks, motivating further research on this channel.

*Source: Appendix material and text from the specified content unit in wpiea2024093-print-pdf.*

### References

### References

### Key cited literature
- The references list includes empirical and theoretical studies on monetary policy surprises, high-frequency identification, forward guidance, international spillovers, and monetary transmission in emerging markets. Notable cited works include studies by Acosta (2022); Ahmed, Akinci, and Queralto (2021); Altavilla et al. (2019); Andrade and Ferroni (2021); Andrews, Stock, and Sun (2018); Aruoba et al. (2021); Bauer and Swanson (2023a, 2023b); Bernanke and Kuttner (2005); Brandão-Marques et al. (2021); Bräuning and Ivashina (2020); Campbell et al. (2017); Cieslak (2018); Cieslak and Schrimpf (2019); Cloyne et al. (2023); Cochrane and Piazzesi (2002); Cook and Hahn (1989); Deb et al. (2023); Dedola et al. (2017); De Leo, Gopinath, and Kalemli-Özcan (2022); Frankel (2010); Gertler and Karadi (2015); Gilchrist, López-Salido, and Zakrajšek (2015); Gruss and Kebhaj (2019); Gürkaynak, Sack, and Swanson (2005); Hanson and Stein (2015); Hnatkovska, Lahiri, and Vegh (2016); Iacoviello and Navarro (2019); Jarociński and Karadi (2020); Jeenas (2019); Kalemli-Özcan (2019); Kohlscheen (2014); Kuttner (2001); Meyer (2006); Miranda Agrippino and Ricco (2021); Miranda-Agrippino and Rey (2020); Montiel Olea and Pflueger (2013); Montiel Olea and Plagborg-Møller (2021); Nakamura and Steinsson (2018); Ottonello and Winberry (2020); Pescatori (2018); Ramey (2016); Rey (2015); Romer and Romer (1994, 2000); Schmeling, Schrimpf, and Steffensen (2022); Stock and Watson (2018); Swanson (2021, 2023); Vegh and Vuletin (2012, 2014).

### Appendix: Descriptive statistics and data sources (contents and key elements)
- Figure A.1: Cross-country distributions comparing the sample of emerging markets with advanced economies (IMF classification) for:
  - Per capita GDP (PPP, log)
  - Financial development (Index; IMF Financial Development Index Database)
  - Foreign assets and liabilities (Percent of GDP)
  - Notes: Country values are averages over the period of analysis; countries with population less than 2 million are excluded.
- Table A.1: Stock market indexes used for each country (mapping of country to index name).
- Table A.2: Monetary policy rates in emerging markets (Percent). Sample restricted to meetings with analysts’ forecasts. Columns reported: Obs., Mean, SD, Min, Max, Share≤0.50. Selected entries (exact values as reported):
  - Brazil: Obs. 213; Mean 13.2; SD 15.4; Min 12.00; Max 26.50; Share≤0.50 0.0
  - Chile: Obs. 237; Mean 3.74; SD 2.04; Min 0.50; Max 11.25; Share≤0.50 4.6
  - Colombia: Obs. 214; Mean 5.46; SD 2.21; Min 1.75; Max 12.00; Share≤0.50 0.0
  - Egypt: Obs. 92; Mean 11.23; SD 3.45; Min 8.25; Max 18.75; Share≤0.50 0.0
  - Hungary: Obs. 234; Mean 4.95; SD 3.67; Min 0.60; Max 13.00; Share≤0.50 0.0
  - India: Obs. 93; Mean 6.54; SD 1.38; Min 4.00; Max 9.00; Share≤0.50 0.0
  - Indonesia: Obs. 81; Mean 4.56; SD 0.82; Min 3.50; Max 6.00; Share≤0.50 0.0
  - Malaysia: Obs. 111; Mean 2.88; SD 0.56; Min 1.75; Max 3.50; Share≤0.50 0.0
  - Mexico: Obs. 156; Mean 5.82; SD 1.85; Min 3.00; Max 10.50; Share≤0.50 0.0
  - Nigeria: Obs. 71; Mean 12.61; SD 1.39; Min 8.00; Max 16.50; Share≤0.50 0.0
  - Peru: Obs. 196; Mean 3.54; SD 1.57; Min 0.25; Max 7.50; Share≤0.50 8.2
  - Philippines: Obs. 145; Mean 4.28; SD 1.53; Min 2.00; Max 7.50; Share≤0.50 0.0
  - Poland: Obs. 209; Mean 3.19; SD 1.83; Min 0.10; Max 6.75; Share≤0.50 8.1
  - Romania: Obs. 119; Mean 4.64; SD 2.73; Min 1.25; Max 10.25; Share≤0.50 0.0
  - Russia: Obs. 76; Mean 8.31; SD 2.77; Min 4.25; Max 20.00; Share≤0.50 0.0
  - South Africa: Obs. 122; Mean 6.84; SD 2.10; Min 3.50; Max 13.50; Share≤0.50 0.0
  - Thailand: Obs. 126; Mean 1.89; SD 0.98; Min 0.50; Max 4.75; Share≤0.50 13.5
  - Türkiye: Obs. 139; Mean 9.88; SD 5.13; Min 4.50; Max 24.00; Share≤0.50 0.0
- Table A.3: Descriptive statistics of firm-level data. Columns: Country; Start; Firms; Leverage ratio; Liquidity ratio; Share of firms paying dividends (percent). Selected entries (exact values as reported):
  - Brazil: Start 2000q1; Firms 417; Leverage ratio 1.3; Liquidity ratio 1.87; Share of firms paying dividends 5.1
  - Chile: Start 2001q1; Firms 2020; Leverage ratio 0.5; Liquidity ratio 3.68; Share of firms paying dividends 9.0
  - India: Start 2000q2; Firms 3,046; Leverage ratio 0.6; Liquidity ratio 3.1; Share of firms paying dividends 65.9
  - Indonesia: Start 2001q1; Firms 6,830; Leverage ratio 0.5; Liquidity ratio 3.1; Share of firms paying dividends 67.5
  - Mexico: Start 2000q1; Firms 1,270; Leverage ratio 0.5; Liquidity ratio 1.9; Share of firms paying dividends 59.2
  - Russia: Start 2004q1; Firms 7,800; Leverage ratio 0.6; Liquidity ratio 1.9; Share of firms paying dividends 62.0
  - Türkiye: Start 2001q1; Firms 3,820; Leverage ratio 0.5; Liquidity ratio 2.2; Share of firms paying dividends 64.3
  - Notes: Leverage and liquidity ratios report the cross-firm average of the firm-specific median leverage. Firms are classified as paying dividends if they ever paid a dividend during their lifetime.
- Section B: Additional results and robustness analysis includes:
  - B.1 Monetary policy surprises: Table B.1 reports correlations of monetary policy surprises based on alternative time windows. Exact correlations reported:
    - Correlation matrix (Last 2 weeks / Last week / Last 3 days): Last 2 weeks vs Last 2 weeks 1.000; Last week vs Last 2 weeks 0.994; Last 3 days vs Last 2 weeks 0.929; Last week vs Last week 1.000; Last 3 days vs Last week 0.934; Last 3 days vs Last 3 days 1.000.
  - B.2 Monetary policy impact on financial markets: Tables B.2 and B.3 report contemporaneous effects and 2-week-ahead effects of monetary policy shocks on financial market variables (1-year, 2-year, 5-year government yields; EMBI spread; exchange rate; stock prices) by country. Regression notes:
    - All regressions include a constant. Robust standard errors reported in parentheses. ***, **, * denote statistical significance at 1, 5, and 10 percent levels.
    - Tables report coefficients, Obs., and R2 for each country-variable pair. Examples of reported coefficients (contemporaneous, Table B.2, exact values as shown for some countries):
      - Brazil: 1-year govt. yield coeff. 0.549***; Obs. 134; R2 0.12. 2-year govt. yield coeff. 0.427**; Obs. 134; R2 0.06. Exchange rate coeff. -1.014**; Obs. 207; R2 0.04.
      - India: 1-year govt. yield coeff. 0.500**; Obs. 39; R2 0.16. 2-year coeff. 0.483***; Obs. 39; R2 0.19. 5-year coeff. 0.570***; Obs. 39; R2 0.21. Exchange rate coeff. -1.385**; Obs. 39; R2 0.10.
      - Poland (Table B.3, 2 weeks after): 1-year govt. yield coeff. 1.056***; Obs. 206; R2 0.18. 2-year coeff. 1.065***; Obs. 206; R2 0.13. 5-year coeff. 0.780***; Obs. 206; R2 0.07.
  - B.3 Monetary policy impact on macroeconomic conditions:
    - Figure B.1: Plots of the effective F-statistic (Montiel Olea and Pflueger (2013)) from IV first-stage estimation for: (a) 1-year govt. bond yield, (b) Industrial production, (c) Unemployment, (d) PPI, (e) Core (solid) and headline (dashed) CPI, (f) Exchange rate. Notes explain shading for thresholds (rule-of-thumb 10; simplified critical value of Montiel Olea and Pflueger (2013)).
    - Figure B.2: Macroeconomic responses using non-orthogonalized monetary policy surprises — effects of a one-pp monetary policy shock on macro variables during 36 months following the shock; regressions include lags, country and time fixed effects; solid lines show point estimates, shaded areas 90 percent confidence intervals clustered at country level; dotted lines denote baseline estimates reported in Figure 4.
    - Figure B.3: Same as Figure B.2 but sample restricted to inflation-targeting countries; presentation analogous (one-pp shock; 36 months; 90 percent confidence intervals).
  - B.4 Monetary policy impact on firm-level data:
    - Figure B.4: Heterogeneous responses of fixed capital across firms by leverage: panels for (a) Manufacturing only, (b) Time-varying leverage, (c) Rescaled leverage. Shows differential effects of a one-pp monetary policy shock on firms’ fixed capital during 12 quarters following the shock. Regressions include lags and country-sector-time fixed effects; point estimates and 90 percent confidence intervals with standard errors clustered at firm and time level.

*Monetary Policy Transmission in Emerging Markets: Proverbial Concerns, Novel Evidence — Working Paper No. WP/2024/093*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024093-print-pdf.pdf_