## wp18267

## Source details

**Canonical URL:** [wp18267](https://www.imf.org/-/media/files/publications/wp/2018/wp18267.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2018/wp18267.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2018/wp18267.pdf.json)

---

### Purpose and research questions
- Objective: Assess the effectiveness of macroprudential policies in reducing firm credit and their impact on firms’ investment and sales growth.
- Key distributional questions:
  - Do macroprudential policies differentially affect smaller and younger firms compared with larger and older firms?
  - Within MSMEs and young firms, are the least creditworthy firms most affected, consistent with stability-enhancing goals?
- Scope: Firm-level evidence across countries to evaluate borrower-targeted versus financial institution-targeted instruments, short-term versus long-term financing, and real outcomes (investment and sales).

### Data, variables, and empirical approach
- Micro sample:
  - Firm-level data from Orbis covering more than 900,000 firms between 2003 and 2011 in 48 countries.
  - Total # Observations: 3,107,242; Total # Firms: 898,653.
  - Estimations weighted with the inverse of the number of firms in each country to address unbalanced sample.
- Firm classifications and definitions:
  - MSMEs: firms with fewer than 250 employees; Micro firms: 1 to 9 employees; SME firms: 10 to 249 employees.
  - Young firms: firms that are three years or younger since incorporation.
  - Firm size classification fixed by median employees across all observations.
- Financing and real variables (ORIBS definitions preserved):
  - Short-term financing: Log change in short-term debt (residual maturity ≤ 1 year).
  - Long-term financing: Log change in long-term debt (residual maturity ≥ 1 year).
  - Overall Financing Growth: Log change in total financing (sum of short- and long-term debt).
  - Investment: Log change in fixed assets.
  - Sales Growth: Log change in operating turnover.
  - Growth truncation: top and bottom 5% outliers dropped.
- Creditworthiness measures:
  - Leverage = total debt / total assets.
  - Interest Coverage dummy = 1 for interest coverage ratio < 1 (interest coverage = EBIT / interest expenses).
  - ROA = return on assets (EBIT / total assets).
- Macroprudential policy measures (Cerutti et al. / GMPI):
  - BOR index (borrower-targeted): LTV_CAP + DTI; BOR index range 0 to 2.
  - FIN index (financial institution-targeted): 10 instruments; FIN index range 0 to 10; descriptive sample observed FIN range: zero to six tools in use.
  - MPI index = BOR + FIN (MPI range 0-12).
  - Instruments coded 1/0 by country-year; intensity generally not captured except LTV intensity for 16 countries.
- Controls and identification:
  - Country-level controls: log change of GDP (GDP growth) and Real Policy Rate (discount rate minus inflation).
  - GFC dummy = 1 for years 2008 and 2009.
  - Main regressions: firm fixed effects (η_i); some specifications include country-year fixed effects (μ_jt) and industry-year fixed effects (δ_kt).
  - Policies and macro variables lagged (jt-1); standard errors clustered at country level.

### Main empirical findings — financing growth (aggregate and policy-type heterogeneity)
- Aggregate association:
  - Macroprudential policies are negatively associated with firm financing growth, with heterogeneous effects by policy type and firm type.
- Policy-type heterogeneity:
  - BOR is robustly and negatively associated with growth in long-term firm financing.
    - Baseline coefficient: BOR = -0.046** (long-term financing growth) in Appendix C baseline (N = 3,107,242; # of countries = 48).
    - Applying one additional borrower-related macroprudential policy (BOR) is associated with a 4.8 percentage points lower long-term financing growth (reported in main text).
  - FIN does not appear significantly correlated with firm financing growth in pooled estimations (e.g., FIN = -0.009 (long-term) in Appendix C, not significant).
  - Interpretation: leakage/avoidance may be easier when policies target institutions rather than borrowers.

### Firm-type heterogeneity (size and age)
- MSMEs (fewer than 250 employees) vs large firms:
  - Implementation of one additional borrower-related tool (BOR) associated with:
    - 8.6 percentage points lower short-term financing growth (MSMEs vs large firms).
    - 3.8 percentage points lower long-term financing growth (MSMEs vs large firms).
    - 5.1 percentage points lower overall financing growth (MSMEs vs large firms).
- Microenterprises vs SMEs:
  - All six interactions of BOR with micro- and SME dummies enter negatively and significantly at least at the 10 percent level; economic effects somewhat higher for microenterprises than for SMEs.
- Young firms (≤ 3 years) vs older firms:
  - All interactions between macroprudential tools and the young firm dummy enter negatively; significance mainly for BOR.
  - Short- and long-term financing growth is 2.3 percentage points lower for younger than for older firms after one additional borrower-targeted tool.
  - Overall financing growth is 1.4 percentage points lower for younger than for older firms after one additional borrower-targeted tool.
  - Long-term financing growth decreases by 4 percentage points for younger firms relative to older firms after one additional financial-institution-targeted tool.
  - Long-term financing growth decreases by 2.5 percentage points for younger firms relative to older firms after any additional macroprudential tool.

### Creditworthiness interactions — who is most affected
- Motivation: Distinguish whether policies restrict weakest borrowers (financial inclusion concern) or risky borrowers (stability objective).
- Leverage interactions (Table 6):
  - Highly levered micro firms are more sensitive:
    - MPI x Leverage: -0.160** (short-term) in micro sample.
    - BOR x Leverage: -0.127** (short-term) in micro sample.
    - FIN x Leverage: -0.191** (short-term) in micro sample.
  - Marginal effects per 1 percentage point higher leverage for a micro firm after one additional instrument:
    - BOR: 0.127 percentage points lower short-term financing growth.
    - FIN: 0.191 percentage points lower short-term financing growth.
    - MPI: 0.160 percentage points lower short-term financing growth.
  - Example: Change in leverage from the 25th percentile to the 75th percentile = 31 percentage point increase → associated additional decreases in short-term financing growth of 3.9, 5.9, and 5.0 percentage points for BOR, FIN, and MPI, respectively.
  - SMEs: BOR associated with lower short-term financing for high-leverage SMEs; FIN associated with lower long-term financing for high-leverage SMEs; BOR, FIN, MPI associated with lower overall financing growth for high-leverage SMEs.
  - Young firms: no differential effect by leverage detected.
- Profitability (ROA) interactions (Table 7):
  - More profitable micro firms experience relatively higher financing growth after borrower-targeted measures:
    - Per 1 percentage point higher ROA, micro firm has on average:
      - 0.24 percentage points higher short-term financing growth after one additional BOR.
      - 0.11 percentage points higher short-term financing growth after one additional FIN.
      - 0.13 percentage points higher short-term financing growth after one additional MPI.
    - Example: An 8 percentage point increase in ROA (25th to 75th percentile) → additional short-term financing growth of 1.9, 0.9, and 1.0 percentage points for BOR, FIN, MPI respectively.
  - More profitable SMEs experience relatively higher long-term and overall funding growth after one additional borrower-targeted instrument.
  - More profitable young firms: BOR and FIN associated with higher short-term financing growth; BOR associated with higher overall financing growth.
- Interest coverage interactions (Table 8):
  - Interest coverage < 1 denotes potential financial distress.
  - Micro firms with interest coverage < 1:
    - BOR, FIN, MPI associated with lower short-term financing growth: 2.5, 3, and 2.7 percentage points lower short-term financing growth, respectively, relative to firms not in financial distress, following implementation of one additional instrument (main-text magnitudes).
    - BOR also reduces long-term and overall financing growth for distressed micro firms.
  - SMEs in financial distress experience lower long-term and overall funding growth with higher BOR.
  - Young firms with interest coverage < 1: BOR associated with a drop in long-term and overall financing growth; FIN weakly negatively associated with overall financing growth for young distressed firms.
- Overall interpretation: Among MSME and young firms, the weakest firms—those with high leverage, low profitability, and interest coverage ratios below one—experience drops in credit growth as macroprudential policies are implemented. These effects are consistent with the stability objective of restricting credit to risky firms but also raise distributional concerns.

### Intensity analysis — Loan-to-Value (LTV) caps (16-country subset)
- Both level and change in cumulative loan-to-value ratio associated with relatively lower long-term financing growth of MSMEs; no significant impact on short-term or overall financing growth.
  - A tightening of the cumulative LTV cap associated with a 10.1 percentage point lower long-term financing growth for MSMEs relative to large firms (level).
  - An increase in the cumulative tightening of the LTV cap associated with a 6.2 percentage point lower long-term financing growth for MSMEs relative to large firms (change).
- LTV effects hold for both microenterprises and SMEs.
- No significant differential effect of LTV level or change on young vs older firms.

### Real economy effects — investment and sales growth
- BOR is the only macroprudential coefficient entering significantly in investment and sales growth regressions:
  - Applying one additional borrower-related tool is associated with a 4.4 percentage point lower investment growth.
  - Applying one additional borrower-related tool is associated with a 3.5 percentage point lower sales growth.
- Differential effects (Table 10):
  - MSMEs: BOR implementation associated with statistically and economically significant lower relative investment and sales growth. Example coefficients:
    - MSME x BOR: -0.074*** (Investment); -0.078*** (Sales) (Panel A).
  - Micro and SME: Both experience relatively lower investment and sales growth after BOR implementation; FIN shows no significant effect for MSMEs.
  - Young firms: Interactions between young dummy and MPI, BOR, and FIN enter negatively and significantly:
    - Young x MPI: -0.011*** (Investment); -0.022*** (Sales).
    - Young x BOR: -0.019*** (Investment); -0.040*** (Sales).
    - Young x FIN: -0.010*** (Investment); -0.021** (Sales).
- Conclusion: Negative association between macroprudential policies (especially borrower-targeted) and financing growth for MSMEs and young firms translates into lower investment and sales growth for these groups.

### Robustness checks and estimation caveats
- Robustness:
  - Results hold across checks: joint inclusion of BOR and FIN, splitting FIN into cyclical vs structural tools, controlling for firm creditworthiness measures, interacting macro shocks with firm size/age, adding industry-year fixed effects.
  - LTV intensity analysis limited to 16 countries due to data availability.
- Limitations and interpretation cautions:
  - GMPI generally does not capture intensity or bindingness of instruments (intensity used only for LTV subset).
  - Combining data across many countries and tools prevents speaking to effectiveness of specific tools.
  - Residual endogeneity concerns remain: changes in aggregate debt growth could induce adoption of macroprudential tools; policies can affect both demand and supply of credit.
  - Causal inference approached cautiously; results described as associations with strategies to mitigate identification concerns (lagging policies, firm fixed effects, country-year fixed effects, within-firm variation).

### Key descriptive statistics and regression magnitudes (preserved)
- Sample coverage: more than 900,000 firms between 2003 and 2011 in 48 countries.
- Financing growth variation:
  - Short-term financing growth range: ‒190% to 200%.
  - Long-term financing growth range: ‒140% to 150%.
  - Median short-term financing growth: 2.6%.
  - Median long-term financing growth: ‒7.7%.
  - Median overall financing growth: ‒3.4%.
- Selected baseline regression coefficients (Appendix C; N = 3,107,242; # of countries = 48):
  - Firm assets: -0.132***; -0.144***; -0.147*** (short-term; long-term; overall).
  - GDP growth: 0.007***; 0.007***; 0.006***.
  - GFC: -0.046***; -0.039***; -0.045***.
  - BOR: -0.016; -0.046**; -0.022 (short-term; long-term; overall).
  - FIN: 0.001; -0.009; -0.004.
- Real effects regression (Table 9; N 19,876,932; # of countries 48):
  - BOR: -0.044*** (Investment); -0.035** (Sales).
  - MPI: -0.010 (Investment); -0.001 (Sales).
  - FIN: 0.005 (Investment); 0.016 (Sales).

*Source: wp18267*

### 1. Introduction

### 1. Introduction

### Purpose and research questions
- Objective: Assess the effectiveness of macroprudential policies in reducing firm credit and their impact on firms’ investment and sales growth.
- Key distributional questions:
  - Do macroprudential policies differentially affect smaller and younger firms compared with larger and older firms?
  - Within MSMEs and young firms, are the least creditworthy firms most affected, consistent with stability-enhancing goals?
- Scope: Firm-level evidence across countries to evaluate borrower-targeted versus financial institution-targeted instruments, short-term versus long-term financing, and real outcomes (investment and sales).

### Data and methods summary
- Micro sample:
  - Firm-level data from Orbis covering more than 900,000 firms between 2003 and 2011 in 48 countries.
  - Analysis restricted to non-financial firms; various cleaning steps (drop duplicates, acquirers post-acquisition, zero/negative total assets or employees, listed MSMEs).
  - Weighting: estimations weighted with the inverse of the number of firms in each country to address unbalanced sample.
- Firm classifications:
  - MSMEs: firms with fewer than 250 employees.
  - Micro firms: 1 to 9 employees.
  - SME firms: 10 to 249 employees.
  - Young firms: firms that are three years or younger since incorporation.
  - Firm size classification fixed by median employees across all observations to avoid reclassification bias.
- Financing and real variables:
  - Short-term financing: growth in short-term debt (residual maturity < 1 year).
  - Long-term financing: growth in long-term debt (residual maturity ≥ 1 year).
  - Overall Financing Growth: growth in total financing (sum of short- and long-term debt).
  - Investment: growth in fixed assets.
  - Sales Growth: growth in operating turnover.
  - Growth defined as annual growth rate via log-differences; top and bottom 5% outliers dropped.
- Creditworthiness measures:
  - Leverage ratio: total debt / total assets.
  - Interest Coverage: dummy = 1 for interest coverage ratio < 1, 0 otherwise (interest coverage = EBIT / interest expenses).
  - Profitability: return on assets (ROA).
- Macroprudential policy measures (from GMPI / Cerutti et al. (2015)):
  - BOR index (borrower-targeted): includes LTV and DSTI ratios; BOR index range 0 to 2.
  - FIN index (financial institution-targeted): 10 instruments (dynamic loan-loss provisioning; countercyclical capital buffers; bank leverage ratio; capital surcharge for SIFIs; limits on interbank exposures; concentration limits; limits on foreign currency loans; limits on domestic currency loans; reserve requirement ratios; taxes/levies on financial institutions); FIN index range 0 to 10.
  - MPI index = BOR + FIN.
  - Instruments coded 1/0 for each country-year when in place; intensity/bindingness not generally captured.
- Controls:
  - Country-level time-varying controls include log change of GDP (GDP Growth) and Real Policy Rate (discount rate minus inflation).
  - GFC dummy for years 2008 and 2009 to control for Global Financial Crisis effects.
- Additional methodological notes:
  - Industry-time fixed effects used in robustness checks to control for industry demand shocks.
  - Within-firm variation and country-year fixed effects exploited to mitigate endogeneity concerns.
  - Sample limitations on firm counts per country noted; estimations only on countries with at least 25 firms initially.

### Key findings
- Aggregate association:
  - Macroprudential policies are negatively associated with firm financing growth, with heterogeneous effects by policy type and firm type.
- Policy-type heterogeneity:
  - BOR (borrower-targeted index) is robustly and negatively associated with growth in long-term firm financing.
  - FIN (financial institution-targeted index) does not appear to be significantly correlated with firm financing growth in pooled estimations.
  - Interpretation: leakage/avoidance may be easier when policies target institutions rather than borrowers.
- Firm-type heterogeneity:
  - MSMEs (firms with fewer than 250 employees) and young firms (three years or less) show stronger negative correlations between financing growth and macroprudential policies than larger and older firms.
  - Possible mechanism: MSMEs and young firms are more opaque and dependent on bank relationship lending; larger firms can substitute with non-bank finance when bank credit is limited.
- Creditworthiness interactions:
  - Among MSMEs and young firms, the negative association between credit growth and macroprudential policies is stronger for the least creditworthy and riskiest firms (as measured by leverage, interest coverage, and ROA), consistent with stability-enhancing objectives.
- Real economy effects:
  - MSMEs in countries with borrower-targeted macroprudential instruments have lower investment and sales growth.
  - Young firms in economies with borrower and/or financial institution-targeted macroprudential instruments have lower investment and sales growth.
  - Conclusion: macroprudential policies affect both financial stability and real activity for smaller and younger firms.

### Key statistics and descriptive ranges (preserved exactly as in the source)
- Sample coverage: more than 900,000 firms between 2003 and 2011 in 48 countries.
- Financing growth variation (summary statistics):
  - Short-term financing growth range: ‒190% to 200%.
  - Long-term financing growth range: ‒140% to 150%.
  - Median short-term financing growth: 2.6%.
  - Median long-term financing growth: ‒7.7%.
  - Median overall financing growth: ‒3.4%.
- Macroprudential indices ranges:
  - BOR index: 0 to 2 (out of 2 possible borrower-targeted instruments).
  - FIN index: 0 to 10 (out of 10 possible financial institution-targeted instruments).
  - Descriptive sample observed FIN range: zero to six tools in use in the sample period for some countries.

### Robustness, limitations, and interpretation cautions
- Advantages of micro-data:
  - Reduces endogeneity concerns compared with aggregate analysis; allows firm-group differential effects and inclusion of country-year fixed effects.
- Limitations and qualifications:
  - Combining data across many countries and many tools prevents speaking to effectiveness of specific tools.
  - GMPI does not capture intensity or bindingness of instruments in general (intensity data used only in subset robustness checks).
  - Residual endogeneity concerns remain (changes in aggregate debt growth could induce adoption of macroprudential tools; policies can affect both demand and supply of credit).
  - Causal inference is approached cautiously; results described as associations with efforts to mitigate identification concerns via within-firm variation and firm-group heterogeneity.

### Paper structure (as described)
- Section 2: Data and empirical methodology.
- Section 3: Results on association between macroprudential policies and firm financing and growth.
- Section 4: Conclusion.

*Source: wp18267 - 1. Introduction*

### Appendix B provides detailed definitions and sources of each of the variables used in the analysis.

### Appendix B provides detailed definitions and sources of each of the variables used in the analysis.

### Empirical specifications for financing growth
- Main regression (firm-level):
  - Financing growth_ijt = β1 Macropru_jt-1 + β2 Firm Assets_ijt + β3 Macro_jt-1 + β4 GFC_t + η_i + ε_ijt  (1)
  - Dependent variables: Short-term financing growth; Long-term financing growth; Overall Financing Growth.
  - Explanatory variables:
    - Macropru: indicator of macroprudential policies.
    - Firm Assets: log of total assets.
    - Macro: vector of macroeconomic variables including the real monetary policy rate and the log change of GDP.
    - GFC: Global Financial Crisis dummy variable for 2008 and 2009.
  - Fixed effects and weighting:
    - η_i: firm fixed effects to control for time-invariant firm characteristics.
    - Observations weighted by the inverse of the number of firms per country and year so that each country has the same weight.
    - Standard errors clustered at the country level.

- Rationale and estimation choices:
  - Macroprudential and macroeconomic variables are lagged (jt-1) to reduce reverse causation and allow for policy implementation lags.
  - Firm fixed effects control for sector and business model factors that are time invariant.

### Identification of within-country heterogeneity (reducing time-varying omitted variables)
- Interaction regression with country-year fixed effects:
  - Financing growth_ijt = β1 Macropru_jt-1 * Firm Characteristic_i + β2 Firm Assets_jit + μ_jt + η_i + ε_ijt  (2)
  - μ_jt: country-year fixed effects to control for any time-varying country factor affecting financing growth for the average firm.
- Firm characteristics and grouping:
  - Firm size groups: micro (1 to 9 employees), SME (10 to 249 employees).
  - Firm age group: firms three years or younger (since incorporation).
  - Expectation: smaller and younger firms — being more bank-dependent and having less diversified external financing — will exhibit stronger responses to macroprudential policies.
- Note on omitted terms:
  - Main effects of Firm Characteristic and Macroprudential Regulation are not included separately because they are subsumed by firm and country-time fixed effects, respectively.

### Empirical specifications for real effects (firm growth)
- Main regression (firm-level):
  - Firm growth_ijt = β1 Macropru_jt-1 + β2 Firm Assets_ijt + β3 Macro_jt-1 + β4 GFC_t + η_i + ε_ijt  (3)
  - Firm growth measured by Investment growth and Sales growth.
- Heterogeneity regression with country-year fixed effects:
  - Firm growth_ijt = β1 Macropru_jt-1 * Firm Characteristic_i + β2 Firm Assets_it + μ_jt + η_i + ε_ijt  (4)
  - As in financing regressions, interaction with Firm Characteristic_i helps mitigate endogeneity concerns by explaining differential effects across firm groups.

### Estimation and inference practices emphasized
- Use of lagged policy and macro variables to mitigate reverse causality.
- Inclusion of firm fixed effects to control for time-invariant heterogeneity.
- Inclusion of country-year fixed effects in interaction specifications to control for time-varying country-level confounders.
- Weighting by inverse firm counts per country-year to equalize country influence.
- Clustering standard errors at the country level to allow intra-country correlation of error terms.

*Source: Appendix B, wp18267.*

### 3. Macroprudential policies and firm financing and growth

### 3. Macroprudential policies and firm financing and growth

### Methodology and identification
- Regressions of firms’ short-term, long-term and overall financing growth on different macroprudential policies, controlling for firm assets and a number of country-level variables.
- Firm-fixed effects included to exploit within-firm financing growth variation.
- Country-year fixed effects used in some specifications and interactions to improve identification and address policy endogeneity.
- Intensity indicator used for loan-to-value (LTV) caps (quarterly index aggregated to yearly cumulative changes); LTV data available for 16 countries.

### Main findings on financing growth (Table 2 and robustness)
- All nine coefficients on macroprudential policies enter negatively, but only BOR is significant in the long-term financing growth regression.
- Applying one additional borrower-related macroprudential policy (BOR) is associated with a 4.8 percentage points lower long-term financing growth.
- Firms reduce financing growth as they grow larger.
- GDP growth is positively and significantly associated with firm financing growth.
- Financing growth was significantly lower during the Global Financial Crisis.
- No significant relationship between the real interest rate and firms’ financing growth over the period analyzed.
- Robustness: Including FIN and BOR jointly yields consistent results (only BOR significant for long-term financing). Splitting FIN into cyclical vs structural tools yields no significant coefficients. Controlling for firm creditworthiness measures (leverage, interest coverage dummy, return on assets) does not alter results (sample reduces).

### Firm heterogeneity (Table 3 and related checks)
- MSMEs: Implementation of one additional borrower-related tool (BOR) is associated with:
  - 8.6 percentage points lower short-term financing growth (MSMEs vs large firms)
  - 3.8 percentage points lower long-term financing growth (MSMEs vs large firms)
  - 5.1 percentage points lower overall financing growth (MSMEs vs large firms)
- Microenterprises vs SMEs (Panel B): All six interactions of BOR with micro- and SME dummies enter negatively and significantly at least at the 10 percent significance level; economic effects somewhat higher for microenterprises than for SMEs.
- FIN (tools targeted at financial institutions) associated with relatively lower financing growth for microenterprises, but not for SMEs. The overall index of macroprudential tools negatively associated with relative financing growth of microenterprises (significant) but not SMEs (not significant).
- Young firms (Panel C):
  - All interaction terms between macroprudential tools and the young firm dummy enter negatively; significance occurs mainly for BOR across financing measures.
  - Short- and long-term financing growth is 2.3 percentage points lower for younger than for older firms after one additional borrower-targeted tool is adopted.
  - Overall financing growth is 1.4 percentage points lower for younger than for older firms after one additional borrower-targeted tool.
  - Long-term financing growth decreases by 4 percentage points for younger firms relative to older firms after one additional financial-institution-targeted macroprudential tool is adopted.
  - Long-term financing growth decreases by 2.5 percentage points for younger firms relative to older firms after any additional macroprudential tool is adopted.
- Robustness: Results hold when interacting size and age dummies with macro variables (Table 4), when adding industry-year fixed effects (Appendix E), and when controlling for firm creditworthiness measures (Appendix F).

### Intensity of borrower-targeted measures: Loan-to-value caps (Table 5)
- Both the level and the change in cumulative loan-to-value ratio are associated with relatively lower long-term financing growth of MSMEs; no significant impact on short-term or overall financing growth.
- A tightening of the cumulative loan-to-value cap is associated with a 10.1 percentage point lower long-term financing growth for MSMEs relative to large firms (level).
- An increase in the cumulative tightening of the loan-to-value cap is associated with a 6.2 percentage point lower long-term financing growth for MSMEs relative to large firms (change).
- These LTV results hold for both microenterprises and SMEs (Panel B).
- No significant differential effect of LTV level or change on young vs older firms (Panel C).

### Differential impact by financial strength (Tables 6–8)
- Motivation: Determine whether macroprudential policies restrict credit to weakest firms (financial inclusion concern) or restrict credit to risky firms (stability objective).

Leverage interactions (Table 6)
- Highly levered micro firms are more sensitive to macroprudential policies:
  - BOR, FIN, and MPI associated with lower short-term financing for highly levered micro borrowers.
  - BOR associated with lower overall financing growth for highly levered micro borrowers relative to less levered micro borrowers.
- Marginal effects per 1 percentage point higher leverage for a micro firm following implementation of one additional instrument:
  - 0.127 percentage points lower short-term financing growth for BOR
  - 0.191 percentage points lower short-term financing growth for FIN
  - 0.160 percentage points lower short-term financing growth for MPI
- Example: Change in leverage from the 25th percentile to the 75th percentile = 31 percentage point increase → associated additional decreases in short-term financing growth of 3.9, 5.9, and 5.0 percentage points for BOR, FIN, and MPI, respectively.
- No effect of leverage on long-term financing for micro borrowers.
- SMEs: BOR associated with lower short-term financing for high-leverage SMEs; FIN associated with lower long-term financing for high-leverage SMEs. BOR, FIN, and MPI associated with lower overall financing growth for high-leverage SMEs.
- Young firms: no differential effect by leverage detected.

Profitability (ROA) interactions (Table 7)
- More profitable micro firms experience relatively higher financing growth after borrower-targeted measures:
  - Per 1 percentage point higher ROA, a micro firm has on average:
    - 0.24 percentage points higher short-term financing growth following one additional BOR instrument
    - 0.11 percentage points higher short-term financing growth following one additional FIN instrument
    - 0.13 percentage points higher short-term financing growth following one additional MPI instrument
  - Example: An 8 percentage point increase in ROA (25th to 75th percentile) → additional short-term financing growth of 1.9, 0.9, and 1.0 percentage points for BOR, FIN, MPI respectively.
- More profitable SMEs experience relatively higher long-term and overall funding growth after one additional borrower-targeted instrument.
- More profitable young firms: BOR and FIN associated with higher short-term financing growth; BOR associated with higher overall financing growth.

Interest coverage interactions (Table 8)
- Interest coverage < 1 denotes potential financial distress.
- Micro firms with interest coverage < 1:
  - BOR, FIN, and MPI associated with lower short-term financing growth: 2.5, 3, and 2.7 percentage points lower short-term financing growth, respectively, relative to firms not in financial distress, following implementation of one additional instrument.
  - BOR also reduces long-term financing and overall financing growth for distressed micro firms.
- SMEs in financial distress experience lower long-term and overall funding growth with higher BOR.
- Young firms with interest coverage < 1:
  - BOR associated with a drop in long-term financing and overall financing growth.
  - FIN weakly negatively associated with overall financing growth for young distressed firms.
- Overall interpretation: Among MSME and young firms, the weakest firms—those with high leverage, low profitability, and interest coverage ratios below one—experience drops in credit growth as macroprudential policies are implemented. These effects can be viewed as unintended distributional effects or consistent with the stability objective of restricting credit to risky firms.

### Real effects on investment and sales (Tables 9–10)
- Regressions of firm investment and sales growth on macroprudential policies (firm-fixed effects).
- BOR is the only macroprudential coefficient entering significantly in investment and sales growth regressions.
  - Applying one additional borrower-related tool is associated with a 4.4 percentage point lower investment growth.
  - Applying one additional borrower-related tool is associated with a 3.5 percentage point lower sales growth.
- Differential effects (Table 10):
  - MSMEs: BOR implementation associated with statistically and economically significant lower relative investment and sales growth.
  - Microenterprises and SMEs: Both experience relatively lower investment and sales growth after BOR implementation; no significant effect for FIN.
  - Young firms: Interaction between young dummy and all three macroprudential indicators enters negatively and significantly — both BOR and FIN associated with relatively lower investment and sales growth of younger firms.
- Economic significance of investment/sales effects aligns with earlier findings on long-term financing growth for MSMEs and young firms.
- Overall conclusion: Negative association between macroprudential policies (especially borrower-targeted) and financing growth for MSMEs and young firms translates into lower investment and sales growth for these groups, underscoring financing constraints and bank dependency.

### Robustness and caveats
- Results robust to multiple checks: joint inclusion of BOR and FIN, splitting FIN into cyclical vs structural components, controlling for firm creditworthiness measures, interacting macro shocks with firm size/age, adding industry-year fixed effects.
- LTV intensity analysis limited to 16 countries due to data availability.
- Even where no significant reduction in measured financing growth is observed, macroprudential policies could still impose real effects via higher financing costs or stricter collateral requirements.

*Source: wp18267 - 3. Macroprudential policies and firm financing and growth.*

### 4. Conclusion

### 4. Conclusion

### Purpose and scope
- Examines the micro-evidence on the impact of macroprudential policies.
- Assesses effects on firms’ financing growth across a broad cross-section of firms and countries, differentiating by firm size and age, and by types of macroprudential policies.

### Key empirical findings
- The smallest firms (those with fewer than 10 employees) and youngest firms (less than or equal to three years since incorporation) are more likely to be affected by macroprudential policies.
- Borrower-targeted policies are more effective than policies targeted at financial institutions.
- Among MSME and young firms, the decline in credit growth is concentrated in the weakest firms:
  - Firms with high leverage,
  - Firms with low profitability,
  - Firms with interest coverage ratios below one.
- These distributional effects on credit are consistent with the stability objective of macroprudential policy tools.

### Real-economy effects
- Implementation of macroprudential policies produced real effects on economic activity:
  - MSMEs experience relatively lower investment and sales growth after implementation of borrower-targeted macroprudential policies.
  - Young firms have lower investment and sales growth after implementation of both borrower-targeted and financial institution-targeted policies.

*Source: wp18267 - 4. Conclusion*

### Appendix B. *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively.

### Appendix B. *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively.

### Panel summaries: Interactions with firm size and age (financing growth)
- Panel A: Interaction with MSME (N 564,858 / 557,964; # of countries 16)
  - MSME x Cumulative Intensity: 0.125 (Short-term financing growth)
  - MSME x Cumulative Intensity: -0.101** (Long-term financing growth)
  - MSME x Cumulative Intensity: -0.011 (Overall financing growth)
  - MSME x ∆Cumulative Intensity: 0.042 (Short-term)
  - MSME x ∆Cumulative Intensity: -0.062* (Long-term)
  - MSME x ∆Cumulative Intensity: -0.009 (Overall)
  - Adj. R-sq: columns show 0.002, 0.003, 0.151, 0.156, 0.178, 0.180

- Panel B: Interaction with Micro and SME (N 564,858 / 557,964; # of countries 16)
  - Micro x Cumulative Intensity: 0.120 (Short-term)
  - Micro x Cumulative Intensity: -0.095** (Long-term)
  - Micro x Cumulative Intensity: -0.011 (Overall)
  - SME x Cumulative Intensity: 0.125 (Short-term)
  - SME x Cumulative Intensity: -0.102** (Long-term)
  - SME x Cumulative Intensity: -0.011 (Overall)
  - Micro x ∆Cumulative Intensity: 0.046 (Short-term)
  - Micro x ∆Cumulative Intensity: -0.052* (Long-term)
  - Micro x ∆Cumulative Intensity: 0.003 (Overall)
  - SME x ∆Cumulative Intensity: 0.041 (Short-term)
  - SME x ∆Cumulative Intensity: -0.064** (Long-term)
  - SME x ∆Cumulative Intensity: -0.010 (Overall)
  - Adj. R-sq: 0.002, 0.003, 0.151, 0.156, 0.178, 0.181

- Panel C: Interaction with Age (Young firms ≤ 3 years; N 607,083 / 598,082; # of countries 16)
  - Young x Cumulative Intensity: -0.036 (Short-term)
  - Young x Cumulative Intensity: 0.018 (Long-term)
  - Young x Cumulative Intensity: 0.002 (Overall)
  - Young x ∆Cumulative Intensity: -0.052 (Short-term)
  - Young x ∆Cumulative Intensity: 0.032 (Long-term)
  - Young x ∆Cumulative Intensity: -0.004 (Overall)
  - Adj. R-sq: -0.006, -0.005, 0.146, 0.150, 0.177, 0.179

### Table 6: Financing growth and macroprudential policies — interactions with leverage
- Specification: Financing growthijt = β1 Macroprumt-1 * Firm Leverageit + β2 Firm Leverageit + β3 Firm Assetsit + μjt + ηi + εijt. Macropru indicators: MPI, BOR, FIN. Firm leverage = debt/assets.
- Panel A: Micro sub-sample (N 1,220,523; # of countries 39)
  - Leverage (Short-term): -0.829*** (s.e. 0.128)
  - Leverage (Long-term): -0.567*** (s.e. 0.059)
  - Leverage (Overall): -0.779*** (s.e. 0.054)
  - MPI x Leverage: -0.160** (Short-term) ; 0.043 (Long-term) ; -0.058* (Overall)
  - BOR x Leverage: -0.127** (Short-term) ; -0.015 (Long-term) ; -0.070* (Overall)
  - FIN x Leverage: -0.191** (Short-term) ; 0.066 (Long-term) ; -0.061 (Overall)
  - Adj. R-sq: Short-term 0.024 (cols 1–3), Long-term 0.153 (cols 4–6), Overall 0.247 (cols 7–9)

- Panel B: SME sub-sample (N 1,292,988; # of countries 43)
  - Leverage (Short-term): -1.068*** (s.e. 0.090)
  - Leverage (Long-term): -0.502*** (s.e. 0.048)
  - Leverage (Overall): -0.907*** (s.e. 0.057)
  - MPI x Leverage: -0.103* (Short-term) ; -0.118*** (Long-term) ; -0.115*** (Overall)
  - BOR x Leverage: -0.217** (Short-term) ; -0.087 (Long-term) ; -0.127* (Overall)
  - FIN x Leverage: -0.087 (Short-term) ; -0.142*** (Long-term) ; -0.128*** (Overall)
  - Adj. R-sq: Short-term 0.114, Long-term 0.189, Overall 0.298

- Panel C: Young sub-sample (N 808,648; # of countries 45)
  - Leverage (Short-term): -0.982*** (s.e. 0.106)
  - Leverage (Long-term): -0.570*** (s.e. 0.051)
  - Leverage (Overall): -0.855*** (s.e. 0.055)
  - MPI x Leverage: -0.042 (Short-term) ; 0.010 (Long-term) ; -0.055 (Overall)
  - BOR x Leverage: -0.049 (Short-term) ; 0.018 (Long-term) ; -0.064 (Overall)
  - FIN x Leverage: -0.054 (Short-term) ; 0.010 (Long-term) ; -0.071 (Overall)
  - Adj. R-sq: Short-term 0.069, Long-term 0.187, Overall 0.303

### Table 7: Financing growth and macroprudential policies — interactions with profitability (ROA)
- Specification: Financing growthijt = β1 Macroprumt-1 * Firm ROAit + β2 Firm ROAit + β3 Firm Assetsit + μjt + ηi + εijt. ROA = return on assets.
- Panel A: Micro sub-sample (N 1,341,249; # of countries 39)
  - ROA (Short-term): 0.137*** (s.e. 0.047)
  - ROA (Long-term): 0.120*** (s.e. 0.027)
  - ROA (Overall): 0.139*** (s.e. 0.023)
  - MPI x ROA: 0.127*** (Short-term) ; -0.019 (Long-term) ; 0.027 (Overall)
  - BOR x ROA: 0.235** (Short-term) ; 0.072** (Long-term) ; 0.167*** (Overall)
  - FIN x ROA: 0.113** (Short-term) ; -0.036 (Long-term) ; 0.004 (Overall)
  - Adj. R-sq: Short-term -0.001, Long-term 0.142, Overall 0.191

- Panel B: SME sub-sample (N 1,338,679; # of countries 42)
  - ROA (Short-term): 0.350*** (s.e. 0.074)
  - ROA (Long-term): 0.123** (s.e. 0.055)
  - ROA (Overall): 0.239*** (s.e. 0.045)
  - MPI x ROA: 0.015 (Short-term) ; -0.009 (Long-term) ; 0.019 (Overall)
  - BOR x ROA: 0.175 (Short-term) ; 0.188* (Long-term) ; 0.168** (Overall)
  - FIN x ROA: -0.013 (Short-term) ; -0.044 (Long-term) ; -0.006 (Overall)
  - Adj. R-sq: Short-term 0.094, Long-term 0.190, Overall 0.232

- Panel C: Young sub-sample (N 888,460; # of countries 44)
  - ROA (Short-term): 0.159*** (s.e. 0.046)
  - ROA (Long-term): 0.129*** (s.e. 0.035)
  - ROA (Overall): 0.141*** (s.e. 0.026)
  - MPI x ROA: 0.079** (Short-term) ; -0.006 (Long-term) ; 0.038 (Overall)
  - BOR x ROA: 0.142** (Short-term) ; 0.066 (Long-term) ; 0.126*** (Overall)
  - FIN x ROA: 0.080** (Short-term) ; -0.027 (Long-term) ; 0.022 (Overall)
  - Adj. R-sq: Short-term 0.039, Long-term 0.193, Overall 0.245

### Table 8: Financing growth and macroprudential policies — interactions with interest coverage
- Specification: Financing growthijt = α1 + β1 Macroprumt-1 * Interest coverageit + β2 Interest coverageit + β3 Firm Assetsit + μjt + ηi + εijt. Interest Coverage dummy = 1 if interest coverage < 1.
- Panel A: Micro sub-sample (N 1,040,466; # of countries 27)
  - Interest coverage coefficients: -0.015 (Short-term); -0.027*** (Long-term); -0.034*** (Overall) across columns
  - MPI x Interest coverage: -0.027** (Short-term) ; -0.002 (Long-term) ; -0.008** (Overall)
  - BOR x Interest coverage: -0.025* (Short-term) ; -0.019** (Long-term) ; -0.018** (Overall)
  - FIN x Interest coverage: -0.030** (Short-term) ; 0.003 (Long-term) ; -0.005 (Overall)
  - Adj. R-sq: Short-term -0.006, Long-term 0.139, Overall 0.165

- Panel B: SME sub-sample (N 1,219,311; # of countries 31)
  - Interest Coverage: -0.063*** (Short-term); -0.018 (Long-term); -0.031*** (Overall)
  - MPI x Interest coverage: -0.012 (Short-term) ; -0.011* (Long-term) ; -0.015* (Overall)
  - BOR x Interest coverage: -0.007 (Short-term) ; -0.052* (Long-term) ; -0.033** (Overall)
  - FIN x Interest coverage: -0.016 (Short-term) ; 0.004 (Long-term) ; -0.009 (Overall)
  - Adj. R-sq: Short-term 0.078, Long-term 0.168, Overall 0.221

- Panel C: Young sub-sample (N 704,860; # of countries 33)
  - Interest coverage: -0.037*** (Short-term); -0.027*** (Long-term); -0.028*** (Overall)
  - MPI x Interest coverage: -0.011 (Short-term) ; -0.013** (Long-term) ; -0.017*** (Overall)
  - BOR x Interest coverage: -0.014 (Short-term) ; -0.024** (Long-term) ; -0.027** (Overall)
  - FIN x Interest coverage: -0.012 (Short-term) ; -0.013 (Long-term) ; -0.018* (Overall)
  - Adj. R-sq: Short-term 0.032, Long-term 0.175, Overall 0.241

### Table 9: Real effects of macroprudential policies (firm growth: Investment and Sales Growth)
- Specification: Firm growthijt = β1 Macroprumt-1 + β2 Firm Assetsit + β3 Macrojt-1 + β4 GFCt + ηi + εijt.
- Sample: N 19,876,932; # of countries 48
- Key coefficients (Investment columns 1–3; Sales Growth columns 4–6):
  - Firm assets: -0.127***; -0.125***; -0.130*** (Investment columns) ; -0.136***; -0.133***; -0.138*** (Sales columns) (s.e. 0.008)
  - GDP growth: 0.006***; 0.006***; 0.006*** (Investment) ; 0.007***; 0.007***; 0.007*** (Sales) (s.e. 0.002)
  - Real policy rate: -0.002; -0.002; -0.002 (Investment) ; 0.001; 0.001; 0.002 (Sales) (s.e. 0.002)
  - GFC: -0.049***; -0.049***; -0.049*** (Investment) ; -0.112***; -0.112***; -0.112*** (Sales) (s.e. 0.008–0.010)
  - MPI: -0.010 (Investment) ; -0.001 (Sales) (s.e. 0.013; 0.014)
  - BOR: -0.044*** (Investment) ; -0.035** (Sales) (s.e. 0.015; 0.013)
  - FIN: 0.005 (Investment) ; 0.016 (Sales) (s.e. 0.016)
  - Adj. R-sq: Investment 0.212–0.213; Sales 0.178–0.179

### Table 10: Real effects of macroprudential policies — allow for firm size/age heterogeneity
- Specification: Firm growthijt = β1 Macroprumt-1 * Firm Characteristici + β2 Firm Assetsit + μjt + ηi + εijt.
- Panel A: Interaction with MSME (N 17,128,303; # of countries 48)
  - MSME x MPI: -0.016 (Investment) ; -0.014 (Sales) (s.e. 0.014; 0.015)
  - MSME x BOR: -0.074*** (Investment) ; -0.078*** (Sales) (s.e. 0.016)
  - MSME x FIN: -0.000 (Investment) ; 0.003 (Sales) (s.e. 0.001; 0.002)
  - Adj. R-sq: Investment 0.250–0.251; Sales 0.266

- Panel B: Interaction with Micro and SME (N 17,128,303; # of countries 48)
  - Micro x MPI: -0.016 (Investment) ; -0.015 (Sales) (s.e. 0.012; 0.015)
  - SME x MPI: -0.016 (Investment) ; -0.014 (Sales) (s.e. 0.014; 0.015)
  - Micro x BOR: -0.070*** (Investment) ; -0.078*** (Sales) (s.e. 0.016; 0.015)
  - SME x BOR: -0.074*** (Investment) ; -0.078*** (Sales) (s.e. 0.016)
  - Micro x FIN: -0.004 (Investment) ; 0.002 (Sales) (s.e. 0.005; 0.003)
  - SME x FIN: 0.000 (Investment) ; 0.003 (Sales) (s.e. 0.001; 0.002)
  - Adj. R-sq: Investment 0.250–0.251; Sales 0.266

- Panel C: Interaction with Age (Young ≤ 3 years; N 19,781,843; # of countries 48)
  - Young x MPI: -0.011*** (Investment) ; -0.022*** (Sales) (s.e. 0.003; 0.005)
  - Young x BOR: -0.019*** (Investment) ; -0.040*** (Sales) (s.e. 0.007; 0.005)
  - Young x FIN: -0.010*** (Investment) ; -0.021** (Sales) (s.e. 0.003; 0.008)
  - Adj. R-sq: Investment 0.262; Sales 0.269–0.268

### Appendix A: Country and firm coverage (Orbis sample totals)
- Total # Observations: 3,107,242
- Total # Firms: 898,653
- Country-level examples (MPI / FIN / BOR increase/decrease years reported where applicable):
  - Bulgaria: # Observations 10,739; # Firms 3,762; MPI Increase 2005, 2006, 2007; MPI Decrease 2008; FIN Increase 2005, 2007; FIN Decrease 2008; BOR Increase 2006
  - Canada*: # Observations 1,418; # Firms 382; MPI Increase 2008; FIN Increase 2008
  - Colombia*: # Observations 1,859; # Firms 583; MPI Increase 2007; BOR Increase 2007
  - France: # Observations 659,089; # Firms 224,786; MPI Increase 2011; FIN Increase 2011
  - Germany: # Observations 66,755; # Firms 20,000; MPI Increase 2010; FIN Increase 2010
  - Hungary: # Observations 14,289; # Firms 4,064; MPI Increase 2010, 2011; FIN Increase 2010
  - Republic of Korea*: # Observations 111,647; # Firms 28,727; MPI Increase 2005, 2007, 2011; MPI Decrease 2007, 2011; FIN Increase 2005
  - Turkey: # Observations 4,251; # Firms 1,581; MPI Increase 2007, 2009, 2010, 2011; MPI Decrease 2009, 2010; FIN Increase 2007, 2011
  - United States of America: # Observations 20,393; # Firms 6,935
- Several countries are marked with an asterisk in the original listing.

*Source: Appendix B and Appendix A tables in wp18267 - Appendix B. *, **, and *** represent significance at 10%, 5%, and 1% levels, respectively.*

### Appendix B. Variable definitions and sources

### Appendix B. Variable definitions and sources

### Variable definitions and data sources
- Short-term financing: Log change in short-term debt (with maturity less or equal than a year) — ORBIS
- Long-term financing: Log change in long-term debt (with maturity greater than a year) — ORBIS
- Overall financing growth: Log change in total financing (defined as the sum of short- and long-term debt) — ORBIS
- Investment: Log change in fixed assets — ORBIS
- Sales growth: Growth in turnover computed as the log change in operating turnover — ORBIS
- Firm Assets: Log of total assets — ORBIS
- Leverage: Ratio of loans and long-term debt to total assets — ORBIS
- ROA: Return on assets defined as the EBIT to total assets ratio — ORBIS
- Interest Coverage (<1): Dummy taking value 1 for companies with <1 interest coverage (defined as the ratio of EBIT to interest expense) — ORBIS
- MSME: Dummy taking value 1 for micro, small, and medium companies (those with less than 250 employees) and 0 otherwise — ORBIS
- Young: Dummy taking value 1 for companies that are <=3 years of age when they appear in our sample — ORBIS
- MPI: Macroprudential Index (0-12) = LTV_CAP + DTI + DP + CTC + LEV + SIFI + INTER + CONC + FC + RR_REV + CG + TAX — Cerutti, Claessens, and Laeven (2015)
- BOR: Borrower-Targeted Instruments (0-2) = LTV_CAP + DTI — Cerutti, Claessens, and Laeven (2015)
- FIN: Financial Institution-Targeted Instruments (0-10) = DP + CTC + LEV + SIFI + INTER + CONC + FC + RR_REV + CG + TAX — Cerutti, Claessens, and Laeven (2015)
- LTV: Loan-to-Value Ratio — Cerutti, Claessens, and Laeven (2015)
- DTI: Debt-to-Income Ratio — Cerutti, Claessens, and Laeven (2015)
- DP: Time-Varying/Dynamic Loan-Loss Provisioning — Cerutti, Claessens, and Laeven (2015)
- CTC: General Countercyclical Capital Buffer/Requirement — Cerutti, Claessens, and Laeven (2015)
- LEV: Leverage Ratio — Cerutti, Claessens, and Laeven (2015)
- SIFI: Capital Surcharges on SIFIs — Cerutti, Claessens, and Laeven (2015)
- INTER: Limits on Interbank Exposures — Cerutti, Claessens, and Laeven (2015)
- CONC: Concentration Limits — Cerutti, Claessens, and Laeven (2015)
- FC: Limits on Foreign Currency Loans — Cerutti, Claessens, and Laeven (2015)
- RR: Reserve Requirement Ratios — Cerutti, Claessens, and Laeven (2015)
- CG: Limits on Domestic Currency Loans — Cerutti, Claessens, and Laeven (2015)
- TAX: Levy/Tax on Financial Institutions — Cerutti, Claessens, and Laeven (2015)
- LTV_CAP: Loan-to-Value Ratio Caps — Cerutti, Claessens, and Laeven (2015)
- RR_REV: FX and/or Countercyclical Reserve Requirements (specific survey-based definition) — Cerutti, Correa, Fiorentino and Segalla (2017)
- GDP growth: GDP growth rate (annual %) — World Development Indicators
- Real policy rate: Real monetary policy rate (%), defined as the discount rate minus the inflation rate. IFS Central Bank Policy Rate when available, otherwise Discount Rate or Repurchase Agreement Rate. ECB deposit facility rate for Eurozone countries.
- GFC: Global Financial Crisis dummy – takes value 1 for years 2008 and 2009 and 0 otherwise.
- Cumulative Intensity: Cumulative change in the loan-to-value ratio cap from 2000-Q1, computed Q4 every year. Missing if the instrument is not available in the country — Cerutti, Correa, Fiorentino and Segalla (2017)
- ∆Cumulative Intensity: Year-to-year change in the Q4 values of cumulative intensity — Cerutti, Correa, Fiorentino and Segalla (2017)

### Regression specifications (key equations and estimation)
- Baseline specification (Appendix C): Financing growthijt = β1 Macroprujt-1 + β2 Firm Assetsit + β3 Macrojt-1 + β4 GFCt + ηi + εijt.
  - Dependent variables: Short-term financing growth; Long-term financing growth; Overall financing growth.
  - Macropru: MPI, BOR, or FIN.
  - Controls: Firm Assets (log total assets); macro vector including real monetary policy rate and log change of GDP; GFC dummy for 2008 and 2009.
  - Estimation: OLS weighted by number of observations in each country; standard errors clustered at country level; firm fixed effects.

- Specification controlling for firm creditworthiness (Appendix D): Financing growthijt = β1 Macroprujt-1 + β2 Firm Assetsit + β3 Firm Creditworthinessjt-1 + β4 Macrojt-1 + β5 GFCt + ηi + εijt.
  - Firm Creditworthiness: Leverage, ROA, Interest coverage (<1).

- Heterogeneity/specifications with industry-year fixed effects and interactions (Appendix E and F):
  - Financing growthijt = β1 Macroprujt-1 * Firm Characteristici + β2 Firm Assetsit + μjt + δkt + ηi + εijt (Appendix E).
  - Financing growthijt = β1 Macroprujt-1 * Firm Characteristici + β2 Firm Assetsit + β3 Firm Creditworthinessit + μjt + ηi + εijt (Appendix F).
  - Firm characteristics: MSME, Micro, SME, Young (<=3 years).
  - Fixed effects: firm; country x year; industry x year (in many specifications).

### Key regression estimates and coefficients (selected)
- Appendix C — Baseline (coefficients reported for three dependent variables: Short-term, Long-term, Overall financing growth; N = 3,107,242; # of countries = 48)
  - Firm assets: -0.132*** (0.013); -0.144*** (0.011); -0.147*** (0.010)
  - GDP growth: 0.007*** (0.002); 0.007*** (0.001); 0.006*** (0.001)
  - Real policy rate: -0.000 (0.002); -0.001 (0.002); -0.001 (0.002)
  - GFC: -0.046*** (0.010); -0.039*** (0.011); -0.045*** (0.009)
  - BOR: -0.016 (0.020); -0.046** (0.018); -0.022 (0.015)
  - FIN: 0.001 (0.017); -0.009 (0.019); -0.004 (0.014)
  - Adj. R-sq: 0.000; 0.122; 0.144

- Appendix D — Controlling for firm creditworthiness (N = 2,287,667; # of countries = 36)
  - Short-term financing growth (columns with MPI, BOR, FIN)
    - Firm assets: -0.168*** (0.022); -0.170*** (0.020); -0.168*** (0.022)
    - Leverage: -1.039*** (0.076); -1.038*** (0.076); -1.039*** (0.076)
    - ROA: 0.111* (0.055); 0.112** (0.054); 0.111** (0.055)
    - Interest coverage: -0.038 (0.023) across columns
    - GDP growth: 0.006*** (0.002) across columns
    - GFC: -0.050*** (0.009) across columns
    - MPI: -0.003 (0.010) in MPI specification
    - BOR: 0.003 (0.023) in BOR specification
    - FIN: -0.006 (0.014) in FIN specification
    - Adj. R-sq: 0.014

  - Long-term financing growth
    - Firm assets: -0.168*** (0.011); -0.181*** (0.010); -0.181*** (0.011)
    - Leverage: -0.546*** (0.058) across columns
    - ROA: 0.043 (0.040); 0.040 (0.040); 0.046 (0.040)
    - Interest coverage: -0.026*** (0.009); -0.026** (0.010); -0.026*** (0.009)
    - MPI: -0.019 (0.015)
    - BOR: -0.040* (0.020)
    - FIN: -0.010 (0.022)
    - Adj. R-sq: 0.121

  - Overall financing growth
    - Firm assets: -0.186*** (0.016); -0.186*** (0.015); -0.187*** (0.016)
    - Leverage: -0.931*** (0.054) across columns
    - ROA: 0.072* (0.041); 0.071* (0.041); 0.073* (0.041)
    - Interest coverage: -0.020** (0.008); -0.021** (0.009); -0.021** (0.008)
    - MPI: -0.006 (0.009)
    - BOR: -0.012 (0.016)
    - FIN: -0.003 (0.014)
    - Adj. R-sq: 0.200

### Heterogeneity and interaction results (selected findings)
- Appendix E — Interaction with MSME, Micro/SME, and Young (industry-year fixed effects; N varies; # of countries = 48)
  - Panel A (MSME interactions; N = 2,821,510)
    - MSME x MPI: 0.002 (0.017) for short-term; -0.010 (0.010) for long-term; -0.005 (0.006) for overall
    - MSME x BOR: -0.088* (0.044) for long-term; -0.021 (0.016) for long-term; -0.032*** (0.011) for overall (note columns vary by dependent variable)
    - MSME x FIN: 0.025*** (0.008) for short-term; -0.009 (0.013) for long-term; 0.001 (0.006) for overall
    - Adj. R-sq: 0.041; 0.170; 0.214 (by dependent variable groups)

  - Panel B (Micro and SME interactions; N = 2,821,510)
    - Micro x MPI: -0.052* (0.029) for short-term; -0.032** (0.013) for long-term; -0.031*** (0.009) for overall
    - SME x MPI: 0.008 (0.020) for short-term; -0.008 (0.011) for long-term; -0.002 (0.007) for overall
    - Micro x BOR: -0.096** (0.044) for long-term; -0.025 (0.016) for long-term; -0.044*** (0.010) for overall
    - SME x BOR: -0.087* (0.044) for long-term; -0.021 (0.016) for long-term; -0.031** (0.012) for overall
    - Micro x FIN: -0.072* (0.038) for short-term; -0.047*** (0.015) for long-term; -0.039*** (0.011) for overall
    - SME x FIN: 0.039*** (0.009) for short-term; -0.003 (0.013) for long-term; 0.006 (0.006) for overall
    - Adj. R-sq: 0.061; 0.182; 0.229 (by dependent variable groups)

  - Panel C (Young interactions; N = 3,093,661)
    - Young x MPI: -0.014 (0.013) for short-term; -0.025*** (0.008) for long-term; -0.006 (0.005) for overall
    - Young x BOR: -0.019** (0.008) for short-term; -0.021*** (0.007) for long-term; -0.012** (0.006) for overall
    - Young x FIN: -0.015 (0.030) for short-term; -0.041** (0.016) for long-term; -0.004 (0.008) for overall
    - Adj. R-sq: 0.050; 0.175; 0.224 (by dependent variable groups)

- Appendix F — Allowing for firm heterogeneity and controlling for firm creditworthiness (firm and country x year fixed effects; selected coefficients; N and # of countries vary)
  - Panel A (MSME interaction; N = 2,129,221; # of countries = 36)
    - Leverage: -1.202*** (0.105); -1.200*** (0.104); -1.201*** (0.105) for short-term; -0.651*** (0.040) for long-term; -1.050*** (0.054) for overall (consistently large negative coefficients)
    - ROA: 0.125 (0.091) to 0.126 (0.092) for short-term; negative coefficients for long-term in some columns (e.g., -0.084 (0.066))
    - Interest Coverage: 0.015 (0.048) for short-term; -0.021* (0.011) for long-term; -0.014 (0.009) for overall
    - MSME x MPI: -0.004 (0.021) for short-term; -0.012 (0.018) for long-term; -0.004 (0.009) for overall
    - MSME x BOR: -0.062 (0.041) for short-term; -0.035** (0.017) for long-term; -0.038** (0.014) for overall
    - MSME x FIN: 0.012 (0.027) for short-term; -0.008 (0.022) for long-term; 0.006 (0.005) for overall
    - Adj. R-sq: 0.062; 0.174; 0.278

  - Panel B (Micro/SME interactions; selected)
    - Micro x MPI: -0.086 (0.060) to -0.031** (0.012) to -0.034*** (0.009)
    - SME x MPI: 0.004 (0.020) to -0.011 (0.019) to -0.000 (0.011)
    - Micro x BOR: -0.070* (0.041) to -0.042** (0.018) to -0.055*** (0.015)
    - SME x BOR: -0.062 (0.041) to -0.035** (0.017) to -0.037** (0.014)
    - Micro x FIN: -0.151** (0.073) for short-term; -0.041*** (0.013) for long-term; -0.044*** (0.008) for overall
    - SME x FIN: 0.031 (0.021) to -0.004 (0.023) to 0.012* (0.007)
    - Adj. R-sq: 0.063; 0.175; 0.279

  - Panel C (Young interactions; N = 2,276,358)
    - Leverage: -1.098*** (0.094) to -0.618*** (0.039) to -1.007*** (0.051)
    - Interest Coverage: -0.023* (0.012) for short-term; -0.014* (0.007) for long-term; -0.015** (0.007) for overall
    - Young x MPI: -0.016 (0.010) for short-term; -0.021*** (0.005) for long-term; -0.009 (0.005) for overall
    - Young x BOR: -0.028** (0.013) for short-term; -0.026*** (0.009) for long-term; -0.018** (0.007) for overall
    - Young x FIN: -0.013 (0.024) for short-term; -0.028*** (0.009) for long-term; -0.004 (0.010) for overall
    - Adj. R-sq: 0.059; 0.176; 0.284

### Substantive patterns from the appended regressions
- Firm size (Firm assets) is consistently negatively associated with financing growth across short-term, long-term, and overall measures (negative and statistically significant coefficients in multiple specifications).
- Firm leverage is strongly and negatively associated with financing growth across specifications (e.g., Leverage coefficients around -1.039***, -1.202***, -1.098*** depending on sample and specification).
- GDP growth is positively associated with financing growth (e.g., 0.007***, 0.006***).
- The Global Financial Crisis dummy (GFC) is associated with lower financing growth (e.g., -0.046***, -0.050***, -0.039***).
- Macroprudential indices show heterogeneous effects:
  - BOR (borrower-targeted measures) shows some negative and sometimes statistically significant associations with long-term financing growth (e.g., -0.046** in Appendix C; -0.040* or -0.021*** in interaction tables).
  - FIN (financial institution-targeted measures) often shows small or insignificant coefficients, though some negative interactions appear for Micro firms (e.g., Micro x FIN: -0.151** in Appendix F Panel B).
  - MPI coefficients are generally small and often not statistically significant in the reported specifications.

*Source: Appendix B and subsequent regression appendices (C–F) in the supplied content unit.*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2018/wp18267.pdf_