## _wp05167

## Source details

**Canonical URL:** [_wp05167](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05167.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05167.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05167.pdf.json)

---

### Overview and research focus
- Sample: lower-income countries (World Bank categories: low-income and lower-middle-income); total group 89 countries; regression samples smaller and vary by data availability.
- Performance measures:
  - Depth: deposit-to-GDP ratio; private sector credit-to-GDP ratio; loan-to-asset ratio.
  - Cost-efficiency: overhead costs to total assets (OH); net interest margin to interest-earning assets (NIM).
- Exclusions/limitations:
  - Access (especially by the poor): no cross-country database available.
  - Long-term lending: insufficient cross-country data.
  - Fragility/crises: poor cross-country indicators; inclusion would confound explanatory variables.

### Methodology and empirical approach
- Empirical strategy:
  - Bivariate correlations between explanatory variables and performance indicators.
  - Multivariate OLS regressions with robust standard errors and sequential inclusion of regressors.
  - Dependent variables measured as averages over 1999-2001; many RHS variables measured as averages over 1991-98 where possible.
  - Regressions replicate with GDP per capita in 1970 and excluding countries with fewer than five Bankscope banks for robustness.
  - Settlers’ mortality included in alternate smaller-sample specifications (available for 52 countries).
- Endogeneity guidance:
  - Use of long-term averages for RHS variables where possible.
  - Paper emphasizes correlations, not causation.
- Sample exclusions in regressions: China, Jordan, and Eritrea (outliers for depth variables); very small countries (population < one million).

### Data sources and dependent-variable summary statistics
- Data sources:
  - Depth indicators: IMF IFS.
  - Efficiency measures: bank-level data from Fitch’s Bankscope (country aggregates).
- Number of observations:
  - Deposits/GDP 85; Private Credit/GDP 85; Loans/Assets 81; Net Interest Margin 81; Overhead 81.
- Summary statistics (means, Sd, Min, Max):
  - Deposits/GDP: Mean 21.73; Sd 14.00; Min 2.24; Max 75.57.
  - Private Credit/GDP: Mean 18.09; Sd 15.71; Min 0.86; Max 93.26.
  - Loans/Assets: Mean 44.47; Sd 15.41; Min 8.73; Max 76.53.
  - Net Interest Margin: Mean 6.56; Sd 3.72; Min 0.10; Max 18.19.
  - Overhead: Mean 5.26; Sd 2.60; Min 0.58; Max 14.47.
- Cross-correlations:
  - Deposits and Private Credit correlation: 0.82*** (p-value 0.00).
  - Deposits and NIM correlation: -0.43*** (p-value 0.00).
  - Overhead and NIM correlation: 0.74*** (p-value 0.00).

### Explanatory variables (groups and examples)
- Geography / endowments / legal origin:
  - Rural population density; latitude; settlers’ mortality; ethnic fractionalization; French legal origin dummy; transition (Soviet) dummy.
- Political environment:
  - Political stability; internal conflict; military control of government; freedom from corruption.
- Macroeconomic:
  - Inflation (threshold effects at 15 percent); interest payments on government debt (proxy for fiscal balance); migrant remittances.
- Bank ownership and market structure:
  - Share of state banks (market share of state bank assets); share of foreign banks; concentration (market share of top five banks).
- Investment climate:
  - Time to enforce contracts; availability of credit information; credit registry coverage; costs of collateral and debt recovery.
- Supervisory/regulatory indices (1999 World Bank survey): restrictions on scope, disclosure requirements, supervisory disciplinary powers, accounting standards, auditing requirements; deposit insurance dummy. Indices vary between zero and one.

### Regional patterns and bivariate correlations
- Regional composition of sample:
  - Sub-Saharan Africa 42 percent; Europe and Central Asia (transition) 18 percent; East and South Asia 16 percent; Latin America and the Caribbean 15 percent; Middle East and North Africa (MENA) 9 percent.
- Regional findings (averages):
  - South and East Asia and MENA: relatively more financially developed (deeper and more efficient).
  - Latin America: intermediate depth; larger share of bank assets to private sector.
  - Sub-Saharan Africa and transition economies: least financially developed regions.
  - State banks: largest share of bank assets in MENA and Asia.
  - Foreign banks: largest presence in Sub-Saharan Africa; smallest in MENA and Latin America.
  - Quality of borrower information best in Latin America.
  - Political risk and internal conflict highest in Sub-Saharan Africa and transition countries.
- Selected bivariate correlations:
  - Higher income per capita → deeper financial sector and lower NIM.
  - Political stability and lower inflation → greater depth and efficiency.
  - Less concentrated systems → deeper and allocate larger share to private credit.
  - Larger share of state-owned banks → more deposits and apparently more efficiency but lower loan-to-asset ratio.
  - Greater foreign bank presence → shallower banking systems (deposits and private credit).
  - Credit information sharing and registry coverage → positively correlated with depth and lower NIM.
  - Regulation and supervision variables → not significantly correlated with cross-sectional financial depth in bivariate analysis.

### Main multivariate findings — Financial depth
- Controls: GDP per capita and transition dummy included in each depth regression.
- Geography and institutions:
  - Rural population density: sparser rural population → shallower banking sector; effect pronounced for deposits.
  - Settlers’ mortality: negative and strongly significant for deposits and private credit when included (available for 52 countries); inclusion renders rural density insignificant.
  - Latitude and ethnic fractionalization: generally not significant.
  - French legal origin: negative coefficient for deposits and private credit but not statistically significant; marginally significant positive coefficient for loan-to-asset ratio in some specifications.
- Political environment:
  - Political instability and internal conflict → shallower financial systems.
  - Corruption: strongly negatively associated with deposits and private credit; controlling for corruption weakens political instability significance.
- Macroeconomic:
  - Inflation: negative and significant correlation with private credit and loan-to-asset ratio; threshold effects with much larger disruptions when inflation > 15 percent.
  - Interest on public debt (fiscal proxy): positive sign but not consistently significant.
  - Migrant remittances: positive sign but not statistically significant.
- Bank ownership and market structure:
  - State banks: positively and significantly correlated with bank deposits; negatively but insignificantly correlated with private credit and loan-to-asset ratio.
  - Foreign banks: larger foreign bank presence associated with less deposits and less private sector credit; negative coefficients robust in multivariate regressions.
  - Concentration (top-five share): negatively correlated with private credit and loan-to-asset ratio in some specifications; sensitivity to concentration measure noted.
- Business environment:
  - Time to enforce contracts, availability of credit information, and credit registry coverage have explanatory power for private credit and loan-to-asset ratio even after controls.
  - Other business environment measures (costs of collateral, starting/closing business) not significantly correlated with depth.
- Regulation and supervision:
  - Once other factors controlled, regulation and supervision indices add little explanatory power for depth; in some cases more stringent supervision/auditing associated with reduced depth.
  - Explicit deposit insurance: coefficient negative and marginally significant for deposits.

### Main multivariate findings — Financial sector efficiency (OH and NIM)
- Size and geography:
  - Bank efficiency higher in larger economies (log GDP included; GDP per capita and transition dummy not significant for efficiency).
  - Geographic and institutional characteristics generally limited explanatory power; settlers’ mortality only marginally significant.
- Political environment:
  - Corruption, political instability, and political risk → higher NIM and higher OH (worse efficiency).
- Macroeconomic:
  - Inflation → higher NIM and higher OH; marginal effect more marked at lower inflation rates (contrast with depth where major disruptions appear when inflation > 15 percent).
- Bank ownership and market structure:
  - Foreign bank penetration: no significant relationship with efficiency.
  - State bank presence: larger presence associated with significantly lower OH (more cost-efficient) — robust across specifications.
  - Concentration: in LIC sample, more concentrated markets → lower OH and lower NIM (more efficient), even after controls.
- Business environment and regulation:
  - No robust link between efficiency and business environment indicators.
  - Regulation/supervision generally enter with positive signs (tighter regulation → higher costs), but coefficients not robust across specifications.

### Quantified magnitudes (comparative statics from final specifications)
- Method: predicted changes from a one standard deviation increase in each explanatory variable using final specifications (Table 5 for depth, Table 11 for efficiency); observations: deposits_gdp 56; loans_gdp 56; loans_assets 56; OH 62; NIM 62; R-squared: deposits_gdp 0.62; loans_gdp 0.73; loans_assets 0.58; OH 0.41; NIM 0.37.
- Selected effects (one standard deviation change → implied effect):
  - Corruption (S.D. 0.42):
    - deposits_gdp effect 6.58.
    - loans_gdp effect 8.09.
    - loans_assets effect 4.35.
    - OH effect -0.05.
    - NIM effect 0.11.
  - Inflation (S.D. 1.10):
    - deposits_gdp effect -3.74.
    - loans_gdp effect -6.31.
    - loans_assets effect -9.88.
    - OH effect 0.79.
    - NIM effect 1.12.
  - Interest on public debt (S.D. 2.45):
    - deposits_gdp effect 3.31.
    - loans_gdp effect 2.23.
    - loans_assets effect -1.25.
  - State banks (S.D. 0.31):
    - deposits_gdp effect 2.10.
    - loans_gdp effect -1.45.
    - loans_assets effect -5.31.
    - OH effect -0.68.
    - NIM effect -0.31.
  - Foreign banks (S.D. 0.30):
    - deposits_gdp effect -5.29.
    - loans_gdp effect -5.10.
    - loans_assets effect -1.93.
    - OH effect 0.44.
    - NIM effect 0.24.
  - Concentration (S.D. 0.24; one S.D. ≈ 20 percentage points in top-five share):
    - deposits_gdp effect 0.61.
    - loans_gdp effect -2.26.
    - loans_assets effect -7.44.
    - OH effect -0.91.
    - NIM effect -1.83.
  - Credit registry (S.D. 1.84):
    - deposits_gdp effect 0.09.
    - loans_gdp effect 0.22.
    - loans_assets effect 0.20.

### Selected regression coefficients and significance highlights (Table excerpts)
- Corruption (depth regressions):
  - deposits_gdp coefficient 19.26 (t-statistic [3.61]***).
  - loans_gdp coefficient 20.6 (t-statistic [3.43]***).
- Inflation (depth regressions):
  - deposits_gdp coefficient -3.95 (t-statistic [1.73]*).
  - loans_gdp coefficient -4.73 (t-statistic [2.29]**).
  - loans_assets coefficient -6.4 (t-statistic [2.73]**).
- Foreign banks (depth regressions):
  - deposits_gdp coefficient -22.02 (t-statistic [3.12]***).
  - loans_gdp coefficient -24.39 (t-statistic [3.84]***).
- State banks (depth regressions):
  - deposits_gdp coefficient 1.55 (t-statistic [2.21]** in Panel A); alternative specification deposits_gdp coefficient 11.82 (t-statistic [3.18]***).
- Concentration (loans_assets example):
  - loans_assets coefficient -37.02 (t-statistic [3.20]***).
- Efficiency regressions (Table 11 examples):
  - Size (logGDP) NIM coefficient -0.584*** (t-statistic [2.86]).
  - Corruption OH coefficient -1.848*** (t-statistic [3.01]); NIM coefficient -1.674* (t-statistic [1.71]).
  - Inflation (log) NIM coefficient up to 1.015*** (t-statistic [4.21]).
  - State banks OH coefficient -2.700*** (t-statistic [3.80]).
  - Concentration OH coefficient -3.799*** (t-statistic [2.93]); Concentration NIM coefficient -7.661*** (t-statistic [3.63]).
- Supervision/regulation heterogeneous effects:
  - Example: Disclosure requirements OH coefficient 5.383** (t-statistic [2.58]) in Table 12, indicating disclosure requirements associated with higher OH in that specification.
- Significance notation:
  - * significant at 10%; ** significant at 5%; *** significant at 1%.

### Conclusions and policy implications
- Empirical summary:
  - High inflation, unstable and corrupt political environment, and high settlers’ mortality are associated with poorer financial performance in LICs.
  - Contract enforcement delays and limited creditor information availability associated with shallower credit markets.
  - French legal origin not associated with less private credit in LICs.
  - Larger share of state-owned banks in LICs associated with lower OH (more cost-efficient), greater deposit mobilization, but smaller share of credit allocated to the private sector.
  - Larger foreign bank presence robustly associated with shallower systems (less deposits and less private credit) in the LIC cross-section.
  - Regulatory and supervisory framework characteristics not significantly correlated with financial sector performance in LICs once other factors controlled for.
- Policy lessons (from cross-country regressions):
  - Reduce political instability and corruption to improve depth and efficiency.
  - Keep inflation under control to improve bank efficiency and development.
  - Strengthening prudential regulation and supervision may not yield immediate benefits in LICs; other binding obstacles or weak implementation may limit impact.
- Caveats and research needs:
  - Direction of causality ambiguous for several findings (notably state bank presence and foreign bank penetration); further targeted research and alternative methodologies required to identify causal effects.
  - Determinants of financial sector performance in LICs may differ from those in more advanced countries; exercise caution when extrapolating from advanced-country evidence.

*Source: Appendix Tables and Section V concludes of the PDF chapter _wp05167.*

### Appendix Tables

### Appendix Tables

### Overview and research focus
- The paper analyzes indicators of financial sector development and performance in lower-income countries (LICs), defined as those World Bank categories: low-income and lower-middle-income (footnote 2).
- Performance measures used:
  - Depth of the banking system: bank deposit generation and credit to the private sector.
  - Cost-efficiency measures: overhead costs and interest margins.
- Excluded or limited measures (due to data constraints):
  - Access (especially by the poor) — no cross-country database available.
  - Long-term lending — insufficient cross-country data.
  - Fragility — good cross-country indicators are difficult to find; past banking crises would confound explanatory variables (footnote 5).

### Methodology and data approach
- Explanatory variables considered:
  - Geographic characteristics (some used as proxies for institutional quality).
  - Legal/colonial origin.
  - Political factors.
  - Macroeconomic factors.
  - Banking system structure.
  - Regulatory and business environment features.
- Empirical strategy:
  - Start with bivariate correlations between explanatory variables and performance indicators.
  - Run multivariate specifications with gradually expanding sets of regressors.
  - Where possible, measure RHS variables as long-term averages of years preceding LHS variable measurement to mitigate endogeneity concerns; nevertheless, the paper does not claim causation, only robust correlations.

### Literature context and cited mechanisms
- Prior findings summarized:
  - Institutions are a key element in financial sector performance.
  - Institutional determinants traced to legal origin (La Porta and others, 1998), geographical conditions at colonization (Acemoglu, Johnson, and Robinson, 2001), cultural factors (Stulz and Williamson, 2003).
  - Roles identified for state banks (La Porta and others, 2002; Micco, Panizza, and Yañez, 2004), foreign banks (Claessens, Demirgüç-Kunt, Huizinga, 2001), and inflation (Boyd, Levine, and Smith, 2001).
  - Regulations restricting bank activities hinder performance; those encouraging private sector monitoring help (Barth, Caprio, and Levine, 2004).
  - Compliance with international regulatory/supervisory standards associated with healthier banking systems (Das, Quintyn, and Chenard, 2004; Podpiera, 2004).
  - Better creditor protection and information access increase credit to the private sector (Djankov, McLeish, and Shleifer, 2005).
- Noted heterogeneity in effects across country groups (footnote 3); effects found in broad samples may not hold when restricted to developing countries.

### Main empirical findings (summary)
- Regional differences:
  - LICs in South and East Asia and in the Middle East and North Africa (MENA) are relatively more financially developed.
  - Latin America is in the middle.
  - African countries and the transition economies are the least financially developed regions (this pattern holds even after controlling for per capita income) (paragraph summary).
- Correlates associated with shallower and more inefficient financial systems:
  - Corruption.
  - Inflation.
- Variables with limited or no significant associations in LICs:
  - Legal origin: no significant bearing on financial sector performance in LICs.
  - Regulatory and supervisory system characteristics (disclosure requirements, auditing requirements, supervisory powers to discipline banks): not significantly related to financial performance in LICs.
- Banking structure findings:
  - More foreign bank penetration is associated with a shallower financial sector and is not significantly associated with efficiency. Interpretation is ambiguous because foreign banks may enter "underbanked" markets or lend less to informationally opaque small businesses (citing Clarke and others, 2005).
  - Banking systems with more state-owned banks appear to be better at deposit mobilization and have lower overhead costs (consistent with regional patterns: South and East Asia and MENA have more efficient and deeper systems and have more state-owned banks and smaller foreign bank presence).

### Analytical caveats and interpretation guidance
- The paper emphasizes correlation rather than causation due to potential endogeneity of regressors and measurement limitations.
- Endogeneity mitigation:
  - Use of long-term averages for RHS variables when possible.
  - Acknowledges remaining endogeneity concerns and interprets findings as directions for future work rather than definitive causal claims.

### Structure of the full paper (as reported)
- Section II: Methodology.
- Section III: Overview of the data.
- Section IV: Main regression results.

*Source: Appendix Tables section of the PDF chapter _wp05167 - Appendix Tables*

### Section V concludes.

### _wp05167 - Section V concludes.

### Methodology
- Sample selection
  - Analysis restricted to countries defined by the World Bank as low-income and lower-middle-income countries (Table A1).
  - Total number of countries in group: 89; sample used in regressions smaller and varies by data availability.
  - Exclusions: China, Jordan, and Eritrea excluded from regressions as outliers with respect to the depth variables.
  - Very small countries (population < one million) excluded.
- Data and approach
  - Cross-sectional correlations and regressions (no time-series analysis).
  - Focused on the banking system and on commercial banks only (consistent data for development banks and microfinance institutions not available).
  - Private credit measured from IMF IFS line 22d (deposit money banks); other financial institutions (line 42d) available only for a small subset.
  - Regressions estimated by OLS with robust standard errors.
  - Dependent variables measured as averages over 1999-2001; many right-hand-side variables measured as averages over 1991-98 where possible.

### Dependent variables (financial sector performance indicators)
- Five indicators used as dependent variables:
  - Deposit-to-GDP ratio (deposit mobilization).
  - Private sector credit-to-GDP ratio (bank intermediation to private sector).
  - Loan-to-asset ratio (share of bank funds allocated to private sector loans).
  - Overhead costs to total assets (OH) — banking cost efficiency.
  - Net interest margin to interest-earning assets (NIM) — alternative measure of cost efficiency.
- Data sources:
  - Depth indicators from International Financial Statistics.
  - Efficiency measures from bank-level data in Fitch’s Bankscope; country aggregates constructed over all commercial banks in the Bankscope sample.
- Summary statistics (Table 2)
  - Number of observations: Deposits/GDP 85; Private Credit/GDP 85; Loans/Assets 81; Net Interest Margin 81; Overhead 81.
  - Means and dispersion:
    - Deposits/GDP: Mean 21.73; Sd 14.00; Min 2.24; Max 75.57.
    - Private Credit/GDP: Mean 18.09; Sd 15.71; Min 0.86; Max 93.26.
    - Loans/Assets: Mean 44.47; Sd 15.41; Min 8.73; Max 76.53.
    - Net Interest Margin: Mean 6.56; Sd 3.72; Min 0.10; Max 18.19.
    - Overhead: Mean 5.26; Sd 2.60; Min 0.58; Max 14.47.
  - Cross-correlations:
    - Deposits and Private Credit correlation: 0.82*** (p-value 0.00).
    - Deposits and NIM correlation: -0.43*** (p-value 0.00).
    - Overhead and NIM correlation: 0.74*** (p-value 0.00).

### Explanatory variables (grouped)
- Geography, endowments, and legal origin
  - Rural population density (geographic barriers), latitude, settlers’ mortality, ethnic fractionalization.
  - Legal origin: French legal origin tested (English residual); Soviet legal origin captured by transition dummy.
- Political environment
  - Measures: political stability, internal conflict, military control of government, freedom from corruption.
- Macroeconomic variables
  - Inflation (threshold effects considered at 15 percent and at very high levels); fiscal balance proxied via interest payments on government debt from WEO; migrant remittances.
- Bank ownership and market structure
  - Share of state banks (market share of state bank assets).
  - Share of foreign banks.
  - Market concentration measured by market share of the largest five banks.
- Investment climate indicators (World Bank Business Environment Survey)
  - Time to enforce contracts, availability of credit information, credit registry coverage, costs of collateral and debt recovery.
- Supervisory and regulatory framework
  - 1999 World Bank survey of bank regulators and supervisors (Barth and others, 2001) grouped into six categories: restrictions on scope of bank activities; disclosure requirements; supervisory powers to discipline banks; accounting standards; auditing requirements; plus deposit insurance (zero-one dummy).
  - For each category, an index constructed varying between zero and one.

### Empirical model and estimation strategy
- Sequential model building:
  - Control for GDP per capita and a transition-country dummy in initial specifications.
  - Introduce geography/legal variables, then political variables, then macro, business environment, market structure, and regulation/supervision, retaining robust/significant controls.
  - Same set of specifications used for the first three dependent variables (deposits-to-GDP, private credit-to-GDP, loan-to-assets); slightly different specifications for OH and NIM (efficiency measures), including control for size of financial sector (log GDP).
- Robustness checks:
  - Regressions replicated with GDP per capita in 1970 to address endogeneity; results unchanged.
  - Regressions replicated excluding countries with fewer than five Bankscope banks; efficiency results robust.
  - Settlers’ mortality included in alternate smaller-sample specifications (available for 52 countries); results broadly unchanged.

### Overview of data: regional patterns and bivariate correlations
- Regional composition of sample:
  - Sub-Saharan Africa: 42 percent.
  - Europe and Central Asia (transition): 18 percent.
  - East and South Asia: 16 percent.
  - Latin America and the Caribbean: 15 percent.
  - Middle East and North Africa (MENA): 9 percent.
- Regional findings (averages and patterns)
  - Asia and MENA: on average deeper and more efficient banking systems.
  - Transition countries and Africa: on average shallower financial development.
  - Latin America: intermediate depth; larger share of bank assets goes to finance private sector relative to other regions.
  - State banks: largest share of bank assets in MENA and Asia.
  - Foreign banks: largest presence in Sub-Saharan Africa; smallest in MENA and Latin America.
  - Quality of borrower information best in Latin America.
  - Political risk and internal conflict highest in Sub-Saharan Africa and transition countries.
- Bivariate correlations (selected)
  - Higher income per capita → deeper financial sector and lower NIM.
  - Political stability and lower inflation strongly correlated with greater depth and efficiency.
  - Less concentrated banking systems are deeper and allocate a larger share of assets to private sector credit.
  - Larger share of bank assets held by state-owned banks correlates with more deposits and apparently more efficiency, but with a lower loan-to-asset ratio.
  - Greater foreign bank presence correlates with shallower banking systems (deposits and private credit).
  - Credit information sharing and credit registry coverage positively correlated with depth and efficiency (NIM).
  - Regulation and supervision variables not significantly correlated with cross-sectional financial depth in bivariate analysis.

### Results from multivariate regressions

Financial depth (key multivariate findings)
- Controls
  - GDP per capita and transition dummy included in each regression; GDP per capita captures level-of-development effects; transition dummy captures central-planning legacy.
- Geography and institutions
  - Rural population density: sparser rural population → shallower banking sector; effect pronounced for deposits.
  - Latitude: not significant.
  - French legal origin dummy: negative coefficient for deposits and private credit but not statistically significant; positive (marginally significant) coefficient for loan-to-asset ratio.
  - Settlers’ mortality: negative and strongly significant for deposits and private credit when included (available for 52 countries); when included, rural density becomes insignificant.
  - Ethnic fractionalization: no explanatory power.
- Political environment
  - Political instability and internal conflict associated with shallower financial systems.
  - Corruption index strongly negatively associated with deposits and private credit; when corruption controlled for, political instability measures tend to lose significance.
- Macroeconomic variables
  - Inflation negatively and significantly correlated with private credit and loan-to-asset ratio; not significant for deposits.
  - Allowing for threshold effects at inflation above 15 percent: disruptions to credit markets more severe for higher inflation levels.
  - Fiscal balance proxy (interest payments on government debt/GDP): positive sign but not significant in the reported specification.
  - Migrant remittances: positive sign but not statistically significant.
- Bank ownership and market structure
  - Share of state-owned banks:
    - Positively and significantly correlated with bank deposits.
    - Negatively but insignificantly correlated with private credit and the loan-to-asset ratio.
    - Interpretation ambiguous; potential reverse causality and regional effects.
  - Share of foreign banks:
    - Larger foreign bank presence associated with less deposits and less private sector credit (negative coefficients robust in multivariate regressions).
    - Interpretation ambiguous — possible entry into “underbanked” markets; causality not established.
  - Concentration (top five banks’ share): negatively correlated with private credit and loan-to-asset ratio in some specifications (standard theory view), but result sensitive to alternative concentration measures.
- Business environment indicators
  - Time to enforce contracts, availability of credit information, and credit registry coverage have explanatory power for private credit and loan-to-asset ratio even after controlling for other factors.
  - Other business environment measures (costs of establishing collateral, starting/closing a business) not significantly correlated with depth.
- Regulation and supervision
  - Differences in regulation and supervision add little explanatory power for depth once other factors controlled for.
  - In the few cases with significant relationships, more stringent supervision/auditing requirements sometimes associated with reduced depth rather than increased depth.
  - Presence of an explicit deposit insurance scheme: does not increase deposit mobilization; coefficient negative and marginally significant for deposits.

Financial sector efficiency (NIM and OH)
- Size and geography
  - Bank efficiency higher in larger economies (log GDP included; GDP per capita and transition dummy not significant in efficiency regressions).
  - Geographic and institutional characteristics generally have limited explanatory power for efficiency; settlers’ mortality only marginally significant.
- Political environment
  - Corruption, political instability, and political risk are detrimental to bank efficiency (higher NIM and OH).
- Macroeconomic variables
  - Inflation associated with higher NIM and OH; marginal effect more marked at lower rates of inflation (contrasting with depth results where large effects appear at inflation > 15 percent).
- Bank ownership and market structure
  - Foreign bank penetration: no significant relationship with efficiency.
  - State bank presence: larger presence of state banks associated with significantly lower OH (more cost-efficient) — robust across specifications and samples.
    - This result contrasts with some bank-level panel studies; potential explanations include state banks being larger (scale economies) or sample-specific factors in LICs.
  - Concentration: positively correlated with efficiency in this LIC sample (more concentrated markets → lower OH and NIM), even after controlling for size, business environment, and supervision/regulation. This contrasts with some results for broader samples including developed countries.
- Business environment and regulation
  - No robust link between efficiency and business environment indicators.
  - Regulation and supervision generally enter with positive signs (tighter regulation → higher costs), but coefficients are not robust across specifications and sample sizes drop when these variables included.

Magnitude of effects (illustrative comparative statics from final specifications)
- Effects evaluated as predicted changes resulting from a one standard deviation increase in each explanatory variable (using last specification in Table 5 for depth and Table 11 for efficiency).
- Selected quantified effects:
  - Reducing corruption: yields large benefits for both bank deposits and private sector credit (Table 13 referenced for magnitudes).
  - Lower inflation:
    - Small effect on bank deposits.
    - Strong effect on loan-to-asset ratio, translating into a sizable increase in private sector credit.
  - Increasing share of state-owned banks:
    - Small effect on deposit mobilization.
    - Negative, somewhat large effect on loan-to-asset ratio (coefficient not statistically significant).
  - Reducing foreign bank penetration: affects both deposits and private credit favorably by a similar, sizable margin (i.e., higher foreign bank penetration associated with shallower systems; reduction associated with deeper systems in the cross-section).
  - Concentration and credit registry coverage: statistically significant but small effects on financial depth.
  - Market concentration (one standard deviation increase in top-five banks’ market share ≈ 20 percentage points) leads to:
    - Reduction in OH of 0.9 percentage point.
    - Decline in NIM of 1.83 percentage points.
  - Declines in inflation and changes in state bank penetration: smaller magnitudes on OH and NIM (about 0.7 percentage points for OH for state bank changes reported).

### Conclusions and policy implications
- Summary of main empirical findings
  - Consistent with prior literature: high inflation, unstable and corrupt political environment, and high settlers’ mortality are associated with poor financial performance in LICs.
  - Contract enforcement delays and limited creditor information availability are associated with shallower credit markets.
  - In contrast to some law-and-finance literature findings:
    - French legal origin is not associated with less credit to the private sector in LICs.
    - A larger share of state-owned banks in LICs is associated with more cost-efficient banking sectors (lower OH), greater deposit mobilization, but a smaller share of credit allocated to the private sector.
  - Larger foreign bank presence is robustly associated with shallower financial systems (less deposits and less private credit) in the LIC cross-section.
  - Regulatory and supervisory framework characteristics are not significantly correlated with financial sector performance in LICs once other factors are controlled for.
- Policy lessons (from cross-country regressions)
  - First: political instability and corruption are obstacles to financial development — reducing these should improve depth and efficiency.
  - Second: keeping inflation under control should improve bank efficiency and development.
  - Third: strengthening prudential regulation and supervision may not yield immediate benefits in LICs, possibly because other obstacles are binding or because implementation is weak.
- Caveats and research needs
  - Direction of causality ambiguous for several findings (notably state bank presence and foreign bank penetration); further targeted research and alternative methodologies needed to identify causal effects and to determine which financial sector policies work in LICs.
  - Determinants of financial sector performance in LICs may differ from those in more advanced countries; extrapolation from advanced-country evidence should be done with care.

*Italic: Source — _wp05167 - Section V concludes.*

### 0.18                        Disclosure            requirements

### _wp05167 - 0.18                        Disclosure            requirements

### Supervision and Regulation: summary of empirical associations (Table 9)
- Corruption: positive and statistically significant association with depth measures
  - deposits_gdp: coefficient 19.26 (t-statistic [3.61]***)
  - loans_gdp: coefficient 20.6 (t-statistic [3.43]***)
  - loans_assets: coefficient 8.09 (t-statistic [1.22])
- Inflation: negative and often significant association with depth
  - deposits_gdp: coefficient -3.95 (t-statistic [1.73]*)
  - loans_gdp: coefficient -4.73 (t-statistic [2.29]**)
  - loans_assets: coefficient -6.4 (t-statistic [2.73]**)
- Interest on public debt: mixed effects, some negative on loans/assets
  - examples: loans_assets coefficient -16.92 (t-statistic [2.16]**)
- State banks: small positive associations with depth in some specifications
  - state banks (Panel A) deposits_gdp coefficient 1.55 (t-statistic [2.21]**)
- Foreign banks: negative and significant associations with depth
  - deposits_gdp coefficient -22.02 (t-statistic [3.12]***)
  - loans_gdp coefficient -24.39 (t-statistic [3.84]***)
- Concentration: negative and significant association with loans_assets in multiple specifications
  - loans_assets coefficient -37.02 (t-statistic [3.20]***)
- Supervision/regulation institutional variables (selected)
  - Accounting, Discipline, Disclosure entries appear with coefficients in some panels (e.g., Discipline: 13.68 with [1.97]*; Disclosure: -15.93 with [0.96], -25.45 with [1.77]*), indicating heterogeneous effects across specifications.

Notes:
- Robust t statistics in brackets.
- * significant at 10%; ** significant at 5%; *** significant at 1%.
- Regressions include a constant (not reported).

### Financial Depth: geography, institutions, political variables (Tables 5–8)
- Transition economy dummy: generally associated with lower depth
  - Example Panel A (Table 5): deposits coefficient -9.58 (t-statistic [3.11]***), loans coefficient -7.92 (t-statistic [2.29]**)
- GDP per capita: consistently positive and significant association with financial depth
  - Multiple specifications report GDP per capita coefficients around 9–10 with t-statistics often significant at *** or **
  - Example (Table 6): deposits_gdp coefficient 8.35 (t-statistic [3.62]***)
- Density of rural population: positive in some deposits specifications
  - Example (Table 5): deposits coefficient 6.01 (t-statistic [3.16]***)
- Settlers' mortality and French legal origin: negative associations in some specifications
  - Example (Table 5): Settlers' mortality entries: coefficients reported (e.g., -7.55 with [3.98]***)
  - French legal origin example: loans/assets coefficient 7.17 (t-statistic [1.67]*)
- Corruption: positive and significant association with deposits and loans in many specifications (see Table 6 and 8)
  - Table 6: Corruption deposits_gdp coefficient 10.17 (t-statistic [2.76]***)
  - Table 8: Corruption deposits_gdp coefficient 15.71 (t-statistic [2.93]***)
- Inflation: negative and significant association with depth across many specifications
  - Table 6: loans_assets coefficients -7.92 (t-statistic [4.56]***)
  - Table 8: loans_assets coefficient -9.02 (t-statistic [4.38]***)
- Bank ownership and concentration (Table 7)
  - State banks: positive and significant association with deposits_gdp in several regressions
    - deposits_gdp coefficient 11.82 (t-statistic [3.18]***)
  - Foreign banks: negative and significant association with depth measures
    - Example: deposits_gdp coefficient -11.6 (t-statistic [2.55]**)
  - Concentration: some negative and significant coefficients for depth (e.g., loans_assets -27.25 with [2.24]**)

### Bank efficiency (Overheads (OH) and Net Interest Margin (NIM)) — key results (Tables 10–12)
- Size (log GDP): larger economies associated with lower OH and significantly lower NIM
  - Example (Table 10, Panel A): Size (logGDP) NIM coefficient -0.584*** (t-statistic [2.86])
- Corruption: associated with better measured efficiency (lower OH and lower NIM) in several specifications
  - Table 11: Corruption OH coefficient -1.848*** (t-statistic [3.01]); NIM coefficient -1.674* (t-statistic [1.71])
- Inflation: positively associated with higher OH and higher NIM (worse efficiency)
  - Table 11: Log inflation NIM coefficients up to 1.015*** (t-statistic [4.21])
- Public/state banks: associated with lower OH (improved measured efficiency) in multiple specifications
  - Table 11: State banks OH coefficient -2.700*** (t-statistic [3.80])
- Concentration: significantly associated with lower OH and higher NIM in some models
  - Table 11: Concentration OH coefficient -3.799*** (t-statistic [2.93]); Concentration NIM coefficient -7.661*** (t-statistic [3.63])
- Supervision/regulation and business environment (Table 12)
  - Disclosure requirements: positive coefficient in OH regressions in Table 12 (Disclosure requirements 5.383** with t-statistic [2.58]) — indicating disclosure requirements associated with higher OH in that specification.
  - Concentration continues to show strong and statistically significant coefficients across multiple OH and NIM specifications (see Table 12: many coefficients reported as statistically significant at ** or ***).

### Economic importance of effects (Table 13)
- Summary coefficients (depth and efficiency) and implied effects (columns: deposits_gdp, loans_gdp, loans_assets, OH, NIM)
  - Corruption
    - coefficient: deposits_gdp 15.81; loans_gdp 19.45; loans_assets 10.44; OH -0.123; NIM 0.257
    - S.D.: 0.42 (for listed variables)
    - effect: deposits_gdp 6.58; loans_gdp 8.09; loans_assets 4.35; OH -0.05; NIM 0.11
  - Inflation
    - coefficient: deposits_gdp -3.4; loans_gdp -5.74; loans_assets -8.99; OH 0.717; NIM 1.015
    - S.D.: 1.10
    - effect: deposits_gdp -3.74; loans_gdp -6.31; loans_assets -9.88; OH 0.79; NIM 1.12
  - Interest on public debt
    - coefficient: deposits_gdp 1.35; loans_gdp 0.91; loans_assets -0.51
    - S.D.: 2.45
    - effect: deposits_gdp 3.31; loans_gdp 2.23; loans_assets -1.25
  - State banks
    - coefficient: deposits_gdp 6.83; loans_gdp -4.73; loans_assets -17.27; OH -2.218; NIM -1
    - S.D.: 0.31
    - effect: deposits_gdp 2.10; loans_gdp -1.45; loans_assets -5.31; OH -0.68; NIM -0.31
  - Foreign banks
    - coefficient: deposits_gdp -17.9; loans_gdp -17.28; loans_assets -6.52; OH 1.487; NIM 0.814
    - S.D.: 0.30
    - effect: deposits_gdp -5.29; loans_gdp -5.10; loans_assets -1.93; OH 0.44; NIM 0.24
  - Concentration
    - coefficient: deposits_gdp 2.55; loans_gdp -9.45; loans_assets -31.14; OH -3.799; NIM -7.661
    - S.D.: 0.24
    - effect: deposits_gdp 0.61; loans_gdp -2.26; loans_assets -7.44; OH -0.91; NIM -1.83
  - Credit registry
    - coefficient: deposits_gdp 0.05; loans_gdp 0.12; loans_assets 0.11
    - S.D.: 1.84
    - effect: deposits_gdp 0.09; loans_gdp 0.22; loans_assets 0.20
- Observations and explanatory power for the shown specifications:
  - Observations: deposits_gdp 56; loans_gdp 56; loans_assets 56; OH 62; NIM 62
  - R-squared: deposits_gdp 0.62; loans_gdp 0.73; loans_assets 0.58; OH 0.41; NIM 0.37

Note:
- "Coefficients that are statistically significant at least 10 percent are in italics" (as presented in the table).

### Data appendix: samples and data coverage (Appendix I, Tables A1–A2)
- Country list: Low-Income and Lower Middle-Income Countries enumerated (Table A1) — includes Albania, Algeria, Angola, Argentina, etc., through Zimbabwe (full list provided in source).
- Data sources and variable coverage (Table A2)
  - Example: GDP per capita (logs)
    - Time period: Average 1991-98
    - Obs 87; Mean 6.4; Std. Dev. 0.9; Min 4.4; Max 8.4
    - Source: World Bank, WDI

*Source: _wp05167 - 0.18 Disclosure requirements (Tables and Appendix excerpt).*

### 1.0    La Porta et al. (2002)

### 1.0    La Porta et al. (2002)

### Variables and descriptive statistics (selected)
- French legal origin (dummy)
  - Obs: 90
  - Mean: 0.5
  - Std. Dev.: 0.5
  - Min: 0.0

- Settlers' mortality
  - Obs: 54
  - Mean: 5.1
  - Std. Dev.: 1.0
  - Min: 2.7
  - Max: 8.0
  - Source: Acemoglu, Johnson, and Robinson (2001)

- Latitude
  - Obs: 90
  - Mean: 0.2
  - Std. Dev.: 0.2
  - Min: 0.0

- Density of rural population
  - Obs: 89
  - Mean: 360.6
  - Std. Dev.: 321.1
  - Min: 23.1
  - Max: 1928.7
  - Source: WDI

### Political environment and governance indicators
- Lack of corruption
  - Obs: 90
  - Mean: -0.6
  - Std. Dev.: 0.4
  - Min: -1.6
  - Source: Kaufmann, Kraay et al. (2003)

- Internal stability (Average 1991-93)
  - Obs: 66
  - Mean: 7.8
  - Std. Dev.: 2.2
  - Min: 2.0
  - Max: 11.6
  - Source: International Country Risk Guide

- Political stability (Average 1991-94)
  - Obs: 66
  - Mean: 55.2
  - Std. Dev.: 10.6
  - Min: 24.6
  - Max: 71.4
  - Source: International Country Risk Guide

- Lack of military in the government (Average 1991-95)
  - Obs: 66
  - Mean: 2.8
  - Std. Dev.: 1.5
  - Min: 0.3
  - Max: 6.0
  - Source: International Country Risk Guide

- Ethnic fractionalization
  - Obs: 89
  - Mean: 0.5
  - Std. Dev.: 0.2
  - Min: 0.0
  - Source: Alesina et al. (2003)

### Macroeconomic variables
- Inflation (in logs) (Average 1991-98)
  - Obs: 84
  - Mean: 2.8
  - Std. Dev.: 1.1
  - Min: 1.2
  - Max: 6.2
  - Source: IFS

- Workers' remittances (percent of GDP) (Average 1991-98)
  - Obs: 72
  - Mean: 3.5
  - Std. Dev.: 6.3
  - Min: 0.0
  - Max: 45.3
  - Source: Giuliano-Ruiz Arranz (2005)

- Interest on public debt (percent of GDP) (Average 1991-98)
  - Obs: 85
  - Mean: 3.1
  - Std. Dev.: 2.5
  - Min: 0.0
  - Max: 11.6
  - Source: IFS

- Government balance (percent of GDP) (Average 1991-98)
  - Obs: 87
  - Mean: -4.3
  - Std. Dev.: 3.2
  - Min: -15.0
  - Max: 2.8
  - Source: IFS

### Market structure and banking sector measures
- State-Owned Bank Assets (1995 and 2000-04)
  - Obs: 59
  - Mean: 45.2
  - Std. Dev.: 30.7
  - Min: 0.0
  - Max: 100.0
  - Source: La Porta et al. (2002) and FSAPs

- Foreign Bank Assets (Average 1991-98)
  - Obs: 73
  - Mean: 29.5
  - Std. Dev.: 33.4
  - Min: 0.0
  - Max: 100.0
  - Source: Kodres and Rietti Souto

- Concentration (1998-99)
  - Obs: 75
  - Mean: 65.3
  - Std. Dev.: 23.9
  - Min: 21.7
  - Max: 100.0
  - Source: WB, Financial Structure Database, Barth, Caprio, Levine, and FSAPs

### Business environment (World Bank indicators, selected)
- Days to enforce a contract
  - Obs: 82
  - Mean: 415.4
  - Std. Dev.: 194.6
  - Min: 27.0
  - Max: 1459.0

- Procedures to enforce a contract
  - Obs: 82
  - Mean: 33.8
  - Std. Dev.: 11.2
  - Min: 14.0
  - Max: 58.0

- Cost of enforcement (% debt)
  - Obs: 82
  - Mean: 35.5
  - Std. Dev.: 36.5
  - Min: 8.5
  - Max: 256.8

- Procedures to start a business
  - Obs: 82
  - Mean: 11.0
  - Std. Dev.: 2.8
  - Min: 5.0
  - Max: 19.0

- Days to start a business
  - Obs: 82
  - Mean: 59.9
  - Std. Dev.: 42.0
  - Min: 9.0
  - Max: 203.0

- Cost of starting a business (% GNI per capita)
  - Obs: 82
  - Mean: 124.6
  - Std. Dev.: 193.1
  - Min: 6.7
  - Max: 1268.4

- Minimum capital (% GDP per capita)
  - Obs: 82
  - Mean: 227.8
  - Std. Dev.: 628.4
  - Min: 0.0
  - Max: 5053.9

- Cost of collateral (% GNI per capita)
  - Obs: 74
  - Mean: 24.6
  - Std. Dev.: 33.0
  - Min: 0.0
  - Max: 155.9

- Legal rights of creditors
  - Obs: 75
  - Mean: 4.3
  - Std. Dev.: 1.7
  - Min: 0.0
  - Max: 9.0

- Credit information index
  - Obs: 81
  - Mean: 2.2
  - Std. Dev.: 1.8
  - Min: 0.0
  - Max: 6.0

- Coverage of credit registries
  - Obs: 78
  - Mean: 15.4
  - Std. Dev.: 37.6
  - Min: 0.0
  - Max: 198.0

- Coverage of private credit bureaus
  - Obs: 81
  - Mean: 48.0
  - Std. Dev.: 141.6
  - Min: 0.0
  - Max: 823.0

- Time to close a business (years)
  - Obs: 78
  - Mean: 3.7
  - Std. Dev.: 1.8
  - Min: 1.0
  - Max: 10.0

- Cost of closing business
  - Obs: 78
  - Mean: 16.9
  - Std. Dev.: 15.6
  - Min: 1.0
  - Max: 76.0

- Recovery rate after default
  - Obs: 82
  - Mean: 20.8
  - Std. Dev.: 14.7
  - Min: 0.0
  - Max: 63.5

- Procedures to recover credit
  - Obs: 79
  - Mean: 6.9
  - Std. Dev.: 3.0
  - Min: 2.0
  - Max: 21.0

- Days to recover
  - Obs: 79
  - Mean: 87.2
  - Std. Dev.: 85.3
  - Min: 2.0
  - Max: 382.0

- Cost of recovery (percent of property price)
  - Obs: 79
  - Mean: 8.6
  - Std. Dev.: 7.0
  - Min: 0.2
  - Max: 34.0

- Cost of enforcement (% of credit)
  - Obs: 82
  - Mean: 35.5
  - Std. Dev.: 36.5
  - Min: 8.5
  - Max: 256.8

- Procedures to recover credit (duplicate listing)
  - Obs: 82
  - Mean: 11.0
  - Std. Dev.: 2.8
  - Min: 5.0
  - Max: 19.0

### Supervision and regulation indicators (1998-99, Barth, Caprio, Levine)
- Restrictions to bank activity
  - Obs: 70
  - Mean: 0.75
  - Std. Dev.: 0.40
  - Min: 0.00
  - Max: 1.67

- Auditing requirements
  - Obs: 70
  - Mean: 0.81
  - Std. Dev.: 0.20
  - Min: 0.38
  - Max: 1.00

- Asset diversification requirements
  - Obs: 43
  - Mean: 0.26
  - Std. Dev.: 0.44
  - Min: 0.00
  - Max: 1.00

- Disclosure requirements
  - Obs: 68
  - Mean: 0.61
  - Std. Dev.: 0.15
  - Min: 0.14
  - Max: 0.86

- Supervisors' disciplinary powers
  - Obs: 70
  - Mean: 0.65
  - Std. Dev.: 0.30
  - Min: 0.00
  - Max: 1.00

- Accounting requirements
  - Obs: 62
  - Mean: 0.87
  - Std. Dev.: 0.38
  - Min: 0.00
  - Max: 1.00

- Deposit insurance (1998-99)
  - Max: 1.0
  - Min: 0.0
  - Source: Barth, Caprio, Levine (2003)

### Key referenced works cited in the source
- Acemoglu, Daron, Simon Johnson, and J. A. Robinson, 2001, “The Colonial Origins of Comparative Development,” American Economic Review, Vol. 91, pp. 1369-1410.
- Acemoglu, Daron, and Simon Johnson, 2004, “Unbundling Institutions.”
- Alesina, Alberto, and others, 2003, “Fractionalization,” Journal of Economic Growth, Vol. 8, No.2, pp. 155-94.
- Barth, James R., Gerard Caprio, and Ross Levine, 2001; 2004 (works on bank regulation and supervision).
- Beck, Thorsten, Aslí Demirgüç-Kunt, and Ross Levine, 2003a; 2003b.
- Claessons, Stijn, Aslí Demirgüç-Kunt, and Harry Huizinga, 2001.
- Demirgüç-Kunt, Aslí, and Enrica Detragiache, 2002.
- Demirgüç-Kunt, Aslí, Luc Laeven, and Ross Levine, 2003.
- La Porta, Rafael, Florenciò Lopez de Silanes, and Andrei Shleifer, 2002, “Government Ownership of Commercial Banks,” Journal of Finance, Vol. 57, No. 1, pp. 265-301.
- Additional references include works by Beck & Levine, Berger et al., Bonin & Wachtel, Boyd et al., Cecchetti & Krause, Claessons et al., Clarke et al., Creane et al., Das et al., De Nicolo et al., Detragiache & Gupta, Djankov et al., Focarelli & Pozzolo, Gelbard & Leite, Goldsmith, Huybens & Smith, Kaufmann et al., Levy-Yeyati & Micco, Marquez, Micco et al., Petersen & Rajan, Podpiera, Stulz & Williamson, and von Hagen & Dinger.

*Source: _wp05167 - 1.0    La Porta et al. (2002)*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2005/_wp05167.pdf_