## _wp08215 — References (excerpt)

## Source details

**Canonical URL:** [_wp08215 — References (excerpt)](https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08215.pdf)

## Other formats

- [Markdown version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08215.pdf.md)
- [Structured JSON version](/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08215.pdf.json)

---

### I. Introduction and context
- Scope:
  - Studies credit growth in the Central and Eastern European (CEE) countries from two perspectives: (1) the role of foreign-owned banks; (2) the dependence of foreign-owned CEE subsidiaries on developments at their parent banks.
- Countries covered: Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia.
- Time span of empirical analysis: 1988-2005.
- Sample:
  - Bank-level data for the 72 largest banks of the ten CEE countries; the banks in the sample account for at least 60 percent of the total banking assets in any of the countries.
- Key contextual points:
  - Domestic credit as a share of GDP has been on a steep upward path in the CEE countries since the 1990s.
  - Around 70 percent of the total CEE banking assets are controlled by foreigners by the 2000s (ECB, eurostat).

### II. Hypotheses tested
- Four main hypotheses tested with panel data:
  1. Credit grows more rapidly in foreign-owned banks.
  2. Foreign banks increase their lending through borrowing from other banks (i.e., not restricted by domestic market funding).
  3. Foreign bank entry reduces the cost of borrowing through increased competition.
  4. Economic growth and high interest rates are the two main motivations for foreign banks to operate in the CEE markets.

### III. Data, transformations, and sample handling
- Dataset: Unbalanced panel, 18 years (1988–2005), bank-level and macroeconomic variables.
- Transformations and treatments:
  - All bank-level and macroeconomic variables converted into millions of euro.
  - Bank-level variables transformed into ratios; macroeconomic variables used either in ratios or in real terms.
  - Inflation-adjusted (exchange rate adjusted) dependent variable used as robustness check; correlation with original measure: 0.85, p-value 0.00 (1990s); 0.97, p-value 0.00 (2000s).
  - Outlier handling: any bank-year data with credit growth higher than 500 percent treated as an outlier and excluded.
- Coverage: banks in the sample account for at least 60 percent of total banking assets in any country.
- Figure note: The drop in the number of banks in year 2005 is due to the lag of reporting to the BankScope database.

### IV. Methodology and estimation
- Estimator: Fixed-effect within estimator for static panel data model.
- Model specification:
  - Dependent variable: credit growth = growth rate in net loans of bank i in country j at time t.
  - Core equation (notation preserved):
    ijtjtijtijtsijtfijt uZXsfC +′ +′ +++= − φβγγμ 1
    where ijtiijt vuε+=
  - Ownership dummies: ijt f and ijt s (foreign and state).
  - Bank-specific regressors Xij,t-1 lagged one period.
  - Country-time (time-specific-country) dummies d_jt introduced to capture year-specific host-country macro shocks; these dummies are collinear with host-country macro variables and used as substitutes.
- Endogeneity and multicollinearity:
  - Fixed effects applied to address endogeneity of foreign ownership.
  - Instrumental variable TSLS within estimator also implemented; instruments chosen via panel probit (Model 6 selected).
  - Hausman specification tests indicate no systematic difference between TSLS and exogenous-treatment estimates (whole sample and 2000s).

### V. Main empirical findings — ownership, funding, competition
- Ownership and credit growth:
  - Credit growth is higher in foreign than in domestic banks on average, but when domestic banks are separated, private domestically-owned banks have even higher credit growth than foreign banks (especially in the 2000s).
  - Regression evidence: foreign ownership dummy is significant and negative for the 2000s relative to private domestic banks (see Table 3).
- Funding sources and evolution:
  - Customer Deposits and Interbank Liabilities constitute almost 80 percent of an average CEE bank’s total assets in this dataset.
  - 1990s: Customer Deposits were very important; Interbank Liabilities coefficient significant and negative for private banks.
  - 2000s: Customer Deposits diminished in importance; Interbank Liabilities increased for foreign-owned banks. Interbank Liabilities over Total Assets significant and positive only for foreign-owned banks in the 2000s.
  - Interpretation: foreign banks increasingly used parent bank resources and interbank borrowing; not restricted by domestic financing constraints.
- Competition, spreads, margins, efficiency:
  - Declining spreads associated with intensified competition and higher credit equilibrium.
  - Spread between Lending and Borrowing Rate and Real Interest Rate coefficients are significant and negative in recent years.
  - Net Interest Margin significant and negative for foreign-owned banks (lower margins associated with higher credit growth).
  - Efficiency indicators:
    - Insignificant for private banks during the 2000s.
    - For state-owned banks in the second half: ROA significant and negative; Cost to Income Ratio significant and positive.
- Macroeconomic drivers:
  - Economic growth and relatively high CEE interest rates (much higher than euro-area rates) are driving forces behind lending by foreign banks.
  - Economic growth insignificant in the 1990s but significant and positive for foreign-owned banks in the 2000s.
  - Spread between Domestic and Euro Area Nominal Interest Rates: negative and significant for domestic banks (integration), positive for foreign banks (entry incentive).

### VI. Parent-bank linkages and subsidiary credit growth
- Parent-bank characteristics that matter:
  - Parent bank size and profitability affect positively the amount of credit supplied by the CEE subsidiary.
  - Parent bank Cost to Income Ratio affects subsidiary credit growth negatively.
- Business-cycle alignment:
  - Credit growth in subsidiaries is procyclical with host country Real GDP Growth Rate (positive and significant in all regressions).
  - Home country Real GDP Growth Rate is insignificant.
- Funding and EMU-related effects:
  - Interbank Liabilities over Total Assets of subsidiaries significant and positive in recent years.
  - Euro-Area M3 Growth coefficient is positive and significant in some specifications.
  - Euro Area Short-Term Nominal Interest Rates coefficient significant and negative in specifications — falling EMU rates motivate parent banks to expand lending in higher-rate host countries.
- Heterogeneity across parent banks:
  - Equation (3.2) estimates allow coefficients to vary across parent banks owning at least three subsidiaries.
  - Wald tests fail to reject equality of coefficients across parent banks for key regressors, indicating limited evidence of differing managerial strategies across examined parent-bank groups.

### VII. Robustness checks and diagnostics
- Inflation-adjusted credit growth:
  - Correlation with original credit growth: 0.85 (p-value 0.00) in the 1990s; 0.97 (p-value 0.00) in the 2000s.
  - Tables 5 and 6 (inflation-adjusted dependent variable) produce results very close to Tables 3 and 4; minor changes in significance for ROA and real interest rate in specific periods.
- Instrumental variables and specification tests:
  - Panel data probit (Table 7) used to select instruments for TSLS fixed-effect implementation.
  - Hausman tests (Table 8) for whole sample and 2001–2005 indicate no systematic difference between IV and no-IV estimates (Prob>chi2 = 0.5801 and 0.71 respectively).

### VIII. Key numeric facts and sample statistics (exact values preserved)
- Number of banks in sample: 72.
- Outlier cutoff for credit growth: higher than 500 percent.
- Panel period: 1988-2005.
- Coverage criterion: at least 60 percent of total banking assets in any country.
- Parent-subsidiary panel: 16 years (1990-2005), 59 banks comprised of 18 parent banks and their 41 CEE subsidiaries.
- Ownership shares and changes:
  - 66 percent of the banks became foreign (Table 14 statement).
  - State ownership declined from 52 percent to 10 percent over time (Table 14 statement).
- Macroeconomic shift reported: Real GDP Growth increased from 3 percent to 5 percent during the 2000s (Table 14 statement).
- Selected market-share time series of foreign-owned bank assets (Table 1 entries preserved exactly):
  - Slovenia: 2000 15.0, 2001 16.6, 2002 18.5, 2003 20.0, 2004 22.5
  - Latvia: 2000 45.2, 2001 41.9, 2002 45.7, 2003 48.6, 2004 58.5
  - Hungary: 2000 n.a., 2001 59.2, 2002 60.1, 2003 56.7, 2004 59.0, 2005 58.8
  - Poland: 2000 n.a., 2001 68.9, 2002 67.3, 2003 67.7, 2004 67.7, 2005 67.1
  - Lithuania: 2000 n.a., 2001 75.6, 2002 84.4, 2003 84.1, 2004 83.9, 2005 84.1
  - Czech Republic: 2000 n.a., 2001 77.1, 2002 93.2, 2003 96.0, 2004 96.1, 2005 93.4
  - Slovakia: 2000 85.0, 2001 92.4, 2002 95.6, 2003 96.3, 2004 96.7, 2005 99.5
  - Estonia: 2000 n.a., 2001 97.9, 2002 97.4, 2003 97.4, 2004 98.1, 2005 99.2
- Descriptive sample statistics (selected exact figures from Table 14):
  - Net Loan Growth (Obs 648): Mean 29.85; Std. Dev 43.72.
  - Foreign Ownership Dummy (Obs 648): Mean 0.46; Std. Dev 0.50.
  - State Dummy (Obs 648): Mean 0.33; Std. Dev 0.47.
  - Total Assets over GDP (-1) (Obs 648): Mean 9.01; Std. Dev 15.29.
  - Customer Deposits / Total Assets (-1) (Obs 644): Mean 62.22; Std. Dev 18.93.
  - Interbank Liabilities / Total Assets (-1) (Obs 636): Mean 15.68; Std. Dev 15.85.
  - ROA (-1) (Obs 640): Mean 1.42; Std. Dev 3.70.
  - Net Interest Margin (-1) (Obs 635): Mean 5.94; Std. Dev 5.16.
  - Euro Area M3 Growth (Obs 657): Mean 6.01; Std. Dev 2.11.
  - Euro Area S-T Interest Rate (3 Month) (Obs 657): Mean 4.23; Std. Dev 2.00.

### IX. Policy-relevant conclusions and recommendations
- Ownership and sustainability:
  - Foreign-owned banks on average have higher loan growth than domestic banks; in the 2000s foreign bank growth is higher than state-owned banks but significantly lower than domestically-owned private banks.
  - As CEE financial markets mature and growth decelerates, parent banks’ incentives to expand lending may change.
- Funding vulnerabilities:
  - Foreign-owned banks shifted toward reliance on parent-bank resources and external interbank borrowing over time, raising vulnerabilities to shocks in developed-market funding.
- Competition and efficiency:
  - Increased competition reduced spreads and margins; efficiency and cost indicators lost importance for private banks by the 2000s but remained important for state-owned banks.
  - Parent banks are cost-sensitive: higher group-wide operational costs can reduce subsidiary lending.
  - Policy advice: increase information sharing between CEE subsidiaries and their foreign parent banks.
- Prudential implications:
  - Business-cycle sensitivity implies that an economic downturn will negatively affect credit growth.
  - Recommendation: CEE supervisors should ensure prudential banking regulations and supervision that require extra buffers to manage risky lending during downturns.

*Source: _wp08215 — References (excerpt) — IMF working paper content provided.*

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

### _wp08215 - References.........................................................................................................................

### I. Introduction and context
- Scope: Studies credit growth in the Central and Eastern European (CEE) countries from two perspectives: (1) the role of foreign-owned banks; (2) the dependence of foreign-owned CEE subsidiaries on developments at their parent banks.
- Countries covered: Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia.
- Time span of empirical analysis: 1988-2005.
- Sample: Bank-level data for the 72 largest banks of the ten CEE countries; the banks in the sample account for at least 60 percent of the total banking assets in any of the countries.
- Key contextual points:
  - Domestic credit as a share of GDP has been on a steep upward path in the CEE countries since the 1990s.
  - Around 70 percent of the total CEE banking assets are controlled by foreigners by the 2000s (ECB, eurostat).

### II. Hypotheses tested
- Four main hypotheses tested with panel data:
  1. Credit grows more rapidly in foreign-owned banks.
  2. Foreign banks increase their lending through borrowing from other banks (i.e., not restricted by domestic market funding).
  3. Foreign bank entry reduces the cost of borrowing through increased competition.
  4. Economic growth and high interest rates are the two main motivations for foreign banks to operate in the CEE markets.

### III. Data and sample details
- Dataset: Unbalanced panel, 18 years (1988–2005), bank-level and macroeconomic variables.
- Data transformations and treatments:
  - All bank-level and macroeconomic variables converted into millions of euro.
  - Bank-level variables transformed into ratios; macroeconomic variables used either in ratios or in real terms.
  - To reduce inflation impact and possible exchanges rate jumps, results are also reproduced using inflation adjusted (exchange rate adjusted) dependent variable; coefficients and standard errors are robust to this adjustment.
  - Outlier handling: any bank-year data with credit growth higher than 500 percent treated as an outlier and excluded.
- Coverage table reference: Table 2 provides coverage of dataset as a percent of total banking assets of each CEE country (not reproduced here).

### IV. Methodology
- Estimator: Fixed-effect within estimator for static panel data model.
- Justification:
  - Static model chosen because correlation coefficients between current and lagged bank-level credit growth are insignificant.
  - Fixed effects chosen after analysis of endogeneity and multicollinearity; clustering and robust estimation did not yield smaller standard errors.
- Model specification (as presented):
  - Dependent variable: credit growth = growth rate in net loans of bank i in country j at time t.
  - Equation (2.1):
    ijtjtijtijtsijtfijt uZXsfC +′ +′ +++= − φβγγμ 1
    where ijtiijt vuε+=
  - Variables:
    - ijt f and ijt s: foreign and state ownership dummies (time-specific).
    - Xij,t-1: bank-specific variables lagged one period to reduce simultaneity.
    - Zjt: country-specific macroeconomic variables for country j at time t (in practice, time-specific-country dummies introduced because of collinearity).
    - νi: bank-level fixed effects; εijt: error term.
  - Time-specific-country dummies used to control for unspecified macro effects; macro variables and these dummies are collinear, so dummies introduced separately to observe macro coefficients.

### V. Key empirical findings — differences across ownership types (Section II)
- Four main results (Section II):
  - First: Credit growth is higher in foreign than in domestic banks. However, when domestic banks are separated into private and state-owned banks, private domestically-owned banks have even higher credit growth than foreign banks.
  - Second: Foreign-owned banks relied increasingly on interbank funding during the 2000s, while domestic sources (mainly customer deposits) lost importance over time.
  - Third: As more banks entered CEE markets during the 2000s, interest margins declined and credit supply increased, reducing the cost of borrowing and increasing equilibrium credit in the CEE economies.
  - Fourth: High economic growth and relatively high CEE interest rates, much higher than the euro-area rates, are the driving forces behind lending by foreign banks in the CEE markets.

### VI. Key empirical findings — linkages with parent banks (Section III)
- Three major results (Section III):
  - First: No significant difference across parent banks in terms of their impact on the growth of credit in their CEE subsidiaries.
  - Second: Parent-bank characteristics matter: parent bank size and profitability affect positively, while increasing cost structure of the parent affects negatively, the amount of credit supplied by the CEE subsidiary.
  - Third: Economic growth in the CEE markets is an important driver of credit growth; growth in the home market plays no role. Declining interest margins in the home market motivate parent banks to increase credit in the CEE countries.

### VII. Descriptive statistics and notable numeric details
- Table 1: Market Share of Assets of Foreign-Owned Banks in the CEE Countries (selected values as presented)
  - Slovenia: 2000 15.0, 2001 16.6, 2002 18.5, 2003 20.0, 2004 22.5
  - Latvia: 2000 45.2, 2001 41.9, 2002 45.7, 2003 48.6, 2004 58.5
  - Hungary: 2000 n.a., 2001 59.2, 2002 60.1, 2003 56.7, 2004 59.0, 2005 58.8
  - Poland: 2000 n.a., 2001 68.9, 2002 67.3, 2003 67.7, 2004 67.7, 2005 67.1
  - Lithuania: 2000 n.a., 2001 75.6, 2002 84.4, 2003 84.1, 2004 83.9, 2005 84.1
  - Czech Republic: 2000 n.a., 2001 77.1, 2002 93.2, 2003 96.0, 2004 96.1, 2005 93.4
  - Slovakia: 2000 85.0, 2001 92.4, 2002 95.6, 2003 96.3, 2004 96.7, 2005 99.5
  - Estonia: 2000 n.a., 2001 97.9, 2002 97.4, 2003 97.4, 2004 98.1, 2005 99.2
  - Source: ECB (as presented in the document)
- Other numeric points preserved exactly:
  - Number of banks in sample: 72.
  - Outlier cutoff for credit growth: higher than 500 percent.
  - Panel period: 1988-2005.
  - Coverage: banks in the sample account for at least 60 percent of total banking assets in any country.
  - Figure note: The drop in the number of banks in year 2005 is due to the lag of reporting to the BankScope database.

### VIII. Structure of the paper (as stated)
- Section II: Determinants of credit growth considering variations across bank ownership types.
- Section III: Determinants of credit growth for foreign banks operating in the CEE countries, considering differences across their parent banks.
- Section IV: Conclusions and discussion on policy implications.

*Source: _wp08215 - References (excerpt) — IMF working paper content provided*

### introduction of these dummy variables controls the impact of a year-specific macroeconomic

### _wp08215 - introduction of these dummy variables controls the impact of a year-specific macroeconomic

### Model specification and estimation approach
- Equations and fixed-effect representation:
  - Equation (2.1s) introduces time-specific-country dummies d_jt into the credit-growth equation:  
    ijtjtijtijtsijtfijt udXsfC+′+′+++=−φβγγμ~1 (2.1s)
  - Equation (2.2) allows interaction of foreign and state ownership dummies with bank-specific and macroeconomic variables and is expressed as a fixed-effect within estimator (equation displayed in source as (2.2)).
  - A fixed effect within estimator modification of (2.1) is represented as equation (2.1)* in the text to show the impact of bank-level fixed effects on constant and ownership dummies.
- Time-specific-country (time-specific-host-country) dummies:
  - Introduced to capture year-specific macroeconomic shocks peculiar to each CEE country/host country.
  - Both Akaike and Bayesian information criteria indicate improvement of the fixed effect estimation model after introduction of these dummies (footnote 11).
  - These dummies are collinear with host-country macro variables and are therefore used as substitutes for those macro variables.

### Data structure and sample split
- Structural break and sample split:
  - The dataset exhibits a structural break between the 1990s and the 2000s; regression analysis is performed separately for the 1990s and 2000s.
  - Reasons: 1990s characterized by transition, banking crises, recapitalization programs; 2000s characterized by robust economic growth and rapid financial integration.
  - t-tests for mean comparison between 1990s and 2000s (see Appendix, B and Table 14) support differences between samples.
- Panel characteristics (parent-subsidiary dataset):
  - Unbalanced panel covering 16 years (1990-2005), 59 banks comprised of 18 parent banks and their 41 CEE subsidiaries.
  - Table 9 (described) shows geographic clustering of parent banks; major foreign players are Western European banks.

### Concerns about multicollinearity and estimation choices
- Multicollinearity:
  - Many regressors are significantly correlated; coefficients across similar measurement variables (profitability, efficiency, macroeconomic variables) can be large in magnitude.
  - To mitigate multicollinearity, highly and significantly correlated variables are introduced in separate regression equations.
- Endogeneity of foreign ownership:
  - Recognized concern: acquisition of state-owned banks by foreigners may be nonrandom (better-performing state banks more likely to be purchased).
  - Strategies used:
    - Dataset focuses on largest banks (coverage criterion: account for at least 60 percent of country total banking assets).
    - Fixed effect within estimator applied to panel data to eliminate correlation between foreign ownership dummy and bank-specific error component.
    - Inclusion of many time-varying bank-specific controls (profitability, liquidity, size, soundness).
    - Instrumental variable approach: TSLS generalization of fixed effects with instruments selected via panel data probit (Model 6 chosen based on smallest information criteria in Table 7). Hausman specification tests (Table 8) indicate no systematic difference between TSLS and exogenous-treatment estimates (whole sample and 2000s).

### Main empirical results — ownership, funding, and competition
- Ownership and credit growth:
  - Foreign banks have higher credit growth than domestic banks on average across 1992 to 2005, except years 1992, 2003, and 2004 (Figure 2).
  - After distinguishing domestic ownership types, regression analysis shows the foreign ownership dummy is significant and negative for the second half of the dataset (2000s) — indicating foreign banks had lower credit growth rates than private domestically-owned banks during the 2000s (see Table 3).
  - Explanations for this result include: private domestic banks affiliated with conglomerates raising funding via associated firms; longer-term relationships between private domestic banks and domestic firms; potential “cherry picking” by foreign banks (but considered least likely given observed high credit growth in foreign-owned banks).
- Funding sources and evolution:
  - Customer Deposits and Interbank Liabilities constitute almost 80 percent of an average CEE bank’s total assets in this dataset (footnote 14).
  - 1990s: Customer Deposits were a very important financing source for private banks; Interbank Liabilities coefficient significant and negative for private banks (reflecting crises and recapitalization).
  - 2000s: Importance of Customer Deposits diminished; Interbank Liabilities increased for foreign-owned banks. Interbank Liabilities over Total Assets significant and positive only for foreign-owned banks in the 2000s (Table 4).
  - Interpretation: foreign banks increasingly used parent bank resources and interbank borrowing; not restricted by domestic financing constraints.
- Competition, spreads, margins, and efficiency:
  - Declining spreads associated with intensified competition and higher credit equilibrium (historical declines in spreads since mid-1990s).
  - Coefficient estimates for Spread between Lending and Borrowing Rate and Real Interest Rate (Table 4) are significant and negative in recent years.
  - Spread between Domestic and Euro Area Nominal Interest Rates shows negative and significant coefficients in both periods, larger in the second half (evidence of financial integration).
  - Net Interest Margin significant and negative for foreign-owned banks (lower margins associated with higher credit growth).
  - Efficiency indicators (ROA, Cost to Income Ratio):
    - Insignificant for private banks during the 2000s (Table 4).
    - For state-owned banks in the second half of the dataset: ROA significant and negative; Cost to Income Ratio significant and positive — decline in returns and higher cost structure explain loan growth only for state-owned banks.
- Macroeconomic drivers:
  - Economic growth and high interest rates in CEE were motivations for foreign banks’ operations.
  - Economic growth insignificant in first period (1990s) but significant and positive for foreign-owned banks in the second half (2000s).
  - Higher domestic interest rates in CEE relative to home markets also motivate foreign banks to increase lending.
  - During 2000s, Spread between Domestic and Euro Area Nominal Interest Rates: negative sign for domestic banks (financial integration), positive sign for foreign banks (entry for higher-than-home-market interest rates).

### Robustness checks
- Real exchange rate jumps and credit growth measurement:
  - Concern: credit growth measured in euro may be biased by jumps in real exchange rate, especially in 1990s inflation episodes.
  - New dependent variable: credit growth calculated from inflation-adjusted loan data.
  - Correlation between new and initial credit growth variable: 0.85, p-value 0.00 during the 1990s; 0.97, p-value 0.00 during the 2000s (footnote 20).
  - Results using inflation-adjusted credit growth (Tables 5 and 6) are very close to original results (Tables 3 and 4); for 1990s only ROA loses significance in Table 5; for 2000s only real interest rate loses significance in Table 6 (footnote 21). Conclusion: possible real exchange rate jumps not persistent enough to bias original estimations.
- Instrumental variables and Hausman tests:
  - TSLS fixed-effect implementation uses instruments chosen from panel probit (Model 6).
  - Hausman specification tests (Table 8) for whole sample and 2000s indicate no systematic difference between TSLS and exogenous-treatment estimates — supporting assumption that foreign acquisition decisions are not shaped by bank-specific credit growth conditional on bank fixed effects.

### The role of foreign parent banks — empirical findings
- Empirical model (Equation (3.1)):
  - Dependent variable: credit growth in a CEE subsidiary measured by percentage growth rate in net loans.
  - Model includes: subsidiary bank-level variables X1, parent bank-level variables X2, country-specific macro variables Z1 (host) and Z2 (home), time-specific-host-country dummies, and year dummies.
  - Balance sheet variables are lagged one period to minimize simultaneity.
- Key parent-bank related results (Tables 10 and 11):
  - Parent bank size:
    - ln(Total Assets) of parent bank is significant and positive for subsidiary credit growth — larger parent banks enable faster credit growth in CEE subsidiaries.
    - Interbank Liabilities over Total Assets increasingly important source of funding for subsidiaries in recent years; Euro-Area M3 Growth coefficient positive and significant (Table 11) — may indicate financing via EMU markets.
  - Parent bank profitability and costs:
    - Parent bank Net Interest Margin significant and positive for CEE credit growth overall — parent banks lend more in subsidiaries during profitable periods; Net Interest Margin becomes insignificant in last five years.
    - Parent bank Cost to Income Ratio significant and negative — higher parent operational costs are associated with reduced lending in subsidiaries; sensitivity to costs increases in last five years.
  - Business-cycle alignment:
    - Credit growth is procyclical with host country Real GDP Growth Rate (positive and significant in all regressions, Table 10).
    - Credit growth is not responsive to home country Real GDP Growth Rate (Table 11: coefficients insignificant).
  - Home-market spreads and EMU influence:
    - Tight spreads in home market motivate parent banks to increase subsidiary lending (Spread between Lending and Borrowing Rate significant and negative for parent banks in last five years).
    - Short-Term Euro-Area Nominal Interest Rates coefficient significant and negative (Table 11) — falling EMU rates lead parent banks to increase lending in higher-rate host countries.
- Heterogeneity across parent banks (Equation (3.2) and tests):
  - Equation (3.2) allows coefficients to vary across parent banks; applied only to parent banks owning at least three subsidiaries due to data limitations.
  - Restricted models use three key regressors: Total Assets over GDP, Interbank Liabilities over Total Assets, and Real GDP Growth.
  - Wald tests (Table 13) fail to reject equality of coefficients across parent banks for the examined regressors, suggesting limited evidence of differing managerial strategies across these parent-bank groups.

### Summary of policy-relevant conclusions and recommendations
- Ownership and sustainability:
  - Foreign-owned banks on average have higher loan growth than domestic banks; by the 2000s foreign bank growth is higher than state-owned banks but significantly lower than domestically-owned private banks.
  - Sustainability concerns: foreign banks are motivated by high economic growth in CEE and by lower home-market interest rates; as CEE financial markets mature and growth decelerates, parent banks’ incentives may change.
- Funding vulnerabilities:
  - Over time, foreign-owned banks shifted from domestic deposit financing toward reliance on parent-bank resources and external interbank borrowing.
  - Increased dependence on foreign interbank markets raises vulnerabilities (concerned about exposure amid credit crunches in developed markets).
- Competition and efficiency:
  - Increased competition reduced spreads and margins; efficiency and cost indicators lost importance for private banks by the 2000s but remained important for state-owned banks.
  - “Know-how” effect: foreign banks brought operational advantages that reduced the role of cost/efficiency in private bank credit growth.
  - Parent banks are cost-sensitive: higher group-wide operational costs can reduce subsidiary lending.
  - Policy advice: increase information sharing between CEE subsidiaries and their foreign parent banks.
- Prudential implications:
  - High economic growth and financial deepening were key drivers of CEE credit growth; business-cycle effects imply that an economic downturn will negatively affect credit growth.
  - Recommendation: CEE supervisors should ensure prudential banking regulations and supervision that require extra buffers to manage risky lending during downturns.

### Key data and descriptive statistics highlighted in the text
- Sample periods and structural break: 1990s vs 2000s; sample years referenced include 1992 to 2005 and full panels 1988–2005 in Appendix Table 14.
- Parent-subsidiary panel: 16 years (1990-2005), 59 banks, 18 parent banks, 41 CEE subsidiaries.
- Ownership shares and changes:
  - CEE countries are dominated by foreign banks: 66 percent of the banks became foreign (Table 14).
  - State ownership declined from 52 percent to 10 percent over time (Table 14).
- Macroeconomic shifts:
  - Real GDP Growth increased from 3 percent to 5 percent during the 2000s (Table 14).
- Measurement and robustness figures:
  - Correlation between original and inflation-adjusted credit growth: 0.85, p-value 0.00 (1990s); 0.97, p-value 0.00 (2000s).
  - Data cleaning: any bank year with credit growth rates greater than 500 percent is eliminated.
- Dataset coverage rule:
  - Balance sheet coverage criterion: include as many banks as to account for at least 60 percent of the country’s total banking sector assets (Appendix A).
- Financing composition:
  - Customer Deposits and Interbank Liabilities constitute almost 80 percent of an average CEE bank’s total assets in this dataset (footnote 14).

*Italic: IMF Working Paper content unit _wp08215 as provided in the source material.*

### REFERENCES

### _wp08215 - REFERENCES

### Key bibliographic references
- Altunbas, Y., L. Evans and P. Molyneux, 2001, “Bank Ownership and Efficiency,” Journal of Money, Credit and Banking, Vol. 33, No. 4, pp. 926-54.
- Barajas, A., R. Steiner and N. Salazar, 2000, “The Impact of Liberalization and Foreign Investment in Colombia’s Financial Sector,” Journal of Development Economics, Vol. 63, pp. 157-196.
- Baltagi, B. and H., 2003, “Econometric Analysis of Panel Data,” Wiley, Third Edition.
- Berger, A., R. DeYoung, H. Genay and F. Udell, 2000, “Globalization of Financial Institutions: Evidence from Cross-Border Banking Performance,” Brookings Papers on Economic Activity, Vol. 2, pp. 23-158.
- Bonin, J. P., I. Hasan and P. Wachtel, 2005, “Bank performance, efficiency and ownership in transition countries”, Journal of Banking & Finance, 29 pp. 31–53.
- De Nicolò, G., and E. Loukoianova, 2007, “Bank Ownership, Market Structure and Risk,” International Monetary Fund, WP/07/215.
- Detragiache, E., and P. Gupta, 2006, "Foreign banks in emerging market crises: Evidence from Malaysia," Journal of Financial Stability, vol. 2(3), pages 217-242, October.
- International Monetary Fund, 2006, “Financial Soundness Indicators: Compilation Guide,” International Monetary Fund, ISBN/ISSN: 978-1-58906-385-3.
- (Additional references listed in source document; above are representative entries preserved exactly as supplied.)

### Empirical results — Determinants of Bank Credit Growth (Tables 3–6, 10–12)
- Table 3 (Determinants of Bank Credit Growth in the CEE countries): key reported coefficients and p-values (p values in brackets; * significant at 10%; ** significant at 5%; *** significant at 1%)
  - Foreign (-1): -5.1 [0.86]; -1.76 [0.95]; -80.81 [0.01]**; -84.52 [0.00]***; 21.64 [0.96]; 1.11 [0.97]; -79.9 [0.01]**; -87.24 [0.00]***
  - State (-1): -2.22 [0.94]; -3.91 [0.90]; -113.59 [0.00]***; -110.29 [0.00]***; -0.91 [0.98]; -1.52 [0.96]; -109.86 [0.00]***; -114.21 [0.00]***
  - Total Assets over GDP (-1): 0.39 [0.07]*; 0.26 [0.21]; -1.21 [0.23]; -1.92 [0.11]; 0.28 [0.19]; 0.25 [0.23]; -1.77 [0.09]*; -1.32 [0.26]
  - Customer Deposits / Total Assets (-1): 0.49 [0.17]; 0.31 [0.37]; 0.67 [0.10]; 0.55 [0.17]; 0.12 [0.73]; 0.13 [0.70]; 0.67 [0.10]; 0.66 [0.10]
  - Interbank Liabilities / Total Assets (-1): -0.47 [0.23]; -0.52 [0.14]; 1.06 [0.01]***; 1.06 [0.01]**; -0.83 [0.02]**; -0.72 [0.03]**; 1.22 [0.00]***; 1.17 [0.00]***
  - ROA (-1): 1.18 [0.06]*; 1.26 [0.03]**; -3.34 [0.07]*; -4.03 [0.02]**
  - Spread 1 (Lending - Borrowing): -0.07 [0.81]; -3.52 [0.02]**
  - Spread 2 (Domestic - Euro): -0.38 [0.03]**; -1.02 [0.08]*
  - Country-Time Dummies: No / Yes (varies by model)
  - Observations: 328, 345, 280, 275, 337, 343, 281, 276 (by model)
  - Number of bankid: 63, 63, 66, 56, 36, 63, 66, 65 (as listed)
  - R-squared: 0.08, 0.06, 0.17, 0.46, 0.04, 0.04, 0.16, 0.45

- Table 4 (Determinants by decade — selected coefficients and significance)
  - Foreign (-1): examples across models: -29.28 [0.44]; -89.05 [0.01]**; 104.4 [0.03]**; -32.09 [0.58]; -95.74 [0.00]***; -134.39 [0.03]**
  - State (-1): examples: -48.74 [0.20]; -113.44 [0.00]***; 108.64 [0.03]**; -101.03 [0.07]*; -148.31 [0.00]***; -230.82 [0.01]***
  - TA / GDP (-1): -3.41 [0.00]***; -1.51 [0.13]; 0.02 [0.95]; -1.77 [0.09]*; etc.
  - Interactions (* Foreign, * State) reported with coefficients and p-values throughout the table (see source for full matrix).
  - Observations by model: 296, 243, 331, 279, 314, 266, 329, 274.
  - Number of bankid: 57, 58, 63, 65, 60, 65, 63, 65.
  - R-squared: 0.2, 0.24, 0.15, 0.15, 0.13, 0.23, 0.1, 0.22.

- Table 5 and Table 6: Replications of Tables 3 and 4 with Inflation Adjusted Credit Growth as the dependent variable
  - Key coefficients mirror Table 3/Table 4 with differences in magnitude and p-values; examples:
    - Table 5 Foreign (-1): -2.72 [0.92]; -0.72 [0.98]; -79.1 [0.01]***; -84.25 [0.00]***; etc.
    - Table 6 Foreign (-1): -20.5 [0.56]; -87.74 [0.01]**; 93.72 [0.04]**; -32.95 [0.55]; etc.
  - Observations and sample sizes comparable to non-inflation-adjusted specifications; R-squared values reported per model (e.g., Table 5 R-squared: 0.07, 0.05, 0.18, 0.42, 0.05, 0.05, 0.19, 0.39).

- Table 10–11 (Determinants in Foreign-Owned CEE Subsidiaries)
  - Table 10: ln(Total Assets) (-1) coefficients across multiple specifications: -26.65 [0.00]***; -42.9 [0.00]***; -33.67 [0.00]***; -55.75 [0.00]***; -13.61 [0.01]**; -52.34 [0.00]***; -26.15 [0.00]***; -48.27 [0.00]***; -17.15 [0.00]***; -52.01 [0.00]***; -26.1 [0.00]***; -43.8 [0.00]***
  - ln(Total Assets) of Parent (-1): 31.08 [0.02]**; 28.43 [0.03]**; 47.02 [0.00]***; 47.81 [0.00]***
  - Customer Deposits/ T.A. (-1): examples include 0.21 [0.66]; 0.84 [0.06]*; 0.83 [0.16]; 1.18 [0.05]*; and others with p-values listed.
  - Interbank Liabilities/ T.A. (-1): 0.52 [0.23]; 0.8 [0.04]**; 1.51 [0.00]***; 1.49 [0.00]***; etc.
  - Real GDP Growth Rate: numerous positive and significant coefficients (e.g., 5.51 [0.00]***; 4.68 [0.01]***; 4.57 [0.01]**; etc.)
  - Observations: typically in the 130–186 range depending on specification; Number of bankid: 34–37; R-squared examples: 0.25, 0.49, 0.29, 0.35, 0.19, 0.47, 0.21, 0.27, 0.17, 0.47, 0.26, 0.29.

- Table 11 (Foreign-Owned Subsidiaries — alternative specifications)
  - Examples:
    - ln(Total Assets) (-1): -19.46 [0.00]***; -24.60 [0.02]*; -22 [0.00]***; -18 [0.05]*
    - Customer Deposits/ T.A. (-1): 0.68 [0.18]; 1.90 [0.00]**; -0.24 [0.28]; 0.75 [0.07]*
    - Interbank Liabilities/ T.A. (-1): 0.27 [0.51]; 1.88 [0.00]***; -0.03 [0.90]; 1.35 [0.00]***
    - Real GDP Growth Rate: 2.81 [0.08]*; 4.73 [0.04]**; 2.16 [0.00]***; 0.99 [0.63]
    - Euro Area S-T Interest Rate (3 Month): -5.45 [0.00]***; -10.93 [0.05]*
    - Euro Area M3 Growth: 2.47 [0.00]***; 2.96 [0.11]
  - Observations: 192, 139, 558, 269 (by specification); Number of bankid: 39, 38, 70, 65; R-squared: 0.27, 0.22, 0.12, 0.08.

- Table 12–13 (Fixed Effects and tests across affiliated banks)
  - Table 12: FE regression varying coefficients by affiliated bank (H, E, SW, SK denote parent affiliations). Examples:
    - ln(Total Assets) (-1) H: -23.34 [0.00]***; E: 1.85 [0.94]; SW: -37.47 [0.00]***; SK: -48.78 [0.00]***.
    - Real GDP Growth SW: 7.39 [0.02]** in Model 1; other country/affiliation coefficients and p-values reported.
    - Observations: 99, 91, 103 depending on model; Number of bankid: 17, 18, 17; R-squared: 0.4, 0.37, 0.09.
  - Table 13: Tests of equality of coefficients across affiliated banks (F-tests with Prob > F):
    - Example: ln(Total Assets)(-1) equality test: F(  3,    74) = 1.79; Prob > F = 0.1565.
    - Multiple F-tests reported with respective Prob > F values (see source for full set).

### Panel data probit, IV/Hausman tests, and specification diagnostics (Tables 7–8)
- Table 7: Panel Data Probit Estimation Results for Foreign Ownership Variable (models 1–6)
  - Total Assets over GDP (-1): 0.03 [0.01]***; 0.03 [0.03]**; 0.01 [0.35]; 0.02 [0.07]*; 0.02 [0.09]*; 0.01 [0.34]
  - Loan Loss Provision/Total Assets (-1): -0.13 [0.03]**; -0.18 [0.01]***; -0.12 [0.05]*; -0.16 [0.01]***; -0.2 [0.00]***; -0.07 [0.23]
  - Total Capital Ratio (-1): 0.03 [0.06]*; 0.04 [0.01]**; 0.06 [0.00]***; 0.04 [0.01]**; 0.05 [0.00]***; 0.05 [0.00]***
  - Spread bw Domestic and Euro Interest Rates: -0.08 [0.00]***; -0.06 [0.00]***; -0.04 [0.02]**
  - ROA (-1): -0.11 [0.02]**; -0.12 [0.01]**; -0.08 [0.06]*
  - Net Interest Margin (-1): -0.23 [0.00]***; -0.21 [0.00]***
  - Spread between lending and borrowing Rate: -0.23 [0.00]***; -0.2 [0.00]***; -0.21 [0.00]***
  - Constant terms shown with p-values; AIC and BIC reported per model (AIC examples: 342.79, 337.24, 323.31, 336.15, 31.78, 20.85). Observations: 423, 422, 421, 430, 429, 429. Number of bankid: 62 (per model).

- Table 8: Hausman Specification Test for Foreign Ownership Dummy
  - Model 1 (Whole Sample): reported (b), (B), (b-B), diagonal standard errors for IV vs No IV on variables:
    - Foreign: IV = 90.31; No IV = -26.80; Diff = 117.11; Std. Error = 77.84
    - State: 78.70; -31.12; 109.82; 74.30
    - ROE (-1): -0.03; -0.05; 0.02; 0.01
    - Real GDP Growth: 3.54; 2.92; 0.62; 0.37
    - Test statistic: chi2(4) = 2.87; Prob>chi2 = 0.5801 (Ho: difference in coefficients not systematic)
  - Model 2 (Sample 2001-2005): similar layout with variables:
    - Foreign: IV = 173.74; No IV = 33.21; Diff = 140.53; Std. Error = 126.70
    - Customer Dep./Total Assets (-1): 0.85; 0.70; 0.15; 0.32
    - Interbank Liab./Total Assets (-1): 1.04; 1.20; -0.16; 0.18
    - Real GDP Growth: 2.01; 2.32; -0.31; 0.99
    - Test statistic: chi2(4) = 2.15; Prob>chi2 = 0.71
  - Notes: IV uses two-stage least-squares within estimator. Instrumented: Foreign. Instruments listed in source: Total Assets over GDP(-1), Net Interest Margin(-1), Loan Loss Provn/TA(-1), Total Capital Ratio (-1) and Spread.

### Parent banks, affiliated companies, and cross-border relationships (Table 9)
- Table 9: List of Parent Banks and their Affiliated Companies (selected entries preserved exactly as in source)
  - Austria: Ceska Sporitelna — Czech Republic — Last Accounting Date 12/2005
  - Austria: Erste Bank Hungary Rt — Hungary — Last Accounting Date 12/2004
  - Austria: Romanian Commercial Bank — Romania — Last Accounting Date 12/2005
  - Austria: Slovak Savings Bank — Slovakia — Last Accounting Date 12/2005
  - Austria: First Building Savings Bank — Slovakia — Last Accounting Date 12/2004
  - Raiffeisen International (Austria): Raiffeisenbank EAD — Bulgaria — 12/2004; Raiffeisen Bank Zrt — Hungary — 12/2005; Raiffeisen Bank SA — Romania — 12/2004
  - Germany: HVB Bank Biochim ad — Bulgaria — 12/2004; HVB Bank Czech Republic — Czech Republic — 12/2004; HVB Bank Hungary Rt. — Hungary — 12/2004; Bank BPH SA — Poland — 12/2004; UniBanka — Slovakia — 12/2005; HVB Bank Slovakia — Slovakia — 12/2004
  - France: SOCIÉTÉ GÉNÉRALE — Komercni Banka — Czech Republic — 12/2005; BRD-Groupe Societe Generale — Romania — 12/2004; SKB Banka DD — Slovenia — 12/2004
  - Belgium: KBC GROEP NV — Ceskoslovenska Obchodni — Czech Republic — 12/2005; K&H Bank — Hungary — 12/2005
  - Italy: Banca Intesa SPA — CIB Közép — Hungary — 12/2005; Vseobecna Uverova Banka — Slovakia — 12/2005
  - (Table 9 continued with additional parent–subsidiary entries across Sweden, Finland, Norway, Netherlands, US, Japan, Greece, Poland, Bulgaria, Romania, Estonia, Latvia, Lithuania, etc.; last accounting dates preserved as listed.)

### Summary statistics and sample characteristics (Tables 14–15)
- Table 14: Summary Statistics for the CEE Countries (selected variables)
  - Net Loan Growth (Obs 648): Mean 29.85; Std. Dev 43.72. Subsample means and counts reported for <0, ≠0, >0 tests.
  - Foreign Ownership Dummy (Obs 648): Mean 0.46; Std. Dev 0.50.
  - State Dummy (Obs 648): Mean 0.33; Std. Dev 0.47.
  - Total Assets over GDP (-1) (Obs 648): Mean 9.01; Std. Dev 15.29.
  - Customer Deposits / Total Assets (-1) (Obs 644): Mean 62.22; Std. Dev 18.93.
  - Interbank Liabilities / Total Assets (-1) (Obs 636): Mean 15.68; Std. Dev 15.85.
  - ROE (-1) (Obs 639): Mean 18.32; Std. Dev 74.47.
  - ROA (-1) (Obs 640): Mean 1.42; Std. Dev 3.70.
  - Cost to Income Ratio (-1) (Obs 625): Mean 64.30; Std. Dev 38.04.
  - Net Interest Margin (-1) (Obs 635): Mean 5.94; Std. Dev 5.16.
  - Loan Loss Provision/ Total Assets (-1) (Obs 614): Mean 1.22; Std. Dev 2.11.
  - Total Capital Ratio (-1) (Obs 434): Mean 16.94; Std. Dev 10.24.
  - Real GDP Growth Rate (Obs 634): Mean 4.05; Std. Dev 3.77.
  - Domestic Credit / GDP (-1) (Obs 628): Mean 38.85; Std. Dev 23.49.
  - Real Interest Rate (Obs 626): Mean -0.24; Std. Dev 9.42.
  - Spread 1 (Lending and Borrowing) (Obs 638): Mean 7.97; Std. Dev 8.24.
  - Spread 2 (Domestic and Euro) (Obs 626): Mean 9.95; Std. Dev 17.26.
  - Euro Area M3 Growth (Obs 657): Mean 6.01; Std. Dev 2.11.
  - Euro Area S-T Interest Rate (3 Month) (Obs 657): Mean 4.23; Std. Dev 2.00.
  - Notes on hypothesis tests and sample splits (S1: 1988-2000; S2: 2001-2005) are included in the source.

- Table 15: Summary Statistics for Parent Banks and their CEE Subsidiaries (selected entries)
  - Parent Banks (Obs 271): Net Loan Growth Mean 36.15; Std. Dev 39.04; Min -85.4; Max 196.5.
  - Subsidiary Banks (Obs 153): Net Loan Growth Mean 10.22; Std. Dev 24.6; Min -100; Max 94.09.
  - Ln(Total Assets) (-1): Parent Mean 20.95; Std. Dev 1.42; Subsidiary Mean 25.61; Std. Dev 1.17.
  - Customer Deposits / TA (-1): Parent Mean 61.16; Std. Dev 15.43; Subsidiary Mean 15.95? (values preserved as in source table layout).
  - Interbank Liabilities / TA (-1): Parent Mean 17.38; Std. Dev 14.37; Subsidiary Mean 17.65; Std. Dev 8.05.
  - Other indicators for parents and subsidiaries (Loan Loss Provision/ TA (-1), ROA (-1), ROE (-1), Net Interest Margin (-1), Cost to Income Ratio (-1), Total Capital Ratio (-1), Real GDP Growth, Domestic Credit / GDP (-1), Spread) reported with Obs, Mean, Std. Dev, Min, Max in the source table.

### Figures
- Figure 2 (Credit Growth across Bank Ownership Types in the CEE Countries)
  - Time series plotted 1992–2005 for ownership types: Foreign, State, Private, Domestic (BankScope as source).
  - Note: Figure excludes the two years before 1992 since data for domestically owned private banks does not exist for those two years.

*Italic source attribution: REFERENCES section from _wp08215 - REFERENCES (source PDF content provided).*

---


_Source: https://www.imf.org/-/media/websites/imf/imported-full-text-pdf/external/pubs/ft/wp/2008/_wp08215.pdf_