Empirical Banking - Finance and Growth - 2024-2025 PDF

Empirical Banking Finance and Growth KING and LEVINE Stefano Caiazza 2024-2025 King & Levine (K&L) – The Research Question and main findings K&L investigate whether higher levels of financial development are correlated with faster current and future rates of economic growth,...

Empirical Banking Finance and Growth KING and LEVINE Stefano Caiazza 2024-2025 King & Levine (K&L) – The Research Question and main findings K&L investigate whether higher levels of financial development are correlated with faster current and future rates of economic growth, physical capital accumulation, and economic efficiency improvements. Main findings (after controlling for initial conditions and other economic indicators): (1) Higher levels of financial development are positively associated with faster rates of economic growth, physical capital accumulation, and economic efficiency [“CAUSAL INFERENCE”] (2) We find that the predetermined component of financial development is a good predictor of long-run growth over the next 10 to 30 years [PREDICTIONS] To do this, they must empirically define financial development. Four indicators: (1) Ratio of liquid liabilities to GDP (LLY) – M3/GDP – Measure of size (2) Ratio of total bank deposits to Central Bank total asset (BANK) – [bank-dep / (bank-dep + CB tot assets) – Measure of relative importance (3) Ratio of credit to non-financial firms to total credit (excluding credit to banks) (PRIVATE) – Measure of domestic asset distribution (4) Credit issued to non-financial firms to GDP (PRIVY) – Measure of domestic asset distribution  King & Levine – Measuring Growth K&L decompose per capita growth rate into two components: The rate of real per capita physical capital stock, k and Everything else, x. Given the following production function, taking the log and differencing: ∂ ln y y = kα x → ≡ GYP( growth rate of GDP) = α GK + EFF ∂t Where EFF represents the growth rate of all factors that influence GDP growth (technology, human capital, number of hours worked per worker) but capital accumulation. K&L use four growth indicators: The growth rate of GDP (GYP) The growth rate of physical capital (GK) The growth rate of other factors different from GK (EFF) The gross national investment over GDP (INV)  King & Levine – Data They execute: (1) A purely cross-country analysis using data averaged from 1960-1989; (2) A pooled cross-country, time-series study using data averaged over the 1960s, 1970s, and 1980s so that each country has three observations, data permitting. (1) (2) code Country Name btoti btot dbyi dby ARG Argentina 63,8889 74,1596 18,4849 17,8655 code name year growth initial AUS Australia 83,0754 92,6655 38,1558 47,8042 ARG Argentina 1960 0,03 3535,93 AUT Austria 97,318 98,4359 40,5994 81,362 BGD Bangladesh 86,3752 22,3775 ARG Argentina 1970 0,02 4008,68 BRB Barbados 91,8502 43,8783 BEL Belgium 78,4184 92,0092 26,1253 58,1226 ARG Argentina 1980 0,01 4530,64 BOL Bolivia 10,2146 31,5974 1,6661 10,8131 AUT Austria 1960 0,03 5154,89 BRA Brazil 47,4771 61,6758 10,7361 17,7985 CAN Canada 80,7844 88,997 28,40882112 42,34922944 AUT Austria 1970 0,05 6130,24 CHL Chile 58,565 52,8085 17,6089 27,6462 COL Colombia 74,3253 80,1646 10,7657 13,7608 AUT Austria 1980 0,04 7751,57 CRI Costa Rica 97,3479 72,8176 22,6759 20,5687 BEL Belgium 1960 0,04 5866,19 CYP Cyprus 93,0962 92,72 40,5958 55,117 DNK Denmark 86,0352 88,0999 47,1013 49,8734 BEL Belgium 1970 0,04 7228,21 DOM Dominican Republic 73,6196 73,3352 11,6832 16,3839 ECU Ecuador 64,7281 62,1616 13,7053 13,5746 BEL Belgium 1980 0,03 8974,31 SLV El Salvador 90,7292 72,0441 23,2069 23,5761 BOL Bolivia 1960 -0,03 897,74 FJI Fiji 98,9383 96,9463 29,0619 FIN Finland 95,8457 97,2227 36,8145 53,2541 BOL Bolivia 1970 0,00 754,23 FRA France 89,4408 96,5411 24,3025 62,5041 DEU Germany 93,5074 97,571 45,6712 88,8946 BOL Bolivia 1980 0,01 754,98 King & Levine – Descriptive Statistics  King & Levine – Descriptive Statistics Correlation: It measures the strength of the relationship between variables.  Regression Analysis g ( i ) = α + β fd ( i ) + γ C ( i ) + ε ( i ) Where: g is the average growth rate of a country fd is a vector of indicators of the financial development of the country C represents various conditioning variables to control for other factors associated with growth ε is a disturbance term  Regression Analysis g ( i ) = α + β fd ( i ) + γ C ( i ) + ε ( i ) i = 1,....., N Cross-country regression g ( i, t ) = α + β fd ( i, t ) + γ C ( i, t ) + u ( i, t ) i = 1,....., N , t = 1,....., T Pooled regression Where: g is the real per capita growth rate fd measures financial development C are control variables, i.e., the log of initial real per capita GDP (convergence); the log of initial secondary school enrollment rate (proxy of human capital investments): theory and evidence suggest an important link between long-run growth and initial income and investment in human capital accumulation; other macroeconomic variables. K&L does not use White’s heteroskedastic standard errors but sensitivity analysis. Cross-country regressions The Financial System (banking system) matters.  Sensitivity Analyses K&L report that the results are robust to different tests: Subsamples of countries and periods (OECD, Sub-Saharan, Latin America) Estimators (pooled) Estimation techniques (IV regression. Instruments: initial level of financial development ratios) Heteroskedasticity Extreme Bound Analyses (EBA) – Change the right-hand side variables, observing what happens to the core variable Omitting outliers (omitting countries with extremely high or low variable values also does) Cross-section regressions OLS estimates with initial values of growth and banking sector Pooled Data  Conclusion K&L study the empirical link between a range of financial development indicators and economic growth. They find that: (1) Indicators of the level of financial development – the size of the formal financial intermediary sector relative to GDP, the importance of banks relative to the central bank, the percentage of credit allocated to private firms, and the ratio of credit issued to private firms to GDP – are strongly and robustly correlated with growth, the rate of physical capital accumulation, and improvements in the efficiency of capital allocation; and (2) The predetermined components of these financial development indicators significantly predict subsequent values of the growth indicators. The data are consistent with the view that financial services stimulate economic growth by increasing the rate of capital accumulation and by improving the efficiency with which economies use that capital. We do not, however, link specific financial sector policies with long-run growth. We can confidently make a policy recommendation by relating measures of executable government policies to subsequent growth.  Comments Independent variables (1) Ratio of liquid liabilities to GDP (LLY) – M3/GDP – Measure of size (2) Ratio of total bank deposits to Central Bank total asset (BANK) – [bank-dep / (bank-dep + CB tot assets) – Measure of relative importance (3) Ratio of credit to non-financial firms to total credit (excluding credit to banks) (PRIVATE) – Measure of domestic asset distribution (4) Credit issued to non-financial firms to GDP (PRIVY) – Measure of domestic asset distribution The FED Balance Sheet https://www.federalreserve.gov/monetarypolicy/bst_recenttrends.htm The ECB Balance Sheet  Comments (2) We find that the predetermined component of financial development is a good predictor of long-run growth over the next 10 to 30 years [PREDICTIONS] K&L decompose growth into two components: The rate of physical capital accumulation, k and Everything else, x. ∂ ln y α y=k x→ ≡ GYP( growth rate of GDP) = α GK + EFF ∂t Dependent variables : The growth rate of GDP (GYP) The growth rate of physical capital (GK) The growth rate of other factors different from GK (EFF) The gross national investment over GDP (INV)  Comments They run: (1) A purely cross-country analysis using data averaged from 1960-1989; (2) A pooled cross-country, time-series study using data averaged over the 1960s, 1970s, and 1980s so that each country has three observations, data permitting. (1) (2) code Country Name btoti btot dbyi dby ARG Argentina 63,8889 74,1596 18,4849 17,8655 code name year growth initial AUS Australia 83,0754 92,6655 38,1558 47,8042 ARG Argentina 1960 0,03 3535,93 AUT Austria 97,318 98,4359 40,5994 81,362 BGD Bangladesh 86,3752 22,3775 ARG Argentina 1970 0,02 4008,68 BRB Barbados 91,8502 43,8783 BEL Belgium 78,4184 92,0092 26,1253 58,1226 ARG Argentina 1980 0,01 4530,64 BOL Bolivia 10,2146 31,5974 1,6661 10,8131 AUT Austria 1960 0,03 5154,89 BRA Brazil 47,4771 61,6758 10,7361 17,7985 CAN Canada 80,7844 88,997 28,40882112 42,34922944 AUT Austria 1970 0,05 6130,24 CHL Chile 58,565 52,8085 17,6089 27,6462 COL Colombia 74,3253 80,1646 10,7657 13,7608 AUT Austria 1980 0,04 7751,57 CRI Costa Rica 97,3479 72,8176 22,6759 20,5687 BEL Belgium 1960 0,04 5866,19 CYP Cyprus 93,0962 92,72 40,5958 55,117 DNK Denmark 86,0352 88,0999 47,1013 49,8734 BEL Belgium 1970 0,04 7228,21 DOM Dominican Republic 73,6196 73,3352 11,6832 16,3839 ECU Ecuador 64,7281 62,1616 13,7053 13,5746 BEL Belgium 1980 0,03 8974,31 SLV El Salvador 90,7292 72,0441 23,2069 23,5761 BOL Bolivia 1960 -0,03 897,74 FJI Fiji 98,9383 96,9463 29,0619 FIN Finland 95,8457 97,2227 36,8145 53,2541 BOL Bolivia 1970 0,00 754,23 FRA France 89,4408 96,5411 24,3025 62,5041 DEU Germany 93,5074 97,571 45,6712 88,8946 BOL Bolivia 1980 0,01 754,98 Comments  King & Levine – Descriptive Statistics Correlation: It measures the strength of the relationship between variables. PRIVATE  Regression Analysis g ( i ) = α + β fd ( i ) + γ C ( i ) + ε ( i ) i = 1,....., N Cross-country regression g ( i, t ) = α + β fd ( i, t ) + γ C ( i, t ) + u ( i, t ) i = 1,....., N , t = 1,....., T Pooled regression K&L use White’s heteroskedastic standard errors only in the sensitivity analysis. Since the error term is likely to be correlated over time for a given country (in this case, the usual reported standard errors should not be used as they can be greatly downward biased, and OLS are no longer “best”), you are required to use cluster standard errors. Cross-country Regression The Financial System (banking system) matters. Cross-country Regression Analysis with Initial Values of Growth And Banking Sector Cross-country Regression Pooled Data