We find evidence that monetary policy influenced bank lending in Turkey in the period 1991 -2007 both directly through the money lending channel and indirectly through the bank lending channel. The bank lending channel is shown to depend on two bank characteristics, namely liquidity and capital. We also find that both capital and GDP growth have plausible positive and significant long-run effects on bank loan growth, whereas inflation, bank size and, in particular, efficiency do not have a significant equilibrium relationship with loan growth. This latter result is despite our finding that the efficiency of all Turkish banks improved over the period. It is also evident that domestic banks are, unexpectedly, found to be more efficient, on average, than foreign banks. We discovered no evidence of significant dynamics or fixed-effects in the growth of loans and so prefer to use the pooled OLS estimator over the fixed-effects and Arellano and Bond estimators. We therefore caution against assuming the existence of fixed-effects and dynamics in such models as this may adversely affect inference.

We find evidence that monetary policy influenced bank lending in Turkey in the period 1991 -2007 both directly through the money lending channel and indirectly through the bank lending channel. The bank lending channel is shown to depend on two bank characteristics, namely liquidity and capital. We also find that both capital and GDP growth have plausible positive and significant long-run effects on bank loan growth, whereas inflation, bank size and, in particular, efficiency do not have a significant equilibrium relationship with loan growth. This latter result is despite our finding that the efficiency of all Turkish banks improved over the period. It is also evident that domestic banks are, unexpectedly, found to be more efficient, on average, than foreign banks. We discovered no evidence of significant dynamics or fixed-effects in the growth of loans and so prefer to use the pooled OLS estimator over the fixed-effects and Arellano and Bond estimators. We therefore caution against assuming the existence of fixed-effects and dynamics in such models as this may adversely affect inference.