This paper examines the relationship between problem loans and cost efficiency in commercial banks using Granger-causality techniques. The study tests four hypotheses regarding the intertemporal relationships among loan quality, cost efficiency, and bank capital. The data suggest that problem loans precede reductions in measured cost efficiency, and that measured cost efficiency precedes increases in problem loans. Additionally, reductions in capital at thinly capitalized banks precede increases in problem loans. These findings indicate that cost efficiency may be an important indicator of future problem loans and problem banks. The results are ambiguous regarding whether researchers should control for problem loans in efficiency estimation. The paper explores the connection between problem loans and cost efficiency, highlighting that failing banks tend to be located far from the best-practice frontier and have low cost efficiency. Studies also show negative relationships between efficiency and problem loans even among non-failing banks. Cost-inefficient banks may have loan performance problems due to poor management or external events. The study also finds that supervisory examination data show a positive relationship between cost efficiency and management quality ratings. Furthermore, some studies directly include measures of nonperforming loans in cost or production relationships to control for extra costs or underwriting and monitoring expenditures. The paper presents four hypotheses: 'bad luck,' 'bad management,' 'skimping,' and 'moral hazard.' These hypotheses suggest different intertemporal relationships between problem loans and cost efficiency. The study uses Granger-causality analysis to test these hypotheses, finding that increases in nonperforming loans Granger-cause decreases in measured cost efficiency, consistent with the 'bad luck' hypothesis. Conversely, low levels of cost efficiency Granger-cause increases in nonperforming loans, consistent with the 'bad management' hypothesis. The 'skimping' hypothesis suggests that measured cost efficiency positively Granger-causes nonperforming loans, while the 'moral hazard' hypothesis suggests that low capital Granger-causes high nonperforming loans. The study estimates cost efficiency using a stochastic cost frontier approach, finding that the average bank is measured to be about 92% efficient. The results suggest that cost efficiency is negatively associated with nonperforming loans, and that low capital ratios are associated with higher levels of nonperforming loans. The findings have implications for economic policy, suggesting that prudential regulation and supervision could reduce the risk of failure by limiting banks' exposures to external shocks or by better insulating banks from external shocks. The study also highlights the importance of considering cost efficiency in bank supervision and research, as well as the need to measure efficiency accurately in the presence of nonperforming loans. The results suggest that cost efficiency is more appropriately measured using several years of data, and that long-run profit efficiency should be considered when evaluating bank efficiency.This paper examines the relationship between problem loans and cost efficiency in commercial banks using Granger-causality techniques. The study tests four hypotheses regarding the intertemporal relationships among loan quality, cost efficiency, and bank capital. The data suggest that problem loans precede reductions in measured cost efficiency, and that measured cost efficiency precedes increases in problem loans. Additionally, reductions in capital at thinly capitalized banks precede increases in problem loans. These findings indicate that cost efficiency may be an important indicator of future problem loans and problem banks. The results are ambiguous regarding whether researchers should control for problem loans in efficiency estimation. The paper explores the connection between problem loans and cost efficiency, highlighting that failing banks tend to be located far from the best-practice frontier and have low cost efficiency. Studies also show negative relationships between efficiency and problem loans even among non-failing banks. Cost-inefficient banks may have loan performance problems due to poor management or external events. The study also finds that supervisory examination data show a positive relationship between cost efficiency and management quality ratings. Furthermore, some studies directly include measures of nonperforming loans in cost or production relationships to control for extra costs or underwriting and monitoring expenditures. The paper presents four hypotheses: 'bad luck,' 'bad management,' 'skimping,' and 'moral hazard.' These hypotheses suggest different intertemporal relationships between problem loans and cost efficiency. The study uses Granger-causality analysis to test these hypotheses, finding that increases in nonperforming loans Granger-cause decreases in measured cost efficiency, consistent with the 'bad luck' hypothesis. Conversely, low levels of cost efficiency Granger-cause increases in nonperforming loans, consistent with the 'bad management' hypothesis. The 'skimping' hypothesis suggests that measured cost efficiency positively Granger-causes nonperforming loans, while the 'moral hazard' hypothesis suggests that low capital Granger-causes high nonperforming loans. The study estimates cost efficiency using a stochastic cost frontier approach, finding that the average bank is measured to be about 92% efficient. The results suggest that cost efficiency is negatively associated with nonperforming loans, and that low capital ratios are associated with higher levels of nonperforming loans. The findings have implications for economic policy, suggesting that prudential regulation and supervision could reduce the risk of failure by limiting banks' exposures to external shocks or by better insulating banks from external shocks. The study also highlights the importance of considering cost efficiency in bank supervision and research, as well as the need to measure efficiency accurately in the presence of nonperforming loans. The results suggest that cost efficiency is more appropriately measured using several years of data, and that long-run profit efficiency should be considered when evaluating bank efficiency.