DEFAULT PROBABILITY MODELING FOR AGRICULTURAL LOANS OF THE DEVELOPMENT BANK OF ETHIOPIA

Abstract

Credit risk is the most prominent risk facing banks. Its effective management is vital for banks success. Banks are expected to improve their credit risk management system due to increasing financial loss resulting from loan default. Regulators also emphasized the importance of quantification and credit risk modeling. Currently, credit risk management has become an important topic for financial institutions, since the business of financial service is highly associated with uncertainty. However, credit risk model for agricultural loan is still in its infancy stage. The general objective of this study was to model agricultural loan default probability after examining significant factors determining default. The objective was accomplished by conceptualizing a theory of loan default for agricultural borrowers and deriving a model predictive of loan default. About 322 firmyear observations spanning the time period 2007 to 2013, consisting of balance sheet and gain and loss account of a particular firm for a particular year were used in the study. A binary logit model was used to analyze the relationships between historical data available at loan origination time and loan performance. The result indicated a strong and direct relationship between key financial variables and probability of default. Leverage, liquidity, profitability and debt coverage ratio at loan origination were found to be good indicators of the probability of default. However, loan size, loan duration and farm type were not statistically significant in explaining agricultural loan default probability. The derived default probability model is applicable to agricultural loans which could be used as a benchmark for agricultural lending banks when setting internal rating models. Banks can provide special service required to help avoid default among those borrowers considered more likely to default by developing a more sophisticated default model