The estimation of credit loss distribution of Iran's banking industry using stress test

In this paper, we estimate losses due to credit risk using stress test and calculate the minimum capital requirements under distressed and baseline scenarios. We use quarterly data from 2003:Q1 to 2016:Q2 (50 observations) for Iran’s banking industry and the economy. In the first step, the default probabilities are estimated using a series of macroeconomic variables. Then, the default probabilities are simulated for one year time horizon using Monte-Carlo method. And finally, by using Loss Given Default (LGD) values and Exposed at Default (EAD) amounts, portfolio loss distribution is calculated. For this purpose, a hypothetical portfolio will be build. EAD for each loans is drawn randomly and uniformly distributed and LGD is given a fixed value.
To estimate default probability equation, in addition to the Wilson’s linear model, Quantile regressions are used. The results show that the loss distributions for all scenarios are skewed to the right. The amount of loss in the 50th quantile regression is closed to the Wilson model, but the loss in the 10th and 90th quantile regressions are different. In fact, the Wilson’s model overestimate the loss of 10% qunatile regression and underestimate the loss of 90% qunatile regression.