In addition, these models can also be used as effective stress-testing tools for bank supervision. Since these models offer a method to link the bank-specific and macroeconomic factors to the future loan quality, they can also be used to forecast the loan quality in an adverse situation, depending on the macroeconomic factors input into the model. For a severe downturn scenario, such as negative GDP or significant worsening in borrower’s debt serviceability or extreme price increase, the models can offer the estimation of what the loan portfolio will look like and whether banks have enough provision and capital to cushion for the worst case. In addition, these models tend to yield conservative estimates of the quality of loans (or the loan characteristics that are worse than they actually are) because the predicted SM/NPL growth is driven by the bank-specific and macroeconomic factors and has not yet taken into account banks’ risk management actions that tend to improve the loan quality, unless there is an unprecedented structural shift in bank lending behavior, such as the populist policy we see in the Thai banking sector last year.