| Summary: | 碩士 === 輔仁大學 === 金融研究所 === 94 === This thesis is based on Moody’s KMV model. To calculate the volatility of stock returns, we use the exponential weighted moving average model（EWMA）to capture the dynamic feature of volatility with the latest observation carrying the highest weight.
In practical applications, we find a major problem with respect to the KMV model. We should incorporate the expected operating results of companies in the KMV model, in particular, the downside risk. In theory, the most critical assumption of the Black-Scholes model is that it is applied to options in the financial market and thus, a riskless, arbitrage-free portfolio can be constructed. That is, the model provides a risk-neutral solution without having to considering the risk premium or the expected return of the stock. However, when the model is applied to predict corporation default probabilities, the setting is less likely to be risk-neutral and may have to take the expected operating results of corporations into consideration. This thesis adopts a non risk-neutral modification with a simplifying assumption that the company’s operating result will remain negative when it suffered two consecutive annual operating losses. Then, we calculate the trend of losses to estimate the expected loss by adding it to the total borrowing and re-estimate the default distance.
The major results are as follows. First, the parameter of the exponentially weighted moving average model, the decay factor, is 0.928311959. Second, we use Cluster Analysis to analyze the power of these two models. We find that KMV model adjusted for expected loss improves significantly over the original model. Third, we draw the ROC curve and find the same result as Cluster Analysis. Fourth, we find that incorporating the effect of expected loss is essential for the group of high credit risk corporations. Finally, the KMV model adjusted for expected loss has higher explanatory power than the KMV model with respect to TEJ credit risk ratings.
|
|---|
