טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
M.Sc Thesis
M.Sc StudentBroitman Andrey
SubjectApplication of Bankruptcy Prediction Models of Stock
Returns Predictability
DepartmentDepartment of Industrial Engineering and Management
Supervisors Ms. Efrat Tolkowsky
Professor Uri Ben-Zion


Abstract

The relation between the probability of financial distress and stock return - is there relation between the two? According to classical financial theory, if financial distress is systematic risk, we expect to find a posituve relationship between stock return and probability of financial distress. It means that firms with higher probability of financial distress earn higher returns then firms with lower probability. In the current research I find negative relationship between probability of financial distress (expressed in Altman and Ohlson indexes) and stock return. At the same time I show the application of financial distress prediction models to stock returns predictability. The research is based on a sample of stock prices at the last decade of 20th century. The research method is based on Fama and MacBeth (1973) monthly regressions method and Fama and French (1992, 1993) portfolio analysis. Despite the finding of a counter intuitive negative relationship, I find a positive relation between stock returns and probability of financial distress found in the part of “firms that need attention” (Altman, 1968). Number of the “firms that need attention” was meager in the research sample and this demonstrates that the positive relationship exists in the short area of bankruptcy risk only. In addition to the main findings, I show again the declining influence of firm size to stock returns. On the other hand, the book-to-market effect on stock returns was clear and significant. At the end, I check the improved Ohlson model predictability that included regression running of improved Ohlson index and book-to-market value against stock returns. The regression supplied the explanation of 58 percents of the control sample variance.