Εφαρμογή και έλεγχος ισχύος υποδειγμάτων αποτίμησης περιουσιακών στοιχείων στο διεθνές περιβάλλον
Τσαγκάνης, Κωνσταντίνος Σ.
KeywordsΥπόδειγμα αποτίμησης κεφαλαιακών στοιχείων ; Παράγοντες κινδύνου ; Γραμμική παλινδρόμηση ; Μετοχές ; Αποδόσεις μετοχών ; Μαθηματικά μοντέλα ; Capital assets pricing model ; Risk factors ; Linear regression ; Stock market ; Stock returns ; Mathematical models
The traditional Capital Asset Pricing Model (CAPM), where the only explanatory variable is the market factor, has received much criticism because of its failure in several empirical tests. That led Fama and French to the formulation of the three-factor, and, more recently, the five-factor model. Both these models are, in essence, extensions of the traditional CAPM. Specifically, the five-factor model includes, apart from the market factor, four additional risk factors that are related to size, book-to-market equity, operating profitability, and investment, respectively. These four factors are called “anomaly” variables, since, theoretically, they should not be able to explain average stock returns. This thesis includes some empirical tests for the Fama & French five-factor model. A sample of stocks from three major United States stock exchanges (NYSE, AMEX, NASDAQ) is used, for two recent successive seven-year periods, one before and one during the current economic recession (07/2001 – 06/2008 and 07/2008 – 06/2015). The goal is to identify any differences between these two periods, in regard to the general characteristics of the US stock market, as well as to the performance of the model. The methodology that is used is the same as the one in Fama & French (2015), and is based on time-series (first pass) regressions and mimicking portfolios that are proxies for the four additional risk factors. The results show that, during the second, more recent, sevenyear period, the market tends to become more efficient, since the effect of the anomaly variables is mitigated. Apart from that, during the same period, the investors’ risk aversion has increased. However, the model performs equally well for both sub-periods. Moreover, based on the Fama & French findings that the value factor becomes redundant when the profitability and investment factors are added in the five-factor model, besides the complete five-factor model, another four-factor model is tested, which is derived from the complete model when the value factor is omitted. These additional tests are conducted to check whether the Fama & French findings are confirmed for the two sub-periods that are used in this thesis. The results, despite of being a bit vague, partially support the retention of the value factor in the model, especially during the more recent second period.