Factor models and market anomalies - evidence from Borsa Istanbul

In this thesis, we examine the performances of traditional and recent factor models of asset pricing such as Capital Asset Pricing Model (CAPM) of Sharpe (1964), three- and five-factor models of Fama and French (1993 and 2015) and q-factor model of Hou, Xue and Zhang (2016) against market anomalies such as size, value, beta, debt-to-equity, profitability, accruals, corporate investment and momentum for a developing market from July 2006 to December 2015. We are motivated from the CAPM theory, Fama and French (2006) rational pricing equation, q-theory of investment and prominent empirical evidence in choosing models and anomaly variables. For the first time in literature, we test the performance of the five-factor model that is augmented with a cash-based operational profitability based factor, a profitability measure which is completely free from accounting accruals. Recently added profitability and investment factors are motivated from the Dividend Discount Model and these variables are expected to proxy dividends under clean surplus assumption. Whereas, researchers test the performance of the model by using accrual based profitability factors. However, accruals are adjustment by accountants to transform cash earnings to accounting numbers, hence; they are not relevant to dividend. Consistent with the motivation behind the model, we test the five-factor model with a cash-based profitability factor. Initially, we implement univariate portfolio analysis to examine the relationship between the anomaly variables and expected stock returns. Our portfolio analysis reveals positive monthly (risk adjusted) premiums ranging from 0.17% (0.06) to 1.69% (1.65) for zero-investment portfolio strategies. These premiums are economically larger (comparable) than those of developed (developing) markets. However, due to high volatility of stock returns only value and beta effects are statistically significant. In addition, the relationship between average stock returns and profitability and average stock returns and corporate investment variables found as weak. We then investigate the pricing performance of models. Results reveal that the main problem for these models are their failure to explain the pricing behavior on a portfolio of value stocks whose returns behave like those of small and unprofitable firms that invest aggressively. Additionally, consistent with the findings of the portfolio analysis, recently added profitability and investment factors improve neither the pricing nor the economic performances of the three-factor model. We supplement this finding with portfolio spanning tests and factor optimization. Therefore, even with a cash-based profitability factor, we can conclude that the clean surplus relationship cannot represent dividends and fails to explain common stocks' pricing behavior for an emerging market. The Q-theory of investment fails as well.