Determinants of stock prices in the Egyptian stock market: Traditional asset pricing models versus behavioural asset pricing models

Abdou, Rabab

Abstract

The aim of this thesis is to determine a valuation model for stocks in the Egyptian stock market by comparing conventional and behavioural asset pricing models. To achieve this aim, this thesis constructs and tests the following extensions of the Fama and French three-factor model over the time period 2004-2016: (i) time-varying factor loadings; (ii) time-varying risk premia; and (iii) introducing a behavioural risk factor.
The cross-sectional tests applied on both individual stocks and portfolios double-sorted on size and the book-to-market ratio show that the Fama and French three-factor model that captures time-variation in betas using either the rolling regression approach or the DCC-GARCH model cannot fully capture the cross-sectional variation in stock returns as both specifications have high and significant pricing errors. Similarly, scaling the factor loadings in the Fama and French three-factor model using the Treasury bill rate, size, the book-to-market ratio and sentiment does not enable the model to capture some of the prominent anomalies in financial markets such as turnover and short-term momentum effects.
Modelling time-variation in risk premia, based on simple bull and bear regimes identified using a Markov-switching model, along with time-variation in risk using the DCC-GARCH provides a modest improvement to the results of the model that only captures the time-variation in risk. Specifically, although the hypothesis of time-varying risk premia is never rejected, the model is still weakened by the negative weighted average risk premia of the market factor and the high pricing errors. Finally, the results show that augmenting the Fama and French three-factor model with an additional behavioural factor does not lead to major changes in the performance of the model and that the sentiment risk factor is not significantly priced in the Egyptian stock market. However, by investigating the characteristics of stocks that are most sensitive to changes in sentiment, the results reveal that small and highly volatile are the most sensitive stocks which imply that sentiment is a non-diversifiable risk factor.

• Thesis Type: Thesis
• Deposit Date: Mar 13, 2019
• Publicly Available Date: Mar 13, 2019
• Keywords: Conditional Fama and French Three Factor Model, Behavioural Asset Pricing Models
• Public URL: https://uwe-repository.worktribe.com/output/850577
• Contract Date: Mar 13, 2019
• Award Date: Mar 13, 2019