All Theses and Dissertations
Degree
Doctor of Philosophy in Economics
Faculty / School
School of Economics and Social Sciences (SESS)
Department
Department of Economics
Date of Award
Winter 2024
Advisor
Dr. Wali Ullah, Professor, Department of Economics, School of Economics and Social Sciences (SESS)
Committee Member 1
Dr. Syed Kalim Hyder Bukhari, Examiner-I, State Bank of Pakistan
Committee Member 2
Dr. Amena Urooj, Examiner-II, PIDE Islamabad
Project Type
Thesis
Access Type
Restricted Access
Pages
xiii, 127
Keywords
Time Varying Betas, Sector Beta, Regime switching, Kalman filter, Macroeconomic factors, Spike and slab
Subjects
Economics, Financial economics
Abstract
The theory of asset pricing is based on present value calculations and the efficient capital markets hypothesis. It suggests that an asset's price, not just stocks, depends on the expected future returns discounted to the present. However, a significant unanswered question in investments is why certain assets earn much higher average returns. To address this, financial economists have developed various models like CAPM, three factor, four factor, five factor models and many other multifactor asset pricing models. Yet, due to empirical limitations, researchers have tried to refine the theoretical basis and improve their performance. All these asset pricing models, assumes stable beta coefficients, but this is not true in reality. Beta instability in asset pricing models can be attributed to several factors. Firstly, the variability in a firm's cash flow during business cycles plays a significant role in causing fluctuations in their stock betas. As business cycles are influenced by technology or taste shocks, there are shifts in the relative shares of various sectors within the economy. Moreover, changes in firm leverage, which are triggered by fluctuations in stock prices, also contribute to alterations in beta values. When a company's capital structure changes, it affects the sensitivity of its stock returns to market movements, leading to changes in beta coefficients. In this thesis, we use different econometric techniques to capture the time variation of betas. Additionally, we propose a new multifactor model that includes both fundamental and macroeconomic variables to account for the information conveyed by the instability of betas. Our analysis focuses on stocks traded on the Pakistan stock exchange, covering an extended period and various sector portfolios. Furthermore, we study equity premium predictability in selected South Asian capital markets and examine the robustness of common predictive variables. In the first chapter, our research focuses on investigating time-varying beta parameters in the Pakistan Stock Exchange (PSX). We employ three distinct econometric techniques, namely rolling, regime switching, and Kalman filter, to estimate time-varying beta values at the sector level and evaluate the performance of these models by using the MAE, MSE, and MCS methods. The main results are as follows: Pakistan's sector betas exhibited high volatility over time, indicating that time-varying beta estimates are more suitable for constructing low risk portfolios and accurate long-term forecasting. The time-varying sector betas are best described by the family of state space estimation techniques. The second chapter uses fundamentals and macroeconomics factors and studies its ability to predict sector returns on using data from 1995–2021. The approach taken in this chapter differs from existing literature is that we consider the time-varying sensitivity of each factor is treated as a series of random processes. By constructing a factor model based on sectors, the potential issue of measurement errors in independent variables, such as factor sensitivities in a cross-sectional context, is minimized. We find that market factors are significant for all sectors except for Travel & Leisure, while fundamental factors are significant at the 1% ,5% and 10% levels across all sectors. This suggests that market and fundamental factors play a crucial role in explaining sector returns. In chapter 3, we study how to predict equity premium (the difference between stock returns and risk-free returns) in selected South Asia capital markets. We look at various factors that could potentially help to predict these premiums, such as the dividend-price ratio, three interest rate variables (short run interest rate, relative interest rate and yield spread), and inflation. To get accurate estimates, we use two methods. First, we analyse all capital markets together as a system of regressions. Second, we take into account uncertainty about which predictors are the most useful by using a technique called "spike-and-slab prior." Our findings indicate that considering uncertainty in the prediction models is crucial, as it significantly improves the utility for investors. Out of all the factors we studied, we found that the dividend price ratio stands out as the most reliable predictor of future stock returns.
Recommended Citation
Hussain, H. (2024). Three essays in Empirical Multifactor Asset Pricing Models (Unpublished doctoral dissertation). Institute of Business Administration, Pakistan. Retrieved from https://ir.iba.edu.pk/etd/91
