Asset pricing on Thailand & Malaysia stock exchange: on the use of macroeconomic and behavioral factors

Abstract

This thesis tests five macroeconomic variables that have been both theorized to affect stock returns and been proven to do so in past empirical research. Those variables are risk premium, industrial production, term structure, expected inflation, and unexpected inflation. The variables are retested for their statistical significance using four years of monthly contemporary data using Thailand and Malaysia as two of the five ASEAN markets (Singapore, Thailand, Philippines, Malaysia, and Indonesia). Contrary to previous studies, this study finds that the macroeconomic factors were not significant in explaining domestic market returns. Furthermore, principal component regressions outperformed cross-sectional ones, with factor analysis as the least statistically significant model. For the countries tested, the arbitrage pricing theory was also found to be a less robust pricing tool than the proposed sentiment model.

The sentiment model relies on investor behavior, using four investor groups (local, foreign, institutional, and dealer’s accounts) on the Stock Exchange of Thailand (SET). The daily net purchases of each group are used as leading indicators for sentiment. The sentiments are examined with relation to each other and market returns. Eight proven macroeconomic factors with known cross-sectional relationships and known to forecast with returns are examined as a benchmark for the newly proposed sentiment factor model. Retesting the factors allows for an apples to apples comparison with the proposed sentiment factors. Using a VAR framework this research finds that dealers predominantly sell to institutional accounts, creating a negative correlation between the two groups, in addition to strong institutional herding which is all indicative of potential agency problems on the exchange. Also find that local individual accounts practice negative feedback trading and the other groups practice positive feedback trading. Of the four groups, the only group that influences the SET is the local individual group of investors. The foreign investor is found to be the least significant group on market returns, provide market liquidity to locals, and be the least responsive to daily market changes-- following the prudent man rule. Lastly, propose a simple model which uses investor behavior to accurately predict the market’s direction for the following day 76 percent of the time with market timing ability. This can be useful for buying and shorting the market.