Abstract
This dissertation examines size, country and sector effects in the structure of international stock returns and the implications of these effects for capital asset pricing. To this end, we describe, in Chapter 1, the construction of a new international stock database containing reliable data on small firms, interest and exchange rates. The database is free of survivorship and size bias and contains stocks, interest and exchange rates from 39 countries, both developed and emerging, from the eighties to the millennium.
In Chapter 2 we replicate the Fama-French CAPM test and show that some design elements of that test can have a significant impact on the test results (e.g. frequency of portfolio updating). Especially crucial are the aspects related to the weight one gives to small, low-reputation stocks when constructing both the factor portfolios and the test or style portfolios whose returns are to be explained. We show that the standard Fama-French CAPM fails to explain the cross-section of stock returns in the presence of small stocks. To fit the observed returns we need to redesign the size and distress factor portfolios into two factor portfolios each, one for extremely small or distressed stocks relative to non-extreme stocks, and one for moderately small or distressed stocks versus larger or growth companies. This alternative six-factor model does a better job in pricing unmanaged test portfolios sorted on size, distress and momentum and two-dimensional test portfolios, both in the US and internationally, than the standard four-factor model with factor portfolios designed following Fama and French (1992, 1993, 1995, 1996a, 1996b, 1998, 2000), Carhart (1997), Jegadeesh and Titman (1993) and Rouwenhorst (1999). Stated differently, the alternative six-factor model gets the alphas closer tot zero than the standard four-factor model.
In Chapter 3 we focus on the Heston-Rouwenhorst (HR) methodology and the relative importance of country and sector effects in international stock returns. We adopt the HR-procedure to estimate world, country and industry factors and check the robustness of this procedure with respect to time period, country coverage, size coverage, sector classification and weighting schemes. We show that the average country-specific volatility in international stock returns is robustly larger than the average sector-specific variance and the world-factor variance. However, the relative magnitude widely depends on the test design. Especially, the country-specific volatility is boosted by adding emerging markets to the dataset. We take a closer look at the size variant of the test design and show why the variance of the country factor rises relative to the sector-specific volatility if one introduces small-caps into the data sample: these stocks have significantly more variability than large-caps when controlling for country and industry effects, and they are significantly less sensitive to their global sector index.
Apart from these data issues, we also consider methodological issues. If one would like to know what factor has the largest impact on the return of a randomly chosen stock, the HR-procedure is no longer appropriate because exposures come into the picture--to the extent that the distribution of the non-zero exposures to the world, country and sector factors are not equal across all stocks, which is statistically demonstrated. In other words, a study of var(??) instead of var(?) is appropriate. Hence we extend the HR-procedure in a Fama-MacBeth way to estimate the factor exposures. However, part of the estimated factor-generated variance var(??) is due to estimation error and needs t
Original language | English |
---|---|
Awarding Institution |
|
Supervisors/Advisors |
|
Publication status | Published - 2005 |