مدل های CAPM بر اساس مصرف: شواهد بین المللی

We examine the performance of several types of the consumption-based CAPM (C-CAPM) models to explore if consumption factors matter for determining excess returns across 17 MSCI country indexes. While the classic world C-CAPM does exhibit some power in explaining cross-sectional variations of expected excess returns, the model seems to require an implausibly large coefficient of risk aversion. The more sophisticated models including the heterogeneous C-CAPM, the world surplus consumption and the habit-formation models provide more reasonable estimates and add substantial explanatory power for the variation in the cross section of excess stock returns. Our results suggest that country-specific consumption risk is not fully diversified thus implying that stock returns are related to idiosyncratic consumption risk.

The consumption-based CAPM (C-CAPM) has become a popular approach in the study of asset pricing since the classic work of Lucas, 1978 and Breeden, 1979, and Grossman and Shiller (1981) developed a simple relation of consumption to asset returns. Cochrane (2001, p. 5) notes, “An investor must decide how much to save and how much to consume, and what portfolios of assets to hold…the asset’s price should equal the expected discounted value of the asset’s payoff, using the investor’s marginal utility to discount the payoff”. Empirical research on the world CAPM, where excess world returns play a role, is typically carried out using international asset pricing (Harvey, 1991, De Santis and Gerard, 1997, Li and Zhong, 2005, Li and Zhong, 2009 and Bekaert et al., 2009). Other studies incorporate features from several models. Cho et al. (1986) consider factors from the international arbitrage pricing theory and find support for three or four factors. Ferson and Harvey (1993) include global economic factors, while Dumas and Solnik (1995) and De Santis and Gerard (1998) take account of exchange rate risk. As discussed in Karolyi and Stulz (2003), the consumption-based asset pricing model appears superior to the world CAPM where differences in consumption baskets across markets is an important feature for determining cross-sectional variations in stock returns. The world CAPM is essentially a special case of the C-CAPM which assumes that goods and financial markets are perfect and investors have similar consumption patterns and access to the same investment opportunity sets. In this study, we explore the issue of whether consumption-based variables are related to market returns internationally, i.e. whether variations in expected market returns across countries can be explained by their differing consumption risk exposures. Our paper contributes to the literature in several ways. First, many prior tests of the C-CAPM are confined to a domestic setting, although theory suggests that the model is equally applicable across countries. We conduct our tests of consumption-based models after collecting the seventeen market indexes of the Morgan Stanley Capital International (MSCI). International finance research (e.g., Bekaert and Harvey, 1995) typically assumes three types of market linkage. They are: (1) complete integration, (2) complete segmentation, and (3) mild segmentation. In a completely integrated market, asset prices are determined by a world stochastic discount factor (SDF). This implies that the SDFs for asset returns in each country should be related to aggregate world factors rather than to country-specific variables. Under complete segmentation, individual markets are perfectly segmented and tests of asset pricing models use domestic data. In turn, under mild segmentations, local markets are partially integrated and thus asset returns are related to both domestic and world factors. Second, this study provides a comprehensive set of tests for the consumption-based CAPM. Research work on the consumption-based CAPM in an international context remains scant although prior studies provide some evidence. For instance, using the Morgan Stanley Capital International (MSCI) market indexes, Cumby (1990) employs a consumption-based international asset pricing model for four developed economies (US, UK, Germany, and Japan). Sarkissian (2003) tests the C-CAPM in foreign exchange markets under the assumption of imperfect risk sharing across countries. Li and Zhong (2005) apply habit-formation models in an international setting to investigate cross-sectional returns and their predictability. More recently, Li and Zhong (2009) incorporate both habit formation and uninsurable idiosyncratic risks into an international C-CAPM to examine returns in equity and currency markets. We begin our analysis with the classic world C-CAPM under the assumption of complete market integration and complete risk sharing. World consumption growth becomes the only factor determining asset returns in the classic world C-CAPM given that representative investor has a power utility. However, with imperfect consumption risk sharing across countries, the investors may face persistent consumption shocks. Consequently, the cross-sectional variance of consumption growth can affect asset pricing (Constantinides and Duffie, 1996). Hence, in the heterogeneous C-CAPM, asset returns are not only related to world consumption growth, but also related to cross-country variations in consumption growth. Sarkissian (2003) uses the heterogeneous C-CAPM to study international currency premiums. In this paper, we investigate whether the heterogeneous C-CAPM has explanatory power for the cross-sectional differences in international equity returns. Moreover, we extend the habit models of Abel (1990) and Campbell and Cochrane (1999) to the international setting and investigate whether the more refined consumption models can improve explanatory power relative to the classic world C-CAPM. We use Hansen’s (1982) GMM to estimate and test our models. Third, the results presented in the paper provide some important implications. The evidence from the test of the heterogeneous consumption model suggests that country-specific consumption risk is not fully diversified. The negative relationship between stock returns and consumption dispersion contributes to the explanation of the equity premium puzzle. Given that consumption dispersion rises during economic downturns, consumption dispersion intensifies the risk faced by the individuals who require higher premium to hold risky assets. Our findings from the cross-sectional tests suggest that the world consumption C-CAPM explains about 11% of the cross-sectional variation in expected excess returns on the MSCI country indexes, indicating that stocks tend to have high returns if their covariance with world consumption growth is also high. Distinct from the classic C-CAPM, the alternative models add substantial explanatory power for variations in the cross section of excess stock returns. In particular, the heterogeneous C-CAPM shows that the estimate of cross-country consumption dispersion is substantially significant indicating that an asset commands a positive risk premium when its returns covary negatively with consumption dispersion. This evidence sheds light on the importance of cross-country consumption dispersions as well as world consumption growth to explain the cross-sectional variations in expected excess returns. Additional tests document that the surplus consumption model and the Abel (1990) habit model provide lowered pricing errors and higher explanatory power. Overall, our evidence implies that the more sophisticated models of the C-CAMP outperform the classic world C-CAPM in the tests of international markets. The remainder of the paper is organised as follows. Section 2 outlines various world consumption models, starting with the classic world C-CAPM and extending to the heterogeneous world C-CAPM. For comparability, we also discuss the world surplus consumption model and the Abel habit model. Section 3 describes the data and the variables used in asset pricing tests. Section 4 discusses the empirical results. Section 5 concludes.

This paper empirically examines the performance of various consumption models in international equity markets. Using data for 17 MSCI country indexes, we investigate whether consumption variables matter in determining excess returns across countries. We begin with the classic world C-CAPM which assumes that there is a world representative investor whose consumption growth is the only factor in determining asset returns. Alternatively, we consider the heterogeneous world C-CAPM to explore if incomplete consumption risk sharing influences asset pricing. Under this assumption, country specific factors cannot be diversified away, implying that idiosyncratic consumption risk must be related to asset returns. We also extend the world C-CAPM to consider world habit models. The Hansen’s (1982) GMM is employed to analyse the covariance representations of linear consumption factor models. The results suggest that the world CAPM exhibits little if any power to explain cross-sectional variations of expected excess returns on the MSCI country indexes. The world C-CAPM explains about 11% of the variations in stock returns across countries, indicating that stock returns tend to be high when their covariance with world consumption growth is high. However, an economically implausible large coefficient of risk aversion appears required to explain the equity premium due to low volatility of consumption growth. The dispersion of the cross-country consumption growth in the heterogeneous C-CAPM adds considerably more explanatory power for the variation in the cross section of excess stock returns compared to the world C-CAPM. The point estimate of the cross-country consumption dispersion is highly statistically significant, suggesting that an asset commands a positive risk premium when its returns covary negatively with consumption dispersion. The world surplus consumption model and the Abel (1990) habit model also offer a better fit with the data than does the world C-CAPM. Our empirical results provide another piece of evidence supporting consumption-oriented factors for international asset pricing. The results clearly indicate that the heterogeneous world C-CAPM and the world surplus C-CAPM outperform the world CAPM and the simple world C-CAPM for determining excess returns across countries. Under the world representative agent model, which assumes that the market is complete, country-specific consumption risk is perfectly shared among investors and hence only the aggregate world consumption risk matters for asset pricing.8 Our findings that the heterogeneous consumption model, compared to the representative agent model, can generate a lower coefficient of relative risk aversion and that the cross-country consumption dispersion provides some explanatory power for stock returns in international markets, suggest that country-specific consumption risk is not fully diversified implying that stock returns are related to idiosyncratic consumption risk. Our results also suggest that the return on a stock is higher if its return is more negatively correlated with consumption dispersion. This finding may partly explain the equity premium puzzle. Consumption dispersion rises during economic downturns. Rising consumption dispersion intensifies the risk faced by the individual investor who would then require higher premium to hold risky assets.