Multifactor consumption based asset pricing models using the US stock market as a reference: Evidence from a panel of developed economies

a r t i c l e i n f o JEL classification: C52 E44 G12 Keywords: Consumption based asset pricing model Multi-factor model Panel estimation Fixed effects This article considers a panel framework to test consumption based asset pricing models driven by a US stock market reference for a number of developed economies. Specifically, we focus on a linearized form of what might be seen as a consumption-based capital asset pricing model in a pooled cross section panel with two-way error components. The empirical findings of this multifactor model using a range of specifications indicate that there is a significant unobserved heterogeneity captured by crosscountry fixed effects when consumption growth is treated as a common factor. However, the cross-sectional impact of home consumption growth can vary over the countries, where unobserved heterogeneity in the rate of risk aversion can also be addressed by random effects. There is evidence from empirical studies that shows that the conditional covariances between the Intertemporal Marginal Rate of Substitution (IMRS) in consumption and returns cannot satisfy the equilibrium restrictions imposed by the representative agent Consumption-based Capital Asset Pricing Model (C-CAPM) for a range different countries (Kocherlakota, 1996). This has led to a great deal of interest in the capacity of the C-CAPM to take account of heterogeneity and idiosyncratic risk (Lund and Engsted, 1996). The issue of heterogeneous risk in asset pricing was first addressed by Miller (1977), and then revisited by Constantinides and Duffie (1996) and Jacobs and Wang (2004). It has been concluded that heterogeneous risk has a better chance of explaining the data than standard representative-agent C-CAPM models (Jacobs and Wang, 2004). Gregoriou et al. (2009) using monthly data for the US and Hunter and Wu (2009) using similar data for the UK find that via some form of augmentation, consumption growth can be seen as one of the drivers of the riskier component of asset valuation. Gregoriou et al. (2009) develop a fully formulated Vector Autoregressive (VAR) model of US asset prices where excess returns are simultaneously explained by consumption growth, real money growth and inflation with income growth as a single exogenous variable. The results across the system are controlled where appropriate for Autoregressive Conditional Heteroscedasticity (ARCH) and the major financial shocks to the economy. A key explanatory variable in the VAR is US consumption growth. Hunter and Wu (2009) address the importance for the UK market of simultaneous …