Implications of economic conditions on asset returns

This paper empirically investigates asset pricing implications of a reference-dependent preferences model which extends the standard consumption capital asset pricing (C-CAPM) model to include additional factors related to economic and financial market conditions. In this work, we construct two main factors to proxy the macroeconomic and financial fundamentals using the methodology of factor analysis. The (log) surplus consumption ratio in the ensuing model is assumed to be an affine function of macroeconomic and financial factors. We focus on the dynamics of the reference level with an empirical evaluation of the extended asset pricing models in a data rich environment. In this environment, investor preferences have a reference level specification incorporating economic factors contained in a large dataset. Expectations-based reference-dependent preferences, where agents value consumption outcomes relative to a reference point, specify a dynamic model with a surplus consumption ratio directly linked to economic news, or economic conditions (other risk sources investors need to account for in their preferences), embodied in key economic fundamentals.

The paper contributes to the literature which tackles the empirical shortcoming of consumption-based asset pricing models by modifying investor’s preferences to account for other economic risks in the reference level. Our alternative model that accounts for macroeconomic and financial factors as determinants of reference level, uses factor models as a means of dimension reduction.

This exercise allows us to extract a few factors from a large number of macroeconomic and financial series and use these factors to proxy economic conditions (or economic news) facing agents in the economy. We use updated macroeconomic and financial times series data from Ludvigson and Ng (2007, 2010). The key question is whether fluctuations in the financial market have any link with the macroeconomic conditions and whether only a few estimated factors can effectively summarize the information set driven from large numbers of economic time series. The resulting pricing model accounts for a number of interesting properties, such as time varying risk aversion, small relative risk-aversion and an equity premium that is compatible with the actual equity premium, among others, while relying on an admissible range of local relative risk aversion.