The stochastic discount factor, investment, and asset pricing : three essays in macroeconomics and finance

Abstract

Modern economics studies the behaviour of forward-looking entities. Forward-looking decisions require the evaluation today of tomorrow’s possible outcomes. Discounting is therefore a key link in the chain of optimal economic choices, and it is the central theme of this thesis. My PhD dissertation is a collection of three essays, which aim at uncovering data-consistent discount processes, and studying their implications for the optimal behaviour of firms and for asset prices. In the first chapter, I study the performance of the different econometric methodologies that are commonly used in the empirical asset pricing literature to estimate stochastic discount factor models. I rely on a quantitative model with a well-specified stochastic discount factor process to generate artificial data on asset prices. Using the various methodologies, I proceed with the estimation of the stochastic discount factor from the artificial data. I rely on the model from which the data have been generated to assess the relative performance of the various methods by computing the exact values of the discounted prices implied by the estimated stochastic discount factor processes. The second chapter is co-authored with Russell Cooper and Jonathan L. Willis. We investigate the effects of discounting on plant-level and aggregate investment. We emphasize the empirical dependence of the stochastic discount factor on both the level of productivity and on uncertainty about the aggregate state of the economy. Our findings suggest that data-consistent discount processes have strong implication for firm-level and aggregate investment. We revive the debate of the smoothing of nonconvexities at the micro level, and open up an important new channel for the impact of uncertainty shocks on aggregate activity. The last chapter is in the spirit of the second one, but focuses on the implications for prices instead of quantities. I estimate data-consistent discount processes from portfolios of risky and riskless assets. I study their implications for the cross-section of equity returns within the framework of an investment-based asset pricing model. The size and value premia are sensitive to the estimated process for the stochastic discount factor.