The effects of mean reversion on dynamic corporate finance and asset pricing

Abstract



This thesis aims to investigate the effects of mean reversion on dynamic corporate finance decisions and stock pricing.

In Chapter 1, a continuous-time real option model of mature firm that produces product with exogenous mean reverting price is developed to study the firm’s optimal exit and leverage policies. Simulation results show that both liquidation and bankruptcy triggers are negatively related to the long run price levels, while the speed of mean reversion interacts with the long run price level to affect the firm’s exit decisions in two opposite directions depending on the level’s relative magnitude to total operating expenses (the firm’s instantaneous operation costs plus coupon payments). Regarding the leverage policy, apart from showing the static tradeoff result that firm uses more debts when the current revenues are high, the model exhibits at high long run price levels low-debt scenarios that are analogous to the pecking order prediction, suggesting that both static tradeoff and pecking order effects coexist under a mean reversion environment. Because equity values increase more vigorously with prices than debt values do, the tradeoff effect is overwhelmed and the resulting optimal leverage ratios are generally decreasing with the current price levels.

Chapter 2 extends the model in Chapter 1 to derive the closed-form expression of the firm’s equity beta. Because expected stock returns are linearly related to the equity beta by model assumption, several implications to the cross-sectional behaviors of stock returns are obtained. First, it is predicted that firms with mean reverting characteristics should earn lower average returns than others without. The model further reveals the coexistence of positive book-to-market and leverage premiums to stock returns. Most importantly, due to the possession of bankruptcy option by equity holders, high distress risk stocks are expected to earn lower average returns than otherwise similar but low distress risk stocks. This provides an extra dimension to study the ‘distress premium puzzle’. Finally to verify the model predictions, empirical tests using historical market and accounting data from CRSP and COMPUSTAT are conducted, and supportive results are generally obtained.