The macroeconomics, game theory and econometrics of regime changes.

摘要

The formation of expectations and its consequences on agents' actions are critical considerations when analyzing the impact of shocks on economic outcomes. This dissertation thus focuses on incorporating these aspects to understand the effects of occasional large shocks on macroeconomic and financial variables, potentially interpretable as changes in underlying political and economic regimes.;The theoretical part of this dissertation involves the interplay between policy and expectations, when there can be either large shocks to underlying policy regimes or when there can be policy challenges presented by economic mechanisms that bring about large and persistent shifts. In the first case, I am concerned with understanding how policy actions can be best undertaken when private agents have imperfect information about underlying government objectives or type. Chapter 1 and Chapter 2 explore this set of ideas in the context of fiscal and monetary policy, respectively. Through those concrete examples, I provide a theory for the evolution of reputation and its consequences in optimal policy design, which is absent in the existing literature.;In the second case, I am interested in how policy can serve as an information extraction tool for both the government and the agents to learn about latent fundamentals. In Chapter 3, I work on a class of economies where strong strategic complementarity in production and imperfectly observed productivity generate phases of self-fulfilling "optimism" and "pessimism". There, the signal of underlying productivity largely depends on the aggregate activities by the agents, which gives rise to an innovative "information role" of policy. That is, the tax policy can manipulate agents' coordination so as to accelerate or delay information revelation.;When a shock occurs in the economy, the response of agents varies substantially depending on whether the shock has a long-lasting or short-lived effect. Thus, the empirical part of this dissertation deals with how to estimate the occurrence likelihood of such permanent components in the underlying shock process. Chapter 4 estimates a random level shift model for volatility series of popular stock market return indices. The main finding is that permanent shocks occur only infrequently but have disproportionately pronounced influence.