Essays in learning and market dynamics

Many asset markets exhibit a pattern of slow booms and sudden crashes. The first chapter explains this asymmetry using an endogenous speed of learning. In the model, more observable economic activity takes place in good times than in bad times. Since more activity generates more public information about the economy, faster learning takes place in good times. If the state of the economy changes when times are good and learning is fast, asset prices adjust quickly and a sudden crash occurs. When times are bad and learning is slower, agents take longer to discover that the economy has improved, and a more gradual boom ensues. The chapter also presents data from U.S. and developing-country credit markets to support the theory.;The second chapter (joint work with Stijn Van Nieuwerburgh) extends learning asymmetry to dynamic general equilibrium. Adding a disturbance to a standard real business cycle model makes technology unobservable. Agents must solve a filtering problem to learn about technology. When production is higher, technology is more easily observable. This better information in good times generates asymmetry in output changes, as in the first chapter. We use analyst output forecasts to test the hypothesis that uncertainty is negatively correlated with output over the business cycle.;The third chapter explores the source of widespread political opposition to structural adjustment. In the model, a tariff reduction causes labor to shift from import and government sectors to the export sector. Frictions in the labor matching process cause temporary unemployment and uncertainty about individual costs or benefits from reform. Even if the reform benefits a majority of workers ex post, a majority may still oppose reform. Cross-country, event, and case studies of protests, employment, and wages corroborate this explanation.