LAUSR

Abstract I provide a theory of information production and learning that can help account for both the excessive optimism that fueled booms preceding crises and the slow recoveries that followed. In my theory, persistence and the size of expectation errors depend on information production about changes in aggregate fundamentals. In turn information production, via credit screening, tends to fall during both very good and very bad times. The former gives rise to episodes of rational exuberance in which optimistic beliefs may sustain booms even as fundamentals decline. I also document evidence from survey forecasts consistent with the model predictions.