| My dissertation seeks to better understand the empirical relationship between the Phillips curve and factors that vary over the business cycle. In the first chapter we construct measures of the markup using microeconomic data and examines their performance in the Phillips curve relative to the output gap. We highlight issues of aggregation bias and pitfalls of applying microeconomic concepts like markups at the macroeconomic level. Macroeconomic markups bear little relation to their microeconomic counterparts and do not perform well in the Phillips curve. The macroeconomic markup is best thought of as a measure of price pressure relative to cost pressure. In the second chapter we examine whether the Phillips curve is asymmetric across the business cycle using a smooth transition regression. The asymmetry is that late in expansions the Phillips curve tends to shift it and flatten in the postwar period. The chapter implies NAIRU contains a short-run component and monetary policy may have been too "tight" in the latter half of some business cycle expansions. The third chapter investigates the causal relation between inflation, unemployment, and wage growth as well as theoretical justifications of the Phillips curve using a vector autoregression (VAR) that includes Phillips curve variables and accounts for variables related to firm behavior and monetary policy. Rather than the more common Wold causal ordering, the contemporaneous causal order is selected using the graph theoretic approach, which applies statistical tests to the error terms of the unrestricted VAR, letting the data decide the causal order. Further statistical tests are then applied to identify a more complete causal structure. Examining the structural VAR reveals a more complex set of relationships than most theoretical models and can be seen as lending support to and encompassing each of them. The chapter also emphasizes that identifying the correct causal structure is of utmost importance in order to draw the correct inferences from the model. Finally, in the fourth chapter, we unify the first three chapters and place them in context with the broader Phillips curve literature. Also, we suggest future directions for research. |
