New Keynesian theories of inflation and output

Abstract

This thesis examines two important issues in the empirical literature on the new Keynesian Phillips curve (NKPC). First, are inflation expectations consistent with rational expectations? Many researchers find that the old Keynesian Phillips curve with adaptive expectations fits the data better new Keynesian Phillips curve with rational expectations. Second, do real marginal costs drive inflation dynamics? Gali and Gertler (1999) argue that the reason why the NKPC fits the data poorly is because traditional empirical work on the Phillips curve uses some output gap measures as a proxy for real marginal cost rather than labour's share of income. Our results suggest that the pure rational expectations new Keynesian Phillips curve might be misspecified and that the hybrid new Keynesian Phillips curve fits the data best. The relative importance of backward-looking inflation expectations and forward-looking inflation expectations changes over time. Backward-looking inflation expectations dominate forward-looking inflation expectations independent of which measures of real marginal cost are used. Furthermore, we have tested the rationality of various survey measures of inflation expectations; our results indicate that these survey measures of inflation expectations are biased and inefficient. We have also showed that there are Granger causalities from the professional forecasters (as represented by the SPF forecasts and the Greenbook forecasts) to households (as represented by the Michigan forecasts), but no Granger causality in the opposite direction. Our empirical results suggest that the probability of finding a job or job finding probability (JFP) is a better proxy for real marginal cost than the output gap and labour's share of income, at the same time JFP provides a direct link between frictions in the labour market and the Phillips curve relationship. The use of job finding probability as a proxy for real marginal cost is a novel aspect of this thesis. We have also examined the flattening of the slope of the reduced-form Phillips curve for the United States over the last 20 years; this phenomenon is also observed in many other industrialized countries. We proposed the flattening of the slope of the reduced-form Phillips curve is caused by deindustrialization and the computer revolution have shifted employment from the manufacturing sectors to the service sectors; these structural changes in the labour market have changed jobs' skill requirements, increasing heterogeneity (real rigidities) between workers, producing more mismatches in the labour market.

Date of Award	2012
Original language	English