691. A State-Dependent Model for Inflation Forecasting
- Author:
- Andrea Stella and James H. Stock
- Publication Date:
- 11-2012
- Content Type:
- Working Paper
- Institution:
- Board of Governors of the Federal Reserve System
- Abstract:
- We develop a parsimonious bivariate model of inflation and unemployment that allows for persistent variation in trend inflation and in the non-accelerating inflation rate of unemployment. The model, which consists of five unobserved components including the trends) with stochastic volatility, implies a time-varying vector autoregression model for changes in the rates of inflation and unemployment. The implied backwards-looking Phillips curve has a time-varying slope that is steeper in the 1970s than in the 1990s. Pseudo out-of-sample forecasting experiments indicate improvements upon univariate benchmarks. Since 2008, the implied Phillips curve has become steeper and the the non-accelerating inflation rate of unemployment has increased.
- Topic:
- Development, Labor Issues, Inflation, State Building, and Unemployment
- Political Geography:
- Global Focus