Forecasting inflation: Phillips curve effects on services price measures

Abstract

We estimate an empirical model of infl ation that exploits a Phillips curve relationship between a measure of unemployment and a subaggregate measure of infl ation (services). We generate an aggregate infl ation forecast from forecasts of the goods subcomponent separate from the services subcomponent, and compare the aggregated forecast to the leading time-series univariate and standard Phillips curve forecasting models. Our results indicate notable improvements in forecasting accuracy statistics for models that exploit relationships between services infl ation and the unemployment rate. In addition, models of services infl ation using the short-term unemployment rate (less than 27 weeks) as the real economic indicator display additional modest forecast accuracy improvements.

Combining survey long-run forecasts and nowcasts with BVAR forecasts using relative entropy

“…In their empirical exercise, they consider five variables (output growth, hours worked, labor share, inflation, and interest rates), and their methodology detects changes in the steady states for inflation and the interest rate when estimated over the sample 1954.Q3 through 2012.Q3. This finding coincides with our results on tilting VAR forecasts 7 The empirical models that explicitly permit time variation in the dynamic coefficients and intercepts (e.g., Primiceri, 2005; or unobserved components models that allow for timevarying latent components (e.g., Stock and Watson, 2007;Tallman and Zaman, 2017) can address shifts in the mean. In this section we focus only on those studies that have applied innovative techniques to deal with the shifts in the mean in time-invariant VARs.…”

Real-Time Density Nowcasts of US Inflation: A Model-Combination Approach

“…First and foremost, we devise and implement parametric block wild bootstrap algorithms to produce density nowcasts for the DMS framework. When the DMS selects the multivariate regression model that uses disaggregates, the density nowcasts are constructed by 1constructing density estimates for each of the three disaggregates (core inflation, food inflation, and gasoline inflation); and (2) combining the density estimates using the weights in Cs(t) to construct the density nowcast for aggregate inflation, similar to the combination approaches in Ravazzolo and Vahey (2014) and Tallman and Zaman (2017) but in this case with an application to nowcasting. While Knotek and Zaman (2017) estimate the model using short rolling windows, which leads to very flexible parameters and incorporates changing volatility in a parsimonious way, we consider density combinations that allow for a variety of rolling or expanding estimation windows, as discussed below.…”

Is It Time to Reassess the Focal Role of Core PCE Inflation?

“…(The Federal Reserve Bank of San Francisco publishes a cyclical core PCE inflation series. For related research, seeShapiro 2018, Stock and Watson 2019, Tallman and Zaman 2017, and Zaman 2019 ). Furthermore, asBall and Mazumder (2019) and demonstrate, both median PCE inflation and trimmed mean PCE inflation are well-explained using a Phillips curve relationship.28 Finally, as Dolmas and Koenig (2019) note, both trimmed mean inflation and median inflation give a more prominent role to "sticky" prices.29…”