Inflation and professional forecast dynamics: an evaluation of stickiness, persistence, and volatility

BIS Working Papers  |  No 713  | 
04 April 2018
PDF full text
 |  47 pages



Inflation-targeting central banks regularly point to surveys of inflation expectations as indicators of future price pressures, and a guide for the direction of policy. Indeed, previous research has shown that surveys predict inflation well.  However, even professional forecasters have been shown to make predictable errors. In particular, prior research suggests that survey respondents do not always fully update their predictions in light of incoming data: survey forecasts of inflation have been found to be "sticky". But, so far it has been unclear whether stickiness is a given feature of survey forecasts, or whether it could be related to the predictability of inflation and the monetary regime more generally.


We combine a model of inflation with a model of sticky survey forecasts. The key innovation of our model is to estimate changes of stickiness together with changes in the predictability of inflation over time. In the model, the survey forecast is a weighted average of last period's survey forecast and an optimal forecast. We generate optimal forecasts from a model known to be successful at forecasting inflation from prior research. We measure "stickiness" by the weight that the average survey respondent puts on past forecasts rather than the optimal forecast.


We estimate our model from data on U.S. inflation, since the late 1960s, and forecasts from the Federal Reserve Bank of Philadelphia's Survey of Professional Forecasters. While surveys may have become stickier in recent decades, our results suggest that survey respondents continue to pay attention to changes in trend inflation. Prior to the mid-1980s, also known as the "Great Inflation" period, shocks to inflation had long-lasting effects on future inflation. For this period, we find that survey responses tracked optimal forecasts quite closely. Surveys have become sticky only since the late 1990s, when shocks to inflation were much more short-lived and trend inflation had stabilized.



This paper studies the joint dynamics of real-time U.S. inflation and average inflation predictions of the Survey of Professional Forecasters (SPF) based on sample ranging from 1968Q4 to 2017Q2. The joint data generating process (DGP) comprises an unobserved components (UC) model of inflation and a sticky information (SI) prediction mechanism for the SPF predictions. We add drifting gap inflation persistence to a UC model in which stochastic volatility (SV) affects trend and gap inflation. Another innovation puts a time-varying frequency of inflation forecast updating into the SI prediction mechanism. The joint DGP is a nonlinear state space model (SSM). We estimate the SSM using Bayesian tools grounded in a Rao-Blackwellized auxiliary particle filter, particle learning, and a particle smoother. The estimates show that (i) longer horizon average SPF inflation predictions inform estimates of trend inflation; (ii) gap inflation persistence is procyclical and SI inflation updating is frequent before the Volcker disinflation; and (iii) subsequently, gap inflation persistence turns countercyclical and SI inflation updating becomes infrequent.

JEL classification: E31, C11, C32

Keywords: inflation; unobserved components;professional forecasts; sticky information; stochastic volatility; time-varying parameters; Bayesian; particle filter