Learning and Heterogeneity in GDP and Inflation Forecasts

Abstract

We estimate a Bayesian learning model with heterogeneity aimed at explaining the evolution of expert disagreement in forecasting real GDP growth and inflation over 24 monthly horizons for G7 countries during 1990-2007. Professional forecasters are found to begin and have relatively more success in predicting inflation than real GDP at significantly longer horizons; forecasts for real GDP contain little information beyond 6 quarters, but forecasts for inflation have predictive value beyond 24 months and even 36 months for some countries. Forecast disagreement arises from two primary sources in our model: differences in the initial prior beliefs of experts, and differences in the interpretation of new public information. Estimated model parameters, together with two separate case studies on (i) the dynamics of forecast disagreement in the aftermath of the 9/11 terrorist attack in the U.S. and (ii) the successful inflation targeting experience in Italy after 1997, firmly establish the importance of these two pathways to expert disagreement.

Measuring uncertainty and assessing its predictive power in the euro area

“…The time span we consider goes from the second quarter of 1999 (1999Q2) to the second quarter of 2015 (2015Q2) for GDP growth and from September 1999 (1999Q3) to September 2015 (2015Q3) for inflation, as that is the period that we can compare with the forecasts given by the ECB-SPF. There are two additional issues regarding the data: (1) The euro area for the time period analyzed has expanded; while in the year 2000 it was formed by 11 countries, in 2015 it includes 19 countries, and (2) GDP growth data are subject to important revisions and final data may not correspond with the figures that forecasters first observed in real time (see, for instance, Lahiri and Sheng (2010a)). The problem of revisions is smaller when looking at inflation.…”

Inflation uncertainty revisited: a proposal for robust measurement

“…CE reports average annual growth rates of expected inflation for the current and the next calendar year. However, since the forecast horizon varies for each month, the cross-sectional dispersion of forecasts is strongly seasonal and converges towards zero at the end of each year (Lahiri and Sheng, 2010a). To obtain twelve-month-ahead inflation forecasts, we follow Dovern et al (2009) and calculate a weighted moving average of the annual forecasts.…”

Forecast performance, disagreement, and heterogeneous signal-to-noise ratios

“…1 In Lahiri and Sheng (2008) and Lahiri and Sheng (2010a), disagreement across forecasters arises due to differences in individual forecasters' prior beliefs, differences in processing new information, and differences in the relative importance that forecasters attach to their priors and the new information. This framework is especially useful to describe situations in which a sequence of forecasts (with shrinking forecast horizon) is made for one particular random variable, such as an annual growth rate.…”