An evaluation of the forecasts of the federal reserve: a pooled approach

Abstract

The Federal Reserve Greenbook forecasts of real GDP, inflation and unemployment are analysed for the period 1974 to 1997. We consider whether these forecasts exhibit systematic bias, and whether efficient use is made of information, that is, whether revisions to these forecasts over time are predictable. Rather than analyse the forecasts separately for each horizon of interest, we discuss and implement methods that pool information over horizons.We conclude that there is evidence of systematic bias and of forecast smoothing of the inflation forecasts. JEL classification numbers: C12, C53.

Accuracy, unbiasedness and efficiency of professional macroeconomic forecasts: An empirical comparison for the G7

“…These include Keane and Runkle (1990) and Batchelor and Dua (1990), who introduce an analysis in a panel framework using the Generalized Methods of Moments (GMM) method, or Davies and Lahiri (1995), who develop a framework for analyzing three-dimensional panels of survey data, enabling the use of information along all dimensions. To ensure that our results are comparable to existing studies, we closely follow the approach suggested by Davies and Lahiri (1995), and recently used by Clements et al (2007), Boero et al (2008a), and Ager et al (2009), and suggest only minor modifications to the econometric framework.…”

Analyzing Three-Dimensional Panel Data of Forecasts

“…Given that the rational forecaster has correctly incorporated all publicly available information, the only way for an unbiased forecaster i to obtain a forecast error variance less than that of the rational forecaster is for forecaster i to have access to private information. Clements, et al (2007) employ this framework in testing Federal Reserve forecasts of inflation, real GDP growth, and unemployment for rationality. They test the Fed's Greenbook forecasts for each forecast horizon separately and pooling all the horizons together.…”

Projecting the Medium-Term: Outcomes and Errors for GDP Growth

“…For the analysis of forecast errors from the out-of sample experiments, a framework inspired by the work of Brown and Maital (1981), Keane and Runkle (1990), Davies and Lahiri (1995) and Clements et al (2007) is employed to derive the covariance structure of cumulative forecast errors. It is shown that this particular framework has advantages in small samples over the approaches usually employed to inference in forecast error analysis.…”

“…It is assumed that the errors as depicted in equation (8) Davies and Lahiri (1995) use such a model to analyze forecast errors in a panel data setting using professional forecasts. Clements et al (2007) build on this model to test whether forecasts of the Federal Reserve are systematically biased and efficient. The framework allows them to pool information over horizons and represents an analogue application to the forecast errors analysis in the present paper.…”

“…In order to explicitly account for serial correlation in error processes and to perform consistent tests for unbiasedness and accuracy, a simple model of forecast errors is employed to analytically derive the exact covariance matrix of forecast errors and appropriate test statistics. We use a framework for testing forecast unbiasedness which is inspired by the work of Brown and Maital (1981), Keane and Runkel (1990), Davies and Lahiri (1995) and Clements et al (2007), while the accuracy test draws on the contributions of Diebold and Mariano (1995) and Harvey at al. (1997).…”