The state of macroeconomic forecasting

Abstract

Macroeconomic forecasts are used extensively in industry and government The historical accuracy of US and UK forecasts are examined in the light of different approaches to evaluating macro forecasts. Issues discussed include the comparative accuracy of macroeconometric models compared to their time series alternatives, whether the forecasting record has improved over time, the rationality of macroeconomic forecasts and how a forecasting service should be chosen. The role of judgement in producing the forecasts is also considered where the evidence unequivocally favors such interventions. Finally the use of macroeconomic forecasts and their effectiveness is discussed. The conclusion drawn is that researchers have paid too little attention to the issue of improving the forecasting accuracy record. Areas where improvements would be particularly valuable are highlighted.Keywords: econometric models, macroeconomic forecasting, rationality, forecast evaluation, structural breaks, cliometrics, industry structure Considerable intellectual activity within the economics profession is devoted to the production, interpretation and analysis of forecasts of major economic variables. Since such forecasts are important to both government planning and industry, it is material to determine how well we, as a profession, have performed this activity and what lessons may lead to improvements. Given the number of analyses that have included predictions of one economic variable or another and the range and depth of the macroeconomic forecasting industry (as surveyed by Fildes, 1995), it is impossible, in one paper, to review all aspects of the field. We will, therefore, concentrate on surveying and analyzing the predictions of short-run GDP (GNP), with particular emphasis upon real GDP and inflation forecasts for the US and the UK, bringing in data from other countries only when necessary in order to have a common set of information about cyclical conditions, etc. We focus on GDP and inflation forecasts because these much predicted variables are of interest to the entire profession. Unfortunately, this emphasis precludes an analysis of the characteristics of the forecasts of other variables and leaves many questions unanswered, e.g. which GDP components were the hardest to predict and which contributed the most to the inaccuracy of the aggregate forecasts?While many techniques have been used to make short-run GDP forecasts, the focus of this article will be on quantitative methods. In examining methods designed to provide quantitative estimatesThe State of Macroeconomic Forecasting 2 of GDP, we will consider time series analyses, econometric models (as well as forecasts made using judgmental techniques), and the contribution that expert judgment has made to the modeling process. As the survey evidence in Fildes (1995, p.6) showed, this range of methods covers those used by macroeconomic forecasting services. Since other articles have compared the structure of many of the econometric models used in macroeconomic forecasting, f...

Perspectives on Evaluating Macroeconomic Forecasts

“…However, Fildes and Stekler (2002) note that most US and UK recessions were not predicted in advance, but neither did economists make many predictions of peaks that did not occur. 15 More recently, Sinclair et al (forthcoming) analyzed the directional accuracy of the Fed's forecasts of GDP and inflation and showed that the predictions of increases and decreases in the inflation rate were not associated with the actual changes in that rate.…”

“…I have been involved in two survey papers that have summarized and synthesized the results of these studies. One looks at US and UK forecasts (Fildes and Stekler, 2002). The other does a similar analysis of the G7 (excluding the US) predictions.…”

“…Similarly, inflation was underpredicted when it was rising and overpredicted when it was declining. (See the surveys of Fildes and Stekler (2002) and Stekler (2008) for the specific studies from which these results were obtained). Fildes and Stekler concluded: "these errors occurred when the economy was subject to major perturbations, just the time when accurate forecasts were most needed."…”

“…Fildes and Stekler (2002) reported that the Mean Absolute Error of 16 annual US and UK GDP growth forecasts was around 1%. Newer studies found similar results for the G7 countries, but Oller and Barot (2000) had found that the errors were larger for some of the other European countries.…”

Internal Consistency of Survey Respondents' Forecasts: Evidence Based on the Survey of Professional Forecasters*

“…In the economics sphere point forecasts of the conditional mean or some other measure of the central tendency have been the main focus (see, e.g., Mincer and Zarnowitz (1969), Figlewski andWachtel (1981), Zarnowitz (1985), Keane and Runkle (1990), Davies and Lahiri (1995), Stekler (2002) and Fildes and Stekler (2002)),…”