Combine to compete: Improving fiscal forecast accuracy over time

Budget forecasts have become increasingly important as a tool of fiscal management to influence expectations of bond markets and the public at large. Difficulties in projecting macroeconomic variables in volatile economic times—together with political bias—thwart the accuracy of budget forecasts. Po...

Full description

Saved in:
Bibliographic Details
Published inJournal of forecasting Vol. 43; no. 4; pp. 948 - 982
Main Authors Carabotta, Laura, Claeys, Peter
Format Journal Article
LanguageEnglish
Published Chichester Wiley Periodicals Inc 01.07.2024
Subjects
Accuracy
Bias
Bond markets
consensus forecasts
fiscal forecasting
forecast accuracy
forecast bias
forecast combination
Forecasting
Government agencies
public deficit
Tracking
Online AccessGet full text

Cover

More Information
Summary:Budget forecasts have become increasingly important as a tool of fiscal management to influence expectations of bond markets and the public at large. Difficulties in projecting macroeconomic variables in volatile economic times—together with political bias—thwart the accuracy of budget forecasts. Pooling information from many different forecasters can still lead to substantial gains in predictive accuracy when taking into account time variation. We combine the forecasts of both private and public agencies for Italy over the period 1993–2022, and test absolute and relative forecasting performance over time. Although forecast combinations do not necessarily result in less biased or more efficient forecasts, tracking better performing forecasters and combining their budget predictions produces significantly better predictions.
Bibliography:We thank Barbara Rossi, Domenico Giannone, and participants at the 14th IWH‐CIREQ Macroeconometrics Workshop in Halle, the III Time Series Conference in Zaragoza, the Centro Europeo di Ricerche in Rome, the 34th International Symposium on Forecasting in Rotterdam and AQR‐IREA for useful comments and suggestions. The usual disclaimer applies.
ISSN:0277-6693
1099-131X
DOI:10.1002/for.3058