How far ahead can we forecast? Evidence from cross-country surveys

Using monthly forecast data from Consensus Economics Inc. for 18 developed countries reported over 24 different forecast horizons during 1989-2004, we found wide diversity in the quality of forecasts across countries, and the horizons at which forecasts start becoming useful. For only seven countries (viz., Austria, Belgium, Canada, Denmark, Germany, Norway and U.S.) the initial 24-month-ahead forecasts beat the naïve nochange forecast. In terms of the worst performance, Irish, Portuguese, Swiss and Dutch forecasts beat the naïve forecast as late as at horizons 10-13 months. We also found that that generally the survey forecasts do not have much value when the horizon goes beyond 18 months. The flow of new information to annual GDP growth forecasts follows a bellshaped curve over horizons with a peak point when the forecast horizon is around 13 months. Corresponding Address: Professor K.Lahiri, Department of Economics, University at Albany – SUNY, Albany, NY 12222. USA. Email: [email protected]; phone; (518 442 4758; Fax: (518) 442 4736.