A critical analysis of the accuracy of the country forecasts as prepared by the Economist Intelligence Unit (EIU)

This thesis draws attention to the complexities involved in forecasting economic indicators. A literature review examines the general use of forecasts, errors within forecasts and various methods of analysing the accuracy of forecasts. The focus of this paper is on the testing and measuring of forecast accuracy within the Economist Intelligence Unit Country Forecasts, in particular the forecast accuracy of GOP and Inflation. This is carried out through the assessment of four a priori hypotheses 1) High Income Country Forecasts are consistently more accurate than those forecasts made for countries in the Low Income Category. 2) The accuracy of forecasts decreases the more distant the forecast horizon becomes, therefore Current-Year (t) Forecasts will outperform One-Year-Ahead (t+1) Forecasts. 3) The EIU Forecasts outperform No-Change-Forecasts as measured by the Theil's U-Statistic. 4) The EIU can forecast turning points better than a Random Probability method of forecasting can. The Tests used to evaluate the above hypotheses are the Root Mean Squared Error (RMSE), Theil U-Statistic and Turning Point Directional Analysis. The conclusion reached by this thesis is that the accuracy of forecasts decreases the more distant the forecast horizon becomes, therefore it can be said that Current-Year (t) Forecasts will outperform One-Year-Ahead (t+1) Forecasts. Additionally, the EIU Forecasts do outperform No-Change-Forecasts as measured by the Theil's U-Statistic. Therefore the EIU can forecast turning points better than a Random Probability method of forecasting can. Finally, this thesis concludes that there is little evidence to suggest that High Income Country Forecasts are consistently more accurate than those forecasts made for countries in the Low Income Category.

Grantor dc:publisher.institution

Department of Political Studies

Year dc:date.issued

2007

Author dc:creator

• Durham, Kate Saranne

Advisors dc:contributor.advisor

• Van Walbeek, Corne
• Simons, Mary

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/7466

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/7466