IMF forecasts: the problem of unidirectional bias

THE International Monetary Fund (IMF) recently lowered its global growth forecasts by 0.2 percentage point for both 2016 and 2017. The downgrade maintains what is now becoming a trend. Since 2011, initial forecasts have been successively revised down during the year, sometimes every quarter. Medium-term growth forecasts have been steadily revised down for both advanced and emerging market economies. The gap between the preliminary and revised forecasts against actual growth has ranged between 40 and 100 basis points in this period. One basis point is one-hundredth of a percentage point. Given the serious consideration by all to the Fund's economic outlook, its forecasting errors have attracted criticism.

However, it matters less if future predictions turn out wrong. What is significant is that the forecast error bias is unidirectionaldownward. The difference in the actual (real) and predicted growth has always been negative, which suggests an optimistic bias in the Fund's global growth model. Indeed the IMF recognised this propensity towards over prediction of gross domestic product (GDP) growth and in October 2014 showed self-reflection and willingness to analyse the origins of its forecast errors (World Economic Outlook, Box 1.2, October 2014). Its analysis sourced the origins of over prediction in global growth for 201114 to a relatively small number of economies. The BRICs alone accounted for about half of the overall forecast error; a few stressed economies in the West Asia and euro area, Japan, and some Asian advanced economies added to the rest. Average forecasting errors for emerging and developing economies were about two times larger than for advanced ones.

The interesting finding was that the key reason why growth was over predicted was because potential output, or trend growth rates, were themselves overestimated, especially for the BRICs where potential growth estimates were based upon their strong growth performances before and immediately after the global financial crisis. Over predictions of GDP, in turn, largely reflected over prediction of investment, while other components such as net exports and consumption accounted for little. In other words, much of the gap in predicted and actual growth reflected a shortfall in investment growth.