Data Transparency and GDP Growth Forecast Errors

This paper examines the role of a country’s data transparency in explaining gross domesticproduct growth forecast errors. It reports four sets of results that have not been previously reported in theexisting literature. First, forecast errors—the difference between forecasted and realized gross domestic productgrowth—are large. Globally, between 2010 and 2020, the average same-year forecast error was 1.3 percentage pointsfor the World Bank’s forecasts published in January of each year, and 1.5 percentage points for the InternationalMonetary Fund’s January forecasts. Second, the Middle East and North Africa region has the largest forecast errorscompared to other regions. Third, data capacity and transparency significantly explain forecast errors. Onaverage, an improvement in a country’s Statistical Capacity Index, a measure of data capacity and transparency, isassociated with a decline in absolute forecast errors. A one standard deviation increase in the log of the StatisticalCapacity Index is associated with a decline in absolute forecast errors by 0.44 percentage point for World Bankforecasts and 0.49 percentage point for International Monetary Fund forecasts. The results are robust to a batteryof control variables and robustness checks. Fourth, the role of the overall data ecosystem, not just those elementsrelated to gross domestic product growth forecasting, is important for the accuracy of gross domestic product growthforecasts. Finally, gross domestic product growth forecasts from the World Bank are more accurate and less optimisticthan those from the International Monetary Fund and the private sector.