Measuring Uncertainty and Disagreement in the European Survey of Professional Forecasters

Survey data on expectations and economic forecasts play an important role in providing better insights into how economic agents make their own forecasts and why agents disagree in making them. Using data from the European Survey of Professional Forecasters (SPF), we consider measures of uncertainty and disagreement at both aggregate and individual level. We overcome the problem associated with distributional assumptions of probability density forecasts by using an approach that does not assume any functional form for the individual probability densities but just approximates the histogram by a piecewise linear function. We extend earlier works to the European context for the three macroeconomic variables – GDP, inflation and unemployment – and we analyse how these measures perform with respect to different forecasting horizons. There are two main results. First, uncertainty and disagreement are higher for GDP and unemployment than inflation, in particular for the short and medium forecast horizons. Second, the results do not support the evidence that, if uncertainty or disagreement are relatively high for one variable, then it is the same for the others. JEL classification: C53, E37, C83. Keywords: Survey professional forecast, uncertainty, disagreement, probability distribution.