Survey-based indicators vs. hard data: What improves export forecasts in Europe?

In this study, we evaluate whether survey-based indicators produce lower forecast errorsfor export growth than indicators obtained from hard data such as price and costcompetitiveness measures. Our pseudo out-of-sample analyses and forecastencompassingtests reveal that survey-based indicators outperform the benchmarkmodel as well as the indicators from hard data for most of the twenty European statesfocused on in our study and the aggregates EA-18 and EU-28. The most accurate forecastsare on average produced by the confidence indicator in the manufacturing sector,the economic sentiment indicator and the production expectations. However, largecountry differences in the forecast accuracy of survey-based indicators emerge. Thesedifferences are mainly explained by country-specific export compositions. A largershare in raw material or oil exports worsens the accuracy of soft indicators. The accuracyof soft indicators improves if countries have a larger share in exports of machinerygoods. For hard indicators, we find only weak evidence for the export composition toexplain differences in forecast accuracy.