(Re)Constructing the European Economic Sentiment Indicator: An Optimization Approach

The last recession in Europe has shown us that econometric models that factor in the qualitative perceptions and expectations of businesses and consumers—along with commonly used quantitative macroeconomic variables—can produce better results in explaining and forecasting economic activity. The European Commission’s Business and Consumer Surveys (BCS) conducted by the European Commission (EC) are high-quality source for this kind of “soft” variables. One of the composite indicators based on BCS is the economic sentiment indicator (ESI), which is the main leading indicator for overall economic activity. We propose two new models for constructing the ESI. The first model is based on minimizing the sum of absolute values of estimation errors. The second model is based on maximizing the number of correctly predicted directions of change for GDP growth rates. Rather than using the EC’s official standardization procedure for data, our models use “raw” data, thus simplifying the process of preparing the data. The models were tested for various prognostic horizons (up to four quarters in advance), using aggregated quarterly data for the European Union from 1996Q4 to 2019Q2. The results show that our new models significantly improve the ESI’s predictive power, especially in predicting the direction of change of GDP growth rates, which is the main purpose of the BCS indicators. The best results are obtained for predictions made up to one quarter in advance, for which the second model correctly predicts the direction of change of GDP growth rates in 78.89% of cases versus the official ESI’s 65.56%.