Forecasting private consumption by consumer surveys

Survey-based indicators such as the consumer confidence are widely seen as leading indicators for economic activity, especially for the future path of private consumption. Although they receive high attention in the media, their forecasting power appears to be very limited. Therefore, this paper takes a fresh look on the survey data, which serve as a basis for the consumer confidence indicator (CCI) reported by the EU Commission for the euro area and individual member states. Different pooling methods are considered to exploit the information embedded in the consumer survey. Quantitative forecasts are based on Mixed Data Sampling (MIDAS) and bridge equations. While the CCI does not outperform an autoregressive benchmark for the majority of countries, the new indicators increase the forecasting performance. The gains over the CCI are striking for Italy and the entire euro area (20 percent). For Germany and France the gains seem to be lower, but are nevertheless substantial (10 to 15 percent). The best performing indicator should be built upon pre-selection methods, while data-driven aggregation methods should be preferred to determine the weights of the individual ingredients. -- Consumer confidence ; consumption ; nowcasting ; mixed frequency data