A NOVEL APPROACH TO DETERMINING WEIGHTS IN MEASURING EUROPEAN CONSUMER SENTIMENT USING NONLINEAR OPTIMIZATION

Business and Consumer Surveys (BCS) surveys are useful source of data for various economic analysis and forecasts. The general availability of BCS data stimulates new applications of this data in empirical scientific research, as well as continuous methodological improvements in conducting surveys and calculating composite indicators. The Consumer Confidence Indicator (CCI), which is calculated using BCS data, is one of the widely accepted and empirically proven leading indicators of consumer sentiment and economic activity in general. Although the methodology for calculating confidence indicators has been harmonized at European level, suggestions for improving them have always been welcome. Therefore, in this research the CCI component variables defined at EU level have been retained. However, the possibility of improving the prognostic properties of this consumer sentiment indicator has been indicated by changing the CCI calculation methodology. Instead of the standard method of calculating the CCI indicator as a simple arithmetic mean of its four component variables, that is the methodology which considers all four variables as equally significant in the calculation of the indicator, the nonlinear optimization method is used. The hypothesis of the study is that by updating the methodology of the indicator calculation, its prognostic properties can be improved through a new method of determining the weights associated with individual variables of CCI’s components. The results of an empirical study of the relationship between the innovated CCI and personal consumption as a reference series confirm the research hypothesis. Specifically, the results show that, based on changes in innovated CCI, one can successfully predict the direction of change in personal consumption two quarters in advance. The empirical part of the study is based on quarterly BCS data for four standard CCI’s components and annual growth rates of personal consumption at EU aggregate level. The data covers the period from the first quarter of 1996 to the second quarter of 2019. Data sources are the European Commission and Eurostat. This research was conducted only for the EU at the aggregate level, but this limitation will be removed by future research that will focus on examining and spotting potential differences in consumer sentiment across individual EU member states. Different weighting systems are expected in the calculation of CCI for different national economies in the EU. The originality of this paper is reflected in the innovative way of calculating consumer confidence indicators which is based on applying nonlinear optimization, unlike the standard calculation applied by the EU.