| Popis: |
Background: The recent global financial crisis in 2008 has shown that knowing the level and the dynamics of real macroeconomic variables such as GDP, industrial production, employment, private consumption, savings etc. is not enough to explain and let alone forecast these macroeconomic variables. In order to do that, one should also consider perceptions and expectations of business actors and consumers about these real variables. For example, knowing the quantitatively expressed consumer sentiment enables us to explain and forecast the direction of change in private consumption. The incorporation of the psychological variables such as consumer sentiment or economic sentiment in quantitative analysis turned out to be necessary and invaluable. Business and Consumer Surveys (BCS) play an important role in these efforts. Managers’ and consumers’ judgements about their economic surroundings, derived from BCS results, are expressed as empirically confirmed leading indicators, like economic sentiment indicator or consumer sentiment indicator. Harmonized European BCS are widely used sources of essential information because BCS measures are informative, quite intuitive, easy to interpret, very popular and timely presented in media and commented by economic analysts. They are used in scientific and professional studies as well as in general public. The literature about the relationship between consumer sentiment and private consumption is rich. The first investigations of the predictive power of the Index of Consumer Sentiment by the Survey Research Centre of the University of Michigan in forecasting consumption date back to 1960th. The main conclusion is that the consumer sentiment is a significant explanatory variable for consumption in regression model that includes lagged consumption in the model. The contribution of confidence in explaining and predicting consumption expenditures increases during periods of economic uncertainty when high volatility of consumer confidence is present (Lozza et al., 2016 ; Lahiri et al., 2016 etc.). Consumer surveys which are part of BCS are based on empirical research on consumers’ attitudes and expectations on important economic variables. Their assessments and expectations are expressed in the form of balances. Balance is the difference between positive and negative answering options, measured as percentage points of total answers for each question (variable). The next step in aggregating survey information is calculation of the Consumer Confidence Indicator (CCI). In accordance with the last revision, which was made in December 2018, the European CCI is defined as the simple average of seasonally adjusted balances of four variables: Q1 - Financial situation over last 12 months, Q2 - Financial situation over next 12 months, Q4 - General economic situation over next 12 months and Q9 - Major purchases over next 12 months. Purpose: In order to raise the CCI’s forecasting power, we propose a novel CCI which we construct by using a different weighting scheme. The weights in this newly defined CCI are obtained by using the nonlinear optimization approach. Here, the weights are determined by minimizing the root mean square error (RMSE) in simple regression model and by maximizing the correlation coefficient between the CCI and final consumption for the various time lags. Design/methodology/approach: The research presented in this paper has two parts. In the first part, the forecasting power of the revised CCI as a measure of consumer sentiment in EU is empirically examined by means of different quantitative methods. Using confusion matrix and confusion rate approach, we examine the possibility of predicting the direction of change in household final consumption expenditure based on changes in the CCI for different time lags. The (in)stability of correlation between the CCI and the consumption growth rate during the time will be examined by the rolling window correlations coefficient. In the second part of the research we modify the standardized methodology for the calculation of the CCI. The empirical analysis is based on quarterly data for four standard CCI subcomponents and y-o-y growth rate of household final consumption expenditures. The data set covers the period from 1996Q1 to 2019Q2. The data sources are European Commission and Eurostat. Findings: The results reveal that by changing the weights in calculation of the CCI the predictive power of the CCI may rise with respect to year-on-year household final consumption expenditure growth rates. Using CCI we can predict changes in household final consumption expenditure up to two quarters ahead. Correlation between analysed variables is unstable during the analysed period. Research limitations: The future research will be focused on examining differences between consumer sentiments across EU countries. The main hypothesis for the next research is that the weighting scheme will be different for the different national economies in EU. Originality/value: The added value of the paper is novel CCI weighting scheme for the EU constructed by using nonlinear optimisation techniques. |
