Consumer Tendency Survey of Turkey : A Disaggregated Analysis

The ability to measure the predictive power of consumer surveys is very important especially for central banks in order to have a forward-looking perspective about consumer tendencies and expenditures. Particularly, most studies have found that diffusion indices obtained from surveys are linked to aggregate GDP or consumer expenditures. Therefore, the performance of prediction can be assessed at an aggregate level via the diffusion indices. On the other hand, our paper, while restricting itself to the evaluation of Turkish data provided by Consumer Tendency Survey (CTS) for the period 2003-2012, differs from previous studies in looking at disaggregated measures of both consumers’ opinion and household’s expenditures. In particular, various demographic characteristics such as employment type, age and income of the people interviewed are considered in this paper. Moreover, as a reference series, different categories of consumption are used (services, food and non-food items) instead of more aggregate measures of economic activity (total consumption, GDP). First of all, the survey results are analyzed monthly and quarterly. The quarterly series, which shows less volatility compared with monthly series, are used in order to examine the leading/coincident relationships with the related reference series; then, we test the significance of these relationships. We also construct regression models. The disaggregated analysis confirms that the CTS-consumption relationship is stronger for different demographic categories of consumers and some specific groups of expenditures than the aggregated categories of consumers.