Application of measures of entropy, information content and dissimilarity of structures to business tendency survey data

Abstract

This paper evaluates similarities between a priori information supplied by the business tendency surveys (that is, expectations), and a posteriori information (that is, realizations). A priori structure is defined by fractions of respondents expressing expectations, and a posteriori structure – by fractions of respondents declaring observed changes in economic variables (realizations). On the basis of empirical analysis of the business tendency survey data, the following conclusions have been reached: - in case of production, distribution of increase / no change / decrease fractions is relatively uniform, leading to high entropy, - entropy of prices is relatively low; since value of entropy allows to evaluate degree of concentration, in case of prices fractions of survey answers seems to be particularly centered on one of the three options provided in the questionnaire, - entropy of general business conditions exhibits the highest variability which may be interpreted as volatile changes in information content of surveys from one month to another; in contrast, entropy of production is the least variable, - public enterprises exhibit lower entropy (as measured by average) and higher variability (as measured by standard deviation) than private enterprises; that is, for public enterprises concentration of answers to the survey questions is higher and also more variable.