Survey Techniques for Price Indexes: an Application to Prices Data for the Computation of Price Indexes

Price is a fundamental lever of marketing: the knowledge of prices and their movements can determine the success of a firm. Focusing on the services sector and starting from data collected through a National survey, the prices movements are measured using a price index to synthesize the monthly variation of profit margins. First of all the distribution of survey data and the official statistics about the target population are used to generate a simulated population (the relative efficiencies of different methods of generation are compared). Secondarily, a comparative study of some sampling methods (probability proportional to size and others) is carried on. Then the main imputation methods are also introduced: they are useful to face the challenge of missing data. The analysis is developed taking into account, in particular, the bias of the computed price index. Nevertheless, the studied methodological issues are common to the majority of surveys; thus the conclusions of this work could be useful for practitioners (who have to plan surveys or to use prices data to compute price indexes) to increase the quality of both collected data and price indexes.