Winsorització de la despesa turística a Catalunya

Abstract

The tourism sector is given high priority by the Catalan Statistical Office (Idescat) both regarding theoretical developments and production of statistical data. In this context, tourism expenditure can be regarded as a key strategic variable when it comes to assessing such important issues as the impact of tourism on economic growth and satellite accounting of the tourism sector. The Idescat is therefore committed in the development of methodological improvements in the field. In this article a method is suggested to impute missing data and to correct systematic errors in surveys measuring tourism expenditure. The complexity of tourism expenditure makes systematic errors to consist mainly in underreporting. The suggested method consists in asymmetrically windsorizing the residuals for daily expenditure per person in a regression model that uses the characteristics of the trip as predictor. Estimation is done iteratively with a variant of the EM algorithm. Standard errors are computed by means of the jackknife resampling method. The procedure is illustrated with data on trips to Catalonia made by Catalans, other citizens of Spain and foreigners during 1998.