In and out of the capitalia sample: evaluating attrition bias

Abstract

The Capitalia sample survey of manufacturing firms in Italy represents a potentially important source of panel data on Italian firms. To date 8 waves have been released the most recent of which takes the survey upto the year 2000. In each wave, data is collected for a three year period and resampling of the same firms allows the construction of a firm panel over a much longer period. The composition of such a panel for the 5th, 6th 7th and 8th waves covering the period 1989-2000 inclusive was undertaken as part of the PRIN 2003 research project Metodi e applicazioni per la valutazione delle politiche del lavoro e di aiuto alle imprese. In this paper, the characteristics of the sample are considered. In particular, sample entry and exit behaviour are studied in order to gain some understanding of the potential for using the database constructed at the University of Salerno for undertaking panel studies of firms over 2 or more waves of the survey. Such panels would be useful in many contexts, for example, in gaining a better understanding of the impact of industrial policies on indicators of firm success than has thusfar been possible with the analysis of single waves of the sample. Given the unique richness of the database in terms of the range of qualitative as well as quantitative information on Italian firms which it contains, the potential uses and usefulness of a panel dataset constricted from the Capitalia sample are extensive.