Using a labour force survey to derive information on fertility histories
Valeria Bordone, London School of Economics
Francesco C. Billari, Università Bocconi
Gianpiero Dalla Zuanna, University of Padua
Italy has been a forerunner in lowest-low fertility: since the early 1990s Italy has sustained a total fertility rate under the level of 1.3, that defines the lowest-low fertility. Fertility is an individual and private phenomenon, but it also has deep consequences at a social level since choices of the couples about procreation determine the change between generations and lead to economic and social development. Nevertheless, especially during the second part of the 1990s, complete official data collection was discontinued. The absence of sources that provide at the same time data on population (with socio-economic characteristics) and number of children caused a lack of opportunities to study the components and the determinants of fertility. In this paper we discuss an alternative way to gather fertility information, using the own-children method applied to a large dataset: the Italian Labour Force Survey. This method allows to bridge the gap in certain fertility data. By deriving children information on the basis of parents’ characteristics and the population at risk on the basis of the same characteristics, a large-scale dataset for fertility analysis in Italy becomes thus available (including the opportunity to reconstruct event histories). We subsequently provide a quality assessment for our reconstructed fertility dataset, by comparing Total Fertility Rates that we calculated on Labour Force Survey data with official existing Total Fertility Rates that ISTAT and Eurostat calculated at regional and national level. Using an alternative source (the Labour Force Survey) to fill in the gap of demographic Italian data and using the own-children method to match children with their respective interviewed mothers, gives us usable outcomes. The method performs better at a national level as compared with its applicability to single region; and, as expected, the method performs better when one considers a set of years rather than a single year.
Presented in Session 23: Various national and international data