ADMINISTRATIVE RECORDS AND SURVEYS AS BASIS FOR STATISTICS ON INTERNATIONAL LABOR MIGRATION

This paper discusses possible sources for statistics to be used for de scribing and analysing the number, structure, situation, development a nd impact of migrant workers. The discussion is focused on key, intrin sic features of the different sources, important for the understanding of their strengths and weaknesses, and draws the reader's attention t o features which may tend to undermine the quality of the statistics p roduced as well as ways in which the impact of such features can be ev aluated and, if possible, reduced. This discussion thereby provides co ncrete illustrations of many of the methodological issues referred to in (Hoffmann, 1995). The paper is organized around three key groups of migrant workers: (a) Persons who are arriving in a country to work th ere, i.e. the inflow of foreign workers; (b) Persons who are leaving t heir country to find work abroad, i.e. the outflow of migrant workers; (c) Stock of foreign workers in the country. Definitions of these gro ups are given in section 2. Sections 3 to 5 review, for each key group , main possible sources of statistics and the extent to which they can cover the group and identify separately important sub-groups. The dis cussion in each section distinguishes between administrative registrat ions and statistical surveys, and deals with the capacity of a source to provide estimates of the size of the group, and whether it can prov ide estimates of the amount of change in the group over time, or at le ast indications of the direction of such changes. Section 6 outlines p roblems related to other data quality dimensions, such as frequency an d timeliness of the statistics as well as the validity, reliability an d consistency of the registration of variables used to describe migran t workers and where they come from. Section 7 presents strategies for the development of statistics on migrant workers based on two model sc enarios for ''countries'' of different geographic circumstances and in stitutional capacities. Summary recommendations on how to organize coo peration, for the production of statistics on migrant workers, between the relevant agencies and between them and the potential users are pr oposed in the concluding section. The overall conclusion is that tryin g to piece together a coherent statistical picture of any of the above key groups is similar to trying to put together a large puzzle based on photographs of a rapidly changing reality, with important pieces mi ssing and many of the available photographs out of focus.