National panel surveys of household economics have been mounted in many countries in recent years.  The U.S. Panel Study of Income Dynamics (PSID), conducted by the Survey Research Center of the University of Michigan, began in 1968 and has been collecting data on an annual basis since that time (see Hill, 1992, for a description of PSID).  Household panel surveys similar to PSID are in progress or are being planned in most European countries, including Belgium, Britain, France, Germany, Hungary, Ireland, Luxembourg, the Netherlands, Spain and Sweden.  In 1983, the U.S. Bureau of the Census started to conduct the Survey of Income and Program Participation (SIPP) (Nelson et al., 1985; Kasprzyk, 1988; Jabine et al., 1990; Citro and Kalton, 1993). In 1994 Statistics Canada introduced the Survey of Labor and Income Dynamics (SLID) (Lavellee et al., 1993).

A common feature to most of these household panel surveys is that they start with a national sample of households, and then follow all the members of those households for the life of the panel.  Over the course of time, household compositions change in a variety of ways.  Some members of original sampled households leave those households to set up on their own or to join other households.  For example, a daughter may leave her parental household to get married.  New members may join original sampled households, as, for example, when an elderly parent moves in with the family of a child.  In order to be able to describe the economic circumstances of sample individuals at different points of time, household panel surveys usually collect data not only for the sample individuals but also for the individuals living with the sample individuals at the particular point of time.  These latter individuals are termed associated persons in this paper.