Introduction Theoretical and practical issues

2 Data and method of analysis

We have data for all women born in 1950 or later who have been recorded as domiciled in Sweden at least some time during the years 1985 through 1997. The records have been extracted from a number of databases in Statistics Sweden and linked by means of the person number assigned to each individual. The set contains the date of birth of the woman herself, dates of all her births, and additionally

  • her country of birth and possible dates of immigration and emigration;
  • a record of the municipality (kommun) she lived in at the end of each relevant calendar year;
  • her income earned from work in corresponding years, including any amounts of sickness compensation; and
  • any unemployment benefits and educational grants and loans from public sources.
  • We have additional information about the employment situation in our women's home municipalities. For a population of some 8.8 million, Sweden has 288 municipalities, and for our purposes, that is the smallest statistical unit we can get data for.

    In most of our analysis we have used hazard regression. This is now commonplace in studies of the simultaneous impact of several factors on individual behavior that evolves over time, based on event-history data subject to possible censoring before the occurrence of the event in question. The method allows us to include both determinants that are constant and others that may change over the life segment studied. In our case, the event is a woman's first birth. Our time-varying determinants include individual-level covariates like her annual income, and contextual variables such as the labor-market situation that people in her home municipality face. We will describe our covariates further as our account proceeds.


    Introduction Theoretical and practical issues

    logo70.gif (2450 bytes)

    Entry into motherhood in Sweden:
    the influence of economic factors on the rise and fall in fertility, 1986-1997

    Britta Hoem
    © 2000 Max-Planck-Gesellschaft ISSN 1435-9871