Of Marriages and Mortgages: The Second Demographic Transition and the Relationship between Marriage and Homeownership in Sweden

Abstract

Past research has established a positive relationship between transitions to marriage and transitions into ownership. This paper explores how this relationship is changing by following a population as it advances through the Second Demographic Transition. Following a rational choice model for tenure decisions, it is hypothesized that the Second Demographic Transition is likely to affect the relationship between partnership and tenure in two ways. First, the preferences for ownership unique to marriage are likely to decline. Second, the importance of an extra income, especially for men, is likely to increase. Evidence is found supporting both these assertions for the population of Gävle, Sweden, between 1975 and 1990.

Keywords:

• Marriage
• tenure choice
• social change

Acknowledgements

Lars-Erik Borgegård of Uppsala University collaborated on the paper during its early state. Sadly, he passed away in February 2003.

Notes

¹ Despite these changes in labor force participation, the wage gap between men and women remained roughly the same through this time period, with women earning on average 78 per cent of the wages of men (SCB, 2004b).

² If anything, trends in marital delay are understated by these statistics following a change in pension law that motivated many couples to marry in 1989, resulting in a one-time marriage boom right before the 1990 Census (Andersson, 1998; Hoem & Hoem, 1996).

³ Rising from 5.4 rooms in 1975 and 5.8 rooms in 1990.

⁴ The existence is noted of substantial minorities of individuals taking on partners (for both sexes) without adding an extra income. In addition, a substantial minority of single individuals add extra incomes, in the form of roommates or other family members, without adding partners. This helps us avoid collinearity in the modeling. For the 1985–90 time period, where appropriate data are available, models are also run using income, rather than employment as a control. The addition of partner's income is measured directly in these models. The effects measured in both models remain substantively similar in the 1985–90 period, providing further justification for the robustness of the modeling procedure.