Measuring and explaining the marriage boom in the developed world

Abstract

Using aggregated data from 25 developed countries over a lengthy period of time, this article presents a measure of the marriage boom observed in the twentieth century and an explanation for its causes. One of my main conclusions is that even though it basically developed after the Second World War, its origins are to be found before it. I found that, contrary to the views of some scholars, this boom was not a short-lived phenomenon, but actually lasted for 90 years on average. Using panel data analysis techniques, I am able to show that the rise in women’s education, state spending on social benefits, and larger percentages of people employed in the primary sector tended to discourage marriage. I also found a quadratic relationship between the nuptiality index and the per capita income and mortality rates.

Keywords:

• Marriage boom
• marriage bust
• welfare
• gross domestic product
• mortality
• female education
• occupational composition of the population

Acknowledgements

I would like to express my gratitude for the helpful comments made by Alberto Elices-Vallejo and David Reher on a previous version of this article and for the well-informed contributions by the three anonymous referees who helped us to improve the article considerably.

Notes

1. I gathered information for all those countries where reliable historical data could be found: Australia, Austria, Belgium, Canada, Czechoslovakia, Denmark, England and Wales, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, Russian Federation and the United States of America.

2. I also gathered information about another nuptiality index, the crude marriage rate (CMR), presented in Figure 2. This is the ratio of the number of marriages during the year to the average population in that year. The value is expressed per 1000 persons.

3. Hutterites are a Protestant sect (Anabaptists) founded in the sixteenth century. To escape persecution for their beliefs, they fled Western Europe to Russia in the eighteenth century, and then emigrated to the northern mid-west of the USA in the nineteenth century. Hutterite women have high fertility because contraception and abortion are forbidden and mothers only breastfeed for a few months.

4. It is true that I_m may not be the ideal measure of nuptiality behaviour itself. It has two inseparable dimensions: marriage intensity and marriage timing. Ideally, however, one would want to separate these two dimensions (frequency of marriage and age at marriage) to describe and interpret nuptiality trends. This index is a measure of the contribution of marital status to the overall rate of childbearing. Since historically speaking, access to marriage was closely linked to reproduction, it would seem appropriate to use this index as an indicator of the marriage rate in a study like this, which has a markedly historical character. Table A3 in the appendix shows that the coefficients of correlation between the singulate mean age at marriage (SMAM) and index I_m are very high. That is, index I_m works reasonably well when used in a statistical model designed to capture why people marry at a younger age. SMAM is a measure of mean age at marriage derived by Hajnal (1953b) for use on age-specific marital status data commonly available in population censuses.

5. Since in some cases the marriage boom began in the nineteenth century, for some countries (Germany, Netherlands, Switzerland, Portugal, Australia and Czechoslovakia) the date of onset that I identified coincides with the first year for which information was provided. It is highly likely that if I had been able to obtain more historical data, the year of onset would have been earlier. This means that the marriage boom in these countries may have been slightly underestimated.

6. In the vast majority of countries, the nuptiality index I_m for the year 1990 had not yet dropped to the level it was before the marriage boom.

7. If women registered in the census as ‘single’ are actually cohabiting, the values of the nuptiality index I_m must be greatly underestimated.

8. This procedure for measuring a demographic boom was used in my earlier article about the baby boom (Sánchez-Barricarte, 2016).

9. The baby boom started in the 1930s and 1940s (Van Bavel & Reher, 2013; Sánchez-Barricarte, 2016).

10. GDP per capita for each country is inflation-adjusted expressed in 1990 International Geary-Khamis dollars.

11. I chose age 25 because this is roughly the average age at which people marry/have children (and thus leave the household to form their own families) and because I consider that survival to this age could be a good indicator of the way couples perceived the patterns of mortality around them when they took decisions concerning reproduction. The correlation between ₂₅q₀ and life expectancy at birth (e₀) is very high.

12. I computed the public social transfers per capita (STpc) in each country by multiplying the proportions of the GDP dedicated to social transfers calculated by Lindert (1994) by the GDPpc calculated by Maddison (2009). Lindert (1994, 2004) calculated the percentage of the GDP spent on social transfers in various OECD (Organisation for Economic Co-operation and Development) countries from 1880 to 1930 and from 1960 to 1990. To calculate this percentage, Lindert takes into account the sum of transfers dedicated to welfare, unemployment, pensions, health and housing subsidies. He does not include expenditure on education. Since education is left out, Lindert’s estimates basically measure the effort that different countries have made to care for the adult and elderly population (this demographic segment is the main beneficiary of all the sections of the budget taken into consideration). I can use this information to test whether social protection and care for the elderly acts as a deterrent against nuptiality.

13. I tested the linear effect of GDPpc on the nuptiality index (I_m), and found that it gave a worse specification. The variable mortality (₂₅q₀), which lacked a perfect U-shaped causal relationship with nuptiality, was also tested with different specifications before using the squared term.

14. Both tests are significant at 1%.

15. The time-frame of my study only extends as far as 1990 because I consider that after that date, the percentage of unmarried couples increased considerably, nuptiality therefore ceased to be a mechanism for regulating fertility and the index I_m is underestimated. As we can see from Table A4 in the appendix, R² remains high even when I eliminate yearly dummies, so we can say that most of the trends in the dependent variable are explained by my relevant independent variables. It is worth noting how these models face a potential problem of omitted variables or even reverse causality. However, by using control dummies it is possible to include a wide range of variables which, even though they are not explained, avoid endogeneity problems due to omitted variables. Regarding reverse causality, I do not consider this a big issue in my models and it is left for further studies to try to understand the potential causality that the variable marriage (I_m) can have on variables such as GDP per capita or mortality. There is not much point in trying to interpret mortality (₂₅q₀) or GDP per capita because these will depend on their values (as the relations between these two variables and nuptiality are non-linear, the effects are not fixed). For the variable ‘Female Education’ I can say that, in model 2 of Table 2, one more average year of female education will decrease the nuptiality index by 0.26%.

16. Not only in the primary sector, but also in sectors of the economy related to trade, crafts and the first stages of industrialization.

17. The study by Lynch (1986) provides many insights into the historical relationship between industrialization and workers’ ages at marriage.

18. The countries for which historical information for the variable ‘Primary Sector’ is available are: Austria, Belgium, Denmark, England and Wales, France, Germany, Italy, Netherlands, Norway, Sweden and Switzerland.