Labour market uncertainty and the economic foundations of marriage in South Korea

ABSTRACT

Using data from the Korean Labor and Income Panel Study (1998–2014), this study examines how various indicators of socioeconomic status (SES) (i.e. education, employment status, occupation, and earnings) are associated with the transition to first marriage for South Korean men and women. Results from discrete-time event-history analysis show that all SES indicators are strongly associated with the timing of marriage for men and that there is a clear SES gradient for men's marriage entry. In addition, women with high SES (regular, white-collar jobs, and the highest earnings) are more likely to get married than those with lower SES. We also find that securing a regular job is a significant factor for marriage formation regardless of gender. These results indicate that among those with high SES in South Korea, the economic foundations of marriage appear to be characterised by the cooperation model.

KEYWORDS:

• Marriage
• socioeconomic status
• asymmetric gender relations
• labour market
• South Korea

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Some differences exist across these studies. Woo (2009) uses data up to the 10th wave only and utilises retrospective marriage information. The primary focus of Kim's (2017) study is to demonstrate changes in the determinants of marriage formation across birth cohorts. The research of Kim (2017) and Oh and Lim (2016) is also different in terms of their operationalisation of the education variable.

2 The 2015 college entrance rates for South Korean females and males were 74.6 per cent and 67.3 per cent, respectively (The Ministry of Education & Korean Educational Development Institute, 2015).

3 Non-regular workers include temporary workers (permanent, long-term, and short-term temporary), casual workers, seasonal workers, indirect employees, contractors, special independent workers, and family workers (Kim, 2013).

4 The numbers of respondents included in the analytic sample were 4,050 men and 3,281 women. On average, the respondents participated in five surveys.

5 Evidence suggests that family income might affect the transition to marriage, especially for men (Kim, 2020). We evaluated whether family income was associated with marriage entry, but we found no significant association between family income and the transition to marriage for either men or women. Thus, we decided not to include this aspect in the analyses for the models to be parsimonious.

6 We evaluated the extent to which our results were sensitive to how long we lag the SES measures in predicting marriage by comparing the results between the models that lagged one year and those that lagged two years. The substantive conclusions are similar regardless of how long we lag SES measures.

7 We tested the assumption of proportional hazards by including interactions between SES measures and age (Allison, 2010, 2014). None of the interaction coefficients for age*SES and age²*SES were significant, except for women's junior college. We further compared models with main effects and interaction effects for all SES variables using a likelihood ratio test. The test results confirm no significant interaction effects for most models, except the models including women's with junior college education. The substantive conclusions are similar regardless of the inclusion of interaction effects in that the coefficient for junior college becomes stronger in the interaction effect model than in the main effect model. The main effect model is the preferred one based on the Bayesian Information Criterion (BIC). Additionally, we estimated complementary log-log model that is asymmetrical (compared to logit and probit models) and is useful when the probability of an event is very small or very large (Singer & Willett, 2003). Results from complementary log-log model are almost identical to those estimated from the discrete-time logit models presented here. Only exception is that the coefficient for women's junior college education becomes significant when all posited SES indicators are included in complementary log-log model. It is likely due to a small number of women with junior college education and the resultant low probability of an event (first marriage) among them.

8 No significant association between education and women's marriage formation is explained by age. Supplementary analyses revealed that education was positively associated with the transition to first marriage when not controlling for age (and age squared). However, the significant association between the two disappears with the inclusion of age (and age squared). The importance of age in the relationship between Korean women's education and marriage is also documented in prior research (e.g. Oh & Lim, 2016).

9 In supplementary analyses, we estimated the relationship between the various SES measures and marriage in different time periods (i.e. 1998–2007 and 2008–2014). While there was some evidence that the odds of first marriage for men with high SES were greater in the latter period (2008–2014), the results did not differ from the analyses using pooled data.