When one loses a job or is unemployed for a period, there are several options to mitigate the loss or lower income, such as borrowing, delaying consumption of certain goods and services and relying on social benefits. Living together with others presents a unique opportunity to insure against such adverse income shocks. When one is unemployed, another household member can join the labour market, or increase working hours if they are already in employment. This labour supply arrangement within the household is called the added worker effect.

Previous studies show that the strongest dependency in labour market decisions is between partners rather than between parents and their children or other relatives. Couples are also a relevant unit to investigate for labour market policies and targeting of social benefits. Thus, in this study, I look into the relationship between partners’ labour supply decisions in the UK.

In the UK, the traditional family model with established division of labour between genders still holds, as we see women dedicating and investing more time in household chores and child development than their male partners. While both partners in the majority of couples are in the labour market by being employed or searching for a job (blue line in Figure 1), there is a significant number of traditional male-breadwinner couples in the UK (green line).

Figure 1. Share of couples’ labour market outcomes over time

Earlier studies on the added worker effect in the UK find a negative relationship between a woman’s labour force participation and her partner’s unemployment or non-employment. In other words, a woman with an unemployed partner is less likely to enter the labour market than a woman with an employed partner. This finding is unexpected as household labour supply theories suggest an increase in the other partner’s labour supply to a negative income shock as a result of an income and/or substitution effect. However, this negative effect is interesting in itself, and previous research does not explicitly study whether this negative effect comes from the male partner’s unemployment or the couple’s tastes and preferences towards paid work or coordination of their labour supply decisions.

So, once the couple is formed, does an employed partner increase the chances of the other partner’s labour market attachment? Is it a couple’s permanent characteristics that matters for a woman’s labour force participation, or her partner’s labour market outcome? In this study, I aim to answer these questions with a focus on married and cohabiting couples and use the British Household Panel Survey 1991-2009 (BHPS). A couple is defined as a pair of individuals of working age (23-59 for women, 23-64 for male partners) who are not in full-time education and have been together without interruptions.

An important thing about couples is that people do not come together randomly. While the saying goes as “opposites attract”, the evidence suggests that people who share similar characteristics are more likely to be together. In the data, in almost one in every two couples, partners have the same level of education. To understand whether there is variation across education levels, I compare couples to a counterfactual case where individuals would select their partner randomly. I find that the positive matching is strongest for individuals with at least a first degree, who are three times more likely to be with a partner of same education level compared to the randomly matched scenario. However, there is no clear sorting pattern based on parental employment backgrounds and their health.

To overcome issues of selection and coordination in labour supply, I use male partner’s health as an instrument for his non-employment. This helps identify a causal effect of male partner’s non-employment on woman’s participation. For this instrument to work, there are two key conditions: (1) health should to be correlated with partner’s non-employment, and (2) the only channel a partner’s health affects the woman’s participation should be through its effect on his employment.

A male partner’s poor health would lower his probability of being in employment, and this would either increase the need for additional income and/or lower the woman’s opportunity cost for market work. We would expect the woman to participate in the labour force. A potential concern for the validity of this instrumental variable is that a partner’s poor health may influence the woman’s labour force participation directly as she may not hold a job (or withdraw) to help and be with her partner.

To consider this concern, I control for the woman’s informal care provision vision besides controlling for age, education, cohabitation, benefit claims, number of children, house ownership and calendar month and year.

As expected, I find that men with poor health are less likely to work and the probability that he won’t be employed is even higher when his partner (woman) provides care. This health effect is monotonic, i.e. the effect of health on male partner’s employment decreases (in absolute terms) by better health.

The pooled model shows that on average a woman is 30 percentage points less likely to be in the labour market when her partner is not employed. Cohabiting with the partner, rather than being legally married, increase a woman’s chances of being in the labour force by 3 percentage points, all other things being equal. However, claiming unemployment benefits or income support in the last 12 months decreases the probability of her participation by 10 percentage points.

The effect of a partner’s non-employment on a woman’s labour force participation is quite large compared to previous studies. This effect is potentially overestimated as some unobserved factors that affect the male partner’s non-employment may also have played a role in couple formation and affects the woman’s participation. In a separate model, I consider the unobserved heterogeneity, the permanent characteristics of a couple that are not observed to the researcher and potentially remaining endogeneity by using the male partner’s health as an instrumental variable. The effect of a partner’s non-employment on a woman’s labour force participation drops to 10 percentage points, but it is no longer a significant effect.

What appears to characterise the endogeneity in a couple’s labour market outcomes is the permanent unobserved factors such as time-invariant preferences towards work and social norms than the correlation between time-varying factors in partner’s labour supply decisions. In the current context, controlling for couples’ fixed effects, i.e. time-invariant unobserved factors, would eliminate the potential endogeneity. However, further work is required to uncover the true relationship by testing the results with other instrumental variables.



  • This blog post is based on “Partners Match, and then the Couple Decides” presented at the Royal Economic Society Annual Conference 2019.
  • The post gives the views of its authors, not the position of LSE Business Review or the London School of Economics.
  • Featured image by MabelAmber, under a Pixabay licence
  • When you leave a comment, you’re agreeing to our Comment Policy.

Melisa Sayli is a postdoctoral research associate in Department of Economics and TYMS at University of York. She is an applied labour economist with gender focus. Email: melisa.sayli@york.ac.uk