[cf. 2020 for how assortative mating can drive the ‘gender cliff’ without any misogyny.] et al 2015 document that in the United States there is a discontinuity to the right of 0.5 in the distribution of households according to the female share of total earnings, which they attribute to the existence of a gender identity norm. We provide an alternative [longitudinal] explanation for this discontinuity.
Using linked employer-employee data from Finland, we show that the discontinuity emerges as a result of equalization and convergence of earnings in coworking couples, and it is associated with an increase in the relative earnings of women, rather than a decrease as predicted by the norm.
…the existence of a discontinuity to the right of 0.5 in the relative earnings distribution has been widely cited both in the media and in academia as evidence for the relevance of the gender identity norm. Some authors have also pointed out that a substantial part of the discontinuity is due to the existence of a point mass of couples exactly at 0.5 (2015, 2020).1 As shown in Panel A of Figure 1, the discontinuity to the right of 0.5 estimated by et al 2015 becomes smaller if spouses with equal earnings are excluded, with the 2008 estimate dropping 12.3% → 7.4%.2
In this paper, we provide evidence contradicting the social norm interpretation of the discontinuity (and the point mass) at 0.5 and we propose an alternative explanation. We use linked employer-employee data from Finland that has detailed information on the individual employment and earnings history of the entire population of Finnish individuals for the period 1988–262014.
…First, we examine the distribution of relative earnings at the beginning of cohabitation, which provides a better proxy of the time of union formation than marriage. We find no statistically-significant discontinuity at this stage of the relationship, suggesting that the gender norm does not affect the formation of couples in a discontinuous way.
Second, the norm does not seem to play a role for separations either. Separation rates do not exhibit any discontinuity around the 0.5 threshold of relative earnings. Instead, the relationship between the probability of separation and the relative earnings distribution exhibits a U-shape, with higher separation rates among couples with large earnings differentials either in favor of the husband or in favor of the wife.
Third, the discontinuity in the distribution only arises in couples where both spouses are self-employed (around 6% of all employed couples) or work together in the same firm (around 9%). Hereafter, we refer to these 2 groups as coworking couples. For the rest of the population, there is no evidence of any unusual phenomena in the vicinity of the 0.5 point. The pattern looks different for these 2 groups of coworking couples. In the case of self-employed couples, the discontinuity to the right of 0.5 is mainly due to a substantial fraction of couples bunching exactly at 0.5, while among spouses working for the same employer, the distribution exhibits a cliff at 0.5 with only a small fraction of couples having identical earnings.
Fourth, the observed dynamics rules out a more specific formulation of the gender identity norm theory, according to which the norm is activated only when spouses are jointly self-employed or work in the same firm. Theoretically, this may occur if coworking makes the comparison between spouses more salient or if adjustments in accordance with the norm are feasible only in self-employed couples. We find that the discontinuity does not arise as a result of a reduction in the share of couples where women slightly outearn their husbands, as the gender identity norm would predict. Instead, when couples on both sides of the distribution become self-employed, they tend to equalize earnings leading to an excess mass at 0.5. Similarly, when couples start working together in the same firm, there is a compression of earnings toward 0.5. Since initially there are more couples where women earn less than men, this earnings compression creates a larger mass of couples just to the left of 0.5 than to the right of this point, which statistical tests identify as a discontinuity. Moreover, we also observe that coworking leads to an increase in female earnings above the earnings of similar women in non-coworking couples.
Overall, our results contradict the idea that the gender identity norm exhibits a discontinuity at the point of equal earnings. Some couples may prefer that the husband earns more than his wife, but small variations around the 0.5 point do not seem to make that much of a difference.
…However, in the United States, unlike in Finland, there are no legal defaults for income sharing in partnerships, and households can jointly file their income tax declarations.
For couples coworking in the same firm, the impact of earnings compression is likely to have a similar effect as in Finland. To assess the relevance of income convergence in coworking couples, we use the SIPP/SSA/IRS dataset and we proxy whether spouses work together using available information on industry and occupation. It seems reasonable to expect that the share of coworking couples is substantially higher among couples working in the same industry and occupation.14 Instead, couples working in different industries are unlikely to work in the same firm; although some self-employed couples may be included in this group.15
We observe that around 20% of all couples work in the same industry and occupation, while 60% of couples work in different industries. Figure 9 shows the distribution of relative earnings separately for these 2 groups of couples. The drop in the distribution at 0.5 is statistically-significantly larger among couples working in the same industry and occupation. According to the McCrary test, the estimate of the drop is 14%, which is about twice as large as the drop observed in the overall population. This evidence suggests that factors leading to earnings convergence in coworking couples are also likely to play an important role in explaining the existence of a discontinuity in the United States.