I’ve heard a number of solid hypotheses to explain the gendered pay gap: unpaid care work, the motherhood penalty, or just straight-up discrimination. This one is new to me, though.
Sexual harassment is well documented across many fields but women who work in men-dominated occupations and industries experience higher rates (Fitzgerald et al. 1997; Gruber 1998; McLaughlin, Uggen, and Blackstone 2012). The likelihood of harassment also increases with exposure to a wider range of employees (Chamberlain et al. 2008; De Coster, Estes, and Mueller 1999), and is higher among single women (De Coster, Estes, and Mueller 1999; Rosenberg, Perlstadt, and Phillips 1993), highly educated women (De Coster, Estes, and Mueller 1999), and women in positions of authority (Chamberlain et al. 2008; McLaughlin, Uggen, and Blackstone 2012). Because sexual harassment forces some women out of jobs, it likely influences their career attainment (Blackstone, Uggen, and McLaughlin 2009; Lopez, Hodson, and Roscigno 2009). Numerous studies link voluntary and involuntary career interruptions to significant earnings losses (Brand 2015; Couch and Placzek 2010; Theunissen et al. 2011).
I hope you see where they’re going with this. Sexual harassment causes women to switch jobs or leave the workforce, but pay is usually linked to how long you’ve stayed in your job. Professions where women dominate have less of a sexual harassment problem, but are also viewed as “feminized” and thus worth less. Even within a profession approaching gender parity, like lawyers in the UK, women can be marginalized.
However, optimistic prognoses of gender emancipation are somewhat challenged when considering that the mass entry of women to this profession has been characterized by patterns of vertical stratification and horizontal segmentation (Hagan and Kay, 1995; Sommerlad, 2002; Stake et al., 2007). Women solicitors are more likely to be in subordinate salaried positions, to work part–time, to practise in less prestigious and remunerative firms and legal specialisms and, more generally, to attract lesser terms and conditions. There is a clear pattern of vertical stratification whereby a growing cohort of predominantly female subordinates are confined to ‘a (frequently transient) proletarian role’ (Sommerlad, 2002: 217) and deployed to support the earnings and privileges of a relatively prosperous and autonomous elite of predominately male partners. Women, despite representing a growing majority of salaried solicitors (over 55 per cent of associate and assistant solicitors) and new entries to the profession, still constitute less than a quarter (23.2 per cent) of partners and the average female solicitor enjoys markedly less than half the chances of a male colleague to progress to partnership (17.6 per cent of women solicitors are partners against 39.5 per cent of their male peers) (SRU, 2006c).
So if women switch to a career where they’re less likely to face harassment, or even start off there, they’re paid less than men for the same amount of work. It’s a brilliant theory, and that paper does find evidence in support of it.
In bivariate analyses, women who experienced unwanted touching or multiple harassing behaviors in 2003 reported significantly greater financial stress in 2005 (t = –2.664, p ≤ .01). Some of this strain may be due to career disruption, as harassment targets were especially likely to change jobs. As shown in Figure 1, 79 percent of targets as compared to 54 percent of other working women started a new primary job in either 2004 or 2005 (χ2 = 9.53, p ≤ .01). […]
In Model 2 of Table 2, we test whether the increased financial stress reported by harassment targets can be attributed to their greater likelihood of changing jobs. Analyzing consecutive waves of YDS data, we can establish clear temporal order between sexual harassment (2003), job change (2004–2005), and financial stress (2005). In addition to having a strong direct effect on financial stress (β = .582, p ≤ .01), job change reduces the effect of harassment below standard significance levels. Following Baron and Kenny (1986), we calculate that 35 percent of the total effect of sexual harassment on financial stress is mediated through job change (…). Targets of sexual harassment were 6.5 times as likely as nontargets to change jobs in 2004–2005, net of the other variables in our model (…).
They caution that their data source is from a single cohort, thus it may not generalize, but I think the theoretical axioms are strong enough that it probably will. A similar effect could be happening with non-binary and transgender/transsexual people, too.
It’s also worth underlining that it’s foolish to think the gendered pay gap has one and only one cause; an economy with millions or even billions of actors is a highly complex system, so it’s unlikely to have simple explanations for patterns on that scale. A combination of the above factors is likely driving the gendered pay gap, and we’ll need complex solutions to solve it.
 Ferrant, Gaëlle, Luca Maria Pesando, and Keiko Nowacka. Unpaid Care Work: The Missing Link in the Analysis of Gender Gaps in Labour Outcomes. OECD Development Centre Issues Paper, 2014.
 Budlender, Debbie. The Statistical Evidence on Care and Non-Care Work across Six Countries. United Nations Research Institute for Social Development Geneva, 2008.
 Benard, Stephen, and Shelley J. Correll. “Normative Discrimination and the Motherhood Penalty.” Gender & Society 24, no. 5 (2010): 616–646.
 Murphy, Emily, and Daniel Oesch. “The Feminization of Occupations and Change in Wages: A Panel Analysis of Britain, Germany and Switzerland,” 2015.
 McLaughlin, Heather, Christopher Uggen, and Amy Blackstone. “The Economic and Career Effects of Sexual Harassment on Working Women.” Gender & Society 31, no. 3 (2017): 333–358.
 England, Paula, Michelle Budig, and Nancy Folbre. “Wages of Virtue: The Relative Pay of Care Work,” 2001.
 Sharon Bolton, and Daniel Muzio. “The Paradoxical Processes of Feminization in the Professions: The Case of Established, Aspiring and Semi-Professions.” Work, Employment and Society 22, no. 2 (June 1, 2008): 281–99. https://doi.org/10.1177/0950017008089105.