Pink Papers: Reporting, Observability, and Outcomes
Paper Session
Sunday, Jan. 3, 2027 8:00 AM - 10:00 AM (EST)
- Chair: Kevin Carney, University of Michigan
Queer Wage Gaps and Outness: Labor market outcomes of sexual orientation and gender identity minorities in Germany
Abstract
This paper estimates wage gaps for cis lesbians, gays and bisexuals as well as trans* and non-binary individuals and examines the role of ‘outness’ on the wage gap between sexual and gender minority individuals relative to cisgender heterosexual men in Germany. Using the German Socio-Economic Panel (SOEP) and the LGBielefeld2021 survey, we show the largest wage gaps for trans* individuals, cis bisexual women, and non-binary individuals, both in the raw data and when applying decompositions with extensive controls. In general, we find little to no effect of being out in the workplace on the wage gap and are thus unable to support the hypothesis of taste-based discrimination by employers with our data. Finally, the wage gaps for non-binary and trans* individuals appear to be driven by the subgroup with misascribed gender.The Cost of Being Seen: The Impact of (Not) Passing on Hiring Discrimination and Stereotypes About Transgender Women
Abstract
"Passing" refers to being perceived as belonging to a more socially accepted or privileged group; for transgender women, it means being read by observers as cisgender. This paper studies how passing shapes hiring discrimination against, and stereotypes about, transgender women. In a large-scale field experiment in Germany, I send approximately 10,000 fictitious job applications to real job postings in matched pairs. The applications randomly vary along two key dimensions: whether the applicant appears to pass as cisgender in an AI-generated headshot and whether she indirectly discloses her transgender identity through a male-to-female name change. I find evidence of substantial passing privilege. Relative to otherwise similar cisgender women, transgender applicants who pass but disclose a name change receive 9% fewer callbacks; those who do not pass but do not disclose receive 10% fewer callbacks. When transgender applicants both do not pass and disclose a name change, callbacks fall by 32%—statistically larger than the sum of individual effects. To investigate potential mechanisms, I conduct a follow-up survey experiment in Germany and the United States in which respondents evaluate professional profiles assigned the same treatments as in the field experiment. The survey allows me to identify systematic differences in perceptions of transgender and cisgender women, assess the role of passing in shaping stereotypes about transgender women, and compare patterns across Germany and the United States.Covariate-Dependent Reporting Bias: Methods and Application to the LGBQ Earnings Gap
Abstract
This paper studies covariate-dependent reporting bias in binary self-reported traits. I develop a practical method that combines a list experiment with direct measurement to estimate not only the overall extent of underreporting, but also the characteristics and outcomes of individuals who conceal under direct questioning. The key insight is to reinterpret the list experiment through an instrumental-variables framework: under standard no-design-effects assumptions, individuals whose responses increase when assigned the veiled list correspond to the full population with the sensitive trait. This allows me to recover covariate means for indirectly identified respondents and, under reporting monotonicity, to back out the corresponding means for selective non-disclosers. I apply the method in an original U.S. online survey of 2,501 respondents on Prolific studying sexual identity. Direct questioning yields a non-heterosexual share of about 16 percent, while indirect elicitation raises this estimate to about 31 percent, implying substantial underreporting. The results show that misreporting is strongly selective. Relative to direct LGBQ reporters, individuals who do not disclose under direct questioning are older, higher-income, and exhibit substantially lower depression and anxiety. In baseline comparisons using direct reports, LGBQ respondents appear to have markedly worse mental health and lower earnings than heterosexual respondents. Once selective non-disclosure is incorporated, however, these disparities shrink sharply: the mental-health gap falls from roughly 0.48 standard deviations to near zero, and the estimated earnings penalty reverses from about -$19,000 to +$11,000. These findings imply that reporting bias can distort both prevalence estimates and measured group disparities when stigma varies across individuals. More broadly, the paper provides a portable toolkit for studying selective misreporting in other sensitive domains where direct survey responses may be systematically incomplete. A follow-up in the Understanding America Study will test these patterns in a population-based panel and allow for examination over time.Discussant(s)
Marianne Bitler
,
University of California-Davis
Christopher Carpenter
,
Vanderbilt University
Laura Nettuno
,
RAND Corporation
Michael Martell
,
Bard University
JEL Classifications
- J0 - General
- Z0 - General