This is our 13th blog post for our Job Market Paper Series blog for 2024-2025.
Ruchi Mahadeshwar is a Ph.D. candidate in Economics at Brown University. Her primary fields are development and labor economics, with a focus on gender and firms. You can find her JMP here.
There are 52 million sex workers in a $180 billion global market for sex (International Union of Sex Workers, 2024; Scelles Foundation, 2011; Cunningham and Kendall, 2011). Two facts motivate our study. First, prevailing policies on the market for sex is criminalization. However, bans present a tradeoff: while they can decrease the size of the market, studies show they also decrease worker and general welfare via an increase in sexually-transmitted illnesses and crimes like rapes (Cameron et al., 2021; Cunningham and Shah, 2018). This tradeoff motivates providing evidence on alternative policies, but assessing workers’ responses to such alternatives is challenging given the complexity of collecting data, let alone causal evidence, on participation in sex work. As a result, little is known about the basic economics of this market. Second, the typical sex worker is dynamically and discretely combining sex work with other low-wage employment (International Union of Sex Workers, 2024; Scelles Foundation, 2011; Cunningham and Kendall, 2011). This implies there is scope to decrease the size of the market by improving outside work options. Such a strategy has not been tested.
Hence, in my JMP with Alex Zhou, we answer the following question: How does improving outside options in the labor market impact sex work and sex workers?
We conduct a field experiment in Cambodia to study how improving labor market conditions for sex workers affects the choice between sex and non-sex work, as well as subsequent health and financial status and decisions. We partnered with a survey firm and an NGO already servicing this vulnerable population to enroll a sample of sex workers who combine sex work with service work. While the service work is a traditional waitressing or hostessing job, it also provides opportunities to meet potential sex work clients in venues during work hours. Importantly, since sex work is illegal in Cambodia, transactions occur offsite, usually at hotels. Given the sensitivity of the topic and illicit nature of the market for sex, we conducted extensive focus groups and piloting in order to design a rigorous recruitment strategy to enroll a sample of workers who were representative of the service sector and market for sex more broadly.
In a month-long cluster RCT with 600 workers across 300 service venues in Phnom Penh, Cambodia, we randomized workers by service venue (cluster) to receive either service work incentives, an unconditional cash transfer (UCT), or neither. Participants working in venues assigned to the incentives arm were offered a service work incentives program, which offered bonus weekly piece-rate pay for conducting surveys with venue customers in venues during workers’ shifts. This incentives task mimics the duties and earnings structure of the average service work job (McKinsey and Company, 2020; Oxfam, 2019). Participants working in venues assigned to the UCT arm were offered a fixed, unconditional weekly cash transfer, with a total transfer amount of $60. The UCT amount equaled expected average earnings, as per pilot data, in the incentives arm. The control arm had no experimental intervention.
Can we incentivize the outside option?
We find that the incentives increased service work, but surprisingly, not for all. Despite being relatively low-effort, -risk, and -time investment, only 65% of participants in the incentives arm “took-up” the incentive task, i.e., submitted at least one valid customer survey. This result is robust to excluding participants who could not participate for exogenous reasons, suggesting that this take-up rate reflects what we may expect from a scaled program. The 65% take-up rate is surprising given the highly (financially) vulnerable nature of these workers, implying that the incentive task had disutility for at least some workers. As a result, average participant earnings were $80. Relative to the control, this participation translated to an increase in daily service work earnings by 20% and hours by 9%.
Can incentivizing the outside option decrease sex work?
We find that the incentives intervention decreased sex work relative to the control by 13% (ITT). This result is surprising given the common assumption that sex work labor supply is inelastic, the complementarity of service and sex work in our setting, and the limited effects of other financial programs on the supply of sex work (Gong et al., 2019).
Our findings indicate that workers respond to changes in the price premium of sex vs. service work, i.e., that there are price effects to sex work. To assess how the supply of sex work may respond to a service work price subsidy, we use our results to estimate labor supply elasticities for non-sex work and between sex and non-sex work. We estimate a Marshallian own-price labor supply elasticity for service work of 0.45, but a Marshallian cross-price labor supply elasticity for sex work of -0.57. The contrasting signs on these estimates suggest that alternative work can substitute sex work.
Decreases to sex work are driven by both direct sex-to-service substation during potentially overlapping work windows and indirect substitution during inframarginal service work hours. Decreases are also driven by contraceptive users and those with higher levels of and exposure to sex work at baseline. Beyond labor supply, we find that the incentives intervention increased overall income, led to selection away from poorer but not higher-risk clients, and has the potential for modest projected health gains.
Why not just raise wages for the outside option?
In contrast to the incentives intervention, we find that the UCT intervention had null effects on sex work. Consistent with our labor supply model discussed below, we find suggestive evidence that the UCT may decrease service work instead. Given the month-long duration of our RCT, heterogeneity by baseline income levels suggests that participants may be amortizing the transfer; however, a back-of-the-envelope calculation of expected effects based on baseline income levels corroborates that income effects to sex work in this setting are minimal.
What about other outside options?
To translate our findings into a measure of worker preferences between sex and non-sex work beyond our setting, we interpret our findings through the lens of a labor supply model. In the model, a worker chooses levels of effort to devote to sex work versus non-sex work or leisure. We allow the two types of work to have different prices and disutilities; this is the worker’s key trade-off. Using estimates from our experiment, we calibrate the model to quantify the structural parameter for the relative disutility of sex work. Our estimate implies that the disutility of sex work not only outweighs that of non-sex work but is also marginally increasing in sex work. Using this estimation result, we find that, on the margin, one unit of sex work is associated with the disutility-equivalent of an additional 16 times the disutility of one unit of non-sex work. Our estimation result suggests that the significant policy attention focused on decreasing the size of the market for sex appears justified. However, it also underscores the need for evidence on the effects of interventions, like those tested in this paper, which can decrease the size of the market without the costs associated with status quo bans (Cameron et al., 2021; Cunningham and Shah, 2018).
Policy Implications
Overall, our findings help to inform protective legislation on the market for sex (Tertilt et al., 2022). Prevailing bans often trade off decreasing market size with worker and general welfare (Bisschop et al., 2017; Cameron et al., 2021; Cunningham and Shah, 2018). Our study shows that this tradeoff can be avoided. Specifically, our experimental results show that decreasing the price premium for sex work encourages substitution to outside options. This is a promising result as our structural estimate shows that workers prefer to earn from their outside option. As policymakers test strategies to improve welfare for sex workers (Khmer Times, 2023), our findings provide timely evidence that incentivizing labor market alternatives may decrease the size of the market for sex without compromising worker welfare.
Feature image taken by Ruchi during preparatory groundwork in Phnom Penh, Cambodia in June 2023
