Resource: Risky business - the market for unprotected commercial sex

Each day over 20,000 people become infected with the human immunodeficiency virus (HIV) worldwide, a large proportion of whom are infected through unprotected sex with sex workers (UNAIDS 2002). Although condoms are an effective defense against the transmission of HIV and other sexually transmitted infections (STIs) and there has been substantial education of sex workers regarding the risk of infection, large numbers of sex workers are not using condoms with their clients (UNAIDS 2002). Indeed, infection rates among sex workers are among the highest of any group, especially in developing countries with widely disseminated epidemics (World Bank 1999). A major question confronting policy makers who design and implement interventions for the prevention of acquired immunodeficiency syndrome (AIDS) and STIs is: Why do sex workers risk infection by not using condoms in their work?

Much of the health policy literature argues that in many cases sex workers engage in unprotected sex because they are uninformed of the risks (World Bank 1999; Lau et al. 2002). And in the cases in which sex workers are aware of the risk, many hypothesize that non-condom use occurs because condoms are either very expensive or not available at all (Negroni et al. 2002) or because sex workers are forced to have unprotected sex (Karim et al. 1995; World Bank 1999; Bronfman, Leyva-Flores, and Negroni 2002).

Ignorance does exist and the forced exploitation of sex workers does occur. However, another possible explanation is that sex workers are willing to risk infection by not using condoms with clients if they are adequately compensated. Indeed, economic theory has long posited the general principle of compensating wage differentials (e.g., Rosen 1986), and a number of authors have documented wage differentials that compensate for risky work activities in other labor sectors (e.g., Viscusi 1992; Siebert and Wei 1998). Similarly, automobiles with antilock brakes, homes with fire sprinklers, and other risk-reducing products are priced higher by the market because consumers are willing to pay for safer products (Viscusi, Vernon, and Harrington 2000). While there is anecdotal evidence that sex workers charge more for sex without a condom (Ahlburg and Jensen 1998), there has been little published work that has tried to test this claim. (1)

Understanding why sex workers do not use condoms is critical for the development of policy that is effective in increasing condom use and consequently in reducing the transmission of STIs including HIV. The usual policy recommendations are to intervene on the supply side (World Bank 1999). These policies include (1) educating sex workers about the risks, (2) increasing access to inexpensive condoms, (3) reducing environmental barriers to condom use by working with gatekeepers such as brothel owners and the police, and (4) empowering sex workers by, for example, improving their negotiating skills and fostering self-help organizations. Additionally, governments are urged to implement and enforce laws against human trafficking, rape, assault, and indentured servitude.

However, if some clients are willing to pay substantially larger sums for unprotected sex, supply-side interventions alone are less likely to sufficiently reduce unprotected commercial sex. Even knowledgeable sex workers with condoms, who are free to turn down clients, might be willing to supply unprotected sex if the price is right. In this case, complementary interventions on the client side that reduce the demand for unprotected sex are also necessary in order to increase condom use. However, client-based interventions are likely to be more difficult and more expensive to implement.

In this paper, we investigate why sex workers may not be using condoms. We begin by constructing a simple bargaining model of commercial sex that has a number of empirically testable predictions. The model predicts that a condom will not be used when the client's maximum willingness to pay not to use a condom is greater than the minimum the sex worker is willing to accept to take the risk. Surprisingly, however, the model also predicts that when the client is worried about the risk of infection from unprotected sex, he may be charged more for using a condom than for unprotected sex. Similarly, when the sex worker prefers not to use a condom, the client is given a discount for not using a condom. The price differential between protected and unprotected sex is a weighted average of the maximum the client is willing to pay for not using a condom and the minimum the sex worker is willing to accept to take the risk of infection by not using a condom. The weights are a function of the relative bargaining power of the client and sex worker.

We test these predictions using a panel data set that we recently collected from the Mexican states of Michoacan and Morelos. We use the panel to control for the likelihood that condom use is not exogenous because of sex worker heterogeneity and client sorting based on sex worker characteristics. Sex workers who have a preference for condom use may also charge higher prices regardless of condom use. For example, if better-educated sex workers have a preference for condom use and are better able to negotiate higher prices, then price and condom use will be positively correlated. However, the price will not reflect compensation for risk taking. Another source of bias comes from the possibility that clients who have preferences for condom use select sex workers who also have preferences for condom use. If these clients were, say, better educated and wealthier, then they would also be willing to pay more for the sex workers' services. This situation again introduces a positive correlation between price and condom use that does not reflect compensation for risk taking. Both of these are similar to the unobserved heterogeneity bias introduced from omitted productivity characteristics in estimating compensating wage differentials (Garen 1988; Hwang, Reed, and Hubbard 1992). To control for the endogeneity of condom use, we collected information on the last three to four transactions for each sex worker to create a panel data set. We then estimated a model with a sex worker fixed effect to control for bias from both unobserved sex worker heterogeneity and client selection.

We find that Mexican sex workers received a 23 percent premium for unprotected sex from clients who requested not to use a condom, and this premium jumped to 46 percent if the sex worker was considered very attractive. We also found that clients who requested condom use paid 9 percent more for protected sex, and sex workers who requested not to use a condom gave clients a 20 percent discount. These results are completely consistent with our theoretical predictions.

Mexico is an interesting country to study these issues because it does not yet have a generalized HIV/AIDS epidemic. In fact, the risk of HIV infection is low: 0.35 percent of sex workers and 0.128 percent of the general population are infected with HIV (Conde et al. 1993). In contrast, the risk of being infected with another STI is much higher since 17 percent of sex workers in our sample report having an STI in the last year. (2) The fact that the STI rate is so much higher than the HIV rate raises the concern of a likely rise in HIV infection in the near future because the STI rate is a marker for sexual risk behavior (Centers for Disease Control and Prevention 2004). The Mexican government and governments in other countries at risk of developing a generalized HIV/ AIDS epidemic are keenly interested in policies that increase condom use, especially among those populations at greatest risk of becoming infected or transmitting the infection (CONASIDA 2001, 2002; UNAIDS 2004).

One reason why the STI and HIV infection rates are relatively low is that sex work is regulated in many Mexican states and massive education campaigns for sex workers have been conducted over the last 10 years. (3) As a result, condom use is fairly high among sex workers in Mexico. Female sex workers in Chiapas used condoms 55 percent of the time (Valin and Egremy 2002), and in our data set described below, 84 percent of the sex workers reported always using condoms. Increasing condom use beyond these levels may be quite difficult, and education alone may not be able to do it.

Discussion

We find that sex workers in Mexico are responding rationally to financial incentives given their risk preferences. Consistent with our bargaining model, we find that sex workers are willing to assume the risks associated with providing unprotected sex to clients who request not using a condom for a 23 percent higher price. This premium increased to 46 percent if the sex worker was considered very attractive, a likely indication of her bargaining power. However, clients who preferred condom use paid a 9 percent premium to use condoms, and sex workers who did not want to use condoms had to reduce the price by 20 percent to compensate clients for taking the risk.

The 23 percent premium charged clients who want unprotected sex is quite large in terms of value of life. In the Appendix we estimate the value of life for Mexican sex workers implied by the estimated price premium for risking infection by having unprotected sex with a client. Specifically, the premium is a function of the probability that the sex worker becomes infected and the value of her loss in health status or life expectancy or both should she become infected. Our approach is similar to the literature on wage-risk premiums (Viscusi 1992, 1993; Miller 2000; Shanmugam 2000) in which the implicit value of a statistical life can be imputed from individual preferences regarding the value of employment with increased risk to health and the compensating wage differential associated with riskier employment.

We estimate that the value of a life year is between $14,760 and $51,832. To put these figures in context, annual earnings for sex workers are about $15,340 on the basis of our survey, and annual earnings for women of the same age range in the same states employed in other occupations are about $9,800 (INEGI labor survey). Thus the value of life implied by the risk premium is about one to five times annual earnings of this demographic group. To put these results in the context of previous efforts to estimate compensating wage differentials, Viscusi (1993) reviews 27 studies and reports the value of a year of life between 1.4 and 28 times annual income.

These findings suggest that the most effective interventions for reducing HIV/STI transmission through commercial sex will be those that target both the supply side (the sex workers and their agents) and the demand side (the clients) of the market. Interventions to educate sex workers about the risks of unprotected sex serve to reduce the number of transactions in which a condom is not used. Interventions to empower sex workers or improve their negotiating skills serve to increase their bargaining power so that they are able to capture more of their clients' willingness to pay and improve their incomes (which can also increase their disutility from non-condom use, further reducing unprotected sex). Making condoms more available or available more cheaply serves to make non-condom use relatively more expensive. All three of these types of interventions would thus be expected to reduce unprotected sex and HIV transmission.

However, if, despite an increase in the sex worker's disutility from non-condom use, clients are still willing to pay more than enough to compensate sex workers for taking the risk (not unlikely if clients are, on average, wealthier than the sex workers they hire), then a significant amount of unprotected commercial sex will continue to occur, albeit at a higher price. This implies that efforts to reduce clients' disutility of condom use, by educating them about the risks of unsafe sex or marketing the "joy of safe sex" to them, may be as important as or more important than interventions designed to change sex workers' disutility from non-condom use--and that the relative importance of client-side interventions increases as the inequality in willingness to pay/accept between clients and sex workers increases. Unfortunately, because clients are both more numerous and usually more difficult to target (especially in sex work sites because health educators who scare away clients are not especially welcome), targeting both the supply and the demand sides of commercial sex is likely to cost much more than supply-side efforts alone.

Source: Journal of Political Economy, June 2005 v113 i3 p 518(33)