Brokers and the Earnings of Female Sex Workers in India

A Service of

Leibniz-Informationszentrum econstor Wirtschaft Leibniz Information Centre Make Your Publications Visible. zbw for Economics

Brady, David; Biradavolu, Monica R.; Blankenship, Kim M.

Article — Published Version Brokers and the Earnings of Female Sex Workers in India

American Sociological Review

Provided in Cooperation with: WZB Berlin Social Science Center

Suggested Citation: Brady, David; Biradavolu, Monica R.; Blankenship, Kim M. (2015) : Brokers and the Earnings of Female Sex Workers in India, American Sociological Review, ISSN 1939-8271, Sage Publications, Thousand Oaks, CA, Vol. 80, Iss. 6, pp. 1123-1149, http://dx.doi.org/10.1177/0003122415609051

This Version is available at: http://hdl.handle.net/10419/190199

Standard-Nutzungsbedingungen: Terms of use:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Documents in EconStor may be saved and copied for your Zwecken und zum Privatgebrauch gespeichert und kopiert werden. personal and scholarly purposes.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle You are not to copy documents for public or commercial Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich purposes, to exhibit the documents publicly, to make them machen, vertreiben oder anderweitig nutzen. publicly available on the internet, or to distribute or otherwise use the documents in public. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, If the documents have been made available under an Open gelten abweichend von diesen Nutzungsbedingungen die in der dort Content Licence (especially Creative Commons Licences), you genannten Lizenz gewährten Nutzungsrechte. may exercise further usage rights as specified in the indicated licence. www.econstor.eu ASRXXX10.1177/0003122415609051American Sociological ReviewBrady et al. 6090512015

American Sociological Review 2015, Vol. 80(6) 1123 –1149 Brokers and the Earnings of © American Sociological Association 2015 DOI: 10.1177/0003122415609051 Female Sex Workers in India http://asr.sagepub.com

David Brady,a, b Monica Biradavolu,c and Kim M. Blankenshipc

Abstract This study examines whether working with a broker increases or reduces the payment received for the last client among female sex workers. Building on research on the informal economy and sex work, we formulate a positive embeddedness hypothesis, expecting a positive association, and an exploitation hypothesis, expecting a negative association. We analyze a large survey combined with intensive interview data on female sex workers in Andhra Pradesh, India. These data uniquely distinguish between the amount the sex worker actually received and the amount the client paid. The analyses show that brokers are associated with significantly lower last payment received. Although brokers are associated with a greater number of clients in the past week, this does not result in significantly higher total earnings in the past week. Further analyses suggest that much of the negative relationship with earnings is due to the fact that brokers lead to a lack of control over the amount clients are charged. At the same time, the results fail to show that brokers actually provide services of value. Ultimately, the results support the exploitation hypothesis. We conclude by encouraging the refinement of theories of embeddedness and exploitation and calling for greater research on workers in the informal economy of developing countries.

Keywords exploitation, embeddedness, informal economy, sex work, brokers

Sex work is widespread throughout the world argue that sex work is simply a low-income and has existed for most of recorded history. survival strategy (George, Vindhya, and Ray Every day, tens of millions exchange sex for 2010; Sanders 2005; Sullivan 2003; Weitzer money (World Health Organization 2006). 2007). Despite the clear contributions of these Traditionally, scholars studied sex work as a literatures, a key limitation of both is the matter of public health or ethics. Focusing on neglect of the labor market aspects of sex public health, much work documents the inci- work—the work of sex work (Weitzer 2009). dence of sexually transmitted infections (STIs) like HIV and studies risk behaviors aUniversity of California-Riverside like condom use (Blankenship et al. 2008; bWZB Berlin Social Science Center Mondal and Gupta 2013; Parrado, Flippen, cAmerican University and McQuiston 2004). Another literature scrutinizes the ethics of whether sex work can Corresponding Author: ever be free labor (see Sanders, O’Neill, and David Brady, School of Public Policy, INTS 4133, 900 University Avenue, University of California- Pitcher 2009). Some argue that sex work is a Riverside, Riverside, CA, 92521 system of domination or coercion; others E-mail: [email protected]

Dieser Beitrag ist mit Zustimmung des Rechteinhabers aufgrund einer (DFG-geförderten) Allianz- bzw. Nationallizenz frei zugänglich / This publication is with permission of the rights owner freely accessible due to an Alliance licence and a national licence (funded by the DFG, German Research Foundation) respectively. 1124 American Sociological Review 80(6)

In an influential review, Vanwesenbeeck remains quite ambivalent about the costs and (2001:242, 279) wrote, “The literature is still benefits of working with brokers. Payments more about sex than it is about work . . . [the] received represent the compensation that working situation (e.g., contexts, routines, FSWs receive, which is fundamental to the relations, conditions) has hardly been stud- quality of any work. Moreover, greater pay- ied.” For the first literature, this neglect sim- ments allow FSWs to meet their economic ply resulted from a different concentration. In needs while working less and having fewer the second, it was sometimes considered clients. Thus, payments received are also rel- problematic to call sex work “work” or a evant to public health and ethics. India pro- “market,” because doing so uncritically vides an appropriate and useful context for frames sex work as a freely chosen vocation. exploring these issues because the majority of Partly in response, economists have begun the world’s informal-sector workers, and to analyze sex work as a market (Arunacha- likely FSWs, are in the developing world. lam and Shah 2008, 2013; Gertler, Shah, We utilize a large survey of FSWs com- and Bertozzi 2005; Logan 2010; Moffatt and bined with intensive interview data. The sur- Peters 2004; Rao et al. 2003; Sahni and vey distinguishes between the amount the Shankar 2008). This literature highlights sex worker actually received and the amount the role of human capital, and the prices for the client paid. Further analyses investigate beauty, condom usage, and particular sex acts. the number of clients, control over working In the past decade or so, the sociological lit- conditions, and other outcomes. We aim to erature on the work of sex work has also make four contributions. First, we provide grown rapidly (e.g., Bernstein 2007; Boris, one of the few large sample analyses of the Gilmore, and Parrenas 2010; Brents, Jackson, earnings of FSWs (cf. Arunachalam and and Hausbeck 2010; Lever and Dolnick 2010; Shah 2013). Second, we scrutinize how one Murphy and Venkatesh 2006; Oselin and salient social relationship—between a broker Blasyak 2013; Rosen and Venkatesh 2008; and a worker—shapes the economic out- Sanders 2005; Sanders et al. 2009). The socio- comes of sex work. Third, we apply and logical literature goes beyond purely eco- refine theories of embeddedness and exploi- nomic factors to investigate gender, social tation. Fourth, we advance understanding of relations, stigma, and legality. Rather than the work of sex work and the informal econ- viewing all sex work as domination and coer- omy more generally. cion, sociological research emphasizes the meaningful heterogeneity in domination and coercion within sex work (Hoang 2011; Mor- Past Research on selli and Savoie-Gargiso 2014; Parrenas 2011; Brokers in the Informal Zhang 2011). Because of these contributions, Economy sex work research now spans a variety of topics. The field has answered Vanwesenbeeck’s Sociologists have long studied the contractual call and now focuses considerable attention on and quasi-contractual arrangements facilitat- work. ing the sale of workers’ services by brokers. This study builds on these literatures to Brokers are especially important in informal examine an important aspect of the work of and criminal economic activity. Informal female sex workers (FSWs) in Andhra and illegal workers depend on suppliers, Pradesh, India. Specifically, we analyze the co-workers channeling customers and collect- relationship between working with a broker ing payments, and people providing corollary and the amount of payment received for the services like security (Bourgois 2003; Portes last client. Brokers are fairly common among and Sassen-Koob 1987; Venkatesh 2006). FSWs, and play a key role in the informal Criminals often form “tutelage relationships,” economy more generally. Yet, the literature where more experienced criminals share Brady et al. 1125 knowledge and opportunities in exchange for On the other hand, pimps might not actually a portion of earnings (Hagan and McCarthy provide useful services (Bernstein 2007; 1997). In these often mutually beneficial tute- Maher 1997). Chapkis (1997:98) writes, “Sex lage relationships, the experienced criminal work can be radically transformed . . . when typically serves as both broker and mentor control passes from a worker to a third party (McCarthy and Hagan 2001; Morselli, Trem- (brothel owner, escort agency, manager or blay, and McCarthy 2006). Conversely, con- pimp).” Brokers are often manipulative, vio- siderable evidence suggests illegal and lent, and abusive to FSWs (Karandikar and informal workers are vulnerable to brokers. Prospero 2010; Morselli and Savoie-Gargiso For example, compared to independent work- 2014; Zheng 2009). Chapkis (2000) shows that ers, undocumented Hispanic immigrants managers and brokers try to remove as much working through subcontractors receive lower decision-making power from FSWs as possi- wages and no benefits, and experience longer ble, including in interactions with clients, spells without work (Flippen 2012). scheduling, and rates of pay (see also May, Recent research on brokers in sex work Harocopos, and Hough 2000; Miller 1995). provides evidence of both benefits and harms Relative to female sex workers, the lower pres- (Morselli and Savoie-Gargiso 2014). On the ence of pimps among male sex workers results one hand, brokers manage pace, schedule, in less coercion and violence, and greater con- and compensation; recruit, screen, and col- trol over working conditions (Weitzer 2009). lect money from clients; provide security; In addition, at least some FSWs are deceived and mediate disputes with clients, police, or trafficked into sex work by brokers (George and other FSWs (Weitzer 2009). Morselli et al. 2010; Sullivan 2003; Weitzer 2007, and Savoie-Gargiso (2014) find that most 2009). Because many FSWs engage in sex interactions between pimps and FSWs involve work out of economic necessity and with lim- management (time, money, and site) or the ited legality, they may be especially vulnerable pimp acquiring goods, services, and informa- to brokers (Bowen et al. 2011; Karandikar and tion for FSWs. Some brothel owners and Prospero 2010; Vanwesenbeeck 2001). managers provide safer working conditions Research on brokers in sex work is thus (Brents and Hausbeck 2005), and madams ambivalent about whether they are beneficial often act as teachers and socializers (Heyl or harmful. Moreover, there have been few 1977). Furthermore, FSWs often elect to rigorous analyses of the relationship between work for brokers to avoid police harassment working with a broker and the payments (Chapkis 2000). Levitt and Dubner (2009) received by FSWs. argue that pimps attract higher-paying clients, and therefore lead to significantly higher earnings for Chicago FSWs. Zhang (2011:526) Positive Embeddedness explains that in Tijuana, “sex trade facilitators or Exploitation? were not much different from merchants of Building on the aforementioned literatures any other type in an illicit market place, treat- and relevant sociological theories, we formu- ing sex workers like a commodity.” Marcus late two hypotheses for the relationship and colleagues (2014) conclude that most between brokers and payments received by brokers are “facilitators,” managers, or “spot FSWs. The positive embeddedness hypothe- pimps” who arrange transactions with clients. sis expects brokers increase the payments They conclude that pimps’ control of FSWs received. We label this hypothesis positive tends to decline over time (see also Zhang embeddedness to make clear that the entire 2011), and they estimate that only 2 percent embeddedness literature does not necessarily of FSWs in Atlantic City and New York were lead to the expectation of positive effects. The in a violent relationship with a pimp (see also exploitation hypothesis expects brokers Sanders et al. 2009). reduce the payments received. 1126 American Sociological Review 80(6)

Sociologists have long contended that eco- literature tends to concentrate on (1) empha- nomic action is embedded in social relations sizing the relevance of social relations to (Krippner 2001; Portes and Sensenbrenner economic action and (2) mapping the formal 1993; Powell and Smith-Doerr 1994; Uzzi structure of networks within markets. In turn, 1996). Applying relational explanations to the literature has been criticized for focusing markets has been one of sociology’s truly too much on those questions, for treating all cumulative research programs (Krippner and social ties as equivalent, and for abstracting Alvarez 2007). Sex work involves many of away the content from social relations the processes that have been studied in the (Fligstein 2001; Krippner 2001; Krippner and embeddedness research program: job search, Alvarez 2007; Powell and Smith-Doerr 1994). recruitment of and interactions with clients, In the sex work literature discussed earlier, reputation, autonomy, hierarchy, and compen- social relations are not usually the idealized sation (Chapkis 2000). Although social rela- solidaristic and reciprocal networks featured tions are often implicit in the sex work in the embeddedness literature. Rather, vul- literature, they have rarely been explicitly nerable low-income workers in the informal theorized for the work of sex work (for an economy, like FSWs, are more likely to have exception, see Kotiswaran 2008). what Desmond (2012) calls “disposable ties,” According to the positive embeddedness which are opportunistically formed with loose hypothesis, brokers should increase payments acquaintances, temporary and without long- received by FSWs. To the extent FSWs are term security or trust, and unsustainably bur- connected to, regularly work with, and are dened with emotions and financial demands. known by brokers, FSWs can be said to be Building on these critiques, we formulate the “embedded with” and have network ties to exploitation hypothesis, which expects brokers brokers. Brokers should coordinate efficient reduce FSWs’ last payment received. Exploita- exchange by bridging clients and FSWs. This tion is routinely invoked by sociologists in is because brokers can provide information or many literatures (Tilly 1998; Tomaskovic-Devey a contact point for clients searching for FSWs, 2014; Wright 1997), but the field has made and steer more lucrative clients to FSWs. insufficient progress in defining and measuring Because brokers have knowledge of where to exploitation (Sørensen 2000). As Sakamoto and find FSWs, this should reduce transaction Liu wrote in 2006, “There is still no published costs for clients and reduce idle time and research in modern sociology that has quantita- boost demand for FSWs. By working together tively measured or empirically studied exploita- routinely over time, a relationship of trust and tion in a systematic manner” (p. 219). To solidarity should emerge such that brokers develop the exploitation hypothesis, we make have an interest in and obligation to paying six points. FSWs a fair amount (Uzzi 1996). As dis- First, building on Sakamoto and colleagues cussed earlier, brokers can provide security (Liu, Sakamoto, and Su 2010; Sakamoto and and help FSWs avoid police harassment. As a Kim 2010; Sakamoto and Liu 2006), we pro- result, brokers should enable FSWs to be pose that brokers exploit FSWs if FSWs are mobile and work in a wider variety of set- underpaid relative to the value of their labor. tings, and to be more visible and less discreet, Second, this raises the question of defining all of which should facilitate attracting clients “value.” Sakamoto and colleagues utilize and higher earnings (Bernstein 2007; Oselin information on workers’ marginal value- and Blasyak 2013; Sanders 2005). Ultimately, added productivity and define exploitation as embeddedness with brokers should increase “being paid less than the value of what one payments received by FSWs. produces” (Sakamoto and Kim 2010:20). It would be unbalanced to claim that the Because it is likely impossible to objectively entire embeddedness literature would expect determine the marginal productivity/value- positive effects. That said, the embeddedness added of an FSW in the informal economy, Brady et al. 1127 the best available approximation is the price effort.” Just like embeddedness, exploitation the market is willing to pay. Indeed, Saka- requires FSWs to be connected to, work with, moto and Kim (2010:20) refer to exploitation and have network ties to brokers. Thus, as being underpaid relative to the “market exploitation has similarities to the embedded- value” of productivity. Economists often ness literature’s concept of the negative side claim price is the only measure of value, and of social relations (Portes and Sensenbrenner such a claim should be even more plausible in 1993). an informal economy with many independent Sixth, the concept of exploitation addresses transactions between many FSWs and many critiques of the embeddedness literature by clients. specifying the content and kind of the social Third, therefore, the market price can be relation. Particularly important, exploitation is estimated concretely from a model predicting facilitated by “adapting” (Tilly 1998) legiti- the last payment received conditional on the mated status hierarchies devaluing women characteristics of the FSW and her working (Ridgeway 2014; Tomaskovic-Devey 2014) conditions. If the broker coefficient is signifi- and gendered power relations (Miller 1995) cantly negative, on average, FSWs receive into the content of broker–FSW relations. less than their value when they work with Indeed, scholars of gender inequality at work brokers. often claim exploitation—arguing that wom- Fourth, consistent with classic theories of en’s labor is devalued or that employers exploit exploitation, while brokers would lose out on women by underpaying them relative to their the rewards of the exchange if FSWs did not value (Folbre 1982; Tilly 1998; Tomaskovic- work with them, FSWs would be better off if Devey 2014). When women have limited alter- they did not engage with brokers (Roemer native employment opportunities, their economic 1982).1 This can be demonstrated if one finds independence and bargaining power is under- the costs from working with a broker exceed mined (Folbre 1982; Miller 1995). Because sex the value of the services that brokers provide, work violates gender norms and stigmatizes such as wage gains due to greater safety and FSWs, this erodes alternative employment less police harassment. More simply, exploi- opportunities further, which makes FSWs more tation occurs if brokers fail to provide any dependent on sex work and more vulnerable to valuable services while extracting resources brokers. Partly as a result, the FSW–broker rela- from FSWs. tion has strong potential to become an exploita- Fifth, we emphasize that exploitation is a tive kind of relation. Some FSWs do not work social relation (Tomaskovic-Devey 2014; freely, and some are trafficked or coerced by Wright 1997). Rather than simply saying brokers. The prevalent threat of violence also FSWs are oppressed or disadvantaged, exploi- empowers brokers over FSWs (Bourgois 2003; tation means there is an identifiable actor Miller 1995). In the marginal legality of sex receiving disproportionate rewards relative to work in India, brokers can insert themselves the value of brokering. As Sakamoto and Kim into transactions between clients and FSWs, (2010:20–21) explain: “Someone else obtains because clients may be cautious about solicit- the difference without providing adequate ing FSWs for fear of being arrested. With their compensation . . . part of it is being diverted bridging location between clients and FSWs, to another group of people who are benefiting brokers have leverage in the market for sex from the appropriation.” This is consistent work. The market for sex work also has many with Tilly’s (1998:86–87) definition of imperfections (e.g., breach of payment con- exploitation as “[s]ome well-connected group tracts without legal recourse and opportunities of actors controls a valuable, labor-demanding to bribe police) and these imperfections create resource from which they can extract returns advantages for brokers. Such leverage and only by harnessing the effort of others, whom advantages enhance brokers’ power over they exclude from the full value added by that FSWs and facilitate exploitation. 1128 American Sociological Review 80(6)

In sum, we formulate two hypotheses. The of our respondents reported being originally positive embeddedness hypothesis expects “lured, cheated, or forced into sex work,” and brokers will increase the last payment 16.2 percent started sex work when they were received by FSWs. The exploitation hypoth- under age 18. Most FSWs reported entering esis expects brokers will reduce the last pay- sex work because of economic necessity and ment received by FSWs. because sex work offers better earnings than alternative employment opportunities. One of our interview respondents, a mid-40s street- The Context of Sex Work based FSW with 15 years of experience, in Rajahmundry explained: The data were gathered from 2006 to 2010, as part of a multi-method study of FSWs in the My husband died and the children were Rajahmundry area of the East Godavari dis- young. When I went for other work like trict in Andhra Pradesh, India. Rajahmundry’s daily labor, the men there asked me to sleep metropolitan area has a population of about with them and gave me work only if I did half a million and is one of several large that. I used to work as a laborer for 30 rupees urban areas in Andhra Pradesh. It is one of and had to bear with all these things after two major towns in the largely rural, agricul- the hard work. So I thought it would be bet- turally prosperous East Godavari district. It is ter if I did this work alone. I met a man and an informative site because it is located on I decided it is the same mistake if it is with National Highway 5, a major north-south one or many. So I took a woman’s help and route for truckers, and is known as a location came to the [city] Center. I started earning for FSWs. Although Rajahmundry is worthy some money and brought up my children. I of study in its own right, it is also likely a felt this was providing food for my children typical case of and plausibly representative of and felt this was better. When I went as a India’s medium/smaller cities, towns, and vil- laborer, I had to work hard for 40 rupees and lages (Bowen et al. 2011). Because the sam- had to sleep with the men there. The money ple demonstrates considerable heterogeneity, was not sufficient. Now I can work for an Rajahmundry likely captures much of the hour or two and get 100 and look after my diversity of sex work that can be found in children well. India and other developing countries (Bira- davolu et al. 2009; Blankenship et al. 2008; Similarly, a street-based 30-year-old FSW George et al. 2010; Kotiswaran 2008; Sahni, with five years of experience explained, Shankar, and Apte 2008). That said, Rajah- “After the birth of four children, he [husband] mundry is likely different from large metro- died in an accident because of excess alcohol- politan centers in developing countries, such ism. . . . My in-laws looked after me comfort- as Mumbai, which have closer connections to ably for a year and then they forced me out of the global economy and tourism (Zhang the family. We had no other source of income 2011; Zheng 2009).2 to survive and I decided to enter this profes- Biradavolu and colleagues (2009:1542) sion to survive.” estimate there were approximately 1,500 About 15 percent of the women in our FSWs in the area at the time of the first survey. sample worked with a broker (see Table A1 in In our sample, less than a quarter of FSWs can the Appendix). Respondents worked an aver- read or write, their average age was 31, about age of 3.95 out of the past seven days and had two-thirds were separated/divorced/widowed, an average of 10 clients in that period. They and they had an average of 1.8 children (see estimated that nearly half of their clients in Table A1 in the Appendix). the past seven days were truckers and about a Although most reported having entered third were college students or rickshaw sex work without trafficking, about 8 percent drivers.3 Almost a third of FSWs worked at Brady et al. 1129 home, followed by a quarter on the highway, proven useful for recruiting hidden popula- and smaller shares in the street, brothels, tions like FSWs, and there is evidence that it lodges, and multiple venues. produces more representative samples than do In India, and much of the world, the legal- place-based sampling or other feasible alterna- ity of sex work is quite ambiguous. This gives tives (Magnani et al. 2005). police arbitrary authority to extract resources To be eligible, respondents had to report from and harass FSWs (Biradavolu et al. being at least 18 years old and having 2009). Kotiswaran (2008:589) explains it is exchanged sex for money at least once in the legal for “a sole sex worker to sell sex for her prior year. The surveys lasted 90 to 120 min- own benefit in a discrete manner in a place utes and were conducted in Telugu by trained that is not in or near any public place.” How- interviewers after confirming participants’ ever, solicitation in public is illegal, and it is informed consent. Respondents received a illegal to profit from or depend on the sex modest compensation for participation and work of others (e.g., running a brothel or for recruiting other FSWs. We merged the working as a broker). This enables police to three survey waves and filtered out respond- raid brothels and arrest FSWs arbitrarily on ents who participated in earlier waves.7 After public nuisance charges, and in response to accounting for missing data, this yielded a complaints from the general public, neigh- sample of 1,669 women. Table A1 in the bors, and shopkeepers, or under the accusa- Appendix contains descriptive statistics and tion of making money from other FSWs detailed definitions for the variables. Table (Biradavolu et al. 2009; Kotiswaran 2008).4 A2 shows the correlation matrix among inde- In India (Biradavolu et al. 2009) and other pendent variables. contexts (Levitt and Dubner 2009), FSWs In addition, we utilize 27 intensive inter- may be pressured to provide sex and bribes to views with FSWs who were recruited from the police in exchange for the opportunity to among the survey respondents. Respondents work. The police often know who FSWs are, were drawn from a combination of purposive and they use the threat of arrest and shame to and convenience sampling; they were selected demand bribes and sex. to reflect the diversity of the quantitative sample in terms of demographics, venue, and other characteristics. The interviews were semi- Methods structured, covered life history and several Our principal data are three cross-sectional topics, and included opportunities to discuss surveys, administered in 2006, 2007, and 2009 brokers.8 The interviews lasted approximately to 2010, with samples of 812, 673, and 850, two hours, were conducted in 2006 and 2007 respectively.5 Participants for all three surveys in Telugu, and then were translated and tran- were recruited using respondent driven sam- scribed into English. Respondents received pling (RDS). Five seeds in the first and third, additional compensation for the interview. and 10 in the second, representing different We utilized an embedded design, whereby groups of FSWs, completed the survey and the qualitative data deepens and validates the distributed coupons to up to three members of quantitative analyses (Creswell and Plano their network who met the study’s eligibility Clark 2007). We read the interview tran- criteria (recruits).6 Recruits then had the oppor- scripts in their entirety. Then, we examined tunity to distribute coupons to up to three other all passages relevant to brokers, payments FSWs in their network. To ensure the final received, and several other aspects of work- sample was independent of the initial seeds, ing conditions. We selected examples and we completed at least six recruitment waves quotes that were representative of the broader and generated a sample size of at least 500 in patterns. We used the qualitative interviews to each wave (Magnani et al. 2005). Despite enhance interpretation of and illustrate the some limitations (discussed below), RDS has quantitative results (Plano Clark et al. 2008). 1130 American Sociological Review 80(6)

Dependent Variables and Estimation make the results comparable to the other The key dependent variable is payment models and because the results were similar received for the last client.9 This self-reported with ordinal logit (available upon request). measure is the logged real (2009) rupees actu- In supplementary analyses, we examine ally received by the FSW. To calculate this, seven outcomes measuring services poten- we used survey questions on what the last tially provided by brokers. These models are client paid and how much of the money went estimated with logistic or negative binomial to the respondent (e.g., as opposed to a bro- regression. Paid bribe to police is a binary ker). As discussed below, the payment measure of whether a respondent gave the received for the last client is significantly police something to avoid trouble in the past lower than the amount the last client paid six months. Had sex with police is a binary (t = –5.76 unlogged, t = –9.39 logged). The measure of whether a respondent has ever had mean and median is 4.85 logged real (2009) sex with police to avoid trouble with them. rupees, which translates to about 127 real Number of police raids is a count in the past rupees in 2009. We analyze this dependent six months of police conducting raids in the variable with ordinary least squares (OLS) place the respondent conducts sex work. regression because the logged value is rela- Number of times arrested is a count of the tively normal (skew = .22). number of times in the past six months a In addition to the main dependent variable, respondent has been arrested by the police. we analyze the number of clients and total Number of times experienced violence meas- earnings. The number of clients is a self- ures how often the respondent has been beaten report for the past seven days. The average without a weapon in the past six months number of clients is 10.15 and the median (0 = never, 1 = once, 2 = 2 to 5 times, 3 = 6 to is 6.10 Because it is a count and exhibits over- 10 times, 4 = more than 10 times). Number of dispersion, we analyze this dependent varia- times forced to have sex measures how often ble with negative binomial regression. Total the respondent has been forced to have sex in earnings is the logged product of the payment the past six months (0 = never, 1 = once, 2 = 2 received for the last client and the number of to 5 times, 3 = 6 to 10 times, 4 = more than 10 clients in the past seven days. This assumes times). Number of threats with weapon meas- that payment for last client is representative ures the frequency of having been threatened of all clients in the past week. However, with a weapon or having a weapon used because we do not have a survey question against the respondent in the past six months regarding payments received over time, this is (0 = never, 1 = once, 2 = 2 to 5 times, 3 = 6 to the best available proxy for longer-term total 10 times). compensation. We estimate OLS regression with this dependent variable. The last three dependent variables evalu- Independent Variables ate FSWs’ subjective lack of control over The key independent variable is broker. We work. Respondents were asked how fre- coded an FSW as working with a broker if she quently “you decide the amount you charge reported that someone else collected money for sex with a client,” “you decide the type of from the last client. We also experimented sex you have with a client,” and “you decide with other measures of brokers (e.g., whether the number of clients you see in a week.” We a respondent resides with a “madam/ used these to create lack of control of amount owneramma,” paid someone to bring the last charged, lack of control of type of sex, and client, paid a broker to use a place for the last lack of control of number of clients. We coded client, gets condoms from a madam, or expe- the responses into 0 = always, 1 = usually, 2 = rienced violence from a madam). The results sometimes, 3 = rarely, and 4 = never. We pre- were robust regardless of whether we com- sent OLS regression for these variables to bined or separated these measures. There is Brady et al. 1131 very high overlap, and very few respondents 2005). Hence, we include the number of sex had any relationship with a broker without workers known. reporting someone else collected money from Much literature shows family and kin are the last client.11 This should increase confi- “proximal foreground” priorities for FSWs dence in our measure. However, the high and informal-sector workers (Hagan and overlap inhibits comparison between multiple McCarthy 1997; Venkatesh 2006). On one measures and we cannot be certain our results hand, families provide support and resources generalize to FSWs with brokers who did not (e.g., income and unpaid labor like child- collect money from the last client. Consistent care), which likely raise reservation wages with other research (Marcus et al. 2014; such that sex work is only worthwhile for Sanders et al. 2009; Zhang 2011), there is higher payments. For instance, cohabiting heterogeneity among brokers. For example, a allows criminals to ease up on illegal earn- few interviewees distinguished between kinds ings (Uggen and Thompson 2003), and mar- of brokers. One referred to rickshaw drivers ried FSWs tend to earn more than unmarried as “commissioning agents” who are different FSWs (Arunachalam and Shah 2008). On the from brokers who arrange long-term con- other hand, families create needs and may tracts, for example, at a lodge (cf. Zhang’s pressure women into sex work (Sahni and [2011] “commissions” and “referral fees” in Shankar 2008; Vanwesenbeeck 2001). For Tijuana). Unfortunately, the survey did not example, Davis (1993) finds that FSWs ask about brokers’ sex or characteristics. As receive little monetary or emotional support Table A2 in the Appendix shows, working from families, and women often enter sex with a broker is moderately positively corre- work due to parental rejection or pressure. lated with working in a brothel. Otherwise, Lim (1998) concludes that women turn to sex there are no clear associations between work- work after the departure of a husband.14 We ing with a broker and the other independent thus include five measures of the household. variables. Other contributors in the household is a We begin by adjusting the models for the binary measure for whether another person FSW’s characteristics. We include dummies contributes income. Marital status is meas- for the wave 2 (2007) and wave 3 (2009) sur- ured with dummies for married and sepa- veys (reference = 2006). Because venue is a rated/divorced/widowed (reference = never central stratifier (Murphy and Venkatesh married).15 We also include number of chil- 2006; Sanders et al. 2009; Weitzer 2009), we dren and a binary variable for parent(s) in include dummies indicating if the FSW household. worked in a brothel, street, lodge, or highway We next adjust for working conditions. In (reference = home).12 Much research shows these models, we include the number of cli- that human capital increases wages in the ents. In the final set, we also include meas- informal economy, similar to the formal econ- ures of lack of control over work. We include omy (Hoang 2011; McCarthy and Hagan binary variables for always use condoms and 2001; McKeever 1998). We thus include age usually use condoms in the past seven days in years,13 which incorporates the value of (reference = sometimes, rarely, or never).16 increasing experience, skills, and knowledge Studies in India and elsewhere find FSWs with age and the desirability/beauty attached tend to sacrifice pay if using condoms to younger FSWs. We also include literacy (Arunachalam and Shah 2013; Gertler et al. with a binary measure of whether respondents 2005; Rao et al. 2003). By contrast, recent can read or write (reference = neither). There research in southern India finds there is now is evidence that FSWs utilize social networks to a premium for safe sex (Mondal and Gupta learn how to increase earnings and select cli- 2013). ents (Brents et al. 2010; Hagan and McCarthy The initial entry into sex work is a critical 1997; Murphy and Venkatesh 2006; Sanders juncture that has been shown to have lasting 1132 American Sociological Review 80(6) consequences much later in working condi- about .144 standard deviations, holding all tions (Bowen et al. 2011; Cobbina and Oselin other variables constant at their means. These 2011). At least some women are pressured, results are consistent with the exploitation deceived, or trafficked into sex work, and this hypothesis and contradict the positive embed- often occurs by entering sex work at a young dedness hypothesis. age. Trafficked FSWs experience lower Among FSWs’ characteristics, working in autonomy and greater risk for STIs and sex- the street or on the highway is associated with ual assault (Sullivan 2003). Therefore, we significantly lower payments received than include a binary measure of started sex work working at home. By contrast, working in a at young age, coded one if the respondent lodge is associated with significantly greater reported having begun sex work before the payments. Age and being separated/divorced/ age of 18.17 Extensive literatures show vio- widowed are associated with lower payments lence, abuse, and rape are widespread among received; literacy and having other contribu- FSWs and informal-sector workers (Bourgois tors or parent(s) in the household are associ- 2003; Karandikar and Prospero 2010; Miller ated with higher payments. Among the 1995; Oselin and Blasyak 2013; Zheng 2009). measures of working conditions, the number For instance, Hagan and McCarthy (1997) of clients and having started sex work at a find that violence, sexual assault, and young age are negatively associated with last demands of sex as payment for shelter and payment received. food are common for homeless youths. Mur- The largest coefficients are for literacy phy and Venkatesh (2006) find that even and venue. The coefficient for having a bro- among indoor sex workers in New York, vio- ker is slightly smaller but comparable to a lence, threats, and rape are prevalent. FSWs standard deviation increase in age, having often organize their work to reduce the risk of other contributors or parent(s) in the house- physical harm, and such tactics have ramifi- hold, being separated/divorced/widowed, cations for earnings (Bernstein 2007; Oselin and having started sex work at a young age. and Blasyak 2013; Sanders 2005). Hence, we The coefficient for having a broker is consid- include three of the aforementioned measures erably larger than a standard deviation of experiences with violence and rape: num- increase in the number of clients in the past ber of experiences of violence, number of seven days.19 times forced to have sex, and number of In the intensive interviews, several FSWs threats with weapon.18 reported having madams or pimps, and others were brought clients by intermediaries like rickshaw drivers. Some described working Results “on contract,” meaning they worked in a lodge Last Payment Received or brothel where they were paid a fixed Table 1 shows the OLS models of last pay- amount or rate per client for a time period. ment received. In the first model, we include Some FSWs paid brokers a portion of their the broker variable and the basic characteris- earnings in exchange for delivering the FSW tics. In the second model, we add measures of to a lodge. Across these heterogeneous bro- working conditions (minus the measures of kers, FSWs repeatedly criticized brokers’ high lack of control, which we add later). We fees. One FSW reported that clients pay 110 report standardized or semi-standardized (for rupees at lodges, but she receives only 40. binary independent variables) coefficients Another reported that rickshaw drivers receive and t-scores. 20 to 30 rupees per 100, and the madam and In both models, having a broker is signifi- FSW share the remaining 70 to 80. One FSW cantly negatively associated with last pay- said that rickshaw drivers charge one-third the ment received. In Model 2, having a broker is payment for bringing clients from the rail sta- expected to reduce last payment received by tion. Others explained that the client pays Brady et al. 1133

Table 1. OLS Models of Payment Received for Last Client (Logged Real 2009 Rupees) (N = 1,669): Standardized or Semi-Standardized Coefficients and (T-scores)

Model 1 Model 2

Broker –.136* –.144* (–2.029) (–2.150) Venue Brothel –.096 –.058 (–1.239) (–.746) Street –.276** –.277** (–4.719) (–4.759) Lodge .211** .220** (2.789) (2.907) Highway –.298** –.267** (–5.715) (–5.029) Human Capital Age –.158** –.164** (–6.913) (–7.100) Literacy .352** .356** (6.677) (6.757) Number of Sex Workers Known .018 .035 (.810) (1.543) Household Other Contributors in Household .190** .196** (3.547) (3.671) Married .102 .124 (1.343) (1.640) Separated/Divorced/Widowed –.178** –.199** (–3.055) (–3.386) Number of Children –.035 –.036 (–1.516) (–1.587) Parent(s) in Household .156** .158** (2.957) (2.999) Working Conditions Number of Clients in Past Seven Days –.069** (–2.960) Always Use Condom .014 (.259) Usually Use Condom .283 (1.802) Started Sex Work at Young Age –.164** (–2.643) Number of Times Experienced Violence in Past Six Months –.028 (–1.186) Number of Times Forced to Have Sex in Past Six Months .035 (1.519) Number of Threats with Weapon in Past –.002 (–.098) Wave 2 .229** .225** (4.061) (3.821) (continued) 1134 American Sociological Review 80(6)

Table 1. (continued)

Model 1 Model 2

Wave 3 .580** .548** (10.990) (9.813) R-squared .222 .233

500 to 1,000 rupees to the madam, but the amount. If we have these brokers we have to FSW receives only 200 to 300. give 30 percent to them. Indeed, a widely shared sentiment was that Interviewer: Is there any advantage with them? madams and brokers were extracting an unfair Respondent: No, it is a loss of money. amount. A 56-year-old home-based FSW who had 10 clients in the past week said: Consistent with the exploitation hypothesis, the surveys and interviews provide evidence There is a company owner. We go there if she that working with a broker is associated with sends word. She collects 100 rupees from the lower payments received. client and pays us 25 rupees. And she lies to Given these results, one may ask why us that the client gave her only 50 rupees out FSWs work with brokers. Several FSWs of which she was taking 25 and giving us 25. claimed they do not willingly engage with brokers. Rather, brokers occupied an interme- The respondent in the following exchange diary position that enabled clients to locate was a 30-year-old FSW working in multiple FSWs, and brokers leveraged this position to venues who had 13 clients in the past week: extract a fee regardless of FSWs’ preferences. Still, at least a few interviewees reported vol- Respondent: The rickshaw drivers and the bro- untarily entering into contracts or engaging kers bring customers. If the deal is made for with brokers. We now turn to this issue. 500 we have to give 150 rupees to the broker and with the remaining 350 we have to pay for the room, and give the room boy some Number of Clients and Total Earnings amount. We have some amount left. For all this procedure, we don’t go for the rickshaw One reason FSWs may work with brokers is parties. We talk on our own. Then even if we that brokers bring them clients and thus give for the room we have 300 left. It will increase their number of clients. If such an be difficult to share the amount with three increase in clients is large, it could over- people. They just bring customers and it is whelm the negative effect on the last payment us that do the hard work. So we don’t keep received and result in greater total earnings. agreements with the rickshaws. Table 2 examines these issues. Interviewer: If a broker gets a 500 rupees deal, The first model in Table 2 examines the how much will you get? number of clients in the past seven days. This Respondent:We get 150 or 200. If a friend does model shows that having a broker significantly not get any customers then we give the 50 to increases the number of clients. However, the them and keep the 150 for us. influence of brokers is not terribly large. Hav- Interviewer: Are there any benefits because of ing a broker is expected to increase the number these brokers? of clients by a factor of 1.16. Having a broker Respondent: They take commission. If they has a comparable influence to being married or come directly to us we will get the full having started sex work at a young age, and a Brady et al. 1135

Table 2. Negative Binomial Model of Number of Clients and OLS Model of Total Earnings in Past Seven Days (Logged Real 2009 Rupees) (N = 1,669): Coefficients and (Z-scores or T-scores)

Number of Total Earnings Clients in Past in Past Seven Seven Days Days

Broker .135* –.042 (2.186) (–.516) Venue Brothel .301** .167 (4.282) (1.771) Street .035 –.204** (.661) (–2.901) Lodge .337** .387** (5.002) (4.237) Highway .452** .0678 (9.653) (1.074) Human Capital Age –.007* –.0253** (–2.414) (–7.181) Literacy –.082 .284** (–1.701) (4.471) Number of Sex Workers Known .004** .005** (4.007) (3.299) Household Other Contributors in Household –.059 .154* (–1.227) (2.384) Married .165* .194* (2.404) (2.123) Separated/Divorced/Widowed .119* –.093 (2.222) (–1.308) Number of Children .052* –.011 (2.461) (–.406) Parent(s) in Household .039 .202** (.829) (3.170) Working Conditions Always Use Condom .491** .400** (10.360) (6.414) Usually Use Condom .537** .790** (3.885) (4.170) Started Sex Work at Young Age .140* –.046 (2.565) (–.618) Number of Times Experienced Violence in Past Six Months .061* .033 (2.559) (1.054) Number of Times Forced to Have Sex in Past Six Months .108** .153** (3.652) (3.764) Number of Threats with Weapon in Past Six Months –.016 .005 (–.276) (.062) Wave 2 –.057 .164* (–1.049) (2.309)

(continued) 1136 American Sociological Review 80(6)

Table 2. (continued)

Number of Total Earnings Clients in Past in Past Seven Seven Days Days

Wave 3 –.339** .371** (–6.582) (5.509) R2 .202

Note: Constants not shown. References: wave 1, works exclusively at home, cannot read or write, never married, no other contributors in household, no parents in household, never/rarely use condom, and started sex work at older age. *p < .05; **p < .01 (two-tailed tests). larger influence than the number of times one in the afternoon when we are sick. It is very was forced to have sex, being separated/ problematic. divorced/widowed, the number of experiences Interviewer: Do they decide the number of cli- of violence, age, and the number of sex work- ents you have to serve? ers known. Brokers have a smaller influence Respondent: They decide. than always using a condom, or working on the Interviewer: Does she send the clients though highway, in a lodge, or in a brothel. you refuse to serve? Of course, some of these results may reflect Respondent: Yes, she sends them to me though I that the number of clients causes the inde- refused to serve. pendent variable. Because our data are cross- sectional, we cannot convincingly identify the A 36-year-old highway-based FSW with 18 causal order of these relationships. FSWs with years of experience added: a higher number of clients might be more likely to know other FSWs, always use con- They [brokers] even send us to the beggars. doms, and be exposed to greater violence and We had our heart broken and came out to do forced sex. Indeed, FSWs with many clients the business at the highway. They make us may be visible to and pursued by brokers. do sex work all through the night and don’t Conversely, some of the independent varia- give proper food. They take all the money. bles occurred prior to the past week, and thus How much can we bear? We get tired if we precede the number of clients. take two or three “parties,” and the madam That said, the interviews also provide evi- makes us do the business all the night. She dence that FSWs working with brokers have makes us sit on the steps at the entrance of more clients. However, one theme that emerges the house at five in the evening and we have is that this greater number of clients diminishes to call the men to come and take them all autonomy for FSWs and enhances control for through the night until five in the early brokers. For example, a 36-year-old FSW with morning. We were taken to [various cities] 20 years of experience who had 14 clients in the on contract. If it is on the highway [working past week said, “When I was on contract I was without a broker], we earn 200 and come made to work for more, without rest, even at back if we feel tired. It is not possible with midnight they used to wake me up and ask me to the owners. They say, “He took 10,000 for work.” A 19-year-old street-based FSW with one keeping you here: you have to work until the year of experience, who had 30 clients in the advance is cleared.” past week and worked for a broker, explained: Although brokers result in a higher number of Respondent: Once we go there, we have to work clients, this appears to come at the expense of whether we like it or not. She [madam] does autonomy and control over work. Still, if this not even allow us to go if we want to go home higher number of clients results in greater total Brady et al. 1137 earnings, FSWs could be compensated for this significantly positive. In this model, with the greater work effort and diminished autonomy. measures of lack of control, having a broker Do brokers result in greater total earnings? now becomes insignificant. Because having a The second model in Table 2 shows that hav- broker is positively associated with a lack of ing a broker is not associated with greater control of the amount charged, this suggests total earnings in the past seven days (t = –.5). that lack of control of the amount charged Total earnings is positively associated with mediates the relationship between having a working in a lodge, literacy, the number of broker and lower payments received. A plausi- FSWs known, being married, having other ble interpretation is that brokers take control contributors or parent(s) in the household, of the amount charged away from FSWs, and always using a condom, and the number of this loss of control results in FSWs receiving times one was forced to have sex. Total earn- a lower payment. ings is negatively associated with working in Indeed, many interviewees characterized the street and age. However, and despite a brokers in a way consistent with this interpre- positive relationship with the number of cli- tation. A recurring narrative was that brokers ents, FSWs do not receive greater total earn- secured a higher amount from clients, but bro- ings when working with brokers. kers deceived FSWs by saying it was for a lower amount, and then giving FSWs a share of the lower amount. This loss of control is Lack of Control over Work partly due to brokers negotiating with clients As discussed earlier, one of the critiques of over price, and this typically occurs outside the the embeddedness literature is its neglect of view of FSWs. For example, the aforemen- the content of ties. As the interviews illus- tioned 56-year-old FSW explained: trate, a key aspect of the content of the FSW– broker relation is control over working Respondent: We will have whatever the client conditions. To understand why brokers result gave us inside. And the woman who offers us business will take the major amount and will in a lower payment received, Table 3 investi- pay us very little. gates whether control over work mediates the relationship between brokers and lower last Interviewer: How much does she take? payments received. The first three models Respondent: She takes 75 out of 100. . . . predict lack of control of the amount charged, She would give us 25 . . . type of sex, and number of clients. Having a Interviewer: Does she call you when the client broker is positively associated with a lack of approaches her? control of both the amount charged and the Respondent: Yes . . . when we asked him to pay number of clients. That is, brokers undermine for the work, he would tell us that he had control over the amount charged and the num- already paid it to the room owner. Then we ber of clients. Having a broker is not associ- would tell him that it was separate. When the ated with a lack of control of the type of sex. client is gone, the woman would say, “Don- gamunda Koduku [son of a bitch], he gave In the final model, we include these meas- only 50 rupees.” ures of lack of control in a model of payment received for the last client. This model would Interviewer: She says so even after taking 100 [from the client]. suggest a mediating relationship if (1) the broker coefficient is smaller and less signifi- Respondent: Yes, after taking 100. Then we cant, and (2) lack of control significantly would think to ourselves, “She is the real Dongamunda.” influences last payment received. Indeed, lacking control of the amount charged is sig- In sum, brokers are associated with having nificantly negatively associated with the last less control over sex work. This loss of con- payment received. Perhaps because FSWs trol manifests as weakened decision-making exchange control of the type of sex for higher power over the amount charged and the num- payments, lack of control of type of sex is ber of clients. The loss of control over the 1138 American Sociological Review 80(6)

Table 3. OLS Models of Lack of Control of Work Conditions and Payment Received for Last Client (N = 1,669): Coefficients and (T-scores)

Lack of Lack of Control of Lack of Control of Payment Amount Control of Number Received for Charged Type of Sex of Clients Last Client

Broker 1.082** .110 .392** –.069 (8.796) (.919) (3.482) (–1.005) Lack of Control of Amount Charged –.067** (–4.807) Lack of Control of Type of Sex .047** (3.318) Lack of Control of Number of Clients –.022 (–1.477) Venue Brothel .266 –.180 .330* –.025 (1.846) (–1.278) (2.507) (–.322) Street –.085 .114 –.262** –.297** (–.790) (1.094) (–2.674) (–5.075) Lodge –.115 .335* .177 .203** (–.823) (2.460) (1.384) (2.665) Highway –.037 –.044 .106 –.268** (–.380) (–.458) (1.183) (–5.034) Human Capital Age –.004 .011* .003 –.022** (–.723) (2.120) (.685) (–7.390) Literacy –.089 .069 .090 .352** (–.919) (.725) (1.013) (6.672) Number of Sex Workers Known –.006* –.003 –.001 .001 (–2.567) (–1.392) (–.332) (1.342) Household Other Contributors in Household .026 .154 –.130 .190** (.268) (1.605) (–1.450) (3.541) Married –.289* .134 .285* .106 (–2.079) (.990) (2.242) (1.391) Separated/Divorced/Widowed –.005 .050 .166 –.200** (–.045) (.476) (1.681) (–3.396) Number of Children .088* .045 –.003 –.033 (2.110) (1.101) (–.076) (–1.434) Parent(s) in Household –.124 .159 .115 .146** (–1.283) (1.684) (1.292) (2.765) Working Conditions Number of Clients in Past Seven –.006 .004 .003 –.006** Days (–1.778) (1.406) (.951) (–3.282) Always Use Condom –.366** –.086 –.134 –.010 (–3.809) (–.919) (–1.520) (–.186) Usually Use Condom .192 .579* .205 .276 (.667) (2.055) (.777) (1.752) Started Sex Work at Young Age –.298** –.029 .016 –.184** (–2.623) (–.260) (.150) (–2.956)

(continued) Brady et al. 1139

Table 3. (continued)

Lack of Lack of Control of Lack of Control of Payment Amount Control of Number Received for Charged Type of Sex of Clients Last Client

Number of Times Experienced .023 .153** .034 –.036 Violence in Past Six Months (.470) (3.256) (.782) (–1.381) Number of Times Forced to Have –.119 .022 .007 .043 Sex in Past Six Months (–1.929) (.369) (.126) (1.265) Number of Threats with Weapon in .052 .155 –.026 –.011 Past Six Months (.437) (1.328) (–.233) (–.166) Wave 2 –.357** –.768** –.199* .235** (–3.307) (–7.282) (–2.015) (3.923) Wave 3 –.342** –.960** –.269** .571** (–3.336) (–9.579) (–2.860) (9.910) R2 .113 .101 .046 .247

Table 4. Models of Potential Services Provided by Brokers in Past Six Months: Coefficients and (Z-scores)

Number Number Number Had Sex Number Number of of Times of Times of Threats Paid Bribe with of Police Times Experienced Forced to with to Police Police Raids Arrested Violence Have Sex Weapon

Broker .685** .749** .305 .216 .113 .527** .266 (3.43) (3.52) (1.88) (.93) (.86) (2.96) (.91) N 1,669 1,666 1,667 1,667 1,669 1,669 1,669 Technique Logit Logit Negative Negative Negative Negative Negative Binomial Binomial Binomial Binomial Binomial

Note: All models control for all variables in Model 2 of Table 1, except the dependent variable in a given model. *p < .05; **p < .01 (two-tailed tests). with the exploitation hypothesis, the surveys Discussion provide no evidence that brokers protect FSWs from police or violence. This study examines the relationship between The final potential service of value is that working with a broker and the payments brokers may attract higher-paying clients received by a sample of FSWs in Andhra (Levitt and Dubner 2009). We stress that our Pradesh, India. Payment received is a central dependent variable is a more accurate meas- economic outcome for FSWs and speaks to ure of the earnings of FSWs because, as noted well-being and working conditions. Brokers in the Methods section, the last payment are fairly common among FSWs and have received is significantly lower than the been shown to play a key role in the informal amount the last client paid (t = –5.76 unlogged, economy. Yet, the literature remains quite t = –9.39 logged). Still, it is worth considering ambivalent about their costs and benefits. if brokers actually result in higher-paying This study utilizes literatures on embedded- clients. Table 5 summarizes the relevant ness and exploitation to investigate this results, including Model 2 of Table 1 for salient social relationship. In the process, we comparison. provide one of the few large sample analyses In contrast to the first model, the second of the earnings of FSWs. We also deepen model shows that brokers result in a signifi- understanding of the quantitative results with cantly higher amount paid by the last client. intensive interview data. Indeed, working with a broker raises this We provide evidence that working with a amount by .4 standard deviations, which is broker is associated with significantly lower substantively larger than most independent payments received. Working with a broker variables (except the control for wave 2). does significantly increase an FSW’s number Thus, brokers do attract higher-paying cli- of clients in the past seven days. However, ents. However, even if working with a broker working with a broker does not lead to greater raises the amount the client pays, the pre- total earnings in the past seven days. We find mium appears to go entirely to the broker, and that working with a broker is associated with the FSW receives less than if she had not a lack of control over working conditions, and worked with a broker. These results support this partly explains why working with a bro- the exploitation hypothesis and demonstrate it ker is associated with a lower last payment is essential to distinguish between the amount received. Specifically, a lack of control over the FSW received as opposed to the amount the amount the client is charged appears to the client paid. mediate the relationship between working Brady et al. 1141

Table 5. OLS Models of Payment Received for Last Client, Amount Last Client Paid, and the Difference (Logged Real 2009 Rupees): Semi-Standardized Coefficients and (T-scores)

Payment Received for Last Client (Model 2, Table 1) Amount Last Client Paid

Broker –.144* .404** (–2.15) (6.134)

R2 .233 .270 N 1,669 1,664

Note: All models control for all variables in Model 2 of Table 1. *p < .05; **p < .01 (two-tailed tests). with a broker and last payments received. work. For instance, past research shows the Finally, we find little evidence that brokers salience of human capital and venue for a provide valuable services in terms of security, variety of aspects of sex work, including safety, or avoidance of police harassment. violence, autonomy, well-being, and health. Therefore, the analyses mostly confirm the Consistent with that research, in a pattern exploitation hypothesis and contradict the that is robust to omitting the broker variable, positive embeddedness hypothesis. Consist- the most important predictors of FSWs’ ent with the positive embeddedness hypothe- earnings are literacy and venue. sis, working with a broker is associated with Within economic sociology, we bring fur- a greater number of clients. Only analyzing ther attention to informal and illegal workers the amount the client pays, as opposed to the in developing countries. Like the embedded- amount the FSW receives, also would imply ness literature, our results demonstrate the that working with a broker benefits FSWs. salience of social relations above and beyond However, most of the evidence shows that purely economic factors. As noted earlier, working with a broker results in an FSW previous economic research on sex work being paid less than the value of her labor, highlights human capital and the prices for and brokers fail to provide valuable services beauty, condom usage, and particular sex that offset this cost. acts. Our results show that working with a While research on sex work once concen- broker has effects comparable to age (a proxy trated on public health or ethics, economic for beauty) and more important than condom and sociological literatures on the work of usage (which was not significant).21 Thus, a sex work have grown in recent years. We social relation with a broker is at least as sali- advance these literatures in at least three ent as these far more studied economic vari- ways. First, we provide one of the few large ables. This affirms that social relations shape sample analyses of the earnings of FSWs. By economic behavior, and it demonstrates the uniquely distinguishing between what the value of situating economic actors in relation last client paid and what share went to the to other actors. Relatively unique to the FSW (e.g., as opposed to a broker), we more embeddedness literature, we scrutinize the accurately assess the earnings FSWs actu- kind and content of social relations (i.e., ally receive. Second, although social rela- exploitation). Greater attention to the kind tions are often implicit in the literature, our and content of social relations moves us study thoroughly investigates how one key beyond the often implicit, but questionable, social relation—working with a broker— assumption that all social ties are equivalent matters to the work of sex work. Third, our (Krippner and Alvarez 2007). use of both quantitative and qualitative data This study also advances theories of deepens understanding of the work of sex exploitation. Exploitation is routinely 1142 American Sociological Review 80(6) invoked by sociologists in a variety of lit- is useful for reaching hidden populations like eratures, but the field has made insufficient FSWs, RDS performs less well than random progress in defining and measuring exploi- sampling (Goel and Salganik 2010). While tation. Some even argue that sociology random sampling is likely not feasible with should abandon the concept of exploitation FSWs, it would be valuable to compare our because it is mainly normative (Goldthorpe results with those from place-based or other 2000). Instead of abandoning the concept, sampling techniques. Because about 30 per- we build on recent efforts to provide a more cent of our sample works at home (and those concrete and measurable definition (Saka- respondents are significantly different), place- moto and Kim 2010; Tilly 1998; Tomasko- based sampling would likely miss an impor- vic-Devey 2014). In particular, we advance tant segment of FSWs. Second, because our the definition of exploitation as a relation in data are cross-sectional, one should be cau- which there is underpayment relative to tious in drawing causal inferences. While value. Of course, our claim that FSWs are some results may reflect selection, this con- underpaid relative to their value depends on cern should partly be alleviated because last the quality of the measure of value and the payment received measures the most recent quality of the earnings model. Still, our client, and several independent variables study empirically demonstrates how to occur earlier in time (e.g., in the past six model exploitation and the questions to months). answer in order to claim exploitation. Thus, Third, our evidence suggests Rajahmundry our study is one step toward more concrete is representative of India’s medium-sized and theories and measures of exploitation. smaller cities, and towns and villages. Com- Because we focus on one group of exploited parisons with other studies also suggest workers (FSWs) and one group of exploiters Rajahmundry likely captures much of the (brokers), our analyses are simpler than when diversity of sex work in India and developing analyzing multiple groups of workers and countries. Of course, there are probably dif- exploiters within the complexity of modern ferences between our sample and FSWs in firms (Sakamoto and Kim 2010). We propose more cosmopolitan settings like Mumbai, that this simpler case clarifies how to apply Tijuana, or New York. However, such cosmo- theories of exploitation to gender inequality politan settings, especially in rich countries, more generally. Gender inequality scholars are likely to be unrepresentative of the broader often invoke exploitation (Folbre 1982; population of FSWs in developing countries. Tomaskovic-Devey 2014), and our study Because most of the world’s FSWs are in illustrates one way to substantiate such argu- developing countries, Rajahmundry provides ments. When women are compensated below a useful contribution to the literature. Never- their value, arguments for exploitation should theless, the sex work literature could advance identify the actors receiving disproportionate with studies comparing across settings in both rewards (e.g., investors, managers, or other rich and developing countries. employees). It would be productive to demon- Fourth, because we find that brokers do strate how women are not receiving services not lead to higher total earnings or provide of value in exchange for being compensated services of value, it remains unclear why below their value, and how women could be FSWs work with brokers. It is important to better off not working in particular settings. acknowledge that there is often an involun- Finally, scholars should clarify the mecha- tary and coercive aspect of this relationship, nisms by which gender shapes the content and so it is not simply that FSWs “choose” to kind of relations that result in women’s exploi- work with brokers. However, for at least tation (Tomaskovic-Devey 2014). some FSWs, brokers could provide some ben- Further research is needed to address the efit or service that is difficult to measure or limitations of our study. First, although RDS detect, or that is simply not captured by the Brady et al. 1143 outcomes we studied. Future research should and Brady 2007). The FSWs in this study are investigate why FSWs work with brokers. In some of the most vulnerable workers in the addition to the outcomes we examined, bro- world, and their vulnerability is compounded kers could provide some nonpecuniary bene- by their gender and the stigmatized and fits, such as minimizing negotiations with legally marginalized nature of sex work. Nev- clients, maintaining some degree of secrecy, ertheless, even though sociology focuses con- signaling to clients, or increasing social status siderable attention on inequalities, populations among FSWs. like these remain relatively understudied We conclude by encouraging further (Brady and Burton forthcoming). Therefore, research on informal-sector workers in devel- greater attention to vulnerable workers at the oping countries. Such workers represent a margins of the global economy would be large share of the world’s workforce (Martin valuable.

Appendix

Table A1. Definitions and Descriptive Statistics for Variables

Mean SD Definition

Dependent Variables Payment Received for Last Client 4.847 1.011 Logged real (2009) rupees received from last client Number of Clients in Past Seven Days 10.152 13.028 R’s report of number in past week Total Earnings in Past Seven Days 6.927 1.186 Product of payment received and number of clients Lack of Control of Amount Charged 1.893 1.708 R decides amount charged clients: 0 = always, 1 = usually, 2 = sometimes, 3 = rarely, 4 = never Lack of Control of Type of Sex 1.694 1.657 R decides type of sex with clients: 0 = always, 1 = usually, 2 = sometimes, 3 = rarely, 4 = never Lack of Control of Number of Clients 1.149 1.509 R decides number of clients seen in a week: 0 = always, 1 = usually, 2 = sometimes, 3 = rarely, 4 = never Key Independent Variable Broker .150 .358 1 = someone else collected money from last client, reference = client paid money directly to respondent Venue Brothel .107 .309 1 = work in brothel, reference = work at home Street .183 .387 1 = work in street, reference = work at home Lodge .093 .290 1 = work in lodge, reference = work at home Highway .246 .431 1 = work on highway, reference = work at home Human Capital Age 31.522 7.892 Years Literacy .240 .427 1 = read or write, reference = neither Number of Sex Workers Known 9.693 18.916 The number of sex workers that respondent personally knows

(continued) 1144 American Sociological Review 80(6)

Table A1. (continued)

Mean SD Definition

Household Other Contributors in Household .237 .426 1 = someone else contributes to household income, reference = not Married .155 .362 1 = married, reference = never married Separated/Divorced/Widowed .651 .477 1 = separated/divorced/widowed, reference = never married Number of Children 1.787 1.007 Number of children of FSW Parent(s) in Household .236 .425 1 = parent(s) resides in household, reference = not Working Conditions Always Used Condom .295 .456 1 = self-report of always used with clients in past seven days, reference = sometimes, rarely, never Usually Used Condom .020 .141 1 = self-report of usually used with clients in past seven days, reference = sometimes, rarely, never Started Sex Work at Young Age .162 .369 Began sex work before age 16, reference = at or after 16 Number of Experiences of Violence in .493 .906 Number of times someone beat (e.g., hit, Past Six Months slapped, pushed, kicked, choked, or burned) R in past six months Number of Times Forced to Have .268 .685 Number of times someone forced R to Sex in Past Six Months have vaginal, anal, or oral sex against R’s will in past six months Number of Threats with Weapon in .084 .359 Number of times someone has threat- Past Six Months ened R with knife, gun, or other weapon or had weapon used against R Wave 2 .241 .428 1 = 2007, reference = 2006 and 2009 Wave 3 .295 .456 1 = 2009, reference = 2006 and 2007 Supplementary Outcomes Paid Bribe to Police .144 .352 1 = paid bribe in past six months, 0 = reference Had Sex With Police .110 .313 1 = ever had sex with police to avoid trouble with them, 0 = reference Number of Police Raids 2.202 12.163 Number of times place where respondent conducts sex work was raided by police in past six months Number of Times Arrested .358 1.383 Number of times arrested in past six months

Note: Ns for these descriptive statistics are the same as in the tables. Table A2. Correlation Matrix for Independent Variables (N = 1,669)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19)

1) Broker 2) Brothel .38 3) Street –.06 –.08 4) Lodge .06 .00 .01 5) Highway –.07 –.08 –.12 –.01 6) Age –.12 –.06 –.01 –.10 –.00 7) Literacy .07 .07 –.08 .01 –.09 –.02 8) Number of Sex Workers Known –.00 .01 –.04 .02 .06 .05 .13 9) Other Contributors in Household .03 .07 –.03 .04 –.09 –.02 .04 –.02 10) Married –.00 –.01 –.12 .02 –.09 .03 .16 .06 .13 11) Sep./Div./Wid. –.02 –.02 .15 –.02 .15 –.02 –.15 –.04 –.23 –.59 12) Number of Children –.11 –.10 .01 –.04 .00 .24 –.09 –.03 .06 –.09 .04 13) Parent(s) in Household .11 .05 –.04 .03 –.04 –.19 .08 –.00 .05 .05 .00 –.11 14) Number of Clients .10 .14 –.01 .09 .17 –.06 –.00 .17 –.02 .01 .03 –.01 .03 15) Always Used Condom .06 .12 –.00 .06 –.04 –.04 .08 .00 .09 .09 –.07 –.06 .09 .16 16) Usually Used Condom .07 .06 .03 .01 .02 –.03 –.04 –.02 .03 .02 –.03 –.03 .05 .04 –.09 17) Started Sex Work at Young Age .00 .04 –.05 –.04 –.03 –.03 .12 .12 .10 .21 –.24 –.13 .06 .05 .06 .01 18) Number of Experiences of Violence .08 .12 –.01 .05 .09 –.11 .01 .05 .01 –.01 –.08 –.01 .01 .11 –.05 .03 .08 19) Number of Times Forced to Have Sex .12 .07 .02 .03 .00 –.09 .05 .06 –.02 –.03 .03 –.04 .06 .10 .01 .05 .03 .21 20) Number of Threats with Weapon .05 .07 –.01 .05 .05 –.09 –.00 .13 .02 –.01 –.01 –.01 .02 .07 –.01 .00 .06 .30 .27 1145 1146 American Sociological Review 80(6)

Acknowledgments 7. Sensitivity analyses confirm this is a conserva- tive strategy. If we included respondents who par- We are grateful to the respondents and colleagues in ticipated in earlier waves, the N would be 2,306. Project Parivartan. We thank the ASR reviewers and edi- Working with a broker would have an even more tors, the WZB-USP writing workshop, Andy Andrews, significant and larger negative coefficient. Agnes Blome, Irene Boeckmann, Matt Bradshaw, Lane 8. For instance, interviewers were instructed to probe Destro, Ryan Finnigan, Joya Misra, Megan Reynolds, about brokers after, “Tell me about the different Jocelyn Viterna, and audiences at the Southern Socio- ways that you get clients.” “Who decides the rate? logical Society and Society for the Advancement of Who negotiates what the rate will be?” “When Socio-Economics meetings for helpful comments. deciding the final cost, do you take into account other people that you have to pay? What are the Funding other costs for the encounter?” “Do you discuss sex work business issues with other sex workers?” This research was supported by The Bill & Melinda 9. Unfortunately, the survey did not ask about mean Gates Foundation (OPP30183; P.I. Blankenship). The payment per client. However, payment from last views expressed here are those of the authors and do not client is plausibly less prone to recall error and has necessarily reflect the official policy or position of the precedent (Lever and Dolnick 2010). Bill & Melinda Gates Foundation. 10. Because eight respondents reported 100 or more clients, we experimented with top-coding this variable at 100 or the 99th percentile of 65 clients. The Notes results were consistent. 1. Wright (1997) refers to this as the “inverse interde- 11. For example, only one respondent reported living with pendent welfare principle”: the material welfare of a madam but did not have someone else collect money the exploiter depends on the labor of the exploited from the last client. Only five respondents reported but not vice versa. experiencing violence from a madam but did not have 2. Although our respondents exhibit greater age someone else collect money from the last client. variation and some were trafficked, Rajahmundry 12. These venues are not mutually exclusive, and about appears similar to Hoang’s (2011:376, 378) “low- 7.8 percent of respondents reported multiple ven- end sector,” which she describes as “single mothers ues. However, including a dummy for multiple ven- . . . who had no more than a grade school education. ues produced similar results. They were poor, urban women or rural migrants . . . 13. We tested for a nonlinear relationship with age, but many entered and continued to do this work as a we found a linear relationship fits these data best. means to escape poverty.” Obviously, many FSWs are under 18 years old. 3. These do not sum to 100 because respondents were However, a prerequisite for participation in the sur- first asked how many clients they had in the past vey was reporting being at least 18. week, and then asked how many were of each kind. 14. The literature on how husbands and family influ- 4. An experienced, older, street-based FSW reported, ence sex work contradicts an economics literature “They [police] took me and another girl. They com- positing sex work and marriage are incompat- plained that I was doing business with the girls. I said ible. For example, Edlund and Korn (2002:182) I did not need girls as I was doing my own business. claim: “a woman cannot be both a prostitute and The police got angry and beat me. . . . They took us a wife.” Our and others’ analyses demonstrate that to the court and we paid 50 rupees fine. . . . Earlier marriage and partners are common among FSWs it was the cases; wherever they saw us they used to (Arunachalam and Shah 2008). Indeed, several book a case and make us pay a fine. Now they come interviewees said their husbands pressured them to the lodge when the customer is there and demand into sex work. more money. They are taking the 50 or 100 and send- 15. We also tested whether the husband was present in ing us away empty-handed. The customer is also the household (including “legal” and “temporary afraid to come again and we are at a loss. The police husbands” and “boyfriend” or “lover”). These mea- have taken the food from our mouth.” sures were not significant and marital status cap- 5. For further information on the survey, interview tures most of the variation. data, and code, the reader should contact the authors. 16. Although these measures are based on questions for 6. Prior to the first wave, an ethnographic team had both “regular” and “occasional” clients, the results been in place in Rajahmundry for well over a year. were consistent if we decomposed them. Unfortu- One objective was to understand the local context nately, the survey did not ask whether a condom of sex work. To capture the diversity of sex work, was used with the last client. It is also not possible we determined that seed selection had to take into to determine whether the payment received from account the solicitation venue, area of residence the last client refers to a regular or occasional client. or work, and the extent of participation in “Nari- Reassuringly, almost all respondents who reported SAKSHAM” (the local sex worker mobilization). always using a condom also reported that they used Brady et al. 1147

a condom the last time they had sex with a regular Perspectives toward Sex as Work.” Sexualities client and an occasional client. 13(2):131–37. 17. One respondent was missing on age but answered Bourgois, Philippe. 2003. In Search of Respect, 2nd ed. affirmatively to a question about whether she New York: Cambridge University Press. entered sex work before puberty. For that respon- Bowen Kathryn J., Bernice Dzuvichu, Rachel Rungsung, dent, we imputed that she was less than 18 at entry. Alexandra E. Devine, Jane Hocking, and Michelle We also experimented with a question on whether Kermode. 2011. “Life Circumstances of Women respondents reported being “lured, cheated or Entering Sex Work in Nagaland, India.” Asia-Pacific forced into the business.” Only about 8 percent of Journal of Public Health 23(6):843–51. respondents answered yes, and it is not significantly Brady, David, and Linda M. Burton, eds. Forthcoming. associated with payments received (r = .01). “Introduction.” In The Oxford Handbook of the Social 18. If a respondent answered affirmatively about vio- Science of Poverty. New York: Oxford University Press. lence, forced sex, or threats, the survey asked who did Brents Barbara G., and Kathryn Hausbeck. 2005. “Vio- this. For violence, almost half identified clients; less lence and Legalized Brothel Prostitution in Nevada.” than 5 percent identified “madam or other broker.” Journal of Interpersonal Violence 20(3):270–95. For forced sex, the overwhelming majority identi- Brents, Barbara G., Crystal A. Jackson, and Kathryn Haus- fied clients and none identified madams/brokers. For beck. 2010. The State of Sex. New York: Routledge. threats, the typical response was clients, and very few Chapkis, Wendy. 1997. Live Sex Acts. New York: Routledge. identified madams/brokers. After clients, the most Chapkis, Wendy. 2000. “Power and Control in the Com- common response for all three was police. mercial Sex Trade.” Pp. 181–201 in Sex for Sale, 19. In other analyses, we experimented with interaction edited by R. Weitzer. New York: Routledge. effects between broker and the other independent Cobbina, Jennifer E., and Sharon S. Oselin. 2011. “It’s variables. For the most part, the interaction effects Not Only for the Money: An Analysis of Adolescent were insignificant. versus Adult Entry into Street Prostitution.” Socio- 20. This might not seem surprising because police are logical Inquiry 81(3):310–32. more likely to raid brothels, and FSWs in brothels Creswell, John W., and Vicki L. Plano Clark. 2007. are more likely to work with brokers (see Table A2 Designing and Conducting Mixed Methods Research. in the Appendix). However, these models control for Thousand Oaks, CA: Sage. venue, so this result is net of working in brothels. Davis, Nanette J. 1993. Prostitution: An International 21. The insignificance of condom usage (and near sig- Handbook on Trends, Problems, and Policies. West- nificant positive sign of usually using condoms) is port, CT: Greenwood Press. notable because some research finds it lowers price Desmond, Matthew. 2012. “Disposable Ties and the (but see Mondal and Gupta 2013). Only about 2 per- Urban Poor.” American Journal of Sociology cent of respondents usually use condoms (see Table 117(5):1295–1335. A1 in the Appendix). Also, our measures are for the Edlund, Lena, and Evelyn Korn. 2002. “A Theory past seven days, and we do not know if the FSW of Prostitution.” Journal of Political Economy used a condom with the last client (see note 16). 110(1):181–214. Fligstein, Neil. 2001. The Architecture of Markets. Princ- eton, NJ: Princeton University Press. References Flippen, Chenoa. 2012. “Laboring Underground: The Arunachalam, Raj, and Manisha Shah. 2008. “Prostitutes and Employment Patterns of Hispanic Immigrant Men in Brides?” American Economic Review 98(2):516–22. Durham, NC.” Social Problems 59(1):21–42. Arunachalam, Raj, and Manisha Shah. 2013. “Compen- Folbre, Nancy. 1982. “Exploitation Comes Home: A sated for Life: Sex Work and Disease Risk.” Journal Critique of the Marxian Theory of Family Labour.” of Human Resources 48(2):345–69. Cambridge Journal of Economics 6(4):317–29. Bernstein, Elizabeth. 2007. Temporarily Yours. Chicago: George, Annie, U. Vindhya, and Sawmya Ray. 2010. University of Chicago Press. “Sex Trafficking and Sex Work: Definitions, Debates Biradavolu, Monica Rao, Scott Burris, Annie George, and Dynamics – A Review of the Literature.” Eco- Asima Jena, and Kim Blankenship. 2009. “Can Sex nomic & Political Weekly 45(17):64–74. Workers Regulate Police? Learning from an HIV Pre- Gertler, Paul, Manisha Shah, and Stefano Bertozzi. vention Project for Sex Workers in Southern India.” 2005. “Risky Business: The Market for Unprotected Social Science & Medicine 68(8):1541–47. Commercial Sex.” Journal of Political Economy Blankenship, Kim M., Brooke S. West, Trace S. Kershaw, 113(3):518–50. and Monica R. Biradavolu. 2008. “Power, Commu- Goel, Sharad, and Matthew J. Salganik. 2010. nity Mobilization, and Condom Use Practices among “Assessing Respondent-Driven Sampling.” Pro- Female Sex Workers in Andhra Pradesh, India.” AIDS ceedings of the National Academy of Sciences 22(Suppl 5):S109–S116. 107(15):6743–47. Boris, Eileen, Stephanie Gilmore, and Rhacel Goldthorpe, John H. 2000. “Rent, Class Con- Parrenas. 2010. “Sexual Labors: Interdisciplinary flict, and Class Structure: A Commentary on 1148 American Sociological Review 80(6)

Sørensen.” American Journal of Sociology 105(6): Countries.” American Sociological Review 72(4): 1572–82. 562–84. Hagan, John, and Bill McCarthy. 1997. Mean Streets. May, Tiggey, Alex Harocopos, and Michael Hough. 2000. New York: Cambridge University Press. For Love or Money: Pimps and the Management of Heyl, Barbara Sherman. 1977. “The Madam as Teacher: Prostitution. London, UK: Research, Development The Training of House Prostitutes.” Social Problems and Statistics Directorate. 24(5):545–55. McCarthy, Bill, and John Hagan. 2001. “When Crime Hoang, Kimberly Kay. 2011. “‘She’s Not a Low-Class Pays: Capital, Competence, and Criminal Success.” Dirty Girl!’ Sex Work in Ho Chi Minh City, Viet- Social Forces 79(3):1035–59. nam.” Journal of Contemporary Ethnography McKeever, Matthew. 1998. “Reproduced Inequality: Par- 40(4):367–96. ticipation and Success in the South African Informal Karandikar, Sharavi, and Moises Prospero. 2010. “From Economy.” Social Forces 76(4):1209–41. Client to Pimp: Male Violence against Female Miller, Jody. 1995. “Gender and Power on the Streets.” Sex Workers.” Journal of Interpersonal Violence Journal of Contemporary Ethnography 23(4): 25(2):257–73. 427–52. Kotiswaran, Prabha. 2008. “Born unto Brothels: Toward Moffatt, Peter J., and Simon A. Peters. 2004. “Pricing a Legal Ethnography of Sex Work in an Indian Personal Services: An Empirical Study of Earnings Red-Light District Area.” Law & Social Inquiry in the U.K. Prostitution Industry.” Scottish Journal of 33(3):579–629. Political Economy 51(5):675–90. Krippner, Greta R. 2001. “The Elusive Market: Embed- Mondal, Swadhin, and Indrani Gupta. 2013. “How Costly dedness and the Paradigm of Economic Sociology.” Is Safe Sex? An Economic Analysis of the Commer- Theory & Society 30(6):775–810. cial Sex Market in India.” Global Health Perspectives Krippner, Greta R., and Anthony S. Alvarez. 2007. 1(1):25–32. “Embeddedness and the Intellectual Projects of Morselli, Carlo, and Isa Savoie-Gargiso. 2014. “Coer- Economic Sociology.” Annual Review of Sociology cion, Control, and Cooperation in a Prostitution 33:219–40. Ring.” Annals of the American Academy of Political Lever, Janet, and Deanne Dolnick. 2010. “Call Girls and Social Science 653(1):247–65. and Street Prostitutes: Selling Sex and Intimacy.” Morselli, Carlo, Pierre Tremblay, and Bill McCarthy. Pp. 187–203 in Sex for Sale, edited by R. Weitzer. 2006. “Mentors and Criminal Achievement.” Crimi- New York: Routledge. nology 44(1):17–43. Levitt, Steven D., and Stephen J. Dubner. 2009. Super Murphy, Alexandra K., and Sudhir Alladi Venkatesh. 2006. Freakonomics. New York: William Morrow. “Vice Careers: The Changing Contours of Sex Work in Lim, Lin L. 1998. The Sex Sector. Geneva: International New York.” Qualitative Sociology 29(2):129–54. Labour Office. Oselin, Sharon S., and Aaron Blasyak. 2013. “Contending Liu, Jeng, Arthur Sakamoto, and Kuo-Hsien Su. 2010. with Violence: Female Prostitutes’ Strategic Responses “Exploitation in Contemporary Capitalism: An on the Streets.” Deviant Behavior 34(4):274–90. Empirical Analysis of the Case of Taiwan.” Socio- Parrado, Emilio A., Chenoa A. Flippen, and Chris logical Focus 43(3):259–81. McQuiston. 2004. “Use of Commercial Sex Workers Logan, Trevon D. 2010. “Personal Characteristics, among Hispanic Migrants in North Carolina: Implica- Sexual Behaviors, and Male Sex Work: A Quanti- tions for the Spread of HIV.” Perspectives on Sexual tative Approach.” American Sociological Review and Reproductive Health 36(4):150–56. 75(5):679–704. Parrenas, Rhacel Salazar. 2011. Illicit Flirtations. Stan- Magnani, Robert, Keith Sabin, Tobi Saidel, and Douglas ford, CA: Stanford University Press. Heckathorn. 2005. “Review of Sampling Hard-to- Plano Clark, Vicki L., John W. Creswell, Denise O’Neil Reach and Hidden Populations for HIV Surveillance.” Green, and Ronald J. Shope. 2008. “Mixing Quan- AIDS 19(supplement):S67–S72. titative and Qualitative Approaches: An Introduction Maher, Lisa. 1997. Sexed Work: Gender, Race, and to Emergent Mixed Methods Research.” Pp. 363–87 Resistance in a Brooklyn Drug Market. Oxford, UK: in Handbook of Emergent Methods, edited by S. N. Oxford University Press. Hesse-Biber and P. Leavy. New York: The Guilford Marcus, Anthony, Amber Horning, Ric Curtis, Jo Sanson, Press. and Efram Thompson. 2014. “Conflict and Agency Portes, Alejandro, and Saskia Sassen-Koob. 1987. “Mak- among Sex Workers and Pimps: A Closer Look at ing It Underground: Comparative Material on the Domestic Minor Sex Trafficking.” Annals of the Informal Sector in Western Market Economies.” American Academy of Political and Social Science American Journal of Sociology 93(1):30–61. 653(1):225–46. Portes, Alejandro, and Julia Sensenbrenner. 1993. Martin, Nathan D., and David Brady. 2007. “Work- “Embeddedness and Immigration: Notes on the ers of the Less Developed World Unite? A Multi- Social Determinants of Economic Action.” American Level Analysis of Unionization in Less Developed Journal of Sociology 98(6):1320–50. Brady et al. 1149

Powell, Walter, and Laurel Smith-Doerr. 1994. “Networks Venkatesh, Sudhir Alladi. 2006. Off the Books. Chicago: and Economic Life.” Pp. 368–403 in Handbook of University of Chicago Press. Economic Sociology, edited by N. Smelser and R. Weitzer, Ronald. 2007. “Prostitution: Facts and Fictions.” Swedberg. New York: Russell Sage. Contexts 6(4):28–33. Rao, Vijayendra, Indrani Gupta, Michael Lokshin, and Weitzer, Ronald. 2009. “Sociology of Sex Work.” Annual Smarajit Jana. 2003. “Sex Workers and the Cost of Review of Sociology 35:213–34. Safe Sex: The Compensating Differential for Con- World Health Organization. 2006. World Health Organi- dom Use among Calcutta Prostitutes.” Journal of zation, 2006 Report on the Global AIDS Epidemic. Development Economics 71(2):585–603. WHO: Geneva. Ridgeway, Cecilia L. 2014. “Why Status Matters for Wright, Erik Olin. 1997. Class Counts. New York: Cam- Inequality.” American Sociological Review 79(1):1–16. bridge University Press. Roemer, John E. 1982. A General Theory of Exploitation Zhang, Sheldon. 2011. “Woman Pullers: Pimping and and Class. Cambridge, MA: Harvard University Press. Sex Trafficking in a Mexican Border City.” Crime Rosen, Eva, and Sudhir Alladi Venkatesh. 2008. “A ‘Per- Law Social Change 56(5):509–528. version’ of Choice: Sex Work Offers Just Enough in Zheng, Tiantian. 2009. Red Lights. Minneapolis: Univer- Chicago’s Urban Ghetto.” Journal of Contemporary sity of Minnesota Press. Ethnography 37(4):417–41. Sahni, Rohini, and V. Kalyan Shankar. 2008. “Markets, David Brady is Professor in the School of Public Policy, Histories and Grass-Roots Evidences: Economics of and Director of the Blum Initiative on Global and Sex Work in India.” Pp. 169–94 in Prostitution and Regional Poverty at the University of California-River- Beyond, edited by R. Sahni, V. K. Shankar, and H. side. He is also affiliated with the WZB Berlin Social Apte. Los Angeles: Sage. Science Center, where he was Director of the Inequality Sahni, Rohini, V. Kalyan Shankar, and Hemant Apte, eds. and Social Policy department from 2012 to 2015. He 2008. Prostitution and Beyond. Los Angeles: Sage. studies poverty/inequality, social policy, politics/political Sakamoto, Arthur, and Changhwan Kim. 2010. “Is Rising economy, work/labor, health/healthcare, and other topics. Earnings Inequality Associated with Increased Exploi- He recently edited The Oxford Handbook of the Social tation? Evidence for U.S. Manufacturing Industries, Science of Poverty and is author of Rich Democracies, 1971–1996.” Sociological Perspectives 53(1):19–43. Poor People: How Politics Explain Poverty (both with Sakamoto, Arthur, and Jeng Liu. 2006. “A Critique of Oxford University Press). Wright’s Analysis of Exploitation.” Research in Social Stratification and Mobility 24(2):209–221. Monica Biradavolu is the CEO and Founder of Qual- Sanders, Teela. 2005. Sex Work. Portland, OR: Willan Analytics and is a Scholar-in-Residence at American Publishing. University. She holds a PhD in sociology from Duke Sanders, Teela, Maggie O’Neill, and Jane Pitcher. 2009. University. She studies the implementation of develop- Prostitution: Sex Work, Policy and Politics. Los ment projects globally with a sociological lens. Her Angeles: Sage. research projects include HIV/AIDS interventions with Sørensen, Aage B. 2000. “Towards a Sounder Basis for sex workers in India, multisector nutrition programming Class Analysis.” American Journal of Sociology in Nepal, and education interventions to improve teacher 105(6):1523–58. attendance in Haiti. She advocates for mixed-methods Sullivan, Barbara. 2003. “Trafficking in Women.” Inter- evaluations, integrating rigorous qualitative methods national Feminist Journal of Politics 5(1):67–91. with quantitative surveys. Her publications have appeared Tilly, Charles. 1998. Durable Inequality. Berkeley: Uni- in Social Science and Medicine, Journal of Epidemiology versity of California Press. and Community Health, Global Health, AIDS, AIDS Tomaskovic-Devey, Donald. 2014. “The Relational Gen- Care, and Public Health Reports. eration of Workplace Inequalities.” Social Currents 1(1):51–73. Kim M. Blankenship is Professor and Chair in the Uggen, Christopher, and Melissa Thompson. 2003. “The Department of Sociology at American University (AU), Socioeconomic Determinants of Ill-Gotten Gains: Director of the Center on Health, Risk and Society at AU, Within-Person Changes in Drug Use and Illegal Earn- and Director of the Social and Behavioral Sciences Core of ings.” American Journal of Sociology 109(1):146–85. the DC Center for AIDS Research. Her research and pub- Uzzi, Brian. 1996. “The Sources and Consequences of lications focus on the social dimensions of health and Embeddedness for the Economic Performance of structural interventions (especially community mobiliza- Organizations: The Network Effect.” American Soci- tion interventions) to address health, with an emphasis on ological Review 61(4):674–98. HIV/AIDS. She also conducts research on the association Vanwesenbeeck, Ine. 2001. “Another Decade of Social between mass incarceration and race and gender dispari- Scientific Work on Sex Work: A Review of Research ties in HIV/AIDS. She has received funding from the 1990–2000.” Annual Review of Sex Research NIDA, the CDC, and the Bill & Melinda Gates 12(1):242–89. Foundation.