Shah 12 4-13

The Economics of Sex Work: A Developing Country Perspective

Manisha ShahDepartment of Public Policy

UCLA

Motivation

Why should we as economists/social scientists care about sex market?1. Integral role in spread of disease including HIV/AIDS

Unprotected Commercial Sex is a Major HIV Transmission Vector

• Each day 20,000 people become infected with HIV (UNAIDS, 2002)

• More new cases in developing countries– Condoms are effective defense against infection– Large amounts spent on education of SWs– Still many SWs risk infection by not using condoms

• SW HIV infection rates are high, esp in countries with epidemic

HIV Prevalence (Per Hundred) 1999

Country Adult Pregnant Women Sex Workers Benin 1.2 0.4 53.3 Burkina Faso 6.7 12.0 60.4 Cameroon 3.0 1.9 21.2 Congo D.R. 3.7 4.6 30.3 Congo, Rep 7.2 7.1 49.2 Ivory Coast 6.8 11.6 67.6 Ethiopia 2.5 4.9 67.5 Gambia 2.1 1.7 34.7 Ghana 2.3 2.2 30.8 Kenya 8.3 13.7 85.5 Malawi 13.6 32.8 78.0 Mali 1.3 3.5 55.5 Nigeria 2.2 3.8 22.5 Rwanda 7.2 25.3 87.9 Uganda 14.5 21.2 86.0 Zimbabwe 17.4 35.2 86.0 Guyana 1.3 6.9 25.0 Haiti 4.4 8.4 41.9 Honduras 1.6 1.0 20.5 Jamaica 0.9 0.7 24.6 Mexico 0.4 0.0 0.1 Cambodia 1.9 3.2 43.0 India 0.4 0.3 51.0 Myanmar 1.5 1.3 18.2 Thailand 2.1 2.4 18.8

Motivation

Why should we as economists/social scientists care about sex market?1. Integral role in spread of disease including HIV/AIDS 2. Source of employment for many women in poor

countries (micro/macro implications)

COUNTRY Location Area % FSW Year

Africa Benin Cotonou Capital 1.20% 2001 Burkina Faso Ouagadougou Capital 4.30% 2000–03 Cameroon Yaoundé Capital 2.20% 1997 Ivory Coast Abidjan Capital 0.70% 2000 Niger Niamey Capital 2.60% 2004 Ethiopia Addis Ababa Capital 2.10% 2002 Kenya Kisumu Provincial town 3.00% 1997

Busia, Mumias Provincial town 6.90% 1999 Madagascar Diego-Suarez Provincial town 12.00% 2001

Asia India Mumbai Capital of State 0.50% 2001 Nepal Kathmandu District 1.00% 2001 Indonesia Jakarta Province 1.40% 2002 Cambodia Phnom Penh Province 2.80% 2003

Latin America Dom Republic – – 1.80% 2001 Belize – – 7.40% 2001 Haiti – – 2.00% 2001 Bolivia – – 0.20% 2001 Colombia – – 0.70% 2001 Peru – – 0.30% 2001 Venezuela – – 1.50% 2001

Source: J Vandepitte, R Lyerla, G Dallabetta, F Crabbé, M Alary, A Buvé. (2006) "Estimates of the number of female sex workers in different regions of the world," BMJ.

Labor market issues

• Huge source of employment for women in developing countries, and growing (see BMJ table)

• Financial turnover of sex sector is quite large – Indonesian financial turnover of sex sector was estimated at

between U.S 1.2 and 3.3 billion, or between 0.8 and 2.4% of the country's GDP (Lim, 1998).

– Thailand, close to US 300 million was transferred annually from urban SWs to rural areas in the form of remittances (Lim, 1998).

Motivation

Why should we as economists/social scientists care about sex market?1. Integral role in spread of disease including HIV/AIDS 2. Source of employment for many women in poor

countries (micro/macro implications)3. Failure of policy prescriptions

Today we will….

• Discuss three questions from an economists perspective:1. Why do sex workers engage in non-condom use?

(Gertler, Shah, Bertozzi, JPE 2005; Rao et. al JDE 2003)

Today we will….

• Discuss three questions from an economists perspective:1. Why do sex workers engage in non-condom use?

(Gertler, Shah, Bertozzi, JPE 2005; Rao et. al JDE 2003)

2. Why do women enter the sex market?(Robinson and Yeh, 2011; Edlund and Korn, JPE 2002; Arunachalam and Shah, AER 2008)

Today we will….

• Discuss three questions from an economists perspective:1. Why do sex workers engage in non-condom use?

(Gertler, Shah, Bertozzi, JPE 2005; Rao et. al JDE 2003)

2. Why do women enter the sex market?(Robinson and Yeh, 2011; Edlund and Korn, JPE 2002; Arunachalam and Shah, AER 2008)

3. How can public policy/laws/regulations related to sex market impact the spread of disease? (Gertler and Shah JLE2011, Shah and Cunningham 2013)

• Use economic methods to investigate these questions

Question:

• Do you think a sex worker should get more or less money from a client when she does not use a condom? – Why or why not?

1. Why do SWs engage in non-condom use?

Conventional Wisdom: Sex Workers do not use Condoms because …

• Sex workers uninformed of risks– Would protect themselves if understood risks

• Condoms not available or in short supply, especially when needed

• Forced– Physical-economic threats– Psychological & social norms

Alternatively: SWs may be willing to risk infection if compensated

• Could be rational response to client demand– Clients value unprotected sex & are willing to pay for it– SWs take risk if adequately compensated

• Happens in other sectors– Compensating wage differentials for risky work

• Ex: police, firemen

Public Agencies Focus Interventions on Supply Side

• Supply side interventions– Educating SWs about risks and how to protect themselves– Creating safe and supportive work environment – social capital– Creating accessible supply of condoms

• However, supply-side alone will not stop unprotected sex – If clients are willing to pay, SWs will take risk if compensated

• Alternatives– Educate clients & lower demand for unprotected sex as well

We Investigate Whether SWs are “Rationally” Responding to Incentives

• Are Sex Workers charging more to take the risk of providing unprotected services?

Data Source

• Summer of 2001, wrote, piloted and attached economic questionnaire to UNAIDS “Second Generation” study in Mexico

• 2nd generation study tried to map universe of sex workers in cities in 2 states– Used this as a sampling frame– How good was it?

• Sample of about 1034 sex workers• Information on details of last 3-4 transactions for 3,884

observations

Transaction Specific Information

• Price paid by client & received by sex worker• Services: vaginal, oral, anal, talk, dance, strip, massage

• Condom use & who suggested• Non condom use & who suggested• CSW report of client characteristics: appearance,

wealth, education, personality, hygiene• Alcohol & Drug use during transaction• Client abused/hit sex worker

Table 2. Sex Worker Characteristics (N=1034)

Characteristics Mean St. Dev Age 27.82 7.77 Age of first sexual experience 15.65 2.36 Years in sex work 6.04 6.83 Have had STIs/vaginal problems (=1) 0.17 Sex Worker is Very Attractive (=1) 0.21 Have Children (=1) 0.62 Education

Ever gone to school (=1) 0.84 Some secondary school or more (=1) 0.36

Civil Status Single (=1) 0.41 Married or in Partnership (=1) 0.22 Divorced or Widowed (=1) 0.38

Primary Work Site Bar/Club (=1) 0.82 Street (=1) 0.12 Other (=1) 0.06

Table 3. Client Characteristics Reported By Sex Worker (N=3837) Mean Std. Dev.

Regular Client (=1) 0.64 Age 36.04 11.01 Nice or Pleasant Personality (=1) 0.66 Wealth

Poor (=1) 0.17 Average Wealth (=1) 0.70 Above Average Wealth (=1) 0.08 Very Wealthy (=1) 0.05

Attractiveness Handsome (=1) 0.10 Average (=1) 0.66 Ugly (=1) 0.24

Cleanliness Dirty (=1) 0.10 Clean (=1) 0.73 Very Clean (=1) 0.17

Conducted focus groups with SWs & Clients to Describe Market

• Clients– May not know prices or quality– Clients approach SW based on physical characteristics– Obtain information about prices & services– Clients value SW physical & personality characteristics

(e.g. beauty); pay more for these– Client heterogeneity in tastes– High search costs (time)

Sex Workers Negotiate Prices

• High search costs & client heterogeneity able to charge different prices to different clients

• Collects info based on appearance & conversation to determine willingness to pay– Clothes, car, rings, cleanliness,…– Job, married, hotel, etc…– How much client likes SW …– Regular client gets charged more

Negotiation up front & renegotiate as client preferences revealed

• Heterogeneity in timing of negotiation• Some SWs (or agents) try to negotiate everything

up front – prices, services & condom use• Terms almost always renegotiated in room

because clients ask for more or different services • Condom use negotiated by SW and client

– Heterogeneity in client & SW preferences for condoms

A Bargaining Model

• Two agents– A client who we will call “Richard”– A sex worker called “Julia”

• Negotiate over Price & Condom Use– Payoff functions– Recursive solution

• Condom use• Prices

Our Approach is Estimate a Transaction Model

• Data: survey of 1050 SWs in Mexico– Collected information on last 3-4 transactions– Price, services, condom use & client characteristics

• Have SW panel where i indexes Sex Worker and t indexes the transaction

• Estimate SW Fixed Effects models to control for selection on SW characteristics

• Control for client characteristics with SW reports of client looks, wealth, cleanliness, risk preferences

Table 5. Basic Log Price Fixed Effects Regressions

Whole Sample Exclude

SWs Who Never Use Condoms

Exclude SWs Who Always

Use Condoms

Exclude Both Always

& Never Condom

Users

Random Effects

Fixed Effects

Fixed Effects Fixed Effects Fixed

Effects No Condom Used

0.093 (3.91)***

0.131 (5.49)***

0.132 (5.52)***

0.133 (4.19)***

0.135 (4.19)***

Hausman Test 496.51***

F Stat SW FEs 27.86*** 27.72*** 16.09** 15.36**

# of Obs 3,837 3,837 3,753 1,309 1,225

# of SWs 1,029 1,029 1,007 363 341

Policy Implications (1)

• Strong evidence that – SWs are willing to take the risk of providing

unprotected sex for a higher price• Suggests why just educating sex workers has not

stopped HIV transmission thru unprotected sex• Need to educate clients or provide financial

incentives for condom use to offset client WTP

2. Why might women enter the sex market?

• Economic shocks/Poverty (Robinson and Yeh,2011)

• Sex work pays well(Edlund and Korn Marriage market hypothesis)

• Lack of outside option• Force, kidnapping, trafficking (not discussed too

much in economics lit as we tend to assume free choice)

Sex Work as a Response to Risk in Western Kenya(Robinson and Yeh, 2011)

• Collect daily self-reported data on sexual behavior, income shocks, expenditures, and labor supply for sample of 237 women Western Kenya.

• Find significant day-to-day fluctuations in sex worker decisions

• Women engage in sex-for-money transactions in part to deal with unexpected non-labor income shocks.

Pays Well

Sex work puzzle: Female dominated, low skilled, low education—yet it pays really well.

Table 1: Summary statisticsEcuador Ecuador Ecuador Ecuador Ecuador Mexico Mexico

(1) (2) (3) (4) (5) (6) (7)Female Female Domestic Male Male Female FemaleSWs NSWs Worker NSWs SWs NSWs SWs NSWs

Last week’s earnings 113.5 50.7 37.6 80.4 67.1 3886 2117(154.6) (66.3) (44.6) (134.1) (123.2) (9785) (4101)

Age 27.9 36.2 37.1 24.0 36.7 27.7 33.3(8.01) (12.2) (12.8) (6.92) (12.8) (7.6) (11.3)

None/some primary(%) 4.1 2.5 4.1 2.3 1.9 16.2 11.1Completed primary(%) 41.3 23.8 35.8 25.6 32.5 46.8 20.0Secondary(%) 50.4 40.4 45.2 63.2 42.9 28.5 40.4High School(%) 2.2 1.5 0.7 3.8 1.1 6.4 11.5University + (%) 1.2 31.6 14.0 4.5 21.4 1.7 17.0Observations 2782 1872 1020 574 3319 1038 2454

Earnings from Ecuador are in US dollars and Mexican earnings are in pesos. Standard deviations are given in parenthesis.

11

Edlund-Korn (2002)Marriage Market Hypothesis

• First formal model of prostitution in economics• Draws intriguing link between labour and marriage

market that holds for one profession: prostitution• Central assumption of model is that sex workers

cannot marry--in choosing SW, women relinquish compensation otherwise received in marriage.

• Compensating differential due to foregone opportunity to “sell” their fertility in marriage market.

Empirical test of the model

• Arunachalam and Shah, 2008 American Economic Review P&P test the model

• Major findings:– Sizable earnings premium for sex work (around 33%)– Fail to find support for Edlund-Korn explanation– Sex workers are actually more likely to be married than

non-sex workers at younger ages—when the earnings premium for sex work is highest.

050

100

150

200

Wee

kly

earn

ings

10 20 30 40 50 60 70Age

Fitted sex workers 95% CIFitted non−sex workers Sex workersNon−sex workers

Figure 1: Weekly earnings of female workers in Ecuador (U.S. dollars)

0.2

.4.6

.81

Mar

riage

rat

e by

age

10 20 30 40 50 60 70Age

Fitted sex workers 95% CIFitted non−sex workers Sex workersNon−sex workers

Figure 2: Marriage rates of female workers in Ecuador

An alternative hypothesis?

• Data seem to contradict prima facie case for marriage market explanation for high returns to prostitution

• Natural competing explanation is compensating differential due to risk

• Ecuador female sex worker data includes disease results– Calculate DALYs lost due to observed increase in disease burden

from STIs– Implies a compensating differential of at least 8% of sample

average earnings for sex work• Sex work, like policework or other risky professions,

draws hazard pay.

Policy Implications (2)

• Programs to get women out of sex industry will fail if alternatives don’t pay as well (most likely won’t)

• How might we improve women’s outside option in the labor market?

• Access to credit, savings, health insurance (address these market failures) may reduce risky sex and increased sex work labor supply

New Indonesia Project: Promoting Public Health Through Savings for Sex Workers in Indonesia

• Provide mobile banking savings accounts to sex workers in Indonesia

• Randomize into 3 groups: 1. Control (business as usual) 2. T1: Offer savings account3. T2: Offer savings account + financial incentive

• Follow sex workers for year, collecting daily data to test hypotheses like:1. Do formal savings accounts increase savings for FSWs?2. Do formal savings accounts improve strategies for coping with negative income shocks?3. Do formal savings accounts decrease risky behavior among FSWs during commercial sex

transactions?

3. How can public policy/laws/regulations impact the spread of disease?

• Ecuador project: Collected data on 2000 SWs in 8 cities (plus biologicals) and collected data from police about # of enforcement visits of carnet laws

• Increased enforcement in street decreases STI prevalence but increases in brothel sector– Why?

• Marginal woman on street moves to brothel sector (less risky, less disease). Street prices increase, clients decrease

• Marginal woman from brothel moves to street

Rhode Island Study

• Indoor prostitution decriminalized “accidentally” from 2003-2009 in RI

• Indoor sex sector grows—supply increases• Gonorrhea incidence decreases