2026., Sex Work by Numbers: Bridging Economics and Social Sciences to Understand the Hidden World of Commercial Sex, Stef Adriaenssens
Download citation file:
Overview of Studies Documenting Compensating Wage Differentials in Chronological Order.
| Reference | Data | Identification and Results | |
|---|---|---|---|
| 1 | Rao et al. (2003) | Survey from female SW in Kolkata (India). | IV and ML analysis. Risk penalty ≈ 66% (ML) to 79% (IV) |
| 2 | Gertler et al. (2005) | Survey from female SW in Mexico with data for three transactions. | SW-level FE. Risk premium ≈ 23% (FE) |
| 3 | Levitt and Venkatesh (2007) | Survey from female SW in Chicago. | SW-level FE. Risk premium for unsafe vaginal sex ≈ 7.5% |
| 4 | Willman (2008, 2010) | Survey from female SW in Managua (Nicaragua), n = 138, directly asking SW the premium. Reported CWD per segment: top, middle and bottom segments. | Descriptives: 39% (vaginal), 44% (oral) and 118% (with client to unknown location). Vaginal sex premium: 38% (top segment), 55% (middle) and 18% (bottom) |
| 5 | de la Torre et al. (2010) | Survey from female SW in Ciudad Juarez (Mexico), n = 429. Direct question of prices with and without condom. | Paired prices per SW. Risk premium unsafe sex: 31% |
| 6 | Robinson and Yeh (2011) | Diaries from female SW in Busia Town, Western Province, Kenya (n = 248). | FE at the level of SW. Vaginal sex risk premium ≈ 9.3% |
| 7 | Adriaenssens and Hendrickx (2012) | Online data on transactions (n = 24,998) by female SW (n = 7,451) in The Netherlands and Belgium. | FE at the level of SW. Risk premium ≈ 6.5% |
| 8 | Arunachalam and Shah (2012) | Survey from female SW in Ecuador and Mexico (n = 8,382) with data for three transactions. | FE of SW. Risk premium ≈ 7% for unattractive SW, 40% for attractive SW |
| 9 | Chang and Weng (2012) | Survey from female SW in Taipei (Taiwan), n = 140. | OLS regression Risk premium ≈ 5.5% |
| 10 | Islam and Smyth (2012) | Survey from female SW in Bangladesh (n = 240). | IV. Premium between 28 and 113%. |
| 11 | Arunachalam and Shah (2013) | Survey from female SW in Ecuador (n = 2,833) with data for three transactions. | FE of SW. Risk premium ≈ 11% |
| 12 | Shah (2013) | Survey from MSM in Ecuador (n = 1,589) with data for several transactions (n = 8,100). | FE. Risk premium ≈ 16%. |
| 13 | Logan (2013, 2017) | Male escort advertisement data in the United States. | FE of SW and clients. Risk penalty ≈ 15% |
| 14 | Elmes et al. (2014) | Survey from female SW in Zimbabwe (n = 311). | OLS regressions. Risk premium ≈ 43% |
| 15 | Galárraga et al. (2014) | Survey from MSM in Mexico (n = 253). | OLS regression. Risk premium ≈ 41% |
| 16 | Egger and Lindenblatt (2015) | Internet offers (n = 16,583) by female SW (n = 2,517). | IV estimation. Final model risk premium ≈ 191%a |
| 17 | Muravyev and Talavera (2018) | Online data from transactions (n = 3,877) by female SW (n = 1,392) in London. | SW and client FE. Risk premium (oral sex) ≈15% |
| 18 | Quaife et al. (2018) | DCE from female SW (n = 122) in South Africa. | Discrete choice experiment Risk premium ≈ 395% |
| 19 | George et al. (2019) | Survey from female SW in around Bloemfontein (South Africa), n = 36. | Self-reported average prices Risk premium ranges from 22% to 100% depending on gender and sex actsb |
| 20 | Quaife et al. (2019) | Survey from female SW (n = 3,591) in India. | IV. Risk penalty ≈ 65% |
| 21 | Njuguna et al. (2025) | Transactions (n = 2,375) of women in SW (n = 755) and transactions (n = 2,420) from women in transactional sex (n = 753) in Yaoundé, (Cameroon). | OLS regression. SW risk premium ≈ 30% Transactional sex risk penalty ≈ 14% |
| Reference | Data | Identification and Results | |
|---|---|---|---|
| 1 | Survey from female SW in Kolkata (India). | IV and ML analysis. Risk penalty ≈ 66% (ML) to 79% (IV) | |
| 2 | Survey from female SW in Mexico with data for three transactions. | SW-level FE. Risk premium ≈ 23% (FE) | |
| 3 | Survey from female SW in Chicago. | SW-level FE. Risk premium for unsafe vaginal sex ≈ 7.5% | |
| 4 | Survey from female SW in Managua (Nicaragua), Reported CWD per segment: top, middle and bottom segments. | Descriptives: 39% (vaginal), 44% (oral) and 118% (with client to unknown location). Vaginal sex premium: 38% (top segment), 55% (middle) and 18% (bottom) | |
| 5 | Survey from female SW in Ciudad Juarez (Mexico), | Paired prices per SW. Risk premium unsafe sex: 31% | |
| 6 | Diaries from female SW in Busia Town, Western Province, Kenya ( | FE at the level of SW. Vaginal sex risk premium ≈ 9.3% | |
| 7 | Online data on transactions ( | FE at the level of SW. Risk premium ≈ 6.5% | |
| 8 | Survey from female SW in Ecuador and Mexico ( | FE of SW. Risk premium ≈ 7% for unattractive SW, 40% for attractive SW | |
| 9 | Survey from female SW in Taipei (Taiwan), | OLS regression Risk premium ≈ 5.5% | |
| 10 | Survey from female SW in Bangladesh ( | IV. Premium between 28 and 113%. | |
| 11 | Survey from female SW in Ecuador ( | FE of SW. Risk premium ≈ 11% | |
| 12 | Survey from MSM in Ecuador ( | FE. Risk premium ≈ 16%. | |
| 13 | Male escort advertisement data in the United States. | FE of SW and clients. Risk penalty ≈ 15% | |
| 14 | Survey from female SW in Zimbabwe ( | OLS regressions. Risk premium ≈ 43% | |
| 15 | Survey from MSM in Mexico ( | OLS regression. Risk premium ≈ 41% | |
| 16 | Internet offers ( | IV estimation. Final model risk premium ≈ 191% | |
| 17 | Online data from transactions ( | SW and client FE. Risk premium (oral sex) ≈15% | |
| 18 | DCE from female SW ( | Discrete choice experiment Risk premium ≈ 395% | |
| 19 | Survey from female SW in around Bloemfontein (South Africa), | Self-reported average prices Risk premium ranges from 22% to 100% depending on gender and sex acts | |
| 20 | Survey from female SW ( | IV. Risk penalty ≈ 65% | |
| 21 | Transactions ( | OLS regression. SW risk premium ≈ 30% Transactional sex risk penalty ≈ 14% |
SW: sew workers, RE: random effects, FE: fixed effects, IV: instrumental variables, ML: Maximum Likelihood, CWD: compensating wage differential.
aThe article reports a premium of 91%, but seems to have made an error in counting back from the logarithmic transformation: e1.0673–1 = 191%.
bNotwithstanding the small sample size, no inference test was reported on the premiums.
