Audit, correspondence and experimental studies provide abundant evidence of discriminatory behaviours by firms. But some economic models suggest that competition and/or providing the subjects of discrimination with sufficient choice will prevent those subjected to such discriminatory behaviours from being affected adversely. Our study of the pricing behaviour of street sex workers in the Geylang district of Singapore reveals that having many actors on both sides of the market does not eliminate discrimination. Instead, sex workers indulge in both statistical price discrimination based on the client’s ethnicity, a proxy for wealth, and taste discrimination based on the client’s skin tone. Our results establish that competition alone does not eliminate discrimination.
Competition (apparently) eliminates discrimination
In the canonical economic model of discrimination, Becker (1957) shows that competition eliminates discrimination. If existing unprejudiced firms can grow sufficiently or new ones enter the market freely, then eventually the greater profitability of non-discriminating firms will allow them to occupy a sufficiently large portion of the market that the law of one price will prevail. In this context, the best way to eliminate discrimination is to foster competition by ensuring that there are no barriers that prevent the growth or entry of non-discriminating firms.
But what constitutes sufficient competition? Diamond (1971) shows that even if there are large numbers of sellers and buyers, if buyers search sequentially, even a very small search cost creates an equilibrium in which firms set the monopoly price. In this setting, consumers with more inelastic demand may face higher prices. And Black (1995) suggests that competition may be inadequate to drive out discriminating firms. Policy intervention may be required to combat discrimination.
Sex workers in Singapore
Our research (Li et al. 2014) does not approach policy directly, but uses the market for street sex workers in Geyland, Singapore to test whether small transactions costs can counter the competitive effects of large numbers of actors on both sides of the market. We find that sex workers do price discriminate based on client ethnicity, in some cases because of their beliefs about the demand curve and in others due to prejudice.
Audit (e.g. Ayres and Siegelman 1995) and correspondence studies (e.g. Bertrand and Mullainathan 2004) establish that some firms engage in discriminatory behaviours, but not whether this results in a discriminatory equilibrium. Market-based studies have focused on unusual markets (e.g. List’s 2004 study of baseball cards) or markets with relatively little interaction between buyers and sellers (e.g. Bayer et al. 2012 on housing sales, Doleac and Stein 2013, and Pope and Snydor 2011 on internet sales). We examine a large market with significant interaction between buyers and sellers.
Competitive market setting and discrimination
Geylang is a highly competitive market in the sense that there are large numbers of sellers and buyers in a small neighbourhood (we estimate an average of one sex worker per 40 feet of sidewalk) who can freely negotiate price. Clients report that sex workers are highly substitutable and provide homogeneous sexual services. At the same time, search costs are negligible. Clients face no legal risk since purchasing sex is legal. There is also little reputational risk – since Geylang is also a residential, commercial and popular tourist area, there are many reasons to be there. And given the density of sex workers, the time cost of searching is also negligible. Still, search is sequential. Sex workers do not openly bid against each other, and offers ‘explode’ if bargaining ends. Thus Geylang is virtually ideal for asking whether having large numbers of buyers and sellers is sufficient to eliminate discrimination or whether small search costs can, in the presence of sequential search, support continued price discrimination.
Survey and data
Before undertaking our survey of sex workers, we collected extensive quantitative information from over 100 interviews with sex workers, pimps and clients. This revealed that sex workers believe that whites are willing to pay more than Chinese clients who, in turn are willing to pay more than Bangladeshis. When discussing Indians, sex workers expressed dislike, often complaining that they are dark and/or that they smell. Subsequent information suggests that they are similar to Bangladeshis in willingness to pay.
We surveyed 176 sex workers, obtaining recall data about 814 transactions. Our sample approximates the sex worker population well, with sex workers from China, Thailand, Vietnam, and Indonesia and Singaporean Indians. Note that client ethnicities are reported by the sex workers. She may treat an Indian client as a Bangladeshi. Her behaviour towards the client will reflect her belief.
A given sex worker’s (i.e. using sex worker ‘fixed effects’) initial price offer to white clients is about 11% higher than she asks of Chinese clients (the base group), while Bangladeshis are offered around a 13% discount. Thus whites are charged about 25% more than Bangladeshis. In contrast, even though their willingness to pay is similar, Indians are offered a higher price than are Bangladeshis. The differentials in transaction prices are similar, and are robust to controlling for transaction characteristics such as the type of sexual service, transaction venue, transaction time and service duration as well as client characteristics such as age, dress and attractiveness.
Compensating differentials or discrimination?
If sex workers charge, for example, whites more because they believe whites have a higher willingness to pay, they should not raise the price to the point that whites and other ethnicities have the same probability of rejecting the offer – the cost of losing the client is higher when the potential price is high. And because the client is more profitable, she should be more likely to approach him, hoping to make a sale. Thus we predict that sex workers will be more (less) likely to approach and bargain successfully (unsuccessfully) with whites (Bangladeshis) than they are Chinese clients. In contrast, since sex workers dislike Indians, they should be less likely to approach them and to successfully conclude a deal.
What happens in a market-clearing model? If whites pay more because they want longer or more difficult service, in general, sex workers should be neither more nor less likely to approach them although individuals who find the demands relatively costly will not approach, leaving them to those with relatively low cost. And the market-clearing model makes no prediction about bargaining failing. Indeed bargaining should never fail. Sex workers who demand more than the going price simply do not engage.
In fact, sex workers are 21% more likely to approach whites and 8% less likely to approach Bangladeshis (relative to Chinese). Their bargaining with whites almost never fails, while bargaining with Bangladeshi potential clients is substantially more likely to fail. These results are consistent with statistical discrimination based on anticipated willingness to pay.
At the same time, we find robust evidence that sex workers are less likely to approach and to reach an agreement with Indians. Since the initial prices demanded of Indians and the transaction prices were similar to those demanded of the Chinese, but Indians willingness to pay resembles that of Bangladeshis, this points to taste discrimination.
The differential treatment of Bangladeshis and Indians may be surprising given their apparent similarity. Feedback from sex workers and pimps suggests that sex workers use skin tone and, to a lesser extent, accent to distinguish between these two groups and uniformly describe men with lighter skin tone as Bangladeshi.
Implications for policy
We do not care whether certain ethnicities pay more for or find it more difficult to obtain services from sex workers. We expect that most Vox readers will feel similarly. Nevertheless, we believe that our research has important implications for policy. We have established that having many actors on both sides of the market is not sufficient to eliminate discrimination. Doing so requires policy intervention. Our results are likely to extend to other markets in which search is sequential or nearly sequential. Whenever offers explode as soon as the customer leaves the store, our results should apply. Historically, in the US, this included automobiles and mattresses.
It is more difficult to know whether our results apply to more typical labour markets where search is often non-sequential since employed workers can search on-the-job. We note, however, that worker heterogeneity is likely to be more important and search more costly in long-term employment relations than for a brief sexual encounter. Thus our results point to the conclusion that the free market cannot by itself eliminate discrimination and that there is a role for active policy.
Ayres, I, and P Siegelman (1995), “Race and Gender Discrimination in Bargaining for a New Car”, The American Economic Review 85(3): 304–321.
Bayer, P, M D Casey, F Ferreira, and R McMillan (2012), “Price Discrimination in the Housing Market”, National Bureau of Economic Research.
Becker, G S (1957), The Economics of Discrimination, University of Chicago Press, 2nd edition.
Bertrand, M, and S Mullainathan (2004), “Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labour Market Discrimination”, The American Economic Review, 94(4): 991–1013.
Black, D A (1995), “Discrimination in an Equilibrium Search Model,” Journal of Labour Economics 13(2): 309–334.
Diamond, P A (1971), “A Model of Price Adjustment”, Journal of Economic Theory 3: 156–168.
Doleac, J L, and L C Stein (2013), “The Visible Hand: Race and Online Market Outcomes”, Economic Journal 123(572): F469-F492.
Li, H, K Lang and K Leong (2015), “Does Competition Eliminate Discrimination? Evidence from the Commercial Sex Market in Singapore”, NBER Working Paper No. 20911.
List, J A (2004), “The Nature and Extent of Discrimination in the Marketplace: Evidence from the Field”, The Quarterly Journal of Economics 119: 49–89.
Pope, D G, and J R Sydnor (2011): “What’s in a Picture? Evidence of Discrimination from Prosper.com”, Journal of Human Resources 46(1): 53.92.