The Impact of Short-term Financial Incentives on Sexual Behavior and HIV Incidence in Lesotho

Lesotho has the third-highest HIV prevalence in the world at 23.2%. Many HIV prevention initiatives are currently going on in the country. The proposed impact evaluation responds to the urgent need to find innovative approaches to promote safer sexual behaviors among youths in Lesotho. In particular, the project will be implemented through the New Start Voluntary Counseling and Testing (VCT) sites established by PSI in the country.

PSI, the implementing agency, has 5 static New Start VCT sites nationally in 5 out of the 10 districts in Lesotho. They are located in the main camp town of each district. Each of these sites has vehicles and tents that are used for outreach services. Outreach activities aim at targeting more specific groups of the population who are generally at higher risk of HIV infection. For example, rural and peri-urban villages (within 60 km radius from the VCT sites, working places (garment factories, phone companies, etc.)) and schools. The present randomized evaluation aims at targeting youths in rural and peri-urban villages.

PSI Lesotho has an extensive experience of working in the HIV prevention sector and it is currently involved in various HIV prevention activities, including, in particular, provision of VCT services and condom distribution.

1. Research objective and Motivation The result of traditional information campaigns (like ABC-Abstinence, Be Faithful and use Condom) on sexual behavioral changes is not conclusive. Although, anecdotal evidence shows that it has resulted in some changes, for example, interventions focusing on condoms use have resulted in increasing condoms during commercial and sexual encounters, there is no rigorous evidence actually showing a significant reduction in HIV prevalence in most Southern Africa countries.

Perhaps the reason for this resides in some differences in sexual behavior in Africa (e.g. it is much more common to have concurrent partnerships that can overlap for significant time) and so the information campaigns are not focusing on the right messages and thus not affecting behavior where it matters the most. Or maybe, as recent literature suggests people tend to delay activities that are a little bit unpleasant in the present (like protecting themselves) even if they have very large returns in the future. People may not be aware of the long-term benefits of safe sexual behaviors. Alternatively, they give a lot of weight to short-term gains, either because they discount the future heavily or have preferences that might be inconsistent over time, with a strong preference for the present.

What replicable and feasible interventions can affect this trade-off between short and long run returns? Can the introduction of a small but "short term" reward promote safer sexual behaviors? This intervention aims at contrasting the difference between a reward in the next 4 months (a lottery ticket) and a reward 10 years later (being HIV negative). The main issue is the interval between the sexual activity and the lottery. To gauge the extent of participation in the lottery, the correct interval time among lottery draws, and the impact on behavioural changes, the investigators plan to do a short qualitative study a couple of months before the beginning of the project.

The research objective is to test and rigorously evaluate whether the introduction of small financial rewards in the short term can promote safer sexual behaviour in general, and reduce HIV incidence in particular, in a high risk environment.

The results of the evaluation are expected to be an important input in designing effective HIV prevention in Lesotho. Furthermore, a credible impact evaluation is also a global public good in the sense that it can offer reliable guidance to international organizations, governments, donors, and nongovernmental organizations in their ongoing search for effective HIV prevention programs.

1. Methods and Evaluation Design 4.1 Evaluation Design The investigators are proposing to conduct a randomized controlled field trial to rigorously test whether short-term financial incentives have an impact on sexual behavior and the incidence of HIV and a small set of curable STIs of youth who live in villages that are served by outreach sessions of New Start VCT in Lesotho.

The underlying question for the study is why individuals get involved in short-term risky sexual behavior when the potential long-term cost of becoming HIV infected is so high, which clearly is the case in Lesotho. It appears like individuals, to the extent that they have not been coerced into having sex (which unfortunately is also a common problem); put a lot of weight on short-term gains at the expense of long run benefits. If this is the case, short run financial incentives may affect individual's trade-off between short and long run returns. In fact, there exists some preliminary evidence that small (financial) incentives associated with activities that have very high returns in the future can result in significant change in behavior. However, there is little (no) evidence that such an incentive scheme will work when it comes to sexual behavior.

Thus, the main hypothesis to be tested is that a system of rapid feedback and positive reinforcement using cash as an incentive can effectively lower risky sexual activity and reduce rates of HIV transmission. The primary outcomes for evaluating impact will be a sub-set of sexually-transmitted infections (STIs), Syphilis and Trichomoniasis, that are prevalent in the population and have been incontrovertibly linked to risky sexual activity. Each of these STIs is curable. This is a critical point, since enrollees who test positive for an STI can continue to participate in the study after they have been treated and cured of the infection. Thus, learning is encouraged through positive reinforcement, and mistakes can be corrected and overcome.

The investigators propose to test this hypothesis through the introduction of financial incentives in the form of a lottery to remain/become STI-negative, which the researchers believe is both replicable and relatively easy to scale-up. Lotteries are also more valued by risk-takers and these individuals might also be the ones with the most risky sexual behavior, thus providing additional incentives would reach the most important group. The presumption is that the financial incentive will influence individuals' trade-off between short and long run returns with respect to sexual behavior.

The sample population will consist of youth aged 18-30 who live in villages that are served by outreach activities in rural and peri-urban villages implemented by the VCT centers in Lesotho The investigators aim at a study population of about 5,000 (see section 4.3 for details). To the extent demand will exceed supply; the researchers will randomly choose the participants.

In a randomized controlled field trial the sample population is randomly divided into a treatment group (individuals that will have a chance to participate in the lottery) and a comparison (individuals that will not take part in the lottery) group. Here, the randomization will be at the individual level. 30% of the individuals will be assigned to the first treatment group, 30% to the second treatment group, and 40% to the control group. The two treatment groups only differ on the amount of winnings possible in the lottery. The control group will be slightly larger to control for the possibility of attrition. Since treatment has been randomly assigned, and provided that the sample contains a sufficiently large number of individuals, units assigned to the treatment and comparison groups are similar in expectations before the intervention. The causal effect of the intervention can therefore be gauged by comparing mean outcomes in the treatment and comparison groups after the intervention.

The random assignment to the control and treatment groups will be done in a transparent way, for example by drawing from a bag which individual will be in the treatment and in the control groups Not only does randomization guarantee internal validity, it is also typically a fair way to allocate participants across the treatment and the comparison groups.

Individuals both in the treated and in the control groups will receive free sexual health discussion sessions as part of the PSI prevention and treatment program, free HIV and STI testing, free STI treatment (for individuals tested positive), and free counseling before and after each STI/HIV test. PSI also will provide free male and female condoms.

Individuals in the treatment groups will be eligible to participate in a lottery if they are tested STI negative. There will be testing and lottery draw every four months. Participants will know in advance the amount of the lottery reward. The lottery will be drawn among all the STI negative people in each villages. There will be 4 winners (i.e. two men and two women) in each village. The exact details on the type and the size of the reward will be more precisely decided during the pilot of the project. The focus on STI status as a condition for participation in the lottery, rather than on HIV status, is primarily based on ethical consideration. By focusing on STIs that are curable, the investigators can also study how the intervention affects behavior of both HIV-positive and HIV-negative participants. STI-positive individuals, who receive treatment and are cured, will be eligible to participate in future lottery draws.

The proposal calls for seven follow-up surveys, one after a 4-month period. The investigators aim at surveying twice the sample of around 5,000 participants. To generate more precise estimates of the impact of the intervention, a baseline survey will also be implemented. The information from the baseline survey will also assist in stratifying the sample and will be used to study heterogeneous effects. In order to minimize attrition during the project, the researchers will collect data on names, addresses and phone numbers of the participants in a way to preserve confidentiality and not to decrease their willingness to participate (see section 5 for details).

The design of this experiment avoids the usual complications of selection and reporting bias because it randomized individual incentive to learn STIs status. However, there will be self-selection of people who choose to decide to participate in the study, because, for example, they are more motivated and aware and perhaps knowledgeable about their options (in case they are HIV positive) compared to someone who has tested. Study participants will also choose whether to know their HIV status or not. However, counseling will be provided to assist them to make an informed decision and accept their results. Those are mainly problems of external validity. The investigators are aware of these problems but in practice, they are difficult to avoid, since the investigators cannot impose this study on a representative sample of the population.