6

Linking Prevalence to Policy

In his initial remarks for this workshop session, Sheldon Zhang (member, Planning Committee) said that prevalence estimation is important not just for advocacy purposes, but also for evaluation; one needs to know the baseline in order to measure whether or not a program has had any effect. He said that human trafficking prevalence estimation essentially boils down to two basic questions: what to count and how to count it. The first question deals with definitional issues, while the latter concerns sampling and data collection procedures.

Zhang noted how there is no diagnostic manual for human trafficking akin to the DSM-5 for mental disorders, the ICD-10 for diseases, or the Uniform Crime Report for law enforcement. Thus, most researchers in this field construct instruments specific to their research projects, either in response to their own understanding of the issues they are investigating or to their respective agency mandates. He added that while researchers, himself included, develop measures based on well-known legal frameworks such as the Trafficking Victims Protection Act, the Palermo Convention,¹ and the Forced Labour Convention of the International Labour Organization (ILO), in practice most of them have not had the luxury of carefully thinking through what is being counted. Zhang discussed what he called a perennial threshold problem: most crimes are defined and studied as individual incidents, but human trafficking cannot be as easily defined. What, then, is the threshold over which a series of incidents amount to trafficking?

___________________

¹ The U.N. Convention against Transnational Organized Crime, named for the city in which it was adopted.

Human trafficking researchers agree that there are significant conceptual and operational overlaps over key concepts in their work. Zhang said that whether it is referred to as modern slavery, human trafficking, bonded labor, peonage, indentured servitude, or forced labor, it still amounts to human rights violations, representing gross exploitation of human labor under unfavorable conditions. He asked participants to consider the possibility of agreeing on a set of core measures that can be used as the basic threshold over which one can define these types of human rights violations. He offered the caveat that social science is more culturally nuanced than physical sciences, but encouraged participants not to be discouraged from working together to achieve core definitions and collective understanding. Zhang closed by saying that before researchers can assert credible influence on policy making or legislation, they should first develop counting rules or diagnostic criteria; as has been noted throughout the workshop, how trafficking violations are counted will bear significant consequences on prevalence or baseline estimation.

DEVELOPMENTS IN HEALTH CARE DATA

Roy Ahn (member, Planning Committee) spoke on his experience conducting human trafficking research across health and nonhealth sectors. He noted how health care data collection is expansive and includes quality improvement and data-sharing initiatives. The ability to share data across health topics has largely been driven by the advancement of electronic health records. Health care data sharing occurs at the aggregated, anonymized level for public health. Ahn noted that there has been a push in health care to begin collecting data on social determinants of health, which could give way to a better understanding of the vulnerabilities that cause individuals to fall victim to trafficking. In 2018, the ICD-10 added codes for suspected and confirmed cases of labor and sex exploitation, which Ahn said has the potential to increase agencies’ incentives for collecting and using those data. Ahn said he hopes this will enable providers to create better trauma-informed models for health care or for community-level prevention, and he noted that those are both ways of potentially linking prevalence to policy.

OFFICE TO MONITOR AND COMBAT TRAFFICKING IN PERSONS

Abby Long (Office to Monitor and Combat Trafficking in Persons [TIP]) explained that the agency, which is housed in the U. S. Department of State, partners with foreign governments, international organizations, other federal agencies, civil society, the private sector, and survivors of human trafficking to develop and implement effective strategies to confront modern

slavery. She said that prevalence estimates produced from a combination of rigorous research, expert consultation, and peer review can provide an improved picture of the problem and a better understanding of the resources it will take to fight human trafficking. Federal anti-trafficking efforts rely on funding from Congress. Long said members of Congress and their staffs sometimes ask TIP about prevalence in order to better understand the problem and help make budgetary decisions.

An accurate prevalence estimate can help researchers explain funding goals to Congress, stakeholders, and the general public. It could also be used to encourage federal government partners to include trafficking components in their foreign assistance projects where appropriate. Long said her office takes into account prevalence estimates where they exist, but where they do not, they rely on grantees for qualitative information that gives an indication of high or low levels of prevalence. In places where there are no estimates or qualitative evidence of any kind, her office must explain why it does not have evidence and what it would take to get it.

In 2017, TIP started the Program to End Modern Slavery (PEMS), a foreign assistance program that integrates prevalence estimates into TIP’s regular programming. The program seeks to build the capacity of foreign governments to achieve measurable and substantial reduction of modern slavery in targeted populations and regions across the globe. To date, PEMS has received $100 million in funding and has made two awards: a $46 million award to the Global Fund to End Modern Slavery and a $4 million award to the University of Georgia Research Foundation.

Long said she believes the connection between research and policy can be strengthened by focusing on three key components. First, she said, prevalence estimation needs to be done using the guiding principles of rigor, ethical considerations, replicability, and transparency. Research designs should emphasize strong anonymization practices and should be reviewed with survivors to ensure the study does not add undue discomfort to them. Second, prevalence work needs to be paired with robust monitoring and evaluation, which should receive its own funding. Prevalence estimates alone do not give a complete picture of trafficking, but when paired with mixed methods they can paint a fuller picture of victims’ experiences and help design programs to better protect victims or prevent traffickers from committing the crime in the first place. Lastly, Long said, prevalence studies should use an impact evaluation methodology. Observational prevalence studies on their own cannot determine which program activities contributed to the decrease or increase in prevalence, but when prevalence studies use an impact evaluation methodology, they can more effectively identify what factor or combination of factors elicited a decrease or increase in prevalence.

THE ECONOMICS OF SEX MARKETS

Manisha Shah (University of California, Los Angeles) explained that in her work on the economics of sex markets, she spends a lot of time considering the best ways to collect large representative samples of data and longitudinal datasets that can enhance human trafficking research. She regularly conducts impact evaluations and randomized controlled trials to study sex trafficking throughout the world. Recently, she has begun working with administrative data and using web-scraping techniques in her U.S. research. Shah said that while economics research lends itself to causal inference, data constraints prevent that technique from being used in human trafficking research. She mentioned being excited to hear about the emerging innovative opportunities for human trafficking data collection—particularly those that measure prevalence at the state, county, or jurisdictional level—because they can provide insight into the effects of policies.

Shah said estimates from research suggest that approximately 80 percent of women engaging in commercial sex in the United States are using online markets and mobile messaging apps. She noted that the Defense Advanced Research Projects Agency has devoted money to “dark web” mining, and she thinks this work could be a way forward in the fight to combat trafficking. Shah said she uses a significant amount of administrative data in her U.S. work, and she acknowledged that variations in agencies’ definitions of human trafficking, differences in the legality of sex work from country to country, and conflation between the global definitions of sex work and sex trafficking can complicate data analysis. She told the workshop participants that she is a proponent of using a combination of data collection methods to best capture prevalence in varying contexts, and she noted that respondent-driven sampling and longitudinal studies conducted in countries, such as Indonesia and Ecuador, have proven in some cases to be as valuable as such methods as multiple systems estimation.

USING THE PUBLIC JUSTICE SYSTEM

Annick Febrey (Human Trafficking Institute) said her nongovernmental organization tackles impunity for traffickers by working through the public justice system. It was founded by two former human trafficking prosecutors and modeled after the best practices they saw in their work at the U.S. Department of Justice and through international interactions. The institute currently operates in Belize and Uganda, working with local governments to develop specialized police units, within the countries’ existing law enforcement agencies, that focus only on trafficking cases.

The institute’s goal is to reduce human trafficking by 10 percent in these countries by increasing prosecution of human trafficking cases and

potentially deterring future traffickers. The organization uses three indicators to quantify its impact: the number of cases going through the justice system, the country’s ranking on the U.S. Department of State’s TIP report, and prevalence estimates. Febrey said the institute is designing prevalence studies for small geographic regions that it hopes will give insight into the relation between the rate of occurrence and the number of prosecutions. She said the institute also regularly uses prevalence data in its policy advocacy. She added that in her experience, congressional staffers seem to be more motivated by numbers than by stories. However, she said, in her opinion, it is unfair for Congress to ask for numbers on human trafficking when the members do not ask law enforcement to quantify the illegal drug trade or the sale of illegal arms. Febrey noted that more has been spent on drug enforcement in the past month than has been spent on trafficking over the past decade.

The Human Trafficking Institute participates in a policy coalition called the Alliance to End Slavery and Trafficking (ATEST) Coalition. ATEST was founded in 2007 by Humanity United and funded for about 10 years. In 2017, ATEST split off as an independent coalition of 12 organizations that work on sex and labor trafficking issues for adults and children, both domestically and internationally. Febrey said ATEST has established a presence in Congress, and congressional staff recognize the work that has gone into the research it presents.

Febrey noted that approximately 40 percent of the budget of the Human Trafficking Institute’s congressional appropriations is designated specifically for sex trafficking. That does not mean, she noted, that the other 60 percent goes to labor trafficking; rather, the other 60 percent is not designated. She is unsure whether sex trafficking is prioritized over labor trafficking because sex trafficking cases are often more visible and clear-cut (and therefore are detected more often) or whether the dedicated funding and focus are what enable researchers and law enforcement to find more cases. She walked through the informal calculations she uses to estimate the number of cases that should be prosecuted in order to achieve the goal of a 10 percent reduction and emphasized how concise prevalence estimates could improve communications around policy.

Febrey and her team have also been trying to quantify the cost and impact of not doing anything to fight human trafficking. She said she talks with U.S. and international governments about the risk associated with not investing in anti-trafficking programs, including lost tax revenue and lost productivity, and she believes such data would help incentivize and motivate congressional staff to prioritize the issue. She noted that policy implementation boils down to two major steps—authorization and funding—which are handled separately in Congress. Febrey hopes that the estimates the institute generates help bolster legislative and financial support for effective counter--

trafficking programs, and she emphasized the need for estimates that are clear and easy to communicate to nonresearchers.

DEFINITIONS AND RESEARCH APPROACHES

Bernard Silverman (member, Planning Committee) reiterated Shah’s point about using the best estimation method given the question at hand and the information available. He said it is important to understand the caveats and limitations to each approach, including multiple systems estimation, and to combine insights from all the various different methods. He added that the narrative and the figures need to work together to convey the most effective message to policy makers.

Kelly Dore (National Human Trafficking Survivor Coalition) expressed her appreciation for Febrey’s openness and clarity about how challenging it is for researchers to create human trafficking prevalence estimates. She said that from her perspective as a survivor, it seems that funding for counter-trafficking is more reactive than preventive, in part because prevention can be very expensive. At the same time, Dore said, it can be difficult to get survivors to come forward and re-tell their experiences to researchers without any form of compensation; it can begin to feel like exploitation. Because of that, she said, survivors often stick to helping other survivors through organizations and channels of which mainstream researchers may be unaware.

Dore said she has sometimes brought survivors to her house, which she acknowledged is not the safest idea, but in the absence of access to emergency resources it has been, in her opinion, the best solution at the time. She told participants that just that morning she had heard Congresswoman Martha Roby (Alabama) say that $90 million in appropriations had been set aside for human trafficking prevention, which Dore said she hopes will make its way to final approval and to the ground level.

Dore also commented on Shah’s point about varying definitions of sex work and sex trafficking. She said that researchers need to understand the vulnerability and psychological burden that come when a survivor has to reconcile the choices he or she had to make about engaging in sex work. She said she talks with law enforcement personnel to try to remove misperceptions around victimization and improve the way victims are treated when they come forward. She also acknowledged having to be intentional about sorting through victim accounts and information to weed out false statements and help dispel survivor mistrust—two issues that injure survivors.

In response to Dore’s comment about employing and supporting survivors who share their stories and engage in research, Shah said her team has worked very successfully with national sex worker organizations in several countries. She said that, in many cases, the organizations were given the

contract to conduct the data collection and in turn hired current or former sex workers as enumerators. Shah has learned that the data collected from this approach are of higher overall quality than data gathered from other approaches. She added that forging connections with local survivor organizations also aids in conducting long-term research and allows researchers to stay in touch with victims over time. Regarding Dore’s point about reconciling survivors’ choices to engage in sex work, she said that decriminalization is a major debate and acknowledged that even when men and women “choose” to engage in sex work, it is often because the alternative forms of employment available to them do not provide sustainable forms of income—particularly when the sex worker is caring for a family. Shah is aware that the worlds of sex work and sex trafficking intersect, but thinks that the degree to which they overlap varies by country.

James Johndrow (Stanford University) commented on Shah’s discussion about using different research techniques to measure human trafficking populations and how human trafficking does not lend itself to causal inference. He said he thinks that record linkage could be a major component of the research, which is facilitated by having more detailed information about each victim. Without this degree of detail, Johndrow said, the estimates created could carry very large confidence intervals. He said it is important that the people who are gathering the data and the people who are calculating the estimates are communicating with each other.

A webcast viewer noted that the use of public ads for sex commerce varies in popularity among countries and provinces or states. In many regions, commerce takes place through mobile messaging apps such as WhatsApp and in private chat rooms, which makes the comprehensive estimation more difficult than it was when DARPA Memex² was first started and when backpage.com³ was the primary virtual marketplace. He raised the question of what entity resolution would look like, given the nature of these new virtual forums. Shah agreed with the person’s observation and reiterated how researchers should be doing as much as they can, considering that they know so little about the issue—another reason she is hesitant to give any one research method credence over another. She encouraged continued use of web-scraping techniques and noted that DARPA’s Memex program has devoted a lot of its resources to this work. Ahn added that as technology evolves, research will continue to evolve to meet new challenges. Silverman said that the same evolution applies to law enforcement around human trafficking.

___________________

² See https://www.darpa.mil/program/memex.

³ See https://www.justice.gov/opa/pr/justice-department-leads-effort-seize-backpagecominternet-s-leading-forum-prostitution-ads.

This page intentionally left blank.