There are 300 children in Georgia’s prostitution industry. Here’s how we know it.

By Alex R. Trouteaud, Ph.D.

The question of how many juveniles under the age of 18 are caught up in the illegal sex trade is a difficult research question that, in some ways, has only gotten harder to quantify in the past decade or so since advocates, law enforcement, and public policy actors began taking the issue seriously. I can attest to this difficulty firsthand since I led a research team starting in 2007 that observed and systematically recorded the presence of youth in metro Atlanta’s prostitution industry via street-level activity, online classifieds websites, call-in escort services, and more. At that time, and for a couple of years thereafter, few people were on the lookout for youth in street-level prostitution corridors, and almost no one was paying attention to the sexually gratuitous images of youth posted by users of online classifieds websites.

An Effective Methodology, Now Outmoded

Every few months our researcher team reviewed and recorded each of the hundreds of online prostitution ads in metro Atlanta, and then used a set of (empirically-based) assumptions to estimate how many of the total number of youthful ad images likely involved juveniles. Like most research studies, our methodology was imperfect, relying on a set of clearly-articulated assumptions that allowed critics to assess the strengths and weaknesses of our approach. The study results were consistent over time—and frustratingly low to advocates prone to hyperbolic outrage—and the trends in our results were consistent with the obvious trends affecting the prostitution industry at that time.

Just as e-commerce was growing rapidly during that period, so, too was the apparent popularity of online prostitution advertisements…partly because of users’ increased comfort with transacting online, but I suspect also it was because no one—not law enforcement, not advocates, and not the website owners themselves—were actively monitoring these corners of the web. Ten years ago, it was relatively easy to discern images of youth in online prostitution advertisements.

We dutifully recorded research-based estimates of prostituted juveniles in Georgia that grew over time as these web sites grew in popularity with illicit sex buyers. Other areas of youth prostitution activity generally subsided as online activity flourished. When these online classifieds web sites learned that advocates were using our research studies to attack their businesses, the web site owners used a variety of tactics to target members of the research team and discredit our findings altogether. Yet today, the rate at which juveniles are recovered from the prostitution industry due to having been advertised for sale online is bittersweet vindication that our research was right all along.

Our estimates of the number of juveniles involved in the metro Atlanta sex trade grew from about 250, to 500 at the peak of the “unfiltered image” online prostitution era. The prostitution industry has changed a lot since then, so a methodology that produced consistent results in this era is no longer even feasible.

A New, Cross-Validated Approach

For many years since that time, I have answered with a simple “no one really knows” when people ask me “how many juveniles are there in Georgia now?” Recently, though, I wondered if we have the answer after all. A decade ago there were hardly any studies documenting youth involvement in the sex trade. Today there are many excellent examples of such studies, from different methodological traditions, with different focus areas, and from a wide range of respectable researchers. Any experienced researcher will tell you that we don’t really “know” something to be true from research until the findings are corroborated from multiple studies and by different research teams. But with all the great data sources we have now, I set out to determine if we have enough high-quality empirical data to speak with confidence on how many juveniles are involved in Georgia’s sex trade. The answer, I believe, is yes.

We have enough quality data from enough different sources to indicate with confidence how many juveniles are involved in Georgia’s sex trade. There are around 300 juveniles; 250 girls, and 50 boys. Below I explain how I arrived at these estimates.

Step 1: Market Size

The process I use for calculating the number of prostituted youth in Georgia involves looking for alignments among different studies from different research teams. I’ll start with an oft-cited stat from The Urban Institute on the metro Atlanta illicit sex trade market: $290m annually, which includes both prostitution (of which sex trafficking is a subset) and child pornography. The authors of the report say that the paid child pornography industry has basically evaporated with the proliferation of free, peer-to-peer image sharing. To be on the safe side, let’s discount the overall market size to $250m annually so that it reflects only prostitution.

While I have tremendous respect for this research team, I’ve never felt totally comfortable with the results, in large part because the math behind the market size estimates required many assumptions that were often derived from a relatively small number of qualitative interviews. For instance, one value used in the calculations was that the average pimp brings in $32k in revenue per week. I find that hard to believe as an arithmetic mean for two reasons. First, I’ve never met a prosecutor who tries youth trafficking pimping cases who thinks that revenue estimate is accurate. Second, if it were accurate, the offenders would need a substantial money laundering operation that would likely prove far easier to prosecute than the trafficking itself—$1.7m annually is difficult to hide. Numbers like this estimate were plugged into the research equation by necessity, but when your math requires you to use assumptions like this too many times, it tends to lead to accumulated error that produces out-of-the-ordinary results. More on that $250m estimate shortly.

A different set of data I find very instructive comes from analysis of online content scraping on Backpage. From November 2014 to September 2016, we have monthly data on the number of unique phone numbers that appear on Atlanta’s Backpage escorts section. You can see from the figure below that the low is 2,011 and the high is 4,469. The median is 3,536. We also have the data broken down from before the credit card paywall came down in September 2015, and during the free/cryptocurrency period thereafter. The median number of unique phone numbers during the credit card period is 2,405. The median number during the free/cryptocurrency period is 4,110. Ads are cheaper during the free/cryptocurrency period; thus, we expect more individuals to choose to advertise on Backpage during this period. This includes independent workers, pimps, and massage brothels.

Graph of phone numbers in Atlanta's prostitution market

Based on these data, let’s say that there are approximately 4,000 prostitute individuals—FTEs, that is—on Backpage each month in the metro Atlanta market. The possible overestimate here (i.e., a brothel or individual using multiple phone numbers to create different profiles to appeal to a wider segment of the market) is likely offset by the guaranteed underestimate (namely, that a massage brothel has one phone number to represent multiple individuals, which may also be true in a limited number of pimp-controlled settings and, more commonly, escort agencies). Also, surely there are prostitute individuals who are not advertised on the escort section of Backpage, namely higher-end escorts, prostitute males, and homeless individuals engaged in survival sex. It seems reasonable to estimate that the actual number of prostitute individuals (FTEs) across all segments of the market is 5,000, meaning about 80{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of these individuals (an assumption) appear at least once per month on Backpage, which at the time was the dominant advertising platform in Georgia for prostitution.

Now that we have a robust estimate of prostitute individual FTEs, we can cross-validate the Urban Institute market size estimate by multiplying by the “average” earnings (wages) of a prostitute individual FTE per year. If we take the Urban Institute market estimate from 2009 of $250m, then that’s an average annual wage of $50,000. While that’s possible—and certainly some individuals make that or more, most empirical studies suggest it’s quite an implausible estimate of the average experience. In the book SuperFreakonomics, Sudhir Venkatesh presents the results of field data collected from 160 prostitute women in Chicago. He finds that, unsurprisingly, prostitution is mostly low-wage labor, but that it pays better than minimum wage. This finding makes intuitive sense, since prostitution disproportionately lures the poor, who often evaluate it against alternative low-skill, low-wage labor that typically pays at or near minimum wage.

Venkatesh’s research finds weekly wages of between $325 to $410 depending on whether a pimp is involved or not (pimps increase revenues, but decrease “take home” pay). These prostitute individuals average between 7.8 to 6.2 transactions per (pimps decrease the number of transactions because the price per transaction is raised). Taking the average puts weekly wages at $368—which mirrors data from other studies—and weekly transactions at 7, which puts the average transaction value at around $53. These individuals do not typically work every day of the week, but apparently 7 transactions per week is the definition of FTE. This pay rate equates to an annual wage of $19,136, which is not only tax free, but also 27{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} higher than a pre-tax 40-hour work week at the federal minimum wage of $7.25 per hour.

For the sake of simplicity, let’s say $20,000 is the average annual wage of a prostitute individual in Atlanta. With 5,000 FTEs, that puts the total market value at $100m—a far cry from the $250m+ estimate from Urban Institute.

We should also assume that Venkatesh’s methodology, while solid, misses the higher-end of the market. And by his own account, it likely did. At the higher end of the market there are fewer transactions per FTE, but each transaction value is much higher. We have data from our own study of 1,235 buyers in the metro Atlanta market. In those data depicted below, we see that 56{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of buyers were anticipating spending either $50-$99, or $100-$149. From these data, it seems more reasonable to estimate the average transaction value at $100, which would obviously put the annual market closer to $200m, but still far under the Urban League estimate. But because our data exclude lower-value street transactions, which are not “worth” advertising online in a paid platform, $100 per transaction is probably too high for the market average. Thus, for this analysis, we split the difference and use $75 for the average transaction value, which at 7 transactions per week for 5,000 FTEs yields an annual market for illicit paid sex in metro Atlanta of $136,500,000, which we round to $140m for simplicity.

Desired spend per transaction on prostitutionn

We now have two critical data points about the size of Georgia’s prostitution industry to calculate its impact on juveniles:

  • How many individuals are prostituted in Georgia’s illicit sex trade: 5,000 per month
  • How large the market is, in dollar value: $140m annually

Step 2: Buyers

With a reasonable, and perhaps slightly conservative estimate of Georgia’s overall prostitution market from a revenue standpoint, as well as supply-side measures of FTEs and transactions, we are very close to being able to estimate how many buyers are behind it. Here again I strongly favor empirical data over estimates and equations. In 2008, Brewer et al. published a research paper estimating how many men across the U.S. pay for street prostitution, which at the time was a slightly larger portion of the market than online-facilitated prostitution. They estimated 3.5{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of U.S. men were paying for street prostitution annually, but the overall finding from the study is from a series of complex equations, which are built on empirical data to inform assumptions. One of those empirical data points is all we need to complete our estimate of buyers in Georgia. In their paper, they describe a study of 110 buyers who reported an average (mean) number of transactions per year of 10.1 (median=2, range=1-100; an obviously skewed distribution that also serves as reminder that a small number of buyers drives a disproportionately large share of the market).

Again, for simplicity let’s assume the mean is 10 transactions per year per buyer. We also know the total number of transactions annually through 5,000 FTEs x 7 transactions per week x 52 weeks in a year = 1.82m annual transactions in the market, which we can then divide by 10 transactions per year per buyer from the Brewer et al. study to yield an estimate of 182,000 buyers active per year in Georgia. With approximately 3.8m adult men living in Georgia, that’s 4.8{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of the adult male population—not far off the Brewer et al. estimate which included only street-level prostitution activity.

What about people traveling to from out-of-state to pay for sex illegally? This is an important consideration, and based on data we’ve collected from buyers “shopping” online on Atlanta’s Backpage escorts section in the map below, as many as half of “shoppers” are out-of-state when interacting with ad content. Assuming some of these men are Georgia residents on their way back home, and that an out-of-state “shopper” is far less likely to convert to a transaction due to geographic space, I think it’s reasonable to estimate that 80{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of online “shoppers” who end up transacting are locals. For this analysis, however, this travel estimate doesn’t really change the results at all because the safe assumption is that when Georgia men travel, they are just as likely to buy sex “on the road” as other men are. Thus, the traveler effect cancels out, and we can still estimate that 4.8{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of metro Atlanta men buy sex illegally, though not always locally; and similarly, a portion (perhaps 20{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} or so) of the $140m local market is the result of transactions from men who live outside metro Atlanta.

We now have another critical data point in describing the prostitution market:

  • How many buyers transact within that market annually: 182,000 men, or 4.8{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of the adult male population

As a bit of an aside, if we divide $250m / $750 (the average spend per male per year), we would estimate 333,333 men transacting annually, or 8.8{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of Georgia’s adult male population buying illegal paid sex each year. This result is difficult to accept at face value. Also, if we divide $250m / 182,000, we would estimate that each man spends an average $1,374 per year on paid sex, for an average transaction value of $137—which is similarly unsupportable by empirical data. Not to belabor the point, but it is difficult to make a $250m market make sense, arithmetically at least. The Urban Institute study is still an outstanding research effort that brought to light many important and frequently validated findings beyond the 2009 Atlanta market size estimation.

Step 3: Juveniles in Georgia’s Prostitution Market

So how many juveniles are involved in this $140m market? We haven’t had any empirical data collected in metro Atlanta on the number of (non-system-involved) adolescent females prostituted locally in quite a few years, but we do have an empirical estimate of the number of adolescent males: approximately 40-60. We can use 50 for a safe estimate, which for this analysis will include online prostitution and street runaway survival sex.

We also have empirical studies that look at juvenile participation among youth in the sex trade, including the aforementioned study on boys, as well as a 2014 study by Marcus et al. that found 10{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of prostitute individuals up to age 24 were juveniles in a field study in Atlantic City. From our own recent analysis of self-reported ages on Backpage in the figure below, we observe approximately 60{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of “supply” advertised as age 24 or younger, which is 5,000 monthly FTEs x .60 = 3,000 monthly FTEs ages 24 and younger.

Advertised ages on Atlanta's Backpage site

Obviously 10{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of this number is 300 per month who are juveniles, and we have established that 50 of these are boys. Thus, we estimate 250 juvenile females involved in metro Atlanta’s sex trade each month. These 300 juvenile FTEs equate to 9,100 transactions each month, for a total juvenile market size of approximately $8.2m annually.

This allows us to develop our final calculation:

  • Number of prostitute individuals who are juveniles: 250 girls, 50 boys

This estimate is even further strengthened by cross-validating against a large academic study published in 2016 by experienced researchers at the Center for Court Innovation. In a very robust study of the youth sex trade, they use a combination of over 1,000 in-person field interviews and advanced sampling procedures to estimate 10,500 juveniles involved in the illegal sex trade nationwide. While the study did not sample from Georgia directly, we can extrapolate to the state level easily. Georgia is 3.16{71bb150a2bc889730474e0d4d3c3934bdfc3cc805d65fba18ac3d426a1e1afd5} of the U.S. population, so if prostituted adolescents are distributed proportionally by state, we would expect approximately 316 juveniles in Georgia’s sex trade during a “point in time” count, which aligns with our estimate of approximately 300.

Thus, by aligning multiple studies from credible, peer-reviewed sources, we can safely estimate the number of prostituted youth in Georgia. Hopefully these estimates reflect a decrease from several years ago at the peak of the “unfiltered image” online prostitution era. It certainly seems plausible to assume that fewer youth are involved in the trade now that many law enforcement officials, mandated reporters, and child welfare professionals know how to identify adolescent victims of the illicit sex trade. Only time and additional research will tell.