Does sex work really contribute £5.3bn to UK GDP?

That’s the figure the Office of National Statistics (ONS) came up with when asked by the EU to estimate how much prostitution contributes to the UK’s GDP. A more intellectually honest statistician’s answer might have been “Sorry guys, but we haven’t a clue.” Fair play to them for having a go, but if you look at the detail behind the ONS numbers it is clear they are based on some fundamental misunderstandings of how prostitution in the UK actually works, and some very iffy assumptions.

The background is that countries are required to contribute to the EU budget in proportion to the size of their national economies, as measured by something called their Gross Domestic Product (GDP). So it’s important every country calculates GDP in the same way, which is why from this year various businesses on the fringes of respectability and legality will have to be included for the first time.

What the new rules say is that “illegal transactions to which all units involved consent are included within the production boundary and therefore the relevant items in the National Accounts.” That rules out business activities that make money without consent, such as identity fraud, contract killing or human trafficking. The UK figures apparently already include tobacco and alcohol smuggling, and the ONS also have plans to look at illegal gambling, music and software piracy, and fencing stolen property: all business transactions that are illegal but carried out by consenting buyers and sellers.

For now the ONS has concentrated on two sectors: the manufacture, cultivation and sale of illegal drugs; and prostitution. Both are dealt with in the same paper, which you can read here Inclusion of illegal drugs and prostitution in the UK National Accounts May 2014. There is also a spreadsheet with detailed calculation, which for some reason I don’t seem to be able to link to.

My understanding of the drugs culture is limited to having read a few Iain Banks novels, recognising the distinctive smell of my neighbour’s cigarette smoke and being an occasional user of non-prescription migraine tablets. But I do know a bit about prostitution, so I’ll restrict my comments to that. And although I’m not a professional statistician (unlike Dr Brooke Magnanti – Belle de Jour – who has already expressed her doubts about the ONS numbers) I did, admittedly many years ago, manage to graduate with a Cambridge Maths degree so I’m not entirely ignorant about statistics.

Before we get to the detail I need to say something about a fundamental misunderstanding that runs right through the work done by the ONS. And it’s so important I’m going to highlight it in its own little box.

The ONS have been very good at listing and assessing the validity of their key assumptions, but there is one that is only mentioned in passing, and it’s completely wrong. The ONS have assumed that prostitution is illegal; and of course it isn’t. Street work is illegal, as is managing a brothel, or controlling someone else’s prostitution for gain, but it is perfectly legal for anyone over 18 to work as a prostitute. Furthermore a prostitute’s income is taxable, which they would have known had they asked HMRC, who have a short section of their Business Income Manual devoted to the subject.

Instead the ONS make statements like this: “There is no activity of corporations or quasi-corporations in prostitution. This is simply based on the intuition that prostitution is illegal, and the reporting to government required to achieve corporate or quasi-corporate status would be a pointless risk.”

It is certainly true that individual escorts very rarely attempt to set themselves up as limited companies, but that’s not because prostitution is illegal (it isn’t); it’s because of an issue in English contract law, which was explored in the Lindi St Claire (Personal Services) Ltd case.

So let’s look at the ONS calculations and the assumptions behind them. For some reason, that I can’t see explained, they have opted to calculate what the GDP would have been in 2009. Other than that the calculations are very simple – even simplistic – and they go like this:

  • Number of prostitutes in the UK (2004):   58,000
  • Increase in UK population of males aged 16+ from 2004 – 2009:   5%
  • Estimated number of prostitutes in UK (2009):   60,879
  • Clients per prostitute per week:   25
  • Average price per visit (2004):   £55
  • Increase in Consumer Prices Index for personal services from 2004 – 2009:   22.1%
  • Estimated price per visit (2009):   £67.16
  • Sales of prostitute services (2009): 52 weeks x 25 x 60,879 x £67.16 = £5,314 million
  • Expenses – room hire, clothes and condoms: £44 million
  • GDP 5,314 – 44 = £5,270 million

In case you’re scratching your head at some of these estimates (or wild guesses) the ONS do helpfully tell us where they came from.

Number of prostitutes

The main source is an Eaves – Poppy Project report which estimated the number of prostitutes working in London in 2004. It appears that they did this by ringing every massage parlour, working flat and escort agency they could find and asked how many girls they had working. Their main conclusions were:

  • Between 2,972 and 5,860 women were selling sex from 730 flats, parlours and saunas.
  • Between 1,755 and 2,221 escorts were selling sex via 164 escort agencies.

From these two statistics the ONS have concluded there were approximately 7,000 off-street prostitutes working in London in 2004. The Metropolitan Police have given an estimate of 115 working on the street at any time. Giving a total for London of 7,115.

The ONS have then assumed the ratio of prostitutes to the general population will be the same across the country to arrive at a figure of 58,000 for the whole of the UK. They do admit that the ratio is probably higher in London but say that as the London numbers are probably an under-estimate this does not matter.

There are huge, huge problems with these numbers.

  • The Eavis Poppy Project report deliberately ignores male prostitutes. There is almost no market for heterosexual male escorts, but a thriving gay prostitution economy.
  • It appears that trans-sexuals were also excluded from the Eavis research.
  • There was no attempt to calculate numbers of independent escorts; only escorts marketing themselves through agencies are included. This ignores the fact that the internet and the success of sites like Adultwork have made independent escorting the preferred business model for a large proportion of escorts.
  • It is not clear where the boundaries of prostitution lie. For example are dominatrices (many of whom do not offer penetrative sex) to be included? I don’t believe they were counted by Eavis.
  • The Eavis Poppy Project (unlike the ONS) is not an impartial observer. It is a support organisation for trafficked sex-workers and the avowed intention of their research was to demonstrate the scale of prostitution in London and the extent of the trafficking problem. They were unable to quantify the numbers of trafficked prostitutes but concluded that “evidence gathered from this research, from sexual health outreach agencies and interviews with trafficked women indicates that more women, at various stages of the trafficking process, are working in London’s sex industry than has previously been estimated.”
  • The ONS do not appear to have attempted to exclude victims of trafficking from their numbers. They should of course be taken out because only transactions between consenting parties are to be included in GDP. None of us know how many women are trafficked into prostitution. Eavis quote an estimate of 5% in 1999 and say this is an under-estimate.
  • No allowance has been made to exclude those sex-workers who are already included in the national accounts because they are registered for and pay tax. I have no idea how many that would be, but what I do know is that during May 2014 the accountants advertising on this site received enquiries from 12 separate escorts seeking help with their tax affairs. The numbers vary from month to month and 12 is fairly typical. As an aside it would be interesting to know how that compares with the number of trafficking victims identified each month.
  • Prostitutes are certainly not spread evenly among the population of the UK. Partly that’s because of demand: London contains a higher proportion of men eager to spend their banker’s bonus than does, say, Middlesborough; and London also has a regular flow of business men working away from home. But there are, apparently, other less explicable variations. The sainted NIK of UKpunting in his appallingly badly written memoirs points out there are comparatively few working girls in Birmingham but a lot more than you would expect in Milton Keynes.
  • The estimate of growth in numbers from 2004 to 2009 ignores other societal changes including the effect of the banking crisis and recession from 2008, and the impact on student finances following the introduction of tuition fees.

Clients per prostitute

One of the difficulties the ONS have created for themselves is that instead of looking at each business sector separately – escorts, parlour girls, street workers etc – they have opted to lump them all together. The reality is that they work in completely different ways so attempting to estimate the average number of clients per week across all sectors is fraught with difficulties. Their estimate of 25 per week comes from a Netherlands study. Alongside this is an assumption that every prostitute works 52 weeks per year.

  • I’ve never been to Amsterdam, or anywhere else in the Netherlands, but my impression is that the business model and way of working is totally different to the what goes on in the UK.
  • Parlour girls tend to see a lot of clients per shift, but don’t work every day. Escorts typically see few clients per day, unless they are on tour, and street workers are dependent on the weather and time of year.
  • Very few prostitutes work 52 weeks per year. A large proportion appear to be mothers, which means a different working pattern in school holidays. And escorts certainly take holidays. Many also appear to take a break after a busy high earning spell.
  • A lot are part-time and do this work around another job, or in the case of students around studying.

Average price per visit

The figure of £55 is derived from the ONS own research – research into the Punternet Field Report database!

  • Again there is a problem of mix. Different sex-workers operate their businesses in different ways. Is the mix of different types of prostitutes included in the PN reviews representative of all prostitutes?
  • Once again the Punternet reviews exclude male prostitutes.
  • Street workers are also excluded.
  • The average of £55 looks low to me. As it happens I did some research in February 2009 on the PN reviews in order to see if there was a correlation between the average price per visit and the level of popularity of different escorts. I looked at the 133 escorts with at least 20 reviews and found, among other things, that the average price paid per visit to these most popular escorts was £153. I know that is measuring a very different thing, but I am surprised the ONS found an average for all visits to be as low as £55.
  • And what about other income earned by a prostitute? The rise of the internet, and in particular, has allowed sex-workers to earn from subsidiary activities such as selling access to private photo galleries, offering web-cam sessions, selling video clips and selling used underwear. None of these are illegal, but they are likely to be part of the hidden economy.


The ONS have deducted their estimate of a prostitute’s business expenses (which they call intermediate consumption) because expenses will already been measured as part of the output of some other business. The expenses included are rent, clothes and condoms. The calculation of rent is complicated and explained in little detail, but the figures for clothes and condoms come from Dutch research, which evidently found that the average expenditure on clothing was €125 per year and on condoms €0.50 per client.

  • The rent calculation appears not to make any allowance for an escort’s expenditure on hotels, or for the fact that many independents choose to rent a flat separate from their home specifically for work.
  • No allowance has been made for travelling expenses, which can be considerable for anyone offering outcalls.
  • And nothing has been included for marketing. Most independents will spend on photographs and have their own website.
  • It’s not clear how the ONS have dealt with payments to escort agencies. Most are set up as legitimate businesses, and are presumably already included in the GDP figures, although in nearly every case what they are doing is probably illegal.
  • It is practically impossible these days to operate without a mobile phone and a laptop or tablet with internet access. Indeed most escorts will have a separate work mobile.
  • And of course there are accountant’s fees.

And finally

Does the ONS result look reasonable?

Well, no it doesn’t.

  • If 60,879 prostitutes really are generating £5,314,000,000 of income that means an average income before expenses of £87,000 each. OK they may not get all of that because some of it will go to agency owners, parlour managers, and assorted pimps. But I would be very surprised if there are as many as 1,000 turning over as much as £87,000 per year, and of course any that are should be registering for VAT.
  • Looked at from the other end, let’s suppose the average punter spends £1,000 per year. That would mean there are 5.3 million punters, or over 20% of the male adult population. Alternatively if 5% of men use prostitutes their average annual expenditure would need to be £4,000. Neither of those look credible to me.

In her Telegraph article Brooke Magnanti suggests the ONS might have got their sums wrong by a factor of 10. She also recommends they could get more accurate results by surveying some real life sex-workers. It would certainly be good to have some statistically robust figures on the size of the prostitution sector, the true extent of sexual trafficking, and how much tax is contributed to the economy by sex-workers, instead of having to rely on selective data peddled by organisations with an axe to grind.

And the ONS could always ask HM Revenue & Customs?

PS The Irish statisticians are going through the same exercise. Wendy Lyon has written an excellent piece about National Accounts and Prostitution from an Irish perspective.