Sweden and Germany: One Question, Two Opposite Answers, and What Neither Can Prove

Before anything else, a disclosure, because this chamber asks readers to weigh arguments and a reader is entitled to know who is making this one.

Moonlight sells intimate services. It is not the trade this essay is about — the arrangement here is companionship and touch, conducted under Japanese notification requirements, and no part of it turns on the acts the laws below regulate — but it is adjacent enough that we are not a disinterested observer. So we will not endorse a model. What we will do is lay out what two countries actually did, what can and cannot be shown about the results, and where Japan sits, which is somewhere neither camp talks about.

The question both countries were answering is the same one: given that people sell sex, what should a state do about the people who buy it, the people who sell it, and the businesses in between?

In 1999 Sweden answered by criminalising the purchase and decriminalising the sale. In 2002 Germany answered by legalising the trade and bringing it inside labour and contract law. Two wealthy European democracies, comparable in many respects, took opposite roads at almost the same moment.

That is as close to a natural experiment as social policy ever gets, and after two decades the result is not what either side hoped. Neither country can demonstrate the outcome its advocates promised.

The Swedish law of 1999 made it an offence to purchase a sexual service and not an offence to sell one. The reasoning was explicitly framed in terms of gender equality: the sale of sex was understood as a form of violence against women and as an obstacle to equality, and the person selling was positioned as the injured party rather than the offender.

That framing is the model’s genuine innovation and the reason it travelled. Norway and Iceland adopted versions of it, France and Ireland followed later, and it became known internationally as the Nordic or equality model. Whatever one concludes about its effects, it was the first serious attempt to write a prostitution law that did not punish the person with the least power in the transaction.

The claimed results are a reduction in street-based sex work, a reduction in demand, and a deterrent effect on trafficking into Sweden. Swedish official evaluations have reported in that direction.

The claimed harms, reported principally by people selling sex and by some researchers working with them, run the other way: that criminalising buyers pushes the trade into less visible settings, compresses the time available to assess a client, makes screening harder, discourages reporting violence for fear of consequences to third parties, and complicates housing and banking for sellers even though selling itself is legal.

Both of those accounts can be partly true at once, and the honest reading is that they are.

The German law of 2002 removed the legal treatment of prostitution as immoral, making contracts enforceable and opening access to social insurance and employment protections. A further law in 2017 added registration requirements, mandatory health counselling and licensing of premises.

The reasoning was the mirror image of Sweden’s. If people are going to sell sex, the state can either push the activity into the dark or bring it into the light where it can be inspected, taxed, and made subject to the same protections as other work. Illegality, on this view, is what produces the exploitation; regulation is what reduces it.

The claimed results are a legitimate sector with labour rights, health access, and enforcement capacity against the actually coercive part of the trade.

The claimed harms are that legalisation produced a very large, commercially aggressive industry; that the registration and licensing regime is unevenly implemented; that migrant workers — a large share of the sector — are least able to use the protections on paper; and that a legal market may provide cover for coercion rather than removing it.

Germany’s own 2017 reform is itself evidence that the 2002 settlement was not felt to have worked as intended. A country that is satisfied with a law does not overhaul it fifteen years later.

Here is the part that both camps skip, and it is methodological rather than moral.

The outcomes people most want measured are the ones hardest to measure. The size of a partly hidden market cannot be counted; it can only be estimated, and estimates move with the method. Trafficking prosecutions measure enforcement activity as much as underlying prevalence. Violence against sellers is systematically under-reported everywhere, and any legal change that alters willingness to report will change the measured rate without changing the actual rate — in either direction.

Then there is the comparison problem. Sweden and Germany differ in population, borders, migration patterns, welfare systems and enforcement culture. Attributing a difference in outcome to the one policy that differs, when dozens of other things also differ, is not something the data can support.

And the literature itself is unusually politicised. A great deal of the research is produced by people with a prior commitment to one model, funded by bodies with a position, and the reviews reach opposite conclusions with a regularity that should make any reader cautious.

The responsible summary is therefore uncomfortable and short: after more than twenty years, the evidence does not settle which model produces better outcomes for the people selling, and anybody telling you it clearly does is reporting a conviction rather than a finding.

One observation cuts across the whole debate and is the most useful thing in this essay.

The people most publicly certain about which model is right are, overwhelmingly, not the people selling. They are legislators, campaigners, academics, journalists and activists on both sides — and the positions are held with a confidence that the evidence does not license.

Meanwhile, organisations of people who sell sex have tended, in both countries and internationally, to prioritise a narrower and more practical set of demands: the ability to work together rather than alone for safety, the ability to report violence without a third party being prosecuted, access to healthcare and banking without disclosure penalties, and freedom from immigration consequences that make reporting impossible.

Those demands cut across the Sweden-Germany axis. Some are easier under one model and some under the other, and none of them is the abolition-versus-legalisation question that dominates the argument.

We think that is the finding a reader should carry out of this essay. When the people with the most at stake are asking a different question from the one everybody is arguing about, that is usually a sign that the argument has been organised around something other than their welfare.

Now the mirror, and it is genuinely distinctive rather than a variation on the European pair.

Japan’s Anti-Prostitution Law of 1956 prohibits prostitution, defined narrowly as sexual intercourse in exchange for payment. But the structure is unusual in two ways. The law attaches no criminal penalty to the basic act for either the buyer or the seller; what it criminalises is solicitation, procuring, coercion, and operating a business for the purpose. And because the prohibited act is defined narrowly, an enormous adjacent industry of services that are not that act operates legally, licensed and regulated under separate entertainment-business legislation.

The result is a third model that belongs in this comparison and is almost never included in it. Japan prohibits a narrowly defined act without punishing the two people involved, regulates a very large legal industry immediately beside it, and prosecutes primarily the organising and coercive layer.

It is worth being precise about what that does and does not achieve, because Japanese readers will have views. It avoids criminalising the seller, which the Swedish model also achieves and the pre-1999 European norm did not. It creates a licensed sector with premises, rules and a regulator, which resembles the German aim. And it leaves a definitional line that determines an enormous amount — what is inside the law and what is outside it turns on an act rather than on conditions, consent or coercion, which is a strange place for a line to sit if the concern is harm.

The most significant recent development is not in that law at all. In 2024 Japan brought into force a new statute on support for women facing difficulties, which moved the welfare and support provisions out of the 1956 Anti-Prostitution Law framework. That matters more than it sounds: for nearly seventy years, state support for women in difficulty was legally attached to a prostitution statute and framed around protection and rehabilitation. Detaching it is a change in what the state thinks it is doing when it helps a woman.

Three things, and we think all three hold regardless of which model a reader prefers.

First, the least defensible arrangement is the one most countries had before any of this: criminalising the person selling. Sweden and Germany disagree about almost everything and agree about that, and so does Japan’s statute in its own way. Where a reader finds a jurisdiction still punishing the seller, that is the position with no serious constituency left.

Second, a law’s effect depends on enforcement culture more than on its text. The German registration regime and the Swedish purchase ban are both implemented unevenly, and a law that is enforced selectively against the least protected people in the market is a different law from the one that was written.

Third, and most relevant to this Library: every model in this essay is arguing about the transaction, and none of them touches the thing this chamber keeps finding. Whether a particular encounter is safe for the person providing it depends on whether she can set terms, decline without cost, and be believed if something goes wrong. Law shapes the conditions under which those are possible. It does not supply them, and no legislature in any of these three countries has found a way to legislate them into a room.

We are not going to tell you which country got it right, and it is worth saying why rather than treating the refusal as neutrality.

It is because the evidence does not support a conclusion, and because we are an interested party, and because the people with the most at stake do not speak with one voice — there are people in both countries who will tell you the law there saved them and people who will tell you it endangered them, and both are describing their own lives accurately.

What we will say is that the argument is conducted at the wrong altitude. Abolition versus legalisation is a question about the legal status of a category. The people inside the category are mostly asking about the conditions of a Tuesday evening: can I work near somebody else, can I say no, can I call the police, can I have a bank account.

Those are answerable questions with better and worse answers, and they cut across the models rather than following them. A chamber that reads the world as a mirror rather than a league table should say so, because this is the clearest case in the whole matrix of a debate where the moral framing has crowded out the practical one.

You do not owe tradition obedience. You do not owe modernity performance. You are allowed to discover what actually fits your life.

Three, and the first is the reason this essay exists in a chamber rather than in a policy journal.

Do not import either European model as a ready-made position for Japan. Japan’s arrangement is structurally different from both — a narrowly defined prohibited act with no penalty on the parties, beside a large licensed adjacent sector — so arguments built for a purchase ban or for a legalisation regime frequently do not attach to the Japanese structure at all.

Do not read the evidential stalemate as a reason for indifference. That the research cannot settle which model is better does not mean the conditions people work under are unknowable. Those are directly observable, and they are the thing to attend to.

And do not read this essay as being about Moonlight. It is not, and the disclosure at the top exists precisely so that a reader can discount our view where she thinks our interest is showing. Nothing here is legal advice, nothing here describes anybody’s conduct, and anybody making decisions about their own work in any jurisdiction needs qualified advice rather than an essay.

The legal chronology — the Swedish purchase ban of 1999, the German legalisation of 2002 and its 2017 reform, Japan’s Anti-Prostitution Law of 1956 and the 2024 statute on support for women facing difficulties — is stated from memory and must be verified against the statutes before this piece is marked fact-checked. The description of what Japanese law does and does not penalise is compressed and the definitional detail carries real weight; it should be read from the statute rather than from this summary.

Claimed outcomes on both sides are reported as claims, and deliberately so. Swedish official evaluations and German post-implementation reviews are produced by governments with a position, and independent research reaches opposing conclusions with unusual regularity.

No figures appear anywhere in this essay for the size of either market, the number of people selling, or trafficking prevalence. That is not caution for its own sake: those quantities are estimates with very wide ranges, the estimation methods are contested, and citing any of them would lend a precision the field does not have.

The account of what sex worker organisations prioritise is drawn from the public positions of such organisations and from research working with them. Those organisations do not speak for everybody selling sex, and people in the same trade hold opposed views about the law, which the essay says in the text.

The claim that enforcement culture matters more than statutory text is an interpretation supported by implementation research rather than a measured comparison.

And the disclosure at the head of this essay is a statement of interest, not a claim that the interest has been neutralised. A reader who concludes that our position is shaped by what we sell is making a reasonable inference and should weigh the essay accordingly.

Two countries, two opposite laws, twenty-five years, and no demonstrated answer.

That is a genuinely unusual outcome in social policy, and it should do something to a reader beyond frustration. When a question has been argued this hard, by serious people, with this much at stake, and the evidence still will not resolve it, the most likely explanation is that the question is not the one that determines the outcomes anybody cares about.

What determines those, in every account from the people actually inside it, is smaller and more local than a statute: whether she is alone or near somebody, whether she can decline, whether she will be believed, whether she can say what she will and will not do before anything begins and have that hold.

This chamber has arrived at that same short list from Spain, from Denmark, from Korea, from the Philippines and now from a policy comparison between two countries that agree on nothing. The consistency is the finding.

It is also, unavoidably, a description of what an evening with Moonlight is arranged to be, and we would rather point at that plainly than pretend the essay led there by accident: terms set in advance by the person receiving, a boundary that is part of the conversation rather than an interruption, and nothing inferred from a silence. Not because a law says so. Because that is what makes a room safe, in any country, under any statute.