The Time She Bought Back, and Where the Hours Came From

A woman gets her Saturday morning back. She has been trying to get it back for two years, and the way she finally gets it is that somebody else is in the house doing what she would otherwise be doing. That is the ordinary shape of the solution, in every country that has one, and there is nothing sinister in it. Work that has to be done by somebody is done by somebody who is paid for it, which is how almost everything else in an economy works.

This Library has written about that Saturday morning more than once. It has argued that time which belongs to a woman is not the residue left after everyone else has been served, and that only the fourth of the four kinds — the bounded, claimed, pre-arranged kind, during which the responsibility for whatever happens belongs to somebody else — actually restores anything. The argument still holds. What the argument does not do, in any of those essays, is finish the sentence. Responsibility belongs to somebody else. Who.

This essay finishes it. In the most common international version of the answer, the somebody else is a woman who has crossed a border to do the work, and she has children or parents of her own, and the care they need has been handed to a third woman who has not crossed anything and is very often not paid at all. Sociology named the arrangement a global care chain roughly a quarter of a century ago and has been arguing about it ever since.

Two things have to be said at the start, because the piece is unreadable without them. The first is that this is a claim about the structure of a solution and not an accusation against any woman who employs help. A person buying six hours of relief in a week is not the author of an international division of labour, and an essay that treats her as one has mistaken a moral posture for an analysis. The second is that the women at the far end of the chain are not raw material. They make decisions, they send money that changes households permanently, and a great many of them describe what they are doing as a thing they chose. An essay which can only see them as victims has performed its own erasure, in a more flattering register. Both of those are true at once, and this piece is going to hold them without collapsing into either.

The phrase comes from the sociologist Arlie Russell Hochschild, who used it around the turn of the century to describe a series of personal links between people across the globe, based on the paid or unpaid work of caring. The volume she edited with Barbara Ehrenreich in 2003, Global Woman, put the argument in front of a general readership: the energy that had been keeping households running in poorer countries was being redirected into households in richer ones, easing a care deficit at the receiving end by opening one at the sending end. The metaphor that stuck was extraction. Care was being drawn out of one place and consumed in another, like any other export.

The more precise version of the structure came from Rhacel Salazar Parreñas, whose 2000 article in Gender and Society and 2001 book Servants of Globalization set out what she called an international division of reproductive labour: a three-tier transfer running from middle-class women in receiving countries, to migrant domestic workers, to women in sending countries who are too poor to migrate. The third tier is the part that does the analytic work, and it is the part most often dropped when the idea is summarised.

The concept has been worked over hard since. Nicola Yeates, in a review in Global Networks in 2012, treated it sympathetically and then widened it: care chains are not only about domestic workers in private homes but about nurses and institutional care as well, and the framework risks reinscribing care as women’s work simply by studying it where women already do it. Yeates is worth reading precisely because she is not writing against the idea.

What the concept is, then, is a description of a structure. It is not a charge sheet. It says that an arrangement which looks like a private transaction between two people in one city is the visible segment of something longer, and that the rest of the line is where the interesting questions live.

The International Labour Organization estimated in 2021 that at least 75.6 million people aged fifteen and over were doing domestic work in or for private households. About three-quarters of them were women; men made up roughly a quarter of the sector. The figure that matters most for what follows is the informality rate: 81.2 per cent of them were working informally, which is to say without the labour and social protections that attach to employment elsewhere in the same economies.

The migration share is measured less recently and less well. The ILO’s global estimates published in 2015 put the number of domestic workers then counted at 67.1 million, of whom 11.5 million — 17.2 per cent — were international migrants, and about 73.4 per cent of those migrants were women. Those two estimates come from different exercises and should not be arithmetically combined; what they establish jointly is an order of magnitude and a sex ratio, not a trend line.

The money runs the other way and is measured much better, because banks count it. The World Bank estimated that remittances to low- and middle-income countries reached 685 billion dollars in 2024, a flow larger than foreign direct investment and official development assistance combined, and one that grew by 57 per cent over a decade in which foreign direct investment fell by 41. The Philippines was the fourth-largest recipient at about 40 billion dollars. In several small economies remittances are a quarter or more of national income.

On the legal side the record is thinner than the numbers. The Domestic Workers Convention, 2011, known as Convention No. 189, was the first international instrument written directly for this workforce. By June 2025, according to the ILO’s own material, forty states had ratified it and it was in force in thirty-eight. Forty is not many. The great majority of the world’s domestic workers live under national law that was not written with them in mind, and in a good number of countries was written to exclude them by name.

Here is the part of the structure that the statistics cannot reach, and the reason the essay exists. When a woman migrates to do care work, the care she was doing does not evaporate. It is absorbed. In the accounts gathered by the ethnographers, it is absorbed by a grandmother, an aunt, an older sister, a neighbour, a daughter of fourteen — and where it is paid, it is paid at a local rate that is a fraction of the wage being earned at the other end of the chain, and where it is not paid, it is simply added to a woman who was already doing a full day of something.

That is Parreñas’s third tier, and it is where the chain stops. It does not stop in a solution. It stops in a person who has nobody to hire. Every transfer along the line is a purchase except the last one, and the last one is a gift extracted by circumstance from whoever had the least room to refuse.

Set that against this Library’s own vocabulary and the shape of the problem becomes uncomfortably clear. Claimed time — the fourth kind, the only kind that restores anything — was defined here as time during which the responsibility for what might happen belongs to somebody else. That definition is correct and it is also a transfer. Responsibility does not dissolve when it is handed over; it lands. Most of the time it lands on a person who is paid for taking it, which is a decent arrangement and the basis of most of the caring economy. Sometimes it lands on a person three countries away who is not paid at all, and the woman at the near end of the chain has no way of knowing that and no mechanism for finding out.

It is worth being exact about what this does and does not license. It does not license the conclusion that buying help is wrong, which would be an absurd place to end up: the alternative on offer to most women is not a fairer arrangement but an unfairer one, in which the hours are taken out of her instead. It licenses a narrower and more awkward conclusion. When this house, or any other, describes a woman getting her hours back, it is describing one end of a transaction whose other end it has not looked at. That is the entry the ledger is missing.

Japan is directly implicated in all of this, and the implication is documented in its own official record rather than inferred. There are three distinct channels, they were built at different times for different reasons, and the differences between them are the argument.

The first is the economic partnership agreement route, which brings candidate care workers from Indonesia from fiscal 2008, the Philippines from fiscal 2009 and Vietnam from fiscal 2014, coordinated by a single designated body. A candidate works and trains in a Japanese facility and must pass the national care worker examination; passing converts the status into the residence status for care work, which permits a spouse and children to accompany and carries no limit on renewals. The Ministry of Health, Labour and Welfare recorded 3,252 people in Japan on this route as of 1 March 2025, of whom 452 had obtained the qualification.

The second is the specified skilled worker system, opened in April 2019. In the care field the number in Japan went from 19 at the end of December 2019 to 44,367 at the end of December 2024. The terms are the point: the first tier of that status is capped at a cumulative five years, and family accompaniment is, in the ministry’s own wording, not permitted in principle. A woman may spend five years in this country caring for other people’s parents and may not bring her own children with her while she does it. That is not a description of an informal pattern. It is the immigration rule, printed on the ministry’s own slide.

The third is the household-support scheme run through the national strategic special zones, legislated in July 2015 with intake from March 2017. Foreign household-support workers are employed by approved companies, not by the households they serve, under a third-party oversight council; six operators were running in Tokyo, Kanagawa, Aichi, Osaka and Hyogo with Chiba City to follow, and the cumulative number of workers received had passed one thousand nationally. The permitted period was extended from three years to five, and temporarily to eight for those whose careers were interrupted by the border closure. The government’s stated rationale for the scheme, in the Cabinet Secretariat’s own words, is the promotion of women’s active participation and the answering of household-support needs. The thesis of this essay is written into the rationale of the statute. One woman’s participation, purchased with another woman’s hours, under a licence issued for that purpose.

None of this happened because anybody wanted it. The demand is a counting problem, and Japan’s own statistics office states it plainly. The 2022 Employment Structure Survey found 6.29 million people providing care to a family member, of whom 58.0 per cent were also in work. The sexes separate immediately: 67.0 per cent of men who were caring held a job, against 52.7 per cent of women. The same survey found 9.65 million people raising children, of whom 85.2 per cent were in work — 99.0 per cent of the men and 73.4 per cent of the women. In the year to October 2022, 106,000 people left a job for care or nursing reasons.

The projection is worse than the present. Japan had about 2.15 million care workers in fiscal 2022. Aggregating the prefectural estimates made under the ninth round of municipal long-term care plans, the ministry puts the requirement at about 2.40 million by fiscal 2026 and about 2.72 million by fiscal 2040. That is an additional 570,000 people, in a country whose labour force is contracting, for work that is physically hard and has never been well paid.

This chamber has already found the domestic half of the same equation in Italy, where a family that shelters an adult into their thirties presents a second invoice twenty years later, and where the part of that bill which can be bought is bought through a privately employed home carer, a role filled largely by migrant women. That essay’s formulation is exact and this one adopts it: buying the care in does not remove the cost, it moves the cost along a line of gender and, frequently, of migration as well. What this essay adds is the length of that line and the fact that its far end is in another country.

It is important not to turn any of this into blame. A country whose old people outnumber its carers is not behaving badly by recruiting carers. A woman who cannot both hold her job and look after her mother is not failing at anything. What is available for criticism is narrower: the practice of describing the solution without describing the supply, and the design choices made about the people who constitute it.

Japan’s Labour Standards Act was enacted in 1947. Article 116, paragraph 2, says that the Act does not apply to domestic servants. Not that it applies differently, or applies with modifications: it does not apply. Working hours, rest, overtime, the whole architecture of the statute stops at the door of a private home. The Labour Contract Act, by contrast, contains no such exclusion, which produces the odd position that a domestic worker has contractual protection and no regulatory protection at all.

The ministry’s own study group on labour standards legislation, reporting in January 2025, sets out the history without much sympathy for the provision. The original justification was that domestic work looked nothing like work in the enumerated industries. The report observes that live-in service has largely disappeared, that a domestic worker today typically travels to a workplace and performs set duties for set hours like any other employee, and that a research report had already said in 1993 that the special reason for the exclusion was thinning. It recommends reviewing the blanket exclusion and beginning concrete design work early — and it states the difficulty honestly, which is that the employer would be a private individual and the state would be extending inspection into a private household.

There is a distinction here that matters more than it first appears. The exclusion bites the worker hired directly by a household. A worker employed by a service company and dispatched to households is that company’s employee and is covered by the Act in the ordinary way. Japan’s special-zone scheme was deliberately designed on the covered side of that line: the foreign worker’s employer is a licensed company, not the family whose kitchen she is standing in. Compare Hong Kong, where a foreign domestic helper has been required to live in the employer’s home since April 2003 and must ordinarily leave the territory within two weeks of a contract ending. On that axis Japan’s design is the more protective of the two, and the essay says so.

On the other axis it is not. This chamber’s essay on Canada developed a mechanism it is worth reusing here: a right has a content and a price, and a right you can only exercise by leaving is a different thing from a right. A five-year cap with no family accompaniment sets a price on every ordinary decision a woman might make — to stay, to have a child here, to complain about an employer, to move. The Japanese arrangement protects her at work more carefully than several of its neighbours do, and gives her less standing to build a life around the protection. Both halves are true and neither one cancels the other.

The strongest case against this essay is that its central metaphor is wrong, and the case deserves to be made at full strength rather than conceded politely in a final paragraph. Extraction is a word for what is done to a resource. Applied to people, it removes the one thing the ethnographic record most consistently shows: that these are women making decisions, often shrewd ones, inside a set of options they did not design. Nicole Constable’s long fieldwork in Hong Kong documents discipline and constraint and also organisation, resistance, friendship, ambivalence and lives that are not reducible to their worst hours. A frame in which every migrant care worker is a casualty is not a more compassionate frame. It is a less accurate one, and it takes something from her that the labour market has not yet managed to.

The money is not sentiment either. The cleanest causal evidence available comes from Dean Yang’s study in the Economic Journal in 2008, which used the fact that overseas Filipino workers were scattered across dozens of currencies when the 1997 Asian crisis hit, so that some households received a windfall and others did not for reasons nothing to do with themselves. A favourable exchange-rate shock raised remittances with an elasticity of about 0.60, and in the receiving households child schooling and educational spending rose, child labour fell, and families became more likely to start capital-intensive enterprises. Whatever else the chain is, it is also a mechanism by which a household acquires a floor it did not have.

And the zero-sum picture has been challenged on its own theoretical ground. Loretta Baldassar and Laura Merla proposed care circulation in place of care drain: a reciprocal, multidirectional and asymmetrical exchange of care that moves in more than one direction across a transnational family and changes over a life course. Grandparents care for children and are later cared for by them; money, advice, presence and obligation all travel; the family is not a vessel being emptied. Yeates’s warning belongs here too, that studying care as a chain of women can quietly confirm that care is women’s business.

Against all of which stands the largest synthesis of the evidence on the children. Fellmeth and colleagues, in the Lancet in 2018, pooled 111 studies covering 264,967 children and found that children left behind by a migrating parent carried a higher risk of depression, with a relative risk of 1.52, of anxiety at 1.85, and of suicidal ideation at 1.70, along with smaller but measurable elevations in wasting, stunting and substance use. The authors state their own limits with unusual clarity: 43 per cent of the studies were at high or unclear risk of bias across five or more domains, statistical heterogeneity was very high, and most of the studies were cross-sectional, so the direction of causation cannot be established from them.

The essay refuses to resolve these into a verdict, and the refusal is not a failure of nerve. Parreñas’s own finding, which is the most useful thing in the dispute, is that a great deal of the strain inside these households is carried by gender expectation rather than by distance: a father’s absence is read as provision and a mother’s absence is read as abandonment, by neighbours, by relatives and eventually by the children themselves. That points at something correctable in a way that the migration itself is not. The honest position is the uncomfortable one. An arrangement can be the best option a woman has and still be an arrangement that produced her options. Naming the structure is not the same as speaking for the people standing inside it, and the second is what this essay has no standing to do.

This house sells hours to women who are buying time back, which places it inside the arrangement this essay has just described rather than anywhere above it. The product is an evening, its terms fixed in advance, during which the responsibility for what happens belongs to somebody who is being paid to hold it. That is the fourth kind of time, sold by the unit. It is the same transaction as the Saturday morning at the top of this piece, with a different service on the other side of it.

The uncomfortable half goes first, before anything is said in the house’s favour. The value of the product comes from the scarcity of the thing it supplies. A woman who has learned that claimed time is nearly impossible to obtain in her own week is a woman for whom a bounded, guaranteed, defended block of it has an obvious price, and the price is set by the shortage. The house benefits from a care settlement it did not build and has no power to change, and it should say that in plain words rather than describing itself as a contribution to anybody’s liberation.

The second obligation follows from the essay itself. A house that publishes an argument about the provenance of hours is required to know the provenance of the hours it sells. That means knowing who provides them, on what contract, with what protection, with what freedom to stop; it means not being the kind of buyer that treats the other end of its own transaction as somebody else’s question. This is stated here as an obligation the house has taken on in public, which is the only form in which such a statement is worth anything.

What it may not claim is a longer list. It may not claim that buying a bounded evening is a position on care policy, or migration policy, or anything else. It may not claim that the person who provides the evening is thereby in a good arrangement, because that depends on the contract and not on the sentiment. It may not claim to have located, for any reader, the place her own recovered hours came from; that chain is not visible from inside a single transaction and this essay has not made it visible. And it may not claim that having named a transfer is the same as having done anything about it, which is the plain limit of a piece of writing.

It does not claim to have counted a single chain. Nobody has. No statistical office in any country records who took over the care when a domestic worker left, because the person who took over is usually a relative doing it for nothing, and unpaid substitution inside a household leaves no administrative trace. The three-tier structure is a finding from qualitative research, replicated widely and challenged seriously, and this essay treats it as a description of a mechanism rather than as a measured proportion of anything.

It does not narrate any worker, and the omission is deliberate rather than stylistic. An essay that opened with a woman at an airport would be more readable and would be doing exactly what the argument objects to, which is using a person at the far end of the chain as material for a reader at the near end. The figures here are national aggregates and the mechanisms are drawn from published research; no household, employer or worker appears.

It does not claim that its figures are current or commensurable. The 75.6 million estimate, the 2015 migration share, the Japanese route counts and the Japanese projections were each produced by different bodies in different years with different definitions, and the essay has said the date of every one rather than presenting them as a single picture. The special-zone headcount of more than one thousand comes from a Cabinet Secretariat document whose utilisation charts end before the border closure, so the current number is not known here and is not asserted. Japan’s status with respect to Convention No. 189 was checked through secondary Japanese sources rather than directly against the ILO ratification register, which declined machine access, and the essay therefore rests only on the ratification total the ILO itself publishes.

It does not claim that any scholar named here would accept this reading, and it does not endorse any of them as a body of work: Hochschild, Parreñas, Yeates, Constable, Baldassar, Merla, Yang and the Lancet authors are cited for specific findings and specific disagreements, several of which are with each other. It takes no position on whether any country should admit more care workers or fewer, which is a question of politics this Library has no standing in. And it does not claim that the woman who gets her Saturday morning back owes anybody an apology for it. She does not. The ledger entry is not addressed to her; it is addressed to anyone who writes about her hours as though they began when she took them.