Norway: The Weeks That Cannot Be Given Away

Of everything the Nordic countries built, one instrument travels further than the rest. A ministry that will not copy a childcare system, and cannot copy a wage structure, will copy a quota — because a quota is a number of weeks, and a number of weeks can be written into a bill in an afternoon.

Norway wrote the first one. On the first of April 1993 it took four weeks out of the household’s paid parental leave and put the father’s name on them. Not offered to him. Not encouraged, not promoted, not the subject of a campaign. Reserved: if he did not use those weeks, the mother did not get them either. They simply left the household.

What happened next is why the policy is famous. Before the reform, the share of eligible fathers taking any of the leave was a rounding error. Within four years it was past seven in ten — produced by a design property rather than by generosity, exhortation, or any shift in what Norwegian men believed about themselves.

This essay is about the second half of that sentence, which is quoted far less often. The quota moved leave-taking about as sharply as social policy moves anything. What it moved in the rest of the house — the cooking, the laundry, the appointments, the remembering of which child needs which thing on which day — is a smaller, slower and genuinely contested story. The chamber’s interest is in why, and in what a country that copies the number would not be copying.

The mechanism is worth stating precisely, because most misreadings of the Norwegian case come from getting it slightly wrong. Norwegian parental benefit is a household total — today around forty-nine weeks at the full-coverage option, or a longer period at a reduced rate — cut into three pieces: a block reserved to the mother, a block reserved to the father, and a shared middle the parents divide as they like. Since 2019 the reserved blocks have been fifteen weeks each at the full-coverage option. The benefit is earnings-related up to a ceiling, which matters and will come back.

The quota is the father’s block. It is not extra leave, and it is not a permission. It is a carve-out from a pot the household would otherwise have divided, and the operative clause is that it cannot be moved. A father who declines his weeks does not hand them to the mother. Nobody gets them.

That is the whole trick, and it is why this chamber’s essay on Denmark could state the rule in a line: leave that cannot be transferred is taken, and leave that can be transferred is transferred to the mother. Norway is where that line was first tested. Much of the European Union later adopted a version of the same carve-out, and the 2019 work-life balance directive wrote non-transferability into law for the bloc.

Notice what the design assumes. It does not assume that fathers want less time with their children than mothers do. It assumes that when a household with a newborn is handed a divisible resource and a set of unequal pressures — his employer, her employer, the feeding, the recovery, the relative earnings — the resource will flow along the pressures. Reservation does not persuade the household. It removes the option of flowing.

In 1992, the year before the reform, the recorded share of eligible fathers taking paid parental leave was around two and a half in a hundred. By 1997 it was above seventy in a hundred. The population did not change. The culture did not change in four years. The men were the same men, in the same firms, with the same mothers-in-law. What changed was that a block of weeks had been made unusable by anyone else.

The series has kept climbing, more slowly, since. Administrative records put the share of fathers drawing at least one day of parental benefit at ninety-two in a hundred by 2021, and Statistics Norway records the share using the full quota or more rising from fifty-nine in a hundred in 2013 to sixty-eight in a hundred in 2024.

Two things in that series belong together. The first is that it is a genuine behavioural result and should not be talked down. The second is that a very large part of it clusters exactly on the quota: the OECD’s 2025 review of paid leave for fathers records that in Norway around ninety fathers in a hundred took leave in 2022 and around seventy in a hundred took precisely the length of the reserved block — not more, not less. The quota is not a floor from which men climb. For most of them it is the whole of it.

Then Norway ran the experiment backwards, without meaning to. The quota had been expanded repeatedly — four weeks in 1993, five in 2005, six in 2006, ten in 2009, twelve in 2011, fourteen in 2013 — and in July 2014 a new government cut it to ten, intending a first step towards removing the reservation altogether. Fathers’ leave-taking fell. Not their participation, which by then was ordinary, but the quantity: the weeks released into the shared middle went where transferable weeks go. By 2018 the same government had reversed itself and set the quota at fifteen.

That episode disposes of the most common objection — that Nordic fathers take leave because Nordic fathers are the kind of men who take leave, and the quota is a label on a change that had already happened. If that were true, cutting it would have done nothing. Twenty-one years on, in the country with the longest history of involved fatherhood in Europe, moving the line moved the men. It cuts the other way too: a culture that has practised this for two decades and still needs the line is not being taught anything durable by it. Something is being held in place — a real achievement, and a limited one.

The quota is usually discussed as a gift to fathers. The research record says it is also a bill, and that is the part that makes the behavioural result credible.

Rege and Solli, publishing in Demography in 2013, used the 1993 reform as the shock and followed fathers through Norwegian registry data. Four weeks of leave in the child’s first year reduced a father’s later earnings by about one and a half per cent in the intention-to-treat estimate, and by a little over two per cent once adjusted for who actually took it — an effect still visible when the child was five.

Two per cent is small. It is also not nothing, and its shape is the interesting part: durable, specific to him, and arriving through the ordinary machinery of a career rather than through anybody’s disapproval. The state did not make fathers’ leave free. It made it cheap enough, and unavoidable enough in the household sense, that a man could pay the bill without deciding to. A real but survivable cost, imposed by a rule rather than chosen against a spouse, is a different machine from a generous entitlement he has to go and ask for.

Now the counter-thesis, which is why this is an essay and not a note recommending quotas. The quota moved leave-taking sharply. It moved the division of domestic labour beyond the leave period far less, and the literature on how much is genuinely mixed. Saying so is not a hedge; it is the finding.

On one side, Kotsadam and Finseraas, in Social Science Research in 2011, compared households whose last child arrived just before and just after the 1993 reform and found effects still present more than a decade later: about eleven per cent less reported conflict over the division of housework, and couples about half again as likely to share the laundry equally. That is durable, and it is on exactly the terrain the policy hoped to reach. But note what it is — one task, and the temperature of the argument. The same work found no matching effect on cooking or cleaning.

On the other, Cools, Fiva and Kirkebøen, in the Scandinavian Journal of Economics in 2015, found what the quota did not do. It greatly increased fathers’ leave-taking, and improved children’s school performance in some families. It produced no evidence of countering the traditional allocation of the parents’ labour supply. Who reduces hours, whose career absorbs the shock: that looked much as it had.

The long time-use record is a third kind of answer. Norway’s time use survey shows the daily gap in household work between women and men narrowing from around two hours twenty minutes in 1980 to about forty-four minutes in 2022. That is an enormous social change. It is also a forty-year trend running through a childcare expansion, a rise in women’s employment and four different quota lengths, and no honest reading gives the quota the credit. The researchers who assembled it have argued the policy sent a signal that worked on norms; they have not been able to isolate it, and they say so.

The most useful sentence on this is the OECD’s own, in its 2025 review: the extent to which the increased use of fathers’ leave has changed the sharing of care activities in households is an issue for future research. That is an international body, in the year of publication, declining to claim the thing the policy is most often sold on. The same review records that even where leave uptake is close to equally shared — Iceland, Norway, Portugal, Sweden — fathers account for only around a fifth to a third of the total leave days taken.

If the instrument is portable and the result is not, the difference must lie in what the instrument was standing on. Norway supplies four things that do not travel with a bill.

The first is a compressed wage structure. Norwegian wage dispersion is small by international standards, which flattens the arithmetic a couple does when deciding whose weeks to take. Where one partner earns three times the other, a leave decision is not a values question; it is a budget. The second is the benefit ceiling: parental benefit replaces earnings only up to a cap, which truncates the cost for most households and, quietly, makes the quota most expensive for the highest earners.

The third is a childcare system underneath the leave. The chamber’s Denmark essay made the general point: when care is a utility rather than a private negotiation, a woman’s decision about work stops being a referendum on her character. A quota with nowhere for the child to go at the end of it is a bridge to a cliff.

The fourth is the least comfortable. Norway’s labour market is strongly segregated by sector and occupation. Something like two-thirds of employed women work in the public sector — health, care, education — and women’s monthly earnings sat around twelve per cent below men’s in 2022, a gap that stopped closing in the 1990s and has moved slowly since. Part of what makes a Norwegian father’s weeks survivable is that a large share of Norwegian mothers work in employment structures built to absorb exactly this. The quota did not dismantle that arrangement; it operated inside it.

A country can legislate the weeks. It cannot legislate the wage compression, the ceiling, the childcare places or the sectoral structure that make the weeks cost what they cost. The most-copied part of the Nordic settlement may be the least transferable, precisely because it is the part that looks like it stands alone.

Turn the mirror. The comparison usually drawn — generous Norway, ungenerous Japan — is factually backwards, and the correction is where the useful finding sits. On paper, Japan has the most generous father-specific paid leave entitlement in the OECD. The OECD expresses these in full-rate-equivalent weeks — the length the leave would run to if paid at a father’s full earnings — and on that measure Japan’s comes to around thirty-one weeks, the longest of any member country, against an OECD average near eight. It is an individual, non-transferable right held by each parent separately.

And for a long time almost nobody used it. Then the number began to move. The ministry’s annual employment-management survey put men’s take-up at around seventeen in a hundred in the 2022 financial year, thirty in a hundred in 2023, and forty and a half in a hundred in 2024 — published at the end of July 2025, and the first time it passed forty. Women’s take-up in the same survey was around eighty-seven in a hundred. The stated government targets are half by 2025 and eighty-five in a hundred by 2030.

The causes are legible. A post-birth leave usable in the first eight weeks and splittable into two blocks was introduced in October 2022 and is now taken by the large majority of Japanese men who take any leave at all. From April 2023 employers above a thousand staff have had to publish their male take-up annually; from April 2025 that obligation reached employers above three hundred. And from April 2025 a new supplement raises the replacement rate for up to twenty-eight days, from roughly two-thirds of pre-leave wage to roughly four-fifths, which — the benefit being untaxed and social-insurance contributions waived during leave — comes to approximately full take-home pay. Crucially, it is conditional: it is paid when both parents take leave.

Two qualifications belong here and the essay will not bury them. The first is duration. In the 2023 survey, of the men who took leave, around sixteen in a hundred took under five days and a further twenty-two in a hundred took between five days and two weeks. A headline take-up figure and a fortnight are not the same object, and the Norwegian series is measured in months. The second is what the leave costs him: the ministry’s commissioned workplace-harassment survey for the 2023 financial year found that roughly one in four male employees who tried to use a childcare-related measure reported harassment in connection with it.

All of the Japanese position here is recent, some of it months old, and the 2025 measures have produced no outcome series yet. Where it moves, it should be re-checked rather than quoted from here.

The finding is not that Japan should adopt a quota, nor that Japanese entitlements are stingy. The difference between the two systems is not length, and not quite transferability either — Japan’s leave is already an individual right a father cannot hand to a mother. The difference is which direction the household’s total moves when he declines.

In Norway, the father’s weeks are carved out of a household pot. If he does not take them, the household is shorter by fifteen weeks. The loss is immediate, denominated in the thing the family most wants, and it lands on both parents at once. Nobody has to be persuaded of anything; there is simply a hole where the weeks were.

In Japan, the father’s leave is added to the household rather than subtracted from it. Each parent holds a right up to the child’s first birthday, and the joint-leave provision lengthens the household’s total to fourteen months when both take leave. A father who takes nothing does not shorten his wife’s year; the household loses an option it never counted on. Structurally, Norway punishes the absence and Japan rewards the presence — and an unclaimed reward is far easier not to notice than a confiscated fifteen weeks.

That is exactly what the April 2025 supplement changed, and it is the first Japanese instrument that behaves like a quota: because the higher rate is paid only when both parents take leave, a father’s refusal now costs the household money it would otherwise have had. A household-level loss, which is the right shape. It is also denominated in money rather than weeks, and it runs for twenty-eight days rather than fifteen weeks. Whether an instrument that small produces a behavioural break of the Norwegian kind is not yet knowable, and this essay does not claim to know it.

Then the second half of the comparison: what he is risking. A Norwegian father taking his quota risks roughly two per cent of his later earnings, imposed impersonally by a career that moved on without him. A Japanese father taking leave in a firm where tenure, presence and internal standing do much of the work of a career is risking something with no clean number attached — and one in four of the men who tried reported being treated badly for it. Those are not the same decision, and no quota law converts one into the other. This is what the chamber means when it says the mechanism travels and the model does not.

Three readings are available and the essay declines all of them. The first is that Norway solved this and the rest of us are behind. It solved the allocation of a defined block of weeks, unusually well, and the record on what happened in the house afterwards is mixed at best. A reader who takes from Norway that equality at home is a legislative problem awaiting the right bill has learned the wrong thing from the case that most nearly disproves it.

The second is the mirror-image cynicism: that the quota is theatre, that the men take their weeks and hand the child to the grandmother, and that nothing underneath moved. The evidence does not say that either. Something changed, in the temperature of the argument and in at least one task, and it is still detectable more than a decade on. A change smaller than advertised is not a fake one. The third is any reading in which a birth rate is the scoreboard. This Library does not evaluate a family policy by what it does to a demographic series, and will not start with a country whose quota is routinely argued about in those terms by people arguing about something else.

One narrower refusal. Nothing here is a verdict on a man who did not take leave. The point of the Norwegian record is that behaviour at this scale is produced by the structure of an option, not by the quality of the person choosing, and a reader who converts it into a reproach has inverted it.

Against its own interest, then, the house. This house sells hours in which a woman is not managing anything — in which the planning, the noticing and the remembering are somebody else’s job for the length of an evening. That is a product whose market is made of precisely the imbalance this essay has described. Which is why it cannot pretend to be a remedy. The chamber’s essay on Italy made the point that a family arrangement is underwritten by somebody’s unpaid hours and that a time-use table is where you read whose. An evening in which the load is lifted is not a redistribution of the load; it is a few hours in which somebody else is holding it, after which it comes back exactly where it was.

What such an evening can honestly offer is a measurement. A woman who has spent years as the person who remembers things sometimes cannot say how much of her attention that occupies, because there has been no interval in which it stopped. An evening where it stops is a reading of what the load weighs — information she can take back into a conversation with the person she lives with. Nothing here substitutes for that conversation, nothing here teaches anybody to carry the load better, and nothing here is advocacy for or against any country’s leave policy. This chamber compares mechanisms. It does not campaign.

This essay does not claim that the father’s quota caused the narrowing of the Norwegian housework gap. The time-use series runs across four decades and several other reforms, the researchers who assembled it have been explicit that the policy cannot be isolated within it, and the OECD said in 2025 that the question remains open.

It does not claim that the mixed findings mean the quota failed. Kotsadam and Finseraas found a durable effect on conflict and on one task; Cools, Fiva and Kirkebøen found none on the parents’ labour supply. Both are good work on the same country, answering different questions. Mixed means mixed, and a reader who collapses it in either direction is doing something this essay has tried not to do.

It does not claim to know what the Japanese take-up series will do next. The figures cited run to the 2024 financial year and were published in July 2025; the supplement conditioning the higher benefit rate on both parents taking leave began in April 2025 and has no outcome series; the publication duty reached smaller employers in the same month. This is a moving position and it is stated as one.

Nor does it claim that a harassment survey measures how common an experience is across all workplaces; it measures what a sample of people who tried to use a measure reported about it.

And it does not rank the two countries. Norway built a small, hard, well-designed instrument on top of a wage structure and a care system it did not build for this purpose. Japan wrote a long entitlement and has spent a decade and a half discovering that length was never the binding constraint. The useful comparison is the mechanism, not the flag.