The Two-Thirds Country

Essay · economics · September 2026

For thirty years an hour of work in Portugal has produced about two-thirds of what the average European hour produces, and the Portuguese work more of those hours than anyone else in Western Europe except the Greeks. This essay takes Keynes’s question from America to Europe and then home: what the productivity bought, how much of it was taken as time, who got the income, what the crisis took away, and why the rent now eats the raise.

Reading time ≈ 24 min Revolution 1974 → euro 1999 → troika 2011 A companion to “The Fifteen-Hour Week”

From America to Europe

The last essay on this blog asked what happened to the fifteen-hour week that Keynes promised his grandchildren in 1930. The answer, mostly from American data, was that productivity rose about sevenfold, as he said it would; that rich countries took about a fifth of the gain as shorter hours and the rest as income; that in the United States, since about 1980, the typical worker saw little of that income, which went disproportionately to the top; and that the machines never produced the mass unemployment people feared. Europe was mentioned as the place that took more of the time. This essay looks at Europe properly, and at the one European country this blog is written from.

Portugal is an unusually good test of Keynes’s argument, because almost everything in it happened late and fast. In 1974 it was the poorest country in Western Europe, a dictatorship fighting three colonial wars, with a large share of its workforce in farming and a large share of its young men abroad. Within a generation it was a democracy, a member of the European Community and then of the euro. It caught up with the rich countries at a speed few have matched, stopped, fell back in a crisis that sent close to half a million people abroad in four years, and recovered. Each of Keynes’s questions, about productivity, about hours, about who gets the income and about jobs, has a distinct Portuguese answer, and none of them is the American one.

68Portugal’s output per hour, EU = 100, in 1995 and again in 2024

+438 hhours a year the average Portuguese worker puts in beyond a German one

485,000people who left Portugal in 2011–2014, counting temporary emigrants

×2.8house prices since 2013, against ×1.7 for median income since 2015

The catch-up

Two-thirds, and stuck there

Start with the number Keynes cared about, income per head. Figure 1 shows Portugal’s GDP per person as a share of Germany’s, France’s, Spain’s and America’s since 1950.[1] In 1960 a Portuguese produced about 43 per cent of what a German did. The late dictatorship was a period of fast growth, driven by industrialisation, emigrants’ remittances and the opening to European trade, and by 1974 the figure was 63 per cent. The revolution, decolonisation, the return of half a million or more settlers from Africa and two IMF programmes knocked it back to 57 per cent by 1986. Membership of the European Community, with its structural funds and its market, carried it to almost 70 per cent by 1999. Then it stopped. The first decade of the euro was a decade of growth under one per cent a year, the crisis of 2011 to 2014 pushed Portugal back below 60 per cent of Germany, and the recovery since has brought it back to 69. Against Spain the story is flatter and friendlier: from three-quarters in 1960 to nine-tenths now.

Fig. 1 — Portugal’s real GDP per head as a percentage of four other countries’, 1950–2024, in 2020 purchasing-power dollars. Shaded: the revolution of 1974–75, EC accession in 1986, the euro in 1999 and the troika programme of 2011–14. Data: Bergeaud, Cette & Lecat.

Income per head depends on how many people work and how long, as well as on what an hour produces. The purer measure is output per hour, and here the picture is starker. The European Commission’s AMECO database puts Portuguese GDP per hour worked, at purchasing-power parity, at 68 per cent of the EU average in 1995 and at 68 per cent in 2024.[2] Thirty years, one currency, one crisis and several governments later, it is exactly where it was. Spain slid from above the average to 96. Italy fell from 131 to 99, the most striking decline of any large economy. Greece fell from 73 to 54. On the Bergeaud–Cette–Lecat figures, a Portuguese hour of work in 2024 produced what a German hour produced in 1990.

New cars waiting for export at the port of Setúbal, 2019. The Portugal of high productivity exists, in car plants, in software, in parts of tourism and agriculture; the average is pulled down by the much larger number of very small, low-capital firms. Photograph by Diego Delso, CC BY-SA 4.0, via Wikimedia Commons.

Why the catch-up stopped is the central question of Portuguese economics, and the best-known answers are about what happened after the euro. Olivier Blanchard described the early 2000s as a boom financed by cheap foreign borrowing that pushed wages and prices up faster than productivity and left the economy uncompetitive and slow.[3] Ricardo Reis argued that the borrowing went disproportionately into the sheltered, non-traded part of the economy, construction, retail, services for the home market, and within it to firms that were not the most productive, because the banks that channelled it were better at lending against property than at picking winners.[4] Gita Gopinath and her co-authors found the same pattern across southern Europe: when capital became cheap it flowed to firms with collateral rather than to firms with the best returns, and measured productivity fell.[5] Behind those stories sit older ones that the OECD’s reviews of Portugal return to every few years: an economy of very small firms, low qualifications among managers as well as workers by European standards, slow courts, and a stock of capital per worker well below the European average.[6] None of this is fate. But it is why the two-thirds has proved so sticky.

The hours

Working long, producing little

Now Keynes’s second question. If Portugal had taken its productivity growth as time the way Germany or France did, its workers would be working much less than they did in 1970. They are not. Figure 2 puts every European country on one chart: hours worked per worker in 2024 against GDP per hour.

Fig. 2 — Annual hours per person employed against GDP per hour worked (PPS, EU27 = 100), 2024, twenty-nine European countries. Dashed: log-linear fit. Ireland and Luxembourg, whose GDP is inflated by multinational accounting, are left out. Data: European Commission, AMECO.

The relationship is one of the tightest in cross-country economics: the correlation between the logarithms is −0.85. The countries where an hour produces most, Norway, Denmark, the Netherlands, Germany, Switzerland, are the countries where people work least. The countries where an hour produces least, Greece, Poland, Bulgaria, Croatia, Portugal, are where people work most. It is tempting to read causation in both directions, and there is some: tired people are less productive per hour, and rich countries can afford to work less. But the main reading is simpler and it is Keynes’s. Rich countries became rich first and then took part of it as time. Poorer ones have less to take, and they take less of it.

Figure 3 shows how the two groups got where they are. In 1970 Portuguese, Spanish, French, Italian and German workers all put in about two thousand hours a year. Germany has since cut 32 per cent, Denmark 26, France 24, Spain 20 and Italy 16. Portugal cut 10 per cent, and nearly all of it before the late 1980s: its hours have been flat for thirty-five years.[7] The OECD puts the average Portuguese worker at 1,772 hours a year in 2024, 438 more than the average German, which is eleven forty-hour weeks. The European Commission’s own series puts the gap higher still. Labour-force surveys, which ask people directly, give the same ordering: people in work in Portugal usually work 39.7 hours a week against 37.0 across the EU, and the legal week in the private sector is still the forty hours set in 1996.[8]

Fig. 3 — Annual hours actually worked per worker, 1970–2024. The dip in 2020 is the pandemic. Portuguese hours are uncertain before 1990: the OECD series shown here falls, while the European Commission’s and the Penn World Table’s rise from implausibly low levels. Data: OECD.

The Keynes arithmetic for Portugal comes out like this. Output per worker has risen about two and a half times since 1974. At today’s productivity, the output of an average Portuguese worker in the year of the revolution could be produced in about thirteen hours a week, averaged over the year, against the thirty-four actually worked on the same basis. Portugal has the fifteen-hour week available at roughly the living standard of 1974, a year most Portuguese over sixty remember as poor. That is the difference between Portugal and America. In the United States the fifteen-hour week buys the living standard of 1977, a comfortable one; in Portugal it buys a hard one. The Portuguese have not taken their productivity as time because there has been too little of it to take.

Who got the income

No American divergence, but a revolution and two relapses

The American story in the last essay was a divergence: after 1979, output per hour kept rising and the typical wage did not. Portugal’s story is different in shape. Figure 4 shows the adjusted wage share, the part of national income paid as wages and salaries, including an imputed wage for the self-employed.

Fig. 4 — Adjusted wage share, % of GDP at factor cost, 1960–2025 (Germany from 1991). Shaded: the revolution, the IMF programmes of 1977–78 and 1983–85, and the troika programme of 2011–14. Data: European Commission, AMECO.

Nothing in Europe looks like the Portuguese line of the 1970s. In 1973 wages took about 71 per cent of national income. After the revolution, with wage rises decreed, strikes won and firms nationalised, they took 88 per cent in 1975. It could not last, and it did not. Two IMF programmes, devaluations and inflation above twenty per cent brought the share down to 56 per cent by 1986, below where it had been under the dictatorship. Real pay per worker in 1984, at the bottom of the second programme, was lower than in 1975.[2] Since then the share has moved within a narrow band, from 60 per cent in the late 1990s to a low of 53 in 2014 after the troika, and back to 57, which is about the European average.

Over the long run, Portuguese pay has kept up with Portuguese productivity. Figure 5 compares them since 1995. Real compensation per worker, deflated by consumer prices, rose by 34 per cent between 1995 and 2024; output per worker by 31 per cent and output per hour by 30. There is no American gap. What there is instead is a cycle. During the troika years real pay fell by about 8 per cent while output per worker held up; since 2014 it has risen by 21 per cent while output per hour rose by 3. The recent rise in Portuguese wages is real, but it is mostly a recovery of what the crisis took, paid for partly by a rise in the wage share, not by new productivity.

Fig. 5 — Portugal, index 1995 = 100: real GDP per person employed, real GDP per hour worked (dashed), and real compensation per employee deflated by the private-consumption deflator. Data: European Commission, AMECO.

The rest of southern Europe shows the problem the other way round. Spanish real pay per worker rose only 5 per cent between 1995 and 2024 while output per worker rose 12 and output per hour 18: that is a small American-style gap. Italian real pay fell by 6 per cent over the same thirty years, while productivity barely moved. Greek real pay in 2024 was a quarter below its level of 2008. For much of southern Europe, then, the answer to “did incomes plateau?” is yes, but not for the American reason. It is not that productivity rose and the gains went to the top. It is that productivity stopped rising, and there was little to share.

The level, not the share, is Portugal’s problem. The OECD’s average annual wage in Portugal, at constant prices and purchasing power, was lower in 2015 than in 2000; it has risen since, to about 45,000 dollars in 2025, against 58,000 in Spain, 60,000 in France and 76,000 in Germany.[9] Eurostat’s survey of earnings puts the Portuguese median hourly wage, adjusted for prices, at half the EU median.[10]

The minimum and the middle

When the minimum becomes the wage

The most distinctive thing about Portuguese pay in the last decade is the minimum wage. It was created in 1974, at 3,300 escudos a month, one of the revolution’s first acts, and at the time it was very high relative to what most people earned: the OECD puts it at 70 per cent of the median full-time wage in 1975.[11] Inflation and the IMF programmes eroded it. In 2025 euros it fell from about 700 a month in 1975 to under 460 in 1985, and it did not recover its 1975 value until the end of the 2010s. It was frozen at 485 nominal euros from 2011 to 2014 under the troika. Since 2015 it has risen every year, to 870 euros in 2025 and 920 in 2026, and in real terms it rose by about 40 per cent in the decade to 2025, while the average wage rose by 23.[12]

Fig. 6 — Statutory minimum wage as a percentage of the median wage of full-time employees, 1975–2025 (Germany from 2015). Data: OECD.

Figure 6 shows the result in relative terms. The Portuguese minimum is now close to 60 per cent of the median full-time wage on the OECD’s measure, and higher on Eurostat’s, which puts it at 71 per cent of the median in 2023, the highest ratio of any country in the comparison.[10] About a quarter of full-time private-sector employees were paid exactly the minimum in 2016–2019, according to the ministry’s own figures.[13] The Banco de Portugal estimates that in 2025 the minimum was about 90 per cent of the median base wage, and that the ratio between the base wage at the ninetieth percentile and the tenth fell from 3.4 in 2010 to 2.4.[14] The bottom of the distribution rose to meet the middle. The middle stayed where it was.

This is a real achievement and a real warning. The achievement is that the lowest-paid Portuguese workers are much better off than a decade ago, and that research on Portugal’s earlier increases found smaller job losses than simple theory predicts.[15] The warning is that a minimum wage can raise the floor but cannot raise the average: that takes productivity. A country where the minimum is becoming the typical wage is a country whose skilled workers, nurses, engineers, teachers, see little reward for their skill at home, and some of them draw the obvious conclusion.

Jobs and the exit

The crisis and the people who left

On Keynes’s third question, technological unemployment, Portugal’s record supports the conclusion of the last essay: unemployment has been driven by macroeconomics, not machines. Figure 7 shows it since 1970. It was about two per cent in 1974, rose above eleven in the mid-1980s during the second IMF programme, fell to five by 2000, and then climbed through the stagnant euro decade to 17.2 per cent in 2013, when more than a third of Portuguese aged fifteen to twenty-four who wanted a job could not find one.[16] Spain and Greece went higher still, above twenty-five per cent, with youth unemployment near sixty. None of those peaks had anything to do with robots. They were caused by a debt crisis inside a currency union that removed the usual ways out, devaluation and independent monetary policy, and left only the slow one: cutting wages and waiting.[17]

Fig. 7 — Unemployment rate, % of the labour force, 1970–2025 (Germany from 1991). Data: European Commission, AMECO.

Praça de Espanha, Lisbon, 15 September 2012, at one of the largest demonstrations in Portugal since 1974, under the slogan “Que se lixe a troika! Queremos as nossas vidas!”: to hell with the troika, we want our lives back. Photograph by Pedro Ribeiro Simões, CC BY 2.0, via Wikimedia Commons.

The Portuguese response to unemployment was the one it has used for centuries: people left. The official statistics count 199,000 permanent emigrants between 2011 and 2014, shown in Figure 8. Counting those who left for less than a year, the national statistics office’s total for the same four years is 485,000, in a country of ten and a half million; the peak, 135,000 in 2014, was the highest since the great emigration of the 1960s and early 1970s.[18] The United Nations counts about 1.8 million people born in Portugal living abroad, around a sixth of the resident population, one of the highest proportions in Europe.

Fig. 8 — Permanent emigrants from Portugal per year, 2008–2024 (Eurostat definition: leaving for at least a year). Pink: the troika years. Temporary emigrants roughly double the totals. Data: Eurostat, migr_emi2.

Emigrants waiting outside a shipping agency in Porto, about 1905. Photograph by Charles Chusseau-Flaviens; public domain, via Wikimedia Commons.

Emigration shows up in the unemployment rate as a success, because people who leave stop being counted. It is not one. The emigrants of the 2010s were younger and more educated than those of the 1960s, and each one took with them the public investment in their schooling and the taxes they would have paid. It is also the purest example of Keynes’s point that the choice between income and time is rarely a choice. A nurse in Porto facing a frozen salary and a rising rent does not choose between more money and more leisure. She chooses between Porto and London.

The rent

Where the raise goes

The last essay argued that the things which anchor a life, housing above all, have become dearer faster than wages, and that this is why households keep working long hours even as they get richer. Nowhere in Europe is that truer than in Portugal now. Figure 9 shows house prices and median household income, both indexed to 2015.

Fig. 9 — House price index (2015 = 100, all dwellings, nominal) for Portugal, Spain and the EU, and Portugal’s median equivalised disposable household income on the same base. Data: Eurostat, prc_hpi_a and ilc_di03 (EU-SILC survey years).

Portuguese house prices fell after 2008 and bottomed in 2013. Since then they have risen by a factor of 2.8, to 2025, far faster than in Spain or the EU on average. Median household income rose by a factor of 1.7 over the decade from 2015, a very good decade for incomes by Portuguese standards, and still left far behind. Eurostat’s measure of house prices relative to incomes puts Portugal 23 per cent above its long-term average in 2024, the highest among the countries compared. For people who rent at market prices, 27 per cent spent more than 40 per cent of their disposable income on housing in 2025, against 19 per cent across the EU.[19] The median bank valuation of a square metre of housing, which the national statistics office publishes monthly, went from 759 euros in July 2013 to 2,240 in July 2026.[20]

“A moderate rent of €2,300? That’s mocking those who work.” The Casa para Viver march for housing, Lisbon, 21 March 2026. Photograph by Rafael Medeiros for Esquerda.net, CC BY-SA 4.0, via Wikimedia Commons.

The causes are the usual ones, in an unusual combination: low interest rates for most of the decade, tourism and short-term letting in Lisbon, Porto and the Algarve, foreign buyers drawn by residence schemes and tax regimes, and far too little building. The effect is Keynes’s relative needs made concrete. A flat in a city with jobs is a positional good: there are only so many, and their price is set by what the richest bidder can pay, who is increasingly not Portuguese. For a young Portuguese worker, a minimum wage that rose 40 per cent in real terms in ten years has been more than eaten by a rent that rose faster. The raise went to the landlord.

Inequality and the rest

What did improve

Not everything moved the wrong way, and on several measures Portugal did better than the countries it is usually compared with. Income inequality after taxes and transfers fell. The Gini coefficient of disposable income dropped from 34.5 in 2014 to 30.9 in 2025, close to the EU’s 29.2, and the ratio of the income of the richest fifth to the poorest fell from 7.0 in 2004 to 4.9.[21] Before taxes the long-run story is a U, like everywhere else, but a shallow one: the share of pre-tax income going to the richest tenth fell from about 34 per cent around 1970 to 28 in 1980, rose to 38 by 2008, and has eased to 35, below Germany’s (Figure 10).[22] Wealth is concentrated, but nothing like in America: the OECD puts the richest tenth of Portuguese households at about half of all household wealth, against three-quarters in the United States.[23]

Fig. 10 — Share of pre-tax national income received by the top 10% of adults, 1960–2024. Data: World Inequality Database.

Material life improved sharply after the crisis. The share of Portuguese unable to afford several basic items, a week’s holiday, an unexpected bill, heating, fell from 27 per cent in 2014 to 10 per cent in 2025, now below the EU average of 12. Median disposable income, adjusted for inflation, rose by about 45 per cent between the income years 2013 and 2024, faster than in Spain, Germany or France. And on the most basic measure of all, Portugal is a success story that America is not: life expectancy at birth rose from 64 years in 1960 to 82.5 in 2024, above the EU average and above Germany’s.[24] A Portuguese child born this year can expect to outlive an American one by about three years, in a country with little more than half the income per head.

So the answer to “did quality of life improve with productivity?” is, in Portugal, a qualified yes. The things that are provided collectively, health above all, and the things that are protected by redistribution, the incomes of the poorest, improved a great deal. The things left to the market, the price of a home, and the things that depend on productivity, the wages of the middle, did not.

Taking the time

A four-day week in the two-thirds country

If Portugal cannot yet afford Keynes’s fifteen hours, it can still ask whether it takes too little of what it has as time, and in 2023 it ran an experiment to find out. The government-backed pilot of a four-day week, designed by the economists Pedro Gomes and Rita Fontinha, invited private firms to cut hours without cutting pay for six months, with no subsidy. A hundred and twenty firms expressed interest; 21 ran the trial, and another 20 that already worked four days joined the study.[25] Weekly hours in the trial firms fell from 41.6 to 36.5. Of the 21, 12 kept the four-day week or a modified version of it after the trial, five kept a reduced form, such as summers only or Friday afternoons, and four went back to five days. Among workers, 93 per cent wanted to continue; the share reporting very good or excellent mental health doubled, from 15 to 30 per cent; the share finding it hard to reconcile work and family fell from 46 to 17 per cent. About four in five managers judged the effect on the business financially neutral, and none reported harm, though the study collected no financial accounts. The authors call it a proof of concept, and the caveat is the obvious one: the firms chose to take part.

The detail in the results that matters most is not the headline. Firms that reorganised their work in two or more ways, cutting meetings, changing schedules, automating routine tasks, reverted to five days in 8 per cent of cases; firms that changed one thing or nothing reverted in 38 per cent. Taking productivity as time turns out to require producing more of it first, in exactly the way the Portuguese economy as a whole has struggled to. Gomes’s own argument, in his book on the subject, is that the causation can run the other way too: a shorter week forces firms to organise better.[26] The pilot is too small to prove it. It is the first serious test.

Coda

Keynes at the other end of Europe

Keynes wrote about “progressive countries”, by which he meant the industrial economies of his day, and his prediction was about them. Seen from Portugal, the century looks different. The productivity came late, in a rush between 1960 and 2000, and then stalled at two-thirds of the European level, where it has stayed for thirty years. The hours barely fell, because there was little to take. The income was shared more evenly than in America, and a rising minimum wage has done more for the lowest-paid than any American policy of the period, but the middle has been squeezed between a minimum that rose to meet it and a productivity that did not rise to carry it. The crisis sent half a million people abroad. The recovery has been real, and much of it has been paid into rents.

The lesson from both essays together is that the split between income and time, and between the top and the middle, is set by institutions, not by technology. Europe took more of its productivity as time than America did because it legislated and bargained for it. It shared the income more evenly because it taxed and transferred more. And the countries that did best on both, the Netherlands, Denmark, Germany, did so from a base of high productivity. Portugal has the institutions. What it lacks is the productivity for them to work on, and a housing market that does not take back what they deliver. Keynes’s arithmetic still holds: the fifteen-hour week is a matter of compound interest and choice. For Portugal the compound interest has to come first.

What to expect

Fewer workers, more newcomers, and the same arithmetic

The largest force on the Portuguese economy over the next quarter-century is not a technology. It is age. Figure 11 shows Eurostat’s baseline projection of the number of people aged 65 and over for every hundred of working age.[27] Portugal starts at 43, already among the highest in Europe, and reaches 69 by 2050, above Spain, Italy, Germany and the EU average of 55. With no migration at all it would pass 73. Today there are more than two Portuguese of working age for every one over 65; by 2050 there would be about one and a half.

Fig. 11 — People aged 65 and over per hundred aged 20–64, 2025–2070, Eurostat EUROPOP2023 baseline; dotted, Portugal with zero net migration. The baseline assumes far less immigration into Portugal than has actually happened since 2022; see the text. Data: Eurostat, proj_23ndbi.

The same projection has the share of Portugal’s population aged 20 to 64 falling from 57.8 to 49.5 per cent by 2050. If employment rates and hours stayed as they are, that alone would lower GDP per head by about 0.6 per cent a year, against about 0.4 for the EU as a whole. Portuguese output per hour grew by about 0.3 per cent a year between 2014 and 2024. At that pace the country would be running to stand still: demography would take away more each year than productivity added. Longer working lives, higher employment among women and older workers, and immigration can each offset part of it. What cannot offset it for long is working more hours per worker, which Portugal already does.

Longer working lives are already written into the law. Since 2014 the age at which a full state pension can be drawn has been tied to life expectancy at 65: it was 66 in 2014, is 66 years and 9 months in 2026 and will be 66 years and 11 months in 2027, and a worker who retires early outside the exceptions also has the pension cut by a “sustainability factor”, 16.9 per cent for pensions starting in 2025 and 17.6 per cent in 2026.[28] The OECD calculates that a Portuguese entering work at 22 today will reach the normal pension age at 68. Portuguese men already leave work later than almost anyone else in Western Europe: at 66.5 on average in 2024, up from 60.7 in 2000.[29] The question the law does not ask is how healthy those years are. At 65 a Portuguese can expect to live another 21.1 years, as long as a Swede, but only 9.8 of them in good health on Eurostat’s measure, against 14.2 in Sweden, 13.1 in Italy, 12.5 in Spain and 10.3 across the EU.[30] Tying the pension age to total life expectancy, when healthy life expectancy is shorter and more unequal, takes the extra working years mostly from the healthy ones. That trade-off, and whether Portuguese pensions are enough to live on, is the subject of the third essay in this series, The Long Retirement.

And here the projection is already wrong, in a telling direction. It assumed net migration into Portugal of about 16,000 people a year. The real figures were 371,000 in 2022, 307,000 in 2023 and 217,000 in 2024, by far the largest inflows in decades.[31] Portugal’s population on 1 January 2025 was 11.39 million, a million more than the projection expected for that date. The country that exported people for a century and a half has become, in three years, a country of immigration, while about 34,000 people a year still leave it permanently. Most of the newcomers are of working age, and on the arithmetic above they are the main reason the Portuguese labour force is growing at all. They also need somewhere to live, which is one more reason the rent in Figure 9 is unlikely to fall soon.

Wages have a path set in advance. The tripartite agreement signed in October 2024 raises the minimum wage to €970 in 2027 and €1,020 in 2028, and sets a target of €1,890 for the average monthly wage in 2028.[32] If both are met, the minimum stays at a little over half of the average; if the average falls short, as it has before, the compression of Figure 6 continues and the Portuguese middle gets still closer to the floor.

Convergence is a matter of arithmetic too. To close the gap between 68 and 100 on output per hour, Portugal would need to grow faster than the EU average by half a percentage point a year for 76 years, by one point for 38 years, or by two points for 19. For the last thirty years the difference has been zero. It can be done, and the proof is to Portugal’s east. In 1995 Estonia’s output per hour was 31 per cent of the EU average, Lithuania’s 34, Romania’s 24, Hungary’s 51, Slovakia’s 49, Czechia’s 63 and Slovenia’s 68, level with Portugal. By 2024 all seven had overtaken it, Estonia at 73, Slovakia at 80 and Slovenia at 84.[2] They did it with large inflows of foreign investment, rapid upgrading of skills and a steady shift of workers from small, low-capital firms to large, high-capital ones. Portugal has the skills in its younger cohorts. The other two are policy.

Artificial intelligence could cut either way. The IMF estimates that about 60 per cent of jobs in advanced economies are exposed to generative AI.[33] In a country whose productivity problem is partly one of management and organisation in very small firms, cheap tools that do the work of an accountant, a translator or a customer-service team could narrow the gap, if small firms adopt them. If only the large and already productive firms do, the gap within the country widens, and so, perhaps, does the one with Europe.

So the things to watch in Portugal are these. Whether output per hour finally grows faster than the EU’s, which is the precondition for everything else in this essay. Whether the average wage keeps pace with the minimum. Whether the new immigrants are housed and employed in ways that raise productivity rather than just adding hours. Whether housing supply responds. And whether the four-day week spreads beyond the firms that volunteered for it. Keynes’s fifteen-hour week is not on Portugal’s horizon. A thirty-five-hour one, with the productivity to pay for it, could be.

Open threads

Where this could go

A proper Portuguese hours series. The three standard sources disagree about whether Portuguese hours rose or fell between 1970 and 1990, by several hundred hours a year. Reconstructing the series from the labour-force surveys and the old industrial censuses would settle how much of its productivity Portugal took as time, and when.

The rent and the minimum. A household-level calculation, using the income survey and INE’s rent statistics, of what share of a minimum-wage worker’s pay a new lease in each municipality absorbed in 2015 and in 2025, would put a number on the claim that the raise went to the landlord.

On method and tools

This piece was written collaboratively with Claude Opus 5.5 (Anthropic): human specification, editorial direction and critical review; machine data research, analysis and drafting. The numbers are computed by scripts/portugal_work.py from public data: the European Commission’s AMECO database (spring 2026), the Eurostat dissemination API, the OECD Data Explorer, the World Inequality Database, and the Bergeaud–Cette–Lecat productivity database used in the companion essay. Results are in docs/portugal-work-results.json and the figures in docs/portugal-work-figures.html. The outlook section uses Eurostat’s EUROPOP2023 projections, whose baseline migration assumption the text compares with what has since happened; its convergence dates are simple arithmetic, not forecasts. Where sources disagree, as on Portuguese working hours before 1990, the text says so and uses the OECD series for trends and AMECO for cross-country comparisons. EU-SILC incomes refer to the year before the survey year. The photographs are from Wikimedia Commons and credited with their licences in the captions; local copies are in assets/images/2026/.

The cover is computed by the same script: the scatter of Figure 2 without its labels, one dot per European country, with Portugal in magenta.

Authored by: Luis Matos Ferreira — Physicist, Developer, Writer

Related essays on this blog

1. The Fifteen-Hour Week — Keynes’s prediction, productivity, hours and pay, from American data.
2. The Long Retirement — the end of working life in Europe and Portugal: exit ages, health and pensions.
3. O Grande Debate — the argument over markets, states and who gets the output.

Sources

1. Bergeaud, Cette & Lecat, “Productivity Trends in Advanced Countries between 1890 and 2012”, Review of Income and Wealth 62, 420 (2016); Long-Term Productivity Database v2.7, February 2026.
2. European Commission, Directorate-General for Economic and Financial Affairs, AMECO annual macro-economic database, spring 2026: series HVGDHR, NLHA, ALCD0, RWCDC, OVGD, NETD, NLHT, ZUTN, ZCPIN.
3. Blanchard, “Adjustment within the euro. The difficult case of Portugal”, Portuguese Economic Journal 6, 1 (2007); Blanchard & Portugal, “Boom, slump, sudden stops, recovery, and policy options. Portugal and the Euro”, Portuguese Economic Journal 16, 149 (2017).
4. Reis, “The Portuguese Slump and Crash and the Euro Crisis”, Brookings Papers on Economic Activity, Spring 2013, 143.
5. Gopinath, Kalemli-Özcan, Karabarbounis & Villegas-Sanchez, “Capital Allocation and Productivity in South Europe”, Quarterly Journal of Economics 132, 1915 (2017).
6. OECD, OECD Economic Surveys: Portugal 2023, OECD Publishing, Paris, 2023.
7. OECD Data Explorer, average annual hours actually worked per worker (DSD_HW@DF_AVG_ANN_HRS_WKD).
8. Eurostat, Labour Force Survey, average number of usual weekly hours of work in main job (lfsa_ewhun2), 2025; Lei n.º 21/96, de 23 de Julho.
9. OECD Data Explorer, average annual wages, constant prices and 2025 USD PPPs (DSD_EARNINGS@AV_AN_WAGE).
10. Eurostat, Structure of Earnings Survey 2022, median hourly earnings (earn_ses_pub2s); minimum wage as a proportion of median gross monthly earnings (earn_mw_avgr2).
11. OECD Data Explorer, minimum relative to median wages of full-time workers (MIN2AVE) and real minimum wages (RMW).
12. Retribuição Mínima Mensal Garantida, historical values published by DGAEP; deflated with the Portuguese CPI from AMECO.
13. Gabinete de Estratégia e Planeamento, Ministério do Trabalho, Solidariedade e Segurança Social, bulletins on the minimum wage based on Quadros de Pessoal, April 2016–2019.
14. Banco de Portugal, “Salário Mínimo Nacional: Evolução recente e perspetivas”, presentation to the Conselho Superior de Estatística, 1 June 2026.
15. Portugal & Cardoso, “Disentangling the Minimum Wage Puzzle: An Analysis of Worker Accessions and Separations”, Journal of the European Economic Association 4, 988 (2006); Oliveira, “The minimum wage and the wage distribution in Portugal”, Labour Economics 85, 102459 (2023).
16. Eurostat, unemployment by sex and age, annual (une_rt_a).
17. Blanchard & Portugal (2017), as in note 3.
18. Eurostat, emigration by age and sex (migr_emi2); Vidigal, Pereira, Azevedo, Vilhena & Pires, Emigração Portuguesa 2025: Relatório Estatístico, Observatório da Emigração, 2026, table 1.3, using INE’s permanent and temporary emigration estimates and UN population-division stocks.
19. Eurostat, house price index (prc_hpi_a); house price to income ratio (tipsho60); housing cost overburden rate by tenure status (ilc_lvho07c); mean and median income (ilc_di03).
20. Instituto Nacional de Estatística, Inquérito à Avaliação Bancária na Habitação, median bank valuation per square metre, July of each year.
21. Eurostat, EU-SILC: Gini coefficient of equivalised disposable income (ilc_di12); income quintile share ratio S80/S20 (ilc_di11).
22. World Inequality Database, pre-tax national income shares (sptinc992j), bulk download, September 2026; Chancel, Piketty, Saez & Zucman (eds.), World Inequality Report 2022, Harvard University Press, 2022.
23. OECD Wealth Distribution Database, share of wealth held by the top 10% of households, latest available year (Portugal 2020).
24. Eurostat, material and social deprivation rate (ilc_mdsd07); life expectancy by age and sex (demo_mlexpec).
25. Gomes & Fontinha, Semana de Quatro Dias. Projeto-Piloto. Relatório Final, IEFP, June 2024.
26. Gomes, Friday Is the New Saturday: How a Four-Day Working Week Will Save the Economy, Flint, 2021.
27. Eurostat, EUROPOP2023 population projections, demographic balances and indicators by type of projection (proj_23ndbi): old-age dependency ratio, 3rd variant (65+ to 20–64), share of population aged 20–64 and net migration, baseline and no-migration variants.
28. Decreto-Lei n.º 167-E/2013, de 31 de Dezembro, which tied the normal pension age to life expectancy at 65 from 2014; Portaria n.º 358/2024/1, de 30 de Dezembro, which fixes the age for 2026 at 66 years and 9 months and the sustainability factor for 2025 at 0.8307; Portaria n.º 476/2025/1, de 29 de Dezembro, which fixes the age for 2027 at 66 years and 11 months and the sustainability factor for 2026 at 0.8237, a cut of 17.6 per cent.
29. OECD, Pensions at a Glance database (DSD_PAG@DF_PAG): average effective age of labour market exit, men, and future normal retirement age for a labour-market entrant at 22.
30. Eurostat, healthy life years and life expectancy at age 65 (hlth_hlye), 2024. Healthy life years are based on self-reported activity limitation in EU-SILC; the series has breaks, notably for Germany, and comparisons between countries are approximate.
31. Eurostat, population change, demographic balance and crude rates at national level (demo_gind): population on 1 January and net migration plus statistical adjustment.
32. Governo de Portugal, “Governo e parceiros sociais assinam novo acordo de concertação social” (Acordo Tripartido sobre Valorização Salarial e Crescimento Económico 2025–2028), portugal.gov.pt, 1 October 2024.
33. Cazzaniga, Jaumotte, Li, Melina, Panton, Pizzinelli, Rockall & Tavares, “Gen-AI: Artificial Intelligence and the Future of Work”, IMF Staff Discussion Note SDN/2024/001, January 2024.