Off the Treadmill

Essay · economics · September 2026

If jobs, debts, pensions and politics depend on growth, what would it take to loosen each dependence? There are real policies for all four, some already tried: shorter hours and short-time work, fiscal rules and surpluses, pensions that adjust themselves, taxes on wealth rather than work. Each has evidence behind it and a cost attached. None removes the need to choose who bears the adjustment; growth is what has spared rich countries that choice.

Reading time ≈ 13 min Jobs → debts → pensions → distribution Part III of four

The question

The Treadmill ended with a list. Each dependence on growth could in principle be loosened: shorter working hours instead of unemployment, less debt, pensions designed for a stable wage bill, taxes that share out a fixed income. Whether that could be done without the pain Portugal went through after 2000 is the real question in the debate about growth. This essay takes the list seriously, one item at a time: what has been tried, what it achieved, what it cost, and what it would mean for Portugal.

It does not argue that growth should stop. The case for loosening these dependencies does not require that: growth may slow on its own, as it has in most rich countries, and an economy that can absorb a slowdown without mass unemployment, a debt spiral or a pensions crisis is more robust whatever one thinks of growth.

3.9%of GDP, the primary surplus Portugal would need every year to bring its debt to 60% by 2040 with zero nominal growth

1.9%Portugal’s best five-year average primary surplus since 1995 (2021–2025)

3 of 16years between 2008 and 2023 in which Portugal updated pensions by its own growth-linked rule as written

−4.2%the largest one-year cut in Swedish pensions from its automatic balancing mechanism, in 2011

One

Jobs: share the hours

The first dependence is arithmetic: if output per hour rises and output does not, fewer hours of work are needed. They can be lost as jobs or shared as shorter hours. Rich countries have tried both deliberately.

A shorter legal week. France cut its standard week from 39 to 35 hours in 1998–2002. The labour ministry’s own evaluators estimated about 350,000 jobs created; independent studies comparing firms and workers affected with those not affected found no significant net effect on employment, and a rise of 3.3 points in the share holding second jobs.[1] A study of Germany’s negotiated cuts found that each hour off the standard week raised employment by 0.3 to 0.7 per cent but cut total hours worked by 2 to 3 per cent; a cross-country study found no evidence that work-sharing creates jobs.[2] The lesson is not that shorter hours are harmful, but that they do not by themselves create jobs; they convert productivity into time. Portugal’s own cut, from 44 to 40 hours in 1996, kept monthly pay roughly unchanged and led to no extra job losses among the workers affected.[3]

Short-time work. The more successful tool is temporary. In Germany’s 2009 recession, the Kurzarbeit scheme, which pays part of the wages of workers whose hours are cut, saved an estimated 400,000 jobs, though with wide uncertainty; the rule-based part of the scheme worked and discretionary top-ups did not.[4] In 2020 the EU lent €98 billion to 19 countries to finance such schemes, €6.2 billion of it to Portugal.[4] Short-time work turns a fall in output into fewer hours for everyone rather than no hours for some. It is designed for recessions, not for a permanent slowdown, and it protects those with jobs more than those without.

Part-time by choice. The Netherlands has the shortest average hours in Europe without a short legal week (Figure 1). Dutch full-timers work about 39 hours, only two fewer than Portuguese full-timers; the difference is that 43 per cent of Dutch workers are part-time, against 7 per cent in Portugal, supported since 2000 by a right to request shorter hours.[5] Dutch hours are spread over more people, with one of the highest employment rates in Europe. They are also heavily gendered: 64 per cent of employed Dutch women work part-time.

Fig. 1 — Usual weekly hours of work in the main job, all employed people and full-time workers, 2024; the share of part-time workers in brackets. Data: Eurostat, lfsa_ewhun2 and lfsa_eppga.

Guaranteed work and income. Other proposals target the jobless directly. An Austrian village, Gramatneusiedl, offered a job to every long-term unemployed resident from 2020; long-term unemployment fell and wellbeing rose, at a cost of about €30,000 a participant a year, with no harm to other jobs.[6] Finland’s basic-income trial of 2017–2018 raised employment by about six days over two years, a small effect, while improving wellbeing.[7] Both are local experiments, not tests of what such schemes would cost or do at national scale. The systematic review of degrowth proposals finds that basic income, shorter working time and a job guarantee are the three most often proposed; its authors observe that the choice among them is “more sociological than analytical”.[8]

For Portugal, the relevant facts are long hours, 41 a week for full-timers, very little part-time work, heavy use of fixed-term contracts, and a four-day-week pilot that most firms liked, as The Fifth Day described.[5][9] The Portuguese labour market adjusts to shocks mainly by ending temporary contracts, which puts the cost of a slowdown on the young.

Two

Debts: pay down while growth lasts

The second dependence is the arithmetic of debt: a debt ratio stays stable only if the primary budget balance, before interest, covers the gap between the interest rate and growth. With Portugal’s debt at 90 per cent of GDP and an interest rate of 2.5 per cent, the required balance moves by almost a point of GDP for every point of growth lost (Figure 2). At 3.5 per cent nominal growth, debt stays stable even with a small primary deficit and can fall to 60 per cent by 2040 with a surplus of about 1.3 per cent of GDP. At zero growth, stabilising the debt requires a surplus of 2.2 per cent a year, and reaching 60 per cent by 2040 requires 3.9.[10]

Fig. 2 — Portugal: primary budget balance (before interest) needed every year from 2026 to keep public debt at its 2025 level of 89.7 per cent of GDP, or to bring it to 60 per cent by 2040, at an interest rate of 2.5 per cent and different rates of nominal GDP growth; the dashed line is Portugal’s best five-year average primary surplus since 1995. Ignores stock-flow adjustments. Data: author’s calculation from the EDP debt figure and AMECO.

Portugal’s best five-year average primary surplus since 1995 is 1.9 per cent of GDP, achieved in 2021–2025. Few countries have done better for long: Olivier Blanchard and colleagues note that the best five-year cyclically adjusted surplus since 1980 was 0.9 per cent in France, 1.6 in Germany and 1.5 in Italy.[11] Held at 1.9 per cent with an interest rate of 3 per cent, Portuguese debt would fall to about 56 per cent of GDP by 2040 with 3.5 per cent nominal growth, but would rise to about 104 per cent with none (Figure 3).

The arithmetic has an obvious way out: productivity. Nominal growth is the sum of growth in output per hour, in hours worked and in prices, and with an ageing population hours will struggle to grow. If output per hour rose by 1 per cent a year and prices by the ECB’s target of 2 per cent, Portugal would be close to the comfortable case above, where debt falls with surpluses it has already run. That is the growth that needs no extra hours, and the reason why the stalled productivity described in The Stalled Hour matters for the debt. But it is still growth, and it can be used only once: productivity taken as shorter hours, the first lever above, does not shrink the debt; productivity taken as income does.

Fig. 3 — Portugal: public debt, per cent of GDP, 2025–2040, if the primary surplus is held at 1.9 per cent of GDP every year and the interest rate is 3 per cent, at nominal growth rates of 3.5, 2, 1 and 0 per cent. Data: author’s calculation.

So the first lever is timing: reduce debt while growth lasts, as Portugal has done since 2021, so that a slowdown starts from a lower base. The second is the design of the rules. The EU’s fiscal rules, reformed in 2024, now set a path for net public spending, excluding interest, over four to seven years, based on an analysis of each country’s debt, rather than annual deficit targets; they add a floor of one point of GDP a year of debt reduction for countries above 90 per cent.[12] Portugal’s plan under these rules projects debt at 83 per cent in 2028 and 65 per cent in 2038, on the assumption of potential growth of 1.8 per cent a year plus 2.2 per cent inflation.[13] The rules do not remove the dependence on growth; they make it explicit in the assumptions.

History offers a third lever that Portugal lacks. After 1945 the United States and Britain reduced their debts by holding interest rates below inflation, eroding debt by 3 to 4 per cent of GDP a year.[14] A member of the euro cannot set its own interest rate or inflation. Recent work also narrows Blanchard’s point that interest below growth makes debt cheap: once the fact that interest rates rise with debt is taken into account, the room is smaller than it looks.[15] For private debt, the Banco de Portugal’s limits on loan size relative to income and value since 2018 keep new mortgages from assuming that incomes will grow.[16]

Three

Pensions: let the system adjust itself

The third dependence is that pensions paid from contributions depend on the wage bill. About two-thirds of OECD countries now have some automatic mechanism that adjusts pensions, contributions or retirement ages to demography or the economy, so that governments do not have to legislate each cut.[17] The ones that respond to low growth work in different ways.

Sweden’s system cuts pensions automatically when its liabilities exceed its assets. The brake was triggered after the 2008 crisis: pensions fell by about 3.0 per cent in 2010, 4.2 per cent in 2011 and 2.7 per cent in 2014, and at the worst point pension balances were about 6 per cent below where they would have been without it. The brake works both ways, and by 2018 the loss had been fully restored.[18] Germany adjusts pensions by a quarter of the change in the ratio of pensioners to contributors, but has set a floor on the replacement rate of 48 per cent; Japan’s “macroeconomic slide” of 2004 was first applied in 2015 and stops at a replacement rate of 50 per cent.[17] Every such mechanism has a floor, and several have been suspended or overridden when they bit.

Portugal already has one, though few know it. Since 2006, pensions have been updated each year according to GDP growth. When growth is below 2 per cent, pensions up to twice the social support index keep up with inflation, and larger pensions fall behind it by half a point to three-quarters of a point a year; when growth is above 3 per cent, the smallest pensions rise faster than prices.[19] It is, in effect, a brake that protects the poorest pensioners and slows the rest when growth is weak. But between 2008 and 2023 it was applied as written only three times, in 2008, 2009 and 2016; the rest of the time governments decided the increases themselves. The commission that wrote the 2024 Green Book on Social Security proposed replacing the GDP trigger with an indicator of the system’s own financial balance, a step towards the Swedish or German approach.[19]

The recent literature on pensions without growth makes the underlying point plainly: “The core challenge is distributional: how to allocate income between workers and pensioners when aggregate resources are stagnant or declining.” In a defined-benefit system, workers pay through higher contributions; in a system of notional accounts like Sweden’s, pensioners adjust. The authors suggest ceilings on the largest pensions as one tool.[20] The Portuguese projections already assume that pensions will fall from 53 to 34 per cent of average wages by 2070; that, more than any rule, is how the system is expected to cope.[21] The reserve fund, about €42 billion or 14 per cent of GDP at the end of 2025, is a buffer for the transition, not a solution.[19]

Four

Distribution: tax what accumulates

The fourth dependence is political. When the economy grows, everyone can gain; when it does not, one person’s gain is another’s loss. Thomas Piketty argued that low growth also changes who holds wealth: when growth is slow relative to the return on capital, “past wealth tends to dominate new wealth”, and inheritance matters more than work.[22] If that is right, an economy without growth needs a tax system that reaches accumulated wealth, not only current incomes.

Portugal’s is unusual in that respect. It abolished its inheritance tax in 2004; transfers to spouses, children and parents are exempt from the stamp duty that replaced it. Across the OECD, inheritance and gift taxes raise about 0.15 per cent of GDP, and in France about 0.8.[23] Property is taxed at 0.3 to 0.45 per cent of values last reviewed in 2015, with an extra tax on holdings over €600,000.[9] The OECD recommends shifting the tax burden from labour towards property and pollution.[9] Environmental taxes, at about 2 per cent of Portuguese GDP, are too small a base to replace much labour tax, and they shrink as emissions fall.[24]

Together

What the models say about the package

A few studies have simulated whole economies adjusting to low growth. Tim Jackson and Peter Victor’s model of Canada combined green investment, a gradual cut in working hours from about 1,750 to 1,450 a year by 2067, and large transfers to low incomes. Growth in income per head fell to about 0.4 per cent a year and then to zero; unemployment stayed close to the baseline because hours fell; inequality fell sharply. The cost was a rise in public debt from 55 to more than 80 per cent of GDP, partly because carbon-tax revenue disappeared as emissions did.[25]

A 2026 model of the United Kingdom is less comfortable. Its authors find that without accompanying policies a post-growth transition sends pension and health spending up as a share of a stagnant GDP and public debt to unsustainable levels. With carbon taxes, less generous pension indexation and later retirement, debt in their runs still rose to around 190 to 200 per cent of GDP before easing; only the scenario that also cut working hours by a fifth and kept interest rates low held it below about 135 per cent, ending near 85 per cent in 2070. Unemployment in these runs was between about 11 and 20 per cent.[20] The authors conclude that growth dependencies “are currently being managed by displacing costs onto workers, patients, and future generations”.

Critics point out how thin the evidence base still is. A review of 561 studies of degrowth found that fewer than 2 per cent used a formal theoretical model and few used data.[26] Mainstream economists add that the working-time evidence is weak, that job guarantees are expensive and that the political feasibility of cutting pensions or taxing inheritance is untested. Those are fair objections, and they apply to the more ambitious proposals, not to the modest mechanisms that countries already use.

The balance

What could be done

Put side by side, the options sort themselves. Some are proven and modest: short-time work in recessions, the right to request shorter hours, reducing debt while growth lasts, pension rules that adjust automatically with floors that protect the poorest. Portugal already has versions of several and does not always use them. Some are proven in part: shorter legal weeks convert productivity into time without destroying jobs, but they do not create jobs either, and they cost either pay or productivity. Some are promising but small-scale: job guarantees and basic income. And some are mostly untested at scale: whole economies run without growth.

What none of them does is avoid the question growth has allowed rich countries to postpone: who adjusts. Shorter hours mean less income for someone; pension brakes mean pensioners take the loss; contribution rises mean workers do; debt reduction means spending cuts or taxes; taxes on inherited wealth mean heirs pay. Growth lets a society avoid that choice by giving everyone a little more. Loosening the dependence on growth means making it in the open. The most practical first step for Portugal may be the least dramatic: applying the rules it already has, as written, and deciding in advance who will bear the cost when growth falls short. Part IV, Worth Producing, asks the question underneath: what the growth, and the borrowing, are for.

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 debt calculations use the identity for the debt-stabilising primary balance, pb = (r − g)/(1 + g) × d, with Portugal’s debt of 89.7 per cent of GDP in 2025 and illustrative interest and growth rates; they ignore stock-flow adjustments and the feedback from growth to the budget. The figures are computed by scripts/offtreadmill.py. Several studies were read in working-paper versions (the Jackson–Victor model, the German and French working-time studies, the Marienthal evaluation), and the UK post-growth results are read from the authors’ charts. Results are in docs/offtreadmill-results.json and the figures in docs/offtreadmill-figures.html; the downloaded sources, with a page reference for every number, are kept with the script’s data.

The cover photograph is Jardim da Estrela, Lisbon by Sonse; CC BY 2.0, via Wikimedia Commons, cropped.

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

Related essays on this blog

1. The Treadmill — Part I: why modern economies seem to need growth.
2. Growth of What — Part II: what GDP measures and the alternatives.
3. Worth Producing — Part IV: what productivity measures, waste, and what debt is for.
4. The Fifth Day — Portugal’s four-day-week pilot.
5. The Sustainability Story — the Portuguese pension system and its forecasts.
6. Earned and Unearned — how work and wealth are taxed.

Sources

1. C. Gubian et al., Économie et Statistique 376–377, 2004, abstract, p. 25; M. Chemin and E. Wasmer, study of the 35-hour week using Alsace-Moselle, NBER chapter, 2008, p. 301; M. Estevão and F. Sá, “The 35-hour workweek in France: Straightjacket or welfare improvement?”, IZA Discussion Paper 2459, abstract and pp. 11 and 15.
2. J. Hunt, “Has Work-Sharing Worked in Germany?”, Quarterly Journal of Economics 114(1), 1999, abstract; A. Kapteyn, A. Kalwij and A. Zaidi, “The myth of worksharing”, IZA Discussion Paper 188, abstract.
3. P. S. Raposo and J. C. van Ours, “How Working Time Reduction Affects Employment and Earnings”, Economics Letters 106(1), 2010 (IZA DP 3723, abstract).
4. T. Boeri and H. Brücker, “Short-time work benefits revisited: some lessons from the Great Recession”, IZA Discussion Paper 5635, p. 38 and Table 7; A. Balleer, B. Gehrke, W. Lechthaler and C. Merkl, “Does short-time work save jobs? A business cycle analysis”, working paper, pp. 29–30; European Commission, SURE final evaluation.
5. Eurostat, Labour Force Survey: usual weekly hours (lfsa_ewhun2), part-time employment (lfsa_eppga), employment rates (lfsi_emp_a), 2024.
6. M. Kasy and L. Lehner, “Employing the Unemployed of Marienthal: Evaluation of a Guaranteed Job Program”, IZA Discussion Paper 16088, 2023, pp. 11 and 28.
7. Kela and VATT, Finland’s Basic Income Experiment 2017–2018: results, 2020.
8. N. Fitzpatrick, T. Parrique and I. Cosme, “Exploring degrowth policy proposals: A systematic mapping with thematic synthesis”, Journal of Cleaner Production 365, 2022, p. 8.
9. OECD, OECD Economic Surveys: Portugal 2026, pp. 41, 70 and 118.
10. Author’s calculation: debt-stabilising and target primary balances for Portugal from the EDP notification of April 2026 (debt 89.7 per cent of GDP in 2025).
11. O. Blanchard, A. Leandro and J. Zettelmeyer, “Redesigning EU Fiscal Rules: From Rules to Standards”, PIIE Working Paper 21-1, 2021, p. 7; European Commission, AMECO (primary balance, Portugal).
12. Regulation (EU) 2024/1263 on the effective coordination of economic policies and multilateral budgetary surveillance, arts. 2, 6, 7 and 8.
13. Council of the EU, recommendation endorsing the national medium-term fiscal-structural plan of Portugal, ST 5025/25, January 2025, Table 2.
14. C. M. Reinhart and M. B. Sbrancia, “The liquidation of government debt”, Economic Policy 30, 2015 (working-paper version, p. 7).
15. O. Blanchard, “Public Debt and Low Interest Rates”, American Economic Review 109(4), 2019; A. Mian, L. Straub and A. Sufi, “A Goldilocks Theory of Fiscal Deficits”, NBER Working Paper 29707, 2022, pp. 3–4.
16. Banco de Portugal, macroprudential recommendation on new credit agreements for consumers (2018) and its monitoring reports.
17. OECD, Pensions at a Glance 2021, chapter 2, “Automatic adjustment mechanisms in pension systems”.
18. Swedish Pensions Agency, Orange Report 2019: Annual Report of the Swedish Pension System, pp. 17–19, 44 and 53–54; author’s calculation.
19. Comissão para a Sustentabilidade da Segurança Social, Livro Verde, 2024 (edition of January 2025), Table 4.8 (pp. 91–92) and recommendation 7 (pp. 212–213); European Commission, 2025 Country Report Portugal, pp. 5–6 (reserve fund).
20. M. Leoni and C. Corlet Walker, Making Healthcare and Pensions Growth-Resilient, MAPS project Deliverable D3.2, July 2026, pp. 7–8, 35, 49 and Figure 10.
21. European Commission, 2024 Ageing Report, Portugal country fiche, Table 18.
22. T. Piketty, “On the Long-Run Evolution of Inheritance: France 1820–2050”, Quarterly Journal of Economics 126(3), 2011, pp. 1074–1075.
23. Decreto-Lei 287/2003; Código do Imposto do Selo, art. 6; OECD, Revenue Statistics (tax 4300), 2023.
24. Eurostat, environmental tax revenues (env_ac_tax), Portugal and EU.
25. T. Jackson and P. A. Victor, “LowGrow SFC: a stock-flow-consistent ecological macroeconomic model for Canada”, CUSP Working Paper 16, 2019, pp. 47–55 (published as “The transition to a sustainable prosperity”, Ecological Economics 177, 2020).
26. I. Savin and J. van den Bergh, “Reviewing studies of degrowth: Are claims matched by data, methods and policy analysis?”, Ecological Economics 226, 2024 (figures as reported by the authors; full text not checked).