Work Signals data

AI, jobs and income: the evidence

Four tables and 65 sources on where money comes from, what AI has measurably done to jobs so far, what 25 unconditional cash experiments found, and where Australia stands. Compiled from central banks, statistical agencies and peer-reviewed journals. No forecasts and no policy position.

Data as at 10 November 2026 · Compiled by ApplyAI from the sources listed at the foot of the page · Part of the Work Signals data layer. The compiler's longer evidence review, including policy discussion, is available separately as a PDF and is not part of this data page.

Method. Every figure on this page is taken from a central bank, national statistics agency, government evaluation, peer-reviewed journal or NBER working paper, and carries a numbered reference. Where a share is calculated from two published series (for example, currency as a share of broad money) the calculation is marked "calculated" and both inputs are given. Figures are as published at their source dates and are not adjusted. The page is rechecked against its sources by the weekly accuracy process; a figure that changes at source is changed here with the stamp updated.

What this page does not do. It does not state or imply whether any occupation or person is safe from or at risk from AI. "Exposure" means AI could affect some of a job's tasks; it is not a forecast of job loss. It does not recommend any policy. Limitations are listed at the end.

1. Where money comes from, in five economies

Start with the simplest possible picture. Every dollar is an IOU. A banknote is an IOU from the central bank. The number in your bank account is an IOU from your bank to you. The Bank of England puts it plainly: broad money is made up of bank deposits, which are essentially IOUs from commercial banks to households and companies, and currency, mostly IOUs from the central bank.[1]

Banks do not sit on a pile of savers' deposits and lend them out. It runs the other way. When a bank approves a loan, it types a new deposit into the borrower's account, and that deposit is new money. The Bank of England's 2014 bulletin states that rather than banks receiving deposits when households save and then lending them out, bank lending creates deposits, and that in normal times the central bank does not fix the amount of money in circulation.[2] The Reserve Bank of Australia says the same: money can be created when financial intermediaries make loans, and money and credit are closely linked in a modern economy.[3]

How a bank loan creates new money Borrower Asks for a $500,000 loan Promises to repay with interest Commercial bank Books two entries at once: Asset: loan of $500,000 Liability: new deposit of $500,000 New money now exists Central bank Sets the interest rate Does not count the money loan request deposit appears price of credit What stops banks creating money without limit 1. Profitability and competition: a loan nobody repays is a loss. 2. Prudential regulation: capital rules cap the balance sheet. 3. Monetary policy: the central bank raises the rate, borrowing slows, and so does money creation. The ultimate limit is the inflation target.

Figure 1. Loans create deposits. Source: Bank of England, "Money creation in the modern economy", Quarterly Bulletin 2014 Q1.[2]

This is true everywhere, not just in Britain

The Bank of England's bulletin is the clearest primary statement of the mechanism, which is why it anchors this section, but the proportions are similar in every major economy. The table takes the broadest standard money measure each central bank publishes and the currency component of it. Everything that is not currency is a deposit or deposit-like claim on a bank.

EconomyBroad money measure and latest levelNotes and coin in circulationShare that is bank money (calculated)Source
United KingdomBroad money (M4-type), 20143%97%Bank of England 2014[1]
AustraliaBroad money A$3,462bn, April 2026Currency about A$101bnabout 97%RBA Table D3[160]
JapanM3 about ¥1,640 trillion, Q2 2026Banknotes ¥120.6 trillion, Dec 2025about 93%Bank of Japan[163]
Euro areaM3 €17,717bn, September 2026€1,618bnabout 91%ECB monetary developments[162]
United StatesM2 US$22.97 trillion, May 2026Currency component US$2,345bn, Feb 2026about 90%Federal Reserve H.6 via FRED[161]
Share of money that is bank deposits Australia97% UK97% Japan93% Euro area91% United States90% 3 px per 1%. Remainder is notes and coin. US and euro shares are lower partly because so much of their cash is held abroad.

Figure 2. In every major economy, nine-tenths or more of the money people hold is a claim on a commercial bank, created when the bank lent. Shares calculated from the central-bank series in the table above.[1][160][161][162][163]

Why this matters for the UBI debate

If governments could print money without consequence, UBI would be trivial. They cannot, and the reason is not that money is physically scarce. It is that the central bank aims to keep money creation consistent with low and stable inflation.[2] Money is a claim on real output. Create claims faster than output grows and each claim buys less. A UBI funded by the Treasury is funded by one of three things: taxes now, debt now and taxes later, or central bank money with inflation as the price. Galí's model of a money-financed transfer finds a one-off output lift of about half a percentage point under his scenario, and no author in the literature has shown a stable, non-inflationary way to fund a permanent universal dividend by money creation alone.[66]

2. AI exposure, displacement and the labour share: what has been measured

Here is where the published evidence sat at 5 November 2026. The critical distinction is between exposure (AI can affect some of a job's tasks), displacement (people losing positions), and labour's share of national income. They are different measures and the headline numbers are routinely confused.

SourceEstimateType
IMF, Cazzaniga et al., Jan 2024[70]Nearly 40% of global employment exposed to AI: 60% in advanced economies, 40% emerging, 26% low-income. About half of exposed advanced-economy jobs could gain from complementarity; the other half face reduced labour demand.Exposure, not job loss
Goldman Sachs, Briggs and Kodnani, Mar 2023[71]Generative AI could expose the equivalent of 300 million full-time jobs globally to automation; roughly two-thirds of US and European jobs have some exposure.Exposure, conditional
Goldman Sachs, Briggs, Jul 2026[71]US baseline: 9% of workers, about 15 million, could be displaced from current positions and need reallocation over roughly ten years; unemployment rise under one point in any year if spread out.Forecast
McKinsey Global Institute, Sep 2026[73]By 2035, US automation reduces demand equivalent to 36 million jobs while AI and growth generate demand for more than 40 million; about 11 million workers (range 6 to 16 million) may need to change occupation.Modelled flows
Brynjolfsson, Chandar and Chen, Stanford, Aug 2025 rev. Aug 2026[72]Initial paper: 13% relative employment decline for 22 to 25 year olds in highly exposed occupations. Revision with payroll data to June 2026: employment 19% below the level implied by less-exposed peers. Authors call these descriptive, not causal; no economy-wide displacement found.Observed, descriptive
Stanford HAI, AI Index 2026, Apr 2026[74]Employment of software developers aged 22 to 25 nearly 20% below 2024; about one-third of surveyed employers expected workforce reductions; task-level productivity gains of 14 to 15% in customer support and 26% in software development.Observed plus survey
Anthropic, Massenkoff and McCrory, Mar 2026[75]No systematic rise in unemployment among highly exposed workers since late 2022; tentative 14% decline in job-finding rate for young workers in exposed occupations, only just significant.Observed
Humlum and Vestergaard, NBER, 2025 rev. 2026[76]Danish administrative data: no measurable average chatbot effect on earnings or hours in the first two years; can rule out effects larger than 2%.Observed
Yotzov et al., NBER, Feb 2026 rev. Oct 2026[77]Survey of nearly 9,000 executives in seven countries: essentially no net AI employment effect over three years and little expected over the next three; over 40% expected positive or negative effects that offset in aggregate.Survey
OECD Employment Outlook 2026, Jul 2026[78]Evidence linking language-model diffusion to the youth unemployment gap is limited and mixed; 28% of OECD jobs are in occupations at highest automation risk (not a forecast of disappearance).Review
Acemoglu, "The Simple Macroeconomics of AI", 2024, Economic Policy 2025[79]Total factor productivity gain of 0.66% over ten years (about 0.064 points a year), below 0.53% if hard tasks are included; GDP gain 1.16% or 0.93% with capital deepening. Predicts a wider gap between capital and labour income.Model
Karger et al., NBER, Apr 2026[91]Economists' median US labour-share forecast: 55.0% in 2030 and 52.0% in 2050 under slow AI progress; 52.0% and 45.0% under rapid progress; 2025 baseline 55.48%. Roughly 10 million fewer labour-force participants by 2050 in the rapid scenario.Conditional survey forecast
Jobs and Skills Australia, Aug 2025[100]About 4% of the Australian workforce is in occupations with high automation exposure; 79% have low automation exposure but medium-to-high augmentation potential. Clerical and administrative roles face the highest automation risk.Exposure, Australia
Labour's share of US national income: economists' median forecasts under slow and rapid AI progress 40%45%50%55%60% 202520302050 2025 baseline 55.5% 52% slow 45% rapid Historical context: US labour share fell from 58% in 1981 to 52% in 2016 (Brookings). Australia's fell from over 58% in 1975 to 47.1% in 2018 (Centre for Future Work, ABS data).

Figure 3. Conditional survey forecasts, not measurements. Source: Karger et al., NBER Working Paper 35046, April 2026.[91] Historical lines: Brookings, February 2026[92]; Centre for Future Work on ABS national accounts[101]; RBA Bulletin March 2019.[102]

3. Unconditional cash experiments, 1968 to 2026: what was paid and what was measured

Pilots and natural experiments from 1968 to 2026 with an official evaluation, a peer-reviewed article or an academic working paper behind them. Amounts are nominal. A guarantee is the maximum before earnings-related reductions. One 2026 review counted at least 122 US pilots alone, so this is a selection of the best-evidenced, not a census.[40]

Where and whenDesign and samplePaymentKey findingsEvaluation
New Jersey and Pennsylvania NIT, 1968 to 1972Household RCT; 1,216 families; negative income taxGuarantees 50% to 125% of poverty line; 30% to 70% benefit reduction; mean US$91 per four weeksMale heads' employment about 1.5% lower, hours just over 2% lower; wives worked 23% fewer hours. Negligible psychological effects.Wisconsin IRP and Mathematica[41]
Seattle and Denver SIME/DIME, 1970 to 1978Household RCT; 4,800 families; 3- or 5-year treatmentFamily-of-four guarantees US$3,800, $4,800 or $5,600 (1971 dollars)Year-two hours fell 9% for husbands, 20.1% for wives, 14.3% for female heads. Pooled across four US NIT experiments: about 7% for men, 17% for women.SRI International, US HHS[42]
Manitoba Mincome, 1974 to 1979Saturation site (Dauphin, about 10,000 eligible) plus dispersed experiment60% of low-income cutoff, reduced C$0.50 per C$1 earnedForget: 8.5% fewer hospitalisations, more grade-12 continuation (disputed by Green's reanalysis). Calnitsky and Latner: 11.3 point fall in labour force participation, 3.2 points attributed to social interaction.Forget, Canadian Public Policy 2011; Calnitsky and Latner[43]
Alaska Permanent Fund Dividend, 1982 to presentNatural experiment; synthetic control; whole stateAnnual; US$3,284 in 2022; US$1,000 plus US$200 energy relief in 2026No significant aggregate employment effect; part-time work up 1.8 points (17%). Substance-abuse incidents up 14% the day after payment, property crime down 8%; annual effects small. Price effects suggestive, uncertain.Jones and Marinescu, AEJ: Economic Policy 2022; Watson, Guettabi and Reimer, Review of Economics and Statistics 2020; Jones, National Tax Journal 2022[6][44][52]
Eastern Cherokee casino dividend, 1996 to presentNatural experiment; cohort difference-in-differences; 1,420 childrenAbout US$4,000 per adult member per yearFor the poorest families: about one more year of education by 21 and 22% lower chance of a minor offence at 16 to 17. No significant fall in parental labour participation. Share below poverty line fell from almost 60% to below 25%.Akee et al., AEJ: Applied Economics 2010[45]
Iran subsidy-reform cash transfer, from Dec 2010Natural experiment; over 70 million coveredIRR 455,000 a month, about US$37 to 44; initially 28% of median per-capita incomeNo overall negative effect on participation or hours; positive effects for some women and self-employed men. Cumulative price growth of 136.5% over five years eroded about 40% of the transfer's poverty-reducing effect.Salehi-Isfahani and Mostafavi-Dehzooei, Journal of Development Economics 2018; Enami and Lustig, ERF 2018[46]
Madhya Pradesh, India, 2011 to 2012Cluster randomised; 2,034 households; 8 treatment, 12 comparison villagesRs 200 per adult, Rs 100 per child monthly, later raisedOdds of a minor illness or injury needing treatment 46% lower; no effect on hospitalisation-level illness or vaccination.Beck et al., Journal of Development Effectiveness 2015[47]
Finland, 2017 to 2018National RCT; 2,000 unemployed recipients aged 25 to 58€560 a month, unconditional, replacing unemployment benefitFirst-year employment difference about 0.5 day; 6 more days employed in the later period, confounded by a 2018 activation reform. High stress 17% vs 25% of controls; good health 55% vs 46%.Kela, VATT, Ministry of Social Affairs and Health 2020[10][11]
Kenya, GiveDirectly 12-year UBI, from 2018Cluster RCT, village saturation; 295 villages; about 23,000 treated adults; 12-year, 2-year and lump-sum armsAbout KSh 2,300 (US$22.50) a month per adult; lump sum about US$500Long-term villages: 14% more enterprises, 52% higher aggregate enterprise net revenue; total labour unchanged, shift from wages to self-employment; depression risk 8 points lower against a 47% control mean. Lump sums matched or beat the long-term stream; short-term drip was least effective.Banerjee, Faye, Krueger, Niehaus and Suri, 2023[8]
Kenya, general-equilibrium cash, 2010sCluster RCT; over 10,500 households, 653 villagesAbout US$1,000 once per householdLocal multiplier about 2.5 (corrected 2.44, SE 1.65); average price inflation 0.1%; positive spillovers to non-recipients.Egger et al., Econometrica 2022, corrected 2025[50][51]
Barcelona B-MINCOME, 2017 to 2019RCT; 1,000 treatment householdsMeans-tested, up to €1,676 a monthFood insecurity down 22.4%, going to bed hungry down 12.0% in unconditional arms. Work effects differed by arm.Barcelona City Council, Institut Metròpoli; Campbell review[48]
Stockton SEED, 2019 to 2021Random assignment; 125 recipientsUS$500 a monthFull-time employment rose from 28% to 40% among recipients vs 32% to 37% among controls (a 7 point difference in changes). Less income volatility, better mental health.West and Castro, Tennessee and Penn[53]
OpenResearch, Illinois and Texas, 2020 to 2023Individual RCT; 1,000 treatment, 2,000 controlsUS$1,000 a month for 36 months vs US$50Participation down 4.2 points (pooled administrative and survey data; earlier survey-only estimate 3.9), hours down 1 to 2 a week, individual income excluding transfers down about US$1,900 a year; spending up at least US$300 a month; wellbeing gains faded after year one; no physical-health effect; emergency-department use rose.Vivalt, Rhodes, Bartik, Broockman, Krause and Miller, Quarterly Journal of Economics 2026; Miller et al. health paper[4][5]
Compton Pledge, 2021 to 2023Household RCT; 695 treatment, 1,402 controlsAbout US$500 a month on averageAt 18 months income excluding transfers US$333 a month lower and spending US$302 lower; part-time workers' participation 13 points lower, no effect for full-time workers; housing security improved; no overall wellbeing effect.Balakrishnan et al., NBER working paper[54]
Baby's First Years, US, 2018 onwardIndividual RCT; 1,000 mother-infant pairsUS$333 vs US$20 a month for 76 monthsAge one: higher high-frequency brain activity in an EEG subsample. Age four: no significant effect on four preregistered primary outcomes including language.Noble, Gennetian et al.; NBER[55]
Chelsea Eats, Massachusetts, 2020 to 2021Lottery RCT with linked health records; 2,880 applicantsUp to US$400 a month for nine monthsEmergency visits 217 vs 318 per 1,000 (down 27%); no effect on blood pressure, weight, HbA1c or cholesterol.Agarwal, Cook and Liebman, JAMA 2024[56]
Germany Pilotprojekt Grundeinkommen, 2021 to 2024Preregistered RCT; 122 recipients, about 1,700 participants€1,200 a month for 36 monthsNo significant employment or hours reduction (about 90% employed, 40 hours a week in both groups). Life satisfaction 7.6 vs 7.1 out of 10; mental health up 0.30 SD; over a third of payments saved.DIW Berlin, Bohmann et al. 2025[57]
Ireland Basic Income for the Arts, 2022 to 2026RCT; 2,000 recipients, about 1,000 controls€325 a weekAt one year: almost 8 more hours a week on creative practice; life satisfaction 0.7 points higher; over 8 points less likely to report anxiety; arts income over €500 a month higher, less non-arts work.Irish Government Economic and Evaluation Service; Alma Economics[58]
Wales care leavers, 2022 to 2025Cohort pilot; 644 recipients (97% of eligible)£1,600 a month gross for 24 monthsMid-pilot interviews show mixed education and work experience; causal evaluation due 2027.Cardiff University CASCADE[59]
Ontario basic income pilot, 2017 to 2019Over 4,000 recipients; cancelled 2018Up to C$16,989 a year single, reduced 50% of earningsVoluntary follow-up survey of 217 people only; no causal estimates survive the cancellation.Ontario government; McMaster[61]
Maricá, Brazil, from 2015Observational matching; 5,182 surveyed300 mumbucas (R$300) a month in local currencyHousehold income including transfers up 9%, consumption up 5%; labour income 17% lower among recipients.Balakrishnan, Costa, Haushofer and Waltenberg, NBER WP 33089[62]
Namibia Otjivero, 2008 to 2009Village saturation, before-after, no control villageN$100 a monthProject reports a 36.5% fall in recorded crime; peer-reviewed assessment criticises the design. Descriptive only.BIG Coalition; Osterkamp, Basic Income Studies[63]
Gyeonggi Youth Basic Income, Korea, from 2019Age-restricted programme; 3,500 surveyedKRW 250,000 a quarter for one year at age 2480.6% satisfaction; wellbeing indicators positive; no causal estimate.Gyeonggi Research Institute[64]
Did people work less? Measured labour effects, ordered by payment size relative to local income no change worked less worked more SIME/DIME, US 1970s, near-poverty-line guaranteehours: husbands −9%, wives −20% [42] Mincome, Canada 1970s, 60% of low-income cutoffparticipation −11.3 pp [43] Maricá, Brazil, local-currency UBIlabour income −17%, observational [62] OpenResearch, US, $1,000/mth to ~$30k householdsparticipation −4.2 pp, −1 to 2 hrs/wk [4] Compton, US, ~$500/mthpart-timers −13 pp, full-timers no effect [54] Germany, €1,200/mth, employed recipientsno significant change [57] Alaska, ~$1,000 to $3,300/yr, permanentemployment unchanged; part-time +1.8 pp [6] Iran, ~28% of median income, nationwideno fall; some women and self-employed up [46] Kenya 12-year UBI, ~US$0.75/daytotal unchanged; wage work → self-employment [8] Finland, €560/mth replacing benefit+6 days employed, "small" [11] Stockton, US, $500/mthfull-time employment +7 pp relative [53] Bars are directional and indicative only; studies measure different outcomes on different populations. The pattern, not the lengths, is the finding.

Figure 4. The pattern across fifty years: the larger the payment relative to what people already earn, and the more it replaces rather than supplements income, the larger the reduction in work. Supplements to working households show nothing or small positive effects. Sources as marked in the table above.

Systematic reviews and meta-analyses

ReviewScope and finding
Banerjee, Niehaus and Suri, Annual Review of Economics 2019[8]Cash studies in developing countries show positive welfare effects and little evidence of feared behaviours; very little direct evidence yet on a truly universal, permanent programme and its economy-wide effects.
Gentilini, Grosh, Rigolini and Yemtsov, World Bank 2020[65]Policy guide, not a pooled meta-analysis. Outcomes depend critically on benefit size, taxes and which existing benefits are replaced.
Hasdell, Stanford Basic Income Lab 2020[67]Umbrella review of 16 reviews: generally positive poverty, health and education results, limited aggregate labour effects; most underlying interventions were not full UBIs.
Baird, McKenzie and Özler, World Bank 2018[68]Transfers without an employment objective generally produce little or no change in adult work; elderly recipients and some refugees are exceptions; enterprise grants more often increase work and earnings.
Rizvi et al., Campbell Systematic Reviews 2024[48]27 studies, 10 high-income-country experiments: pooled self-rated mental distress 0.25 SD lower (95% CI −0.37 to −0.13) across two RCTs; low certainty; employment effects not pooled because of heterogeneity.
Pega et al., Cochrane 2022[69]Unconditional transfers in low- and middle-income countries: pooled risk ratio for any illness 0.79 (95% CI 0.67 to 0.92) across six cluster RCTs and 9,367 participants.

Reasonably settled

Modest unconditional cash does not cause mass withdrawal from work.[4][6][8][57] It improves felt financial security and mental health.[10][48][57] People spend it on essentials.[4] Long commitments and lump sums produce more investment than short drips.[8] Paying everyone in a community amplifies local economic effects.[50] In poor countries it cuts illness and hunger.[69]

4. Australia at a glance, November 2026

Current official figures that give the Australian scale for the data above.

28.09m
resident population, ABS, mid 2026.[103] About 22 million adults (estimate: 79% of population, from ABS age structure).
A$2.85tn
nominal GDP, calendar 2025, sum of IMF/ABS quarterly figures.[108]
29.9%
tax to GDP, all levels, 2023 (OECD average 33.7%). Personal income tax is 42.6% of revenue vs OECD 23.7%; GST 11.1% vs 20.5%.[106]
A$308.7bn
social security and welfare in the 2026-27 Budget, 37.1% of A$833.2bn total spending.[109]
$824.90
JobSeeker single basic rate per fortnight from 20 Sep 2026 (A$21,447 a year). Age Pension single maximum A$1,237.70 (A$32,180).[113]
14.2%
of Australians, 3.706 million people, below the poverty line of A$584 a week (A$30,368 a year) in 2022-23; average shortfall A$390 a week.[104]
A$289.7bn
Future Fund at 30 June 2026; A$356bn across all seven funds; 14.8% return in 2025-26.[105]
4.3%
labour productivity uplift from AI over a decade, about A$116bn of GDP, Productivity Commission 2025.[114]

Questions this page answers

How much of the money in Australia is created by banks rather than the government?

About 97%. The Reserve Bank's Table D3 shows broad money of A$3,462 billion in April 2026 against roughly A$101 billion of notes and coin. Bank deposits, created when banks lend, make up the rest. The shares are 97% in the UK, about 93% in Japan, 91% in the euro area and 90% in the United States.

What share of Australian jobs is highly exposed to generative AI?

Jobs and Skills Australia's 2025 Gen AI Capacity Study finds about 4% of the workforce in occupations with high automation exposure and 79% in occupations with low automation exposure but medium-to-high augmentation potential. Exposure describes tasks AI could affect; it is not a forecast of job losses.

Do people stop working when they receive unconditional cash?

In the studies reviewed, modest payments (Alaska, Kenya, Germany, Finland, Iran) produced no significant fall in overall employment. The largest US trial, US$1,000 a month for three years, reduced work by 1 to 2 hours a week and labour force participation by about 4 points. The 1970s US income-maintenance experiments, with guarantees near the poverty line, cut hours by 9 to 20%.

What does a universal basic income cost?

The IMF's 2017 Fiscal Monitor estimates a UBI at 25% of median income costs about 6.5% of GDP gross in advanced economies and 3.75% in emerging economies, before any clawback through tax.

Has anyone tested a payment large enough to live on without working?

No. Every evaluated experiment to 2026 paid a supplement or a near-poverty-line floor. No randomised trial has paid at or above median income.

Limitations

The money shares compare each central bank's broadest standard aggregate with its own currency series; definitions differ between countries and the US and euro figures are lowered by cash held abroad. AI exposure measures are task-based estimates, not observed job losses, and the observed effects cited (Stanford, Anthropic, Denmark) are described by their authors as descriptive. The experiments table reports nominal amounts, and no experiment to date has tested a payment at or above median income; results for supplements and near-poverty floors should not be extrapolated to larger payments. The Australian adult population used in any per-adult figure is an estimate from the ABS age structure. This page compiles evidence; it is not financial, tax or legal advice.

Sources

Grouped by topic. Where a working paper was later published in a journal, both are given.

Money and value

  1. Tier 1 Bank of England (2014). Money in the modern economy: an introduction. Quarterly Bulletin 2014 Q1. bankofengland.co.uk
  2. Tier 1 McLeay, M., Radia, A. and Thomas, R. (2014). Money creation in the modern economy. Bank of England Quarterly Bulletin 54(1), 14 to 27. bankofengland.co.uk; SSRN 2416234
  3. Tier 1 Reserve Bank of Australia (2018). Money in the Australian Economy. Bulletin, September 2018. rba.gov.au. Debelle, G. (2018). Money: Born of Credit? rba.gov.au
  4. Tier 1 Egger, D., Haushofer, J., Miguel, E., Niehaus, P. and Walker, M. (2022). General Equilibrium Effects of Cash Transfers: Experimental Evidence from Kenya. Econometrica 90(6), 2603 to 2643. doi.org/10.3982/ECTA17945
  5. Tier 1 Egger et al. (2025). Correction to General Equilibrium Effects of Cash Transfers, 31 October 2025. econometricsociety.org
  6. Tier 1 Reserve Bank of Australia. Statistical Table D3, Monetary Aggregates: broad money A$3,462.4bn, currency series, April 2026. rba.gov.au/statistics/tables; mirror db.nomics.world/RBA/D3
  7. Tier 1 Board of Governors of the Federal Reserve System. H.6 Money Stock Measures: M2 and currency component of M1, via FRED. fred.stlouisfed.org/series/CURRNS; fred.stlouisfed.org/series/WM2NS
  8. Tier 1 European Central Bank (2026). Monetary developments in the euro area: September 2026, 27 October 2026, annex table 1 (M3 €17,717bn; currency in circulation €1,618bn). ecb.md2609.pdf
  9. Tier 1 Bank of Japan. Money Stock statistics, M3 (about ¥1,640 trillion, Q2 2026); Banknotes in circulation ¥120.6 trillion at end December 2025. boj.or.jp
  10. Tier 1 Galí, J. (2020). The effects of a money-financed fiscal stimulus. Journal of Monetary Economics 115. doi.org/10.1016/j.jmoneco.2019.08.002. Bernanke, B. (2016). What tools does the Fed have left? Part 3: Helicopter money. Brookings. brookings.edu

Core UBI economics and surveys

  1. Tier 1 International Monetary Fund (2017). Fiscal Monitor, October 2017: Tackling Inequality, chapter 1 (universal basic income costing, 6.5% of GDP in advanced economies at 25% of median income). imf.org
  2. Tier 2 Banerjee, A., Faye, M., Krueger, A., Niehaus, P. and Suri, T. (2023). Universal Basic Income: Short-Term Results from a Long-Term Experiment in Kenya. conference.nber.org. Banerjee, Niehaus and Suri (2019). Universal Basic Income in the Developing World. Annual Review of Economics 11, 959 to 983.
  3. Tier 1 Gentilini, U., Grosh, M., Rigolini, J. and Yemtsov, R. (2020). Exploring Universal Basic Income: A Guide to Navigating Concepts, Evidence, and Practices. World Bank. documents.worldbank.org
  4. Tier 2 Hasdell, R. (2020). What We Know About Universal Basic Income: A Cross-Synthesis of Reviews. Stanford Basic Income Lab. basicincome.stanford.edu
  5. Tier 1 Baird, S., McKenzie, D. and Özler, B. (2018). The effects of cash transfers on adult labor market outcomes. IZA Journal of Development and Migration; World Bank Policy Research WP 8404.
  6. Tier 1 Pega, F. et al. (2022). Unconditional cash transfers for reducing poverty and vulnerabilities: effect on use of health services and health outcomes in low- and middle-income countries. Cochrane Database of Systematic Reviews. cochranelibrary.com

Experiments and natural experiments

  1. Tier 1 Vivalt, E., Rhodes, E., Bartik, A.W., Broockman, D.E., Krause, P. and Miller, S. (2026). The Employment Effects of a Guaranteed Income: Experimental Evidence from Two U.S. States. Quarterly Journal of Economics 141(4). academic.oup.com; NBER WP 32719 nber.org
  2. Tier 2 Miller, S. et al. (2024). Does Income Affect Health? Evidence from a Randomized Controlled Trial of a Guaranteed Income. NBER Working Paper. Summary: poverty.umich.edu
  3. Tier 1 Jones, D. and Marinescu, I. (2022). The Labor Market Impacts of Universal and Permanent Cash Transfers: Evidence from the Alaska Permanent Fund. AEJ: Economic Policy 14(2), 315 to 340. aeaweb.org; data openICPSR
  4. Tier 1 Finnish Ministry of Social Affairs and Health and Kela (2020). Results of the basic income experiment. stm.fi; Kela information package
  5. Tier 1 VATT Institute for Economic Research (2020). vatt.fi; Finland Toolbox summary toolbox.finland.fi
  6. Tier 2 University of Chicago Inclusion Economics Lab (2024). Guaranteed Income Literature Review. urbanlabs.uchicago.edu
  7. Tier 2 Institute for Research on Poverty, University of Wisconsin. The Negative Income Tax Experiments archive and Focus summary. irp.wisc.edu; Focus PDF
  8. Tier 1 US Department of Health and Human Services, ASPE. Overview of the Final Report of the Seattle-Denver Income Maintenance Experiment. aspe.hhs.gov. Burtless, G. (1986). The Work Response to a Guaranteed Income. Federal Reserve Bank of Boston. bostonfed.org
  9. Tier 1 Forget, E. (2011). The Town with No Poverty. Canadian Public Policy 37(3), 283 to 305. econpapers; Green, D. reanalysis economics.ubc.ca; Calnitsky and Latner, labour supply.
  10. Tier 1 Watson, B., Guettabi, M. and Reimer, M. (2020). Universal Cash and Crime. Review of Economics and Statistics 102(4), 678 to 689. ideas.repec.org
  11. Tier 1 Akee, R., Copeland, W., Keeler, G., Angold, A. and Costello, E.J. (2010). Parents' Incomes and Children's Outcomes: A Quasi-Experiment Using Transfer Payments from Casino Profits. AEJ: Applied Economics 2(1), 86 to 115. aeaweb.org
  12. Tier 1 Salehi-Isfahani, D. and Mostafavi-Dehzooei, M. (2018). Cash transfers and labor supply: Evidence from a large-scale program in Iran. Journal of Development Economics 135, 349 to 367. doi.org. Enami, A. and Lustig, N. (2018). The Wrecking Force of Inflation. ERF Policy Brief 37. erf.org.eg
  13. Tier 1 Beck, S., Pulkki-Brännström, A.-M. and San Sebastián, M. (2015). Basic income: healthy outcome? Journal of Development Effectiveness 7(1).
  14. Tier 1 Rizvi, A. et al. (2024). The effects of guaranteed basic income interventions in high-income countries. Campbell Systematic Reviews. campbellcollaboration.org. Barcelona B-MINCOME evaluation, Institut Metròpoli.
  15. Tier 1 Jones, D. (2022). Universal Cash Transfers and Inflation. National Tax Journal 75(3). doi.org/10.1086/720770
  16. Tier 2 West, S. and Castro, A. (2021). Preliminary Analysis: SEED's First Year. University of Tennessee and University of Pennsylvania. evidenceforaction.org
  17. Tier 2 Balakrishnan, S. et al. (2025). The Compton Pledge: guaranteed income RCT. NBER working paper; Stanford summary news.stanford.edu
  18. Tier 2 Noble, K., Gennetian, L. et al. Baby's First Years. babysfirstyears.com
  19. Tier 1 Agarwal, S., Cook, B. and Liebman, J. (2024). Effect of Cash Benefits on Health Care Utilization and Health: A Randomized Study. JAMA. jamanetwork.com
  20. Tier 2 Bohmann, S. et al. (2025). Pilotprojekt Grundeinkommen. DIW Berlin Wochenbericht 15/2025. diw.de
  21. Tier 1 Irish Government Economic and Evaluation Service (2024). Basic Income for the Arts: Impact Assessment, First Year. assets.gov.ie; Alma Economics cost-benefit (2025) assets.gov.ie
  22. Tier 1 Welsh Government and Cardiff University CASCADE (2026). Basic Income for Care Leavers in Wales pilot evaluation.
  23. Tier 2 McMaster University Labour Studies (2020). Southern Ontario's Basic Income Experience. labourstudies.mcmaster.ca
  24. Tier 2 Balakrishnan, S., Costa, F., Haushofer, J. and Waltenberg, F. (2024). Maricá basic income. NBER Working Paper 33089.
  25. Tier 1 Osterkamp, R. (2013). The Basic Income Grant Pilot Project in Namibia: A Critical Assessment. Basic Income Studies 8(1).
  26. Tier 2 Gyeonggi Research Institute (2019). Youth Basic Income satisfaction survey and evaluation.

AI, jobs and the labour share

  1. Tier 1 Cazzaniga, M. et al. (2024). Gen-AI: Artificial Intelligence and the Future of Work. IMF Staff Discussion Note 2024/001. imf.org
  2. Tier 2 Briggs, J. and Kodnani, D. (2023). The Potentially Large Effects of Artificial Intelligence on Economic Growth. Goldman Sachs Research. gspublishing.com. Briggs, J. (2026). How Will AI Impact the Labor Market? goldmansachs.com
  3. Tier 2 Brynjolfsson, E., Chandar, B. and Chen, R. (2025, rev. 2026). Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford Digital Economy Lab. digitaleconomy.stanford.edu
  4. Tier 2 McKinsey Global Institute (2026). Workforce in Motion. mckinsey.com
  5. Tier 2 Stanford HAI (2026). AI Index Report 2026, Economy chapter. hai.stanford.edu
  6. Tier 2 Massenkoff, M. and McCrory, P. (2026). Labor Market Impacts of AI. Anthropic. anthropic.com; Anthropic Economic Index, January 2026 anthropic.com
  7. Tier 2 Humlum, A. and Vestergaard, E. (2025, rev. 2026). Still Waters, Rapid Currents. NBER WP 33777. nber.org
  8. Tier 2 Yotzov, I. et al. (2026). Firm Data on AI. NBER WP 34836. nber.org
  9. Tier 1 OECD (2026). Employment Outlook 2026. oecd.org
  10. Tier 1 Acemoglu, D. (2024). The Simple Macroeconomics of AI. MIT working paper; Economic Policy 2025. economics.mit.edu
  11. Tier 2 Karger, E. et al. (2026). Forecasting the Economic Effects of AI. NBER WP 35046. nber.org
  12. Tier 2 Brookings, The Hamilton Project (2026). Building Pro-Worker Artificial Intelligence. brookings.edu

Australia

  1. Tier 1 Jobs and Skills Australia (2025). Our Gen AI Transition: Implications for Work and Skills. Final Overarching Report, 14 August 2025; Analysis Papers, 2 and 30 September 2025. jobsandskills.gov.au; study hub
  2. Tier 2 Centre for Future Work / Australia Institute. The Shrinking Labour Share of GDP (ABS national accounts data). futurework.org.au; Flanagan, Stanford, Peetz and Wilson, Exploring the decline in the labour share of GDP, JAPE symposium futurework.org.au
  3. Tier 1 La Cava, G. (2019). The Labour and Capital Shares of Income in Australia. RBA Bulletin, March 2019. rba.gov.au. ABS (2018). Trends in the Labour Income Share in Australia, 5260.0.55.002 feature article. abs.gov.au
  4. Tier 1 Australian Bureau of Statistics (2026). National, state and territory population, release 18 June 2026; population reached 28 million on 2 June 2026. abs.gov.au. Reported figures 27.59m (2025) to 28.09m (2026) via The Australia Today Tier 3
  5. Tier 2 Davidson, P. and Bradbury, B. (2025). Poverty in Australia 2025: Overview. ACOSS and UNSW Sydney, October 2025. apo.org.au; DSS FLOSSE record flosse.dss.gov.au; ACOSS rent position paper via apo.org.au
  6. Tier 1 Future Fund (2026). Portfolio update at 30 June 2026, released 26 August 2026 (A$289.7bn; A$356.0bn total FUM; 14.8% return). Department of Finance estimates brief, October 2025 (A$252.3bn at 30 June 2025; A$60.5bn endowment). finance.gov.au
  7. Tier 1 OECD (2025). Revenue Statistics 2025: Australia country table and key figures (tax to GDP 29.9% in 2023; PIT 42.6% of revenue; VAT 11.1%). oecd.org; Australia country note
  8. Tier 1 International Monetary Fund, International Financial Statistics via FRED. Nominal GDP for Australia, quarterly, seasonally adjusted, A$ millions: Q1 2025 699,805; Q2 704,909; Q3 718,289; Q4 731,418. fred.stlouisfed.org
  9. Tier 1 Commonwealth of Australia (2026). Budget 2026-27, 12 May 2026: total expenses A$833.2bn; social security and welfare 37.1%, A$308.7bn. Reported via SBS sbs.com.au and GlobalData finance.yahoo.com Tier 3; Services Australia PBS servicesaustralia.gov.au
  10. Tier 1 Department of Social Services (2026). Social Security Payment Parameters, 20 September 2026 indexation. dss.gov.au
  11. Tier 1 Productivity Commission (2025). Harnessing data and digital technology. Interim report 5 August 2025; Inquiry report no. 111, 10 December 2025. assets.pc.gov.au