Oseh Mathias
← Back to Writing

What provisioned macaque populations in Indonesia suggest about UBI

•13 min read

In Ubud, Bali, I watched long-tailed macaques pick up stones and clack them together, turn them over, and strike again.

I had seen that motion before.

As a kid in Plymouth, Massachusetts, I spent summers at a living-history site. The Native Americans there showed us how to chip arrowheads from stone in much the same way.

How did the monkeys arrive at this?

Macaques are sacred in Balinese Hindu culture, so the troop I observed has been fed almost entirely by humans for decades. Relieved of the daily work of finding food, the animals turn to other activities with their hands.

Stone handling as a sustained, inherited tradition appears only in provisioned populations, first documented in Japanese macaques at Arashiyama in 1979 and since in at least four species, every one of them fed by humans (Huffman & Quiatt, 1986) (Leca et al., 2008). On Koshima Island in 1953 a young female named Imo began washing sweet potatoes in a stream; the habit spread through her kin and playmates, and later generations improved on it, switching from fresh water to seawater to salt their food (Kawai, 1965) (Hirata et al., 2001). At a shrine in Lopburi, Thailand, provisioned macaques learned to floss their teeth with human hair, and mothers exaggerated the motion when their infants were watching, which looks a great deal like teaching (Watanabe et al., 2007). At Uluwatu Temple the troop runs a protection racket, stealing tourists' phones and sunglasses and ransoming them back for food, holding out for more when the item is worth more (Brotcorne et al., 2017) (Leca et al., 2021). I have fallen victim to such banditry myself.

All of these behaviors are absent in wild members of the same species. The common factor is provisioning. Remove the pressure of survival and the animals begin to experiment, what one invents the others copy, and what they copy later generations refine. Huffman and Quiatt traced the arc from transmission to tradition to transformation, the same sequence archaeologists imagine for early Homo (Huffman & Quiatt, 1986). Tool use, teaching, the rough valuation of goods. These are rungs of our own ascent, and provisioned monkeys climb them in decades.

Which raises the obvious question. Macaca fascicularis is the pharmaceutical industry's gold standard model for the human body, the most heavily traded primate in the world, a stand-in for us on the strength of our shared DNA. Provisioning ran that stand-in up the path we ourselves climbed. So what does it do to the species already standing at the top? If a guaranteed food supply pushes monkeys toward us, where does it push us?

For nearly sixty years, governments have been running it on people, paying thousands of families to live without the pressure of securing their basic needs and watching what they did.

First Experiments

The United States nearly adopted a guaranteed income in 1970. President Nixon's Family Assistance Plan passed the House of Representatives twice before dying in the Senate, a victim of opposition from both conservatives who thought it too generous and liberals who thought it not generous enough. What killed it politically, however, was data, or rather, a premature interpretation of data from the largest social science experiments ever conducted.

Between 1968 and 1982, the U.S. and Canadian governments spent approximately $225 million (1984 dollars) to answer the question, "If you guarantee people's basic income, will they stop working?" Five major randomized controlled trials enrolled thousands of families across New Jersey, rural Iowa and North Carolina, Gary (Indiana), Seattle, Denver, and the Canadian province of Manitoba.

New Jersey (1968–1972)

The New Jersey experiment was the first large-scale social RCT conducted outside a laboratory. Researchers enrolled 1,216 low-income two-parent families and randomized them to receive guaranteed income at various levels, from 50% to 125% of the poverty line (Kershaw & Fair, 1972). The results, released while the experiment was still running, showed almost no work reduction among men. To the surprise of absolutely nobody, this finding temporarily boosted Nixon's proposal. But later analysis revealed that husbands reduced work by 1–4 weeks per year, and wives somewhat more. The changes were real but modest, nowhere near the mass exodus from employment that critics predicted.

Gary, Indiana (1971–1974)

The Gary experiment focused specifically on Black families, with 59% headed by single women, which was a population absent from other trials (Kehrer, 1977). Single mothers showed larger labor reductions than married couples, but the experiment revealed something else. Birth weights among recipient families increased significantly, children attended school more regularly, and homeownership rose from 23% to 34%. The money wasn't creating dependency; it was creating stability.

SIME/DIME Ended Guaranteed Income

The Seattle-Denver Income Maintenance Experiment was designed to be definitive. It enrolled 4,800 families, ran for up to 20 years for some participants, and tested the most generous benefit levels of any trial, providing up to 140% of the poverty line (Robins & Spiegelman, 1978).

The labor supply findings were larger than previous experiments, showing that husbands reduced work by 9%, wives by 18%, and single mothers by about 20%. More troubling to policymakers, the program appeared to increase divorce rates by 40–60% among some groups. Senator Daniel Patrick Moynihan announced these findings on the Senate floor in 1978, and guaranteed income vanished from serious policy discussion for four decades.

The marital stability findings were later challenged. Glen Cain's 1986 reanalysis found the divorce effects statistically unreliable, because half the coefficients actually showed marriage-stabilizing effects (Cain, 1986). But the political damage was done.

Manitoba's Mincome, which Evelyn Forget remembered

Canada's Mincome experiment included a saturation site, something no American trial attempted. In Dauphin, Manitoba, any resident could apply for benefits, allowing researchers to measure community-wide effects rather than just individual responses (Hum & Simpson, 1993).

Then the political winds shifted. Conservative governments replaced progressive ones, and the experiment was terminated without a final report. 1,800 boxes of documents sat in storage for decades.

In the 2000s, an economist named Evelyn Forget, of all people, remembered the experiment, found boxes, and performed an overdue analysis. What she found was striking. During the Mincome years, hospitalizations in Dauphin fell 8.5% relative to comparison communities (Forget, 2011). Mental health admissions declined. Domestic violence dropped. Automobile accidents decreased. The two groups who did withdraw from work were new mothers (who stayed home longer with infants) and teenage boys (who stayed in school instead of dropping out).

The health improvements exceeded what individual income transfers could explain. Forget identified a social multiplier; when an entire community's floor rises, the effects cascade through reduced stress, improved social cohesion, and changed community norms.

Silence

For thirty years after SIME/DIME, guaranteed income was politically radioactive in North America. Yet, the conversation continued elsewhere.

The Longest-Running Natural Experiment is in Alaska

Since 1982, Alaska has distributed oil revenue to every resident through the Permanent Fund Dividend, typically $1,000 - $2,000 annually, with some years exceeding $3,000 (Goldsmith, 2010). It's not enough to live on, but it's universal, unconditional, and has now run for over four decades.

Jones and Marinescu's rigorous analysis found no effect on employment (Jones & Marinescu, 2022). Part-time work increased modestly, particularly among women. Poverty fell 20–40%. The cash stimulated local spending, offsetting any theoretical work disincentive. If guaranteed income were going to produce societal collapse, Alaska would have shown some sign of it by now.

GiveDirectly Rewrote the Development Playbook

The most rigorous evidence on cash transfers now comes from Kenya, where GiveDirectly has delivered over $900 million to 1.7 million people since 2009.

Short-term results published in the Quarterly Journal of Economics showed consumption increased 22%, assets increased 61%, and psychological wellbeing improved substantially (Haushofer & Shapiro, 2016). The feared increase in alcohol and tobacco spending? It didn't happen. Three-year follow-ups found recipients had preserved 60% of the initial transfer value in durable assets (Haushofer & Shapiro, 2018).

The largest study, published in Econometrica, examined general equilibrium effects across 653 villages. Every dollar transferred generated 2.50–2.50–2.70 in local economic activity, with positive spillovers to non-recipient households and businesses (Egger et al., 2022). Inflation was negligible.

Perhaps most significant, GiveDirectly's ongoing comparison of payment structures found that lump sums outperformed equivalent monthly payments on nearly every financial metric (Banerjee et al., 2023). More businesses started, higher revenues, greater investment. The most common cash transfer design, consisting of small regular payments, may be the least effective for poverty reduction.

A 2025 follow-up found cash transfers produced a 48% reduction in infant mortality, one of the largest reductions ever recorded for any poverty intervention (McIntosh et al., 2025).

Iran Tried Universal Cash at Scale

In 2011, Iran replaced energy subsidies with direct cash payments to over 70 million people, representing 29% of median household income, delivered to nearly the entire population. It was the closest any country has come to implementing UBI at national scale.

The predicted economic catastrophe didn't materialize. Analysis published in the Journal of Development Economics found no negative labor supply effects (Salehi-Isfahani & Mostafavi-Dehzooei, 2018). Women actually increased their work hours. Youth showed slight work reductions, attributed to pursuing education rather than weak labor market attachment.

The New Wave of Experiments (2017–Present)

Finland's Wellbeing Over Employment

Finland's 2017–2018 experiment was the first statutory nationwide basic income RCT. Two thousand unemployed recipients received €560 monthly with no conditions, compared to 173,000 controls (Kangas et al., 2019).

Employment effects were modest in response, resulting in about six additional days of work per year by the second year. But wellbeing effects were substantial and consistent. Life satisfaction rose from 6.8 to 7.3 on a 10-point scale, trust in institutions increased, and recipients showed significantly lower depression, loneliness, and mental strain (Kangas et al., 2020).

The Finnish government, hoping for dramatic employment gains, interpreted modest effects as failure and didn't extend the program. Whether improved wellbeing constitutes success depends on what you're measuring.

Stockton SEED's Counterintuitive Finding

The Stockton Economic Empowerment Demonstration provided $500 monthly to 125 residents in a California city that had recently emerged from bankruptcy (West & Castro Baker, 2021).

The headline finding reversed conventional predictions. Full-time employment increased from 28% to 40% among recipients, a 12 percentage point gain compared to just 5 points for controls. Recipients weren't withdrawing from work; the income floor was enabling them to find better jobs, take risks, and stabilize their lives enough to pursue employment.

Spending tracked carefully. 37% went to food, 22% to home goods and clothing, 11% to utilities. Alcohol and tobacco? Less than 1%.

SAMA's OpenResearch is the Largest American Experiment

Sam Altman's OpenResearch enrolled 1,000 participants receiving $1,000 monthly for three years, with 2,000 controls receiving $50 monthly (Vivalt et al., 2024). At $60 million, it was the largest American basic income experiment ever conducted.

Results were more complex than earlier studies. Employment fell modestly, about 1.3 fewer hours per week. More concerning, the mental health improvements that appeared strongly in Year 1 had faded by Years 2 and 3. Stress reduction, life satisfaction gains, and reduced mental distress didn't persist.

Whether this reflects methodological issues, habituation effects, or the unique context (the study ran during COVID and record inflation) remains unclear. It's the first major experiment to suggest wellbeing gains may not be permanent.

What the Evidence Shows

Across 115 studies covering 72 unconditional cash transfer programs, a 2024 meta-analysis found positive average effects on labor supply, which is the opposite of the dependency narrative (Baird et al., 2024). People don't stop working when given money. In many contexts, they work more, or work differently, or invest in ways that increase future earnings.

Some things are consistent. Temptation goods spending doesn't increase. Health improves. Children stay in school longer. Domestic violence decreases. Mental health improves, at least initially. These findings replicate across populations in Finnish cities and Kenyan villages, across American suburbs and Iranian provinces.

What varies is magnitude. Effects are largest where baseline poverty is deepest. For example, India's SEWA pilots showed savings tripling and business startups doubling because the transfers represented a third of subsistence income (Standing, 2013). Finland's modest employment effects reflected that recipients already had access to generous welfare state benefits.

Program design also matters enormously. SIME/DIME's high benefit reduction rates created strong disincentives, where each dollar earned meant losing 50–80 cents in benefits. Modern programs with lower phase-out rates generate different responses. GiveDirectly's finding that lump sums outperform monthly payments suggests the intuitive approach (small regular amounts) may be backwards.

Unknowns

There is still much to be desired in studies of UBI.

Firstly, I want to see a 2,000 person study where individuals are provisioned from birth to death, over approximately 85 years, and compared to controls. Most experiments run 2–3 years. Only Alaska provides multi-decade evidence, and it's not a true basic income. OpenResearch's fading wellbeing gains raise questions about whether benefits persist or habituate.

Secondly, the optimal design remains uncharacterized. What's the right amount? $2,000 USD / month seems right. How do we know for sure? Also, monthly or lump sum? Universal or means-tested? Taxed back at what rate? Methodological refinement is called for.

Thirdly, while this is not a blocker to privately-funded research, we need to know what variables must be true for UBI to sustain enough political support to show long-lasting outcomes. Mincome was terminated by political change. The Child Tax Credit expansion lasted one year. Finland didn't extend its experiment. Programs require sustained political support that may not survive electoral cycles.

Fourthly, it would be interesting to know what happens when an entire nation implements basic income. Venezuela, if it can emerge from its socialist dictatorship, is a good candidate for this aspect of the study as it has both deep poverty rates and the largest oil reserves in the world. GiveDirectly found positive spillovers in Kenya, but those were village-level interventions in a developing economy. General equilibrium effects on wages, prices, and labor markets remain theoretical. One nation should make it empirical.

Conclusion - UBI is a tool to lift an area out of poverty

If the evidence will bear one conclusion, it is this. UBI is a poverty-exit ramp, most powerful where the hole is deepest.

People do not appear to squander the money. The predicted run on alcohol and tobacco did not happen, in Kenya, in Stockton, or anywhere it was measured with care. Temptation spending is flat or lower. This is among the most replicated results in the field. Give poor people cash and they do not drink it.

The dependency story is also false. In rich countries and poor ones, across sixty years of trials, no population stopped working in any sense resembling collapse.

But the most replicated finding is that the effects are largest, most consistent, and most durable where baseline poverty is deepest. Gary's recipients raised homeownership from 23 to 34 percent and bore heavier, healthier babies. GiveDirectly recipients held 60 percent of their transfer in durable assets three years on. India's SEWA villages tripled savings and doubled business starts. In Kenya, every dollar generated more than two in local activity, with no meaningful inflation.

Drop cash into a poor place and the floor appears to rise.

References

  1. Huffman, M., & Quiatt, D. (1986). Stone Handling by Japanese Macaques (Macaca fuscata): Implications for Tool Use of Stone. Primates, 27, 413-423. doi:10.1007/BF02381887
  2. Leca, J., Gunst, N., & Huffman, M. (2008). Of Stones and Monkeys: Testing Ecological Constraints on Stone Handling, a Behavioral Tradition in Japanese Macaques. American Journal of Physical Anthropology, 135(2), 233-244. doi:10.1002/ajpa.20726
  3. Kawai, M. (1965). Newly-Acquired Pre-Cultural Behavior of the Natural Troop of Japanese Monkeys on Koshima Islet. Primates, 6(1), 1-30. doi:10.1007/BF01794457
  4. Hirata, S., Watanabe, K., & Kawai, M. (2001). "Sweet-Potato Washing" Revisited. Primate Origins of Human Cognition and Behavior, 487-508. doi:10.1007/978-4-431-09423-4_24
  5. Watanabe, K., Urasopon, N., & Malaivijitnond, S. (2007). Long-Tailed Macaques Use Human Hair as Dental Floss. American Journal of Primatology, 69(8), 940-944. doi:10.1002/ajp.20403
  6. Brotcorne, F., Girber, G., Gunst, N., Fuentes, A., & Leca, J. (2017). Intergroup Variation in Robbing and Bartering by Long-Tailed Macaques at Uluwatu Temple (Bali, Indonesia). Primates, 58(4), 505-516. doi:10.1007/s10329-017-0611-1
  7. Leca, J., Gunst, N., & Brotcorne, F. (2021). The Ecology and Evolution of Robbing and Bartering Behaviors in Long-Tailed Macaques. Philosophical Transactions of the Royal Society B, 376(1819). doi:10.1098/rstb.2019.0677
  8. Kershaw, D., & Fair, J. (1972). The New Jersey Income-Maintenance Experiment. [Link]
  9. Kehrer, K. (1977). The Gary Income Maintenance Experiment: Summary of Initial Findings. The Final Report of the Gary Income Maintenance Experiment. [Link]
  10. Robins, P., & Spiegelman, R. (1978). A Review of the Seattle/Denver Income Maintenance Experiment. [Link]
  11. Cain, G. (1986). The Income Maintenance Experiments and the Issues of Marital Stability and Family Composition. Lessons from the Income Maintenance Experiments, 60-93. [Link]
  12. Hum, D., & Simpson, W. (1993). Economic Response to a Guaranteed Annual Income: Experience from Canada and the United States. Journal of Labor Economics, 11(1), S263-S296. doi:10.1086/298335
  13. Forget, E. (2011). The Town with No Poverty: The Health Effects of a Canadian Guaranteed Annual Income Field Experiment. Canadian Public Policy, 37(3), 283-305. doi:10.3138/cpp.37.3.283
  14. Goldsmith, S. (2010). The Alaska Permanent Fund Dividend: A Case Study in Implementation of a Basic Income Guarantee. [Link]
  15. Jones, D., & Marinescu, I. (2022). The Labor Market Impacts of Universal and Permanent Cash Transfers: Evidence from the Alaska Permanent Fund. American Economic Journal: Economic Policy, 14(2), 315-340. doi:10.1257/pol.20190299
  16. Haushofer, J., & Shapiro, J. (2016). The Short-Term Impact of Unconditional Cash Transfers to the Poor: Experimental Evidence from Kenya. Quarterly Journal of Economics, 131(4), 1973-2042. doi:10.1093/qje/qjw025
  17. Haushofer, J., & Shapiro, J. (2018). The Long-Term Impact of Unconditional Cash Transfers: Experimental Evidence from Kenya. [Link]
  18. Egger, D., Haushofer, J., Miguel, E., Niehaus, P., & Walker, M. (2022). General Equilibrium Effects of Cash Transfers: Experimental Evidence from Kenya. Econometrica, 90(6), 2583-2627. doi:10.3982/ECTA17945
  19. Banerjee, A., Faye, M., Krueger, A., Niehaus, P., & Suri, T. (2023). Long-Term Effects of Unconditional Cash Transfers: Experimental Evidence from Kenya. [Link]
  20. McIntosh, C., Haushofer, J., Miguel, E., Niehaus, P., & Walker, M. (2025). Cash Transfers and Infant Mortality: Evidence from a Long-Term Experiment in Kenya. [Link]
  21. Salehi-Isfahani, D., & Mostafavi-Dehzooei, M. (2018). Cash Transfers and Labor Supply: Evidence from a Large-Scale Program in Iran. Journal of Development Economics, 135, 349-367. doi:10.1016/j.jdeveco.2018.08.005
  22. Kangas, O., Jauhiainen, S., Simanainen, M., & Ylikännö, M. (2019). The Basic Income Experiment 2017-2018 in Finland: Preliminary Results. [Link]
  23. Kangas, O., Jauhiainen, S., Simanainen, M., & Ylikännö, M. (2020). Evaluation of the Finnish Basic Income Experiment. [Link]
  24. West, S., & Castro Baker, A. (2021). Stockton Economic Empowerment Demonstration: Year 2 Findings. [Link]
  25. Vivalt, E., Orkin, E., & Malmendier, U. (2024). The Employment Effects of a Guaranteed Income: Experimental Evidence from Two U.S. States. [Link]
  26. Baird, S., McKenzie, D., & McIntosh, C. (2024). Labor Market Effects of Cash Transfers: A Meta-Analysis. [Link]
  27. Standing, G. (2013). India's Experiment in Basic Income Grants. Global Dialogue, 3(5), 24-26. [Link]