Evidence so far 

Past, present, and future evidence on basic income.


The evidence we have on basic income can be grouped into several categories:

  1. An initial wave of negative income tax experiments in the US and Canada
  2. Later experiments in developing countries
  3. Government cash transfer programs in developing countries
  4. Upcoming pilots in several regions
  5. Smaller programs, “natural experiments” and anecdotes from around the world

Across all these categories, we have not found an evaluation of a basic income program that is:

The evidence on cash transfers in general gives plenty of reason to be intrigued about basic income. We summarize this evidence on our website. The Overseas Development Institute has also published a review of 165 studies on cash transfers. We discuss specific examples from that evidence in section 3.


1. Initial negative income tax experiments in the US and Canada

Between 1968 and 1980, the US government conducted negative income tax (NIT) pilots in four cities, and the Canadian government conducted a pilot in Manitoba. All the studies targeted only poor families, and all provided a benefit pegged to the national poverty line (at the time, roughly $4,000 for a family of four in the United States). The benefit varied based on the family’s income level: as a family got closer to the poverty line via income of its own, it would receive less and less assistance, and at a certain point lose it entirely.

Together, these studies suggested three tentative conclusions.

Commentators have continued to debate the relative significance of these conclusions, and to question how much weight should be placed on them given methodological issues with the studies. In particular, the question of how much the programs reduced work hours (and how important that is) has remained contentious. (We should note that this kind of work reduction not been found in cash transfer studies in the developing world.)

Program

Recipients

Universal

Basic

Long-term

RCT

New Jersey, 1968–1972

1216

Seattle/Denver

4800

✗ **

Rural Iowa & North Carolina, 1970–1972

809

Gary, Indiana, 1971–1974

1799

Manitoba (Mincome), 1975–1978

1300

✗ *

*

Included a single saturation site as part of overall study

**

Provided long-term payments to only 169 families

New Jersey Graduated Work Incentive Experiment

Sample size: 1,216
Eligibility: Randomly selected two-parent families in urban areas with a male head aged 18-58 and income below 150% of poverty line
Period: 1968-72

The New Jersey study was the U.S. government’s first attempt to test the relatively novel idea of a negative income tax (which had been championed by economists like Milton Friedman and recently embraced by politicians in both parties). Like the other NIT studies of the time, it focused mostly on whether the new policy discouraged work rather than on the other impacts of an NIT. The official report concluded that there was a “quite small” reduction in hours worked.

Rural Income-Maintenance Experiment, Iowa and North Carolina (RIME)

Sample size: 809
Eligibility: Randomly selected both two-parent families and female-headed households in rural areas with income below 150% of poverty line
Period: 1970-72

RIME was created as a rural complement to the more urban New Jersey study. Its official report found little clear evidence on hours worked (response varied by region, and average productivity sank as average hours increased). It also found participants had more adequate nutrition, improved school performance among children, and “clearly improved” material prosperity.

Seattle/Denver Income-Maintenance Experiments (SIME/DIME)

Sample size: 4,800
Eligibility: Randomly selected families with at least one dependent and incomes below $1,100 for single parents, $13,000 for two-parent families
Period: 1970-76/80

The largest of the early negative income tax studies, SIME/DIME was originally planned to run for six years, but was later authorized to extend to 20 years for roughly 6% of the sample. After initial results suggested a stronger than expected decrease in hours worked, the experiment lost political support and the extension ended early: the 169 recipients promised 20 years of payments received only nine, and experimental records were only kept for seven. The official report issued afterward covered only recipients who had three to five years of payments. It suggested that effects on hours worked varied with benefit amounts, and pointed to increased rates of both schooling and divorce. Subsequent literature refuted the divorce findings and cast general doubts on the methodology of the study, citing low sample size of the long-term study arm and other biasing factors.

Gary, Indiana Income Maintenance Experiment

Sample size: 1,799
Eligibility: Randomly selected households, primarily female-headed, aged 18-58, income below 240% of poverty line
Period: 1971-74

This experiment was complicated by the fact that a large part of the sample stopped participating before the end of the study. This high attrition seemed to stem largely from higher income earners, who did not receive significant benefits from the program and hence had little incentive to remain enrolled. A final, official report of the study was never published widely; subsequent researchers have judged it points to some reduction in hours worked.

Manitoba Basic Annual Income Experiment (Mincome)

Sample size: 1,300
Eligibility: Randomly selected families with head younger than 58 and income below $13,000 for family of four and one saturation site in Dauphin
Period: 1975-78

The last of the major negative income tax experiments, Mincome was canceled suddenly after a political transition in Canada, even before its data was analyzed. In 1993, researchers examined the unused data and concluded that the effect of the program on hours worked was “smaller than would have been expected without experimentation.”  Still later, in 2011, a researcher used health administrative data to revisit the impacts of the Guaranteed Annual Income in the saturation site as compared to the surrounding areas. She found a variety of health-related improvements, including “a significant reduction in hospitalization, especially for admissions related to mental health and to accidents and injuries.”


2. Recent basic income experiments in developing countries

No large-scale basic income studies have taken place in developed countries since this initial wave. However, there have been two recent examples in developing economies with generally positive outcomes.

Program

Recipients

Universal

Basic

Long-term

RCT

Madhya Pradesh Unconditional Cash Transfers Project

6,000

Namibia Basic Income Grant Pilot Project

930

Madhya Pradesh Unconditional Cash Transfers Project

Sample size: 6,000
Eligibility: All residents of eight villages
Period: 2012-2014

This controlled study run by UNICEF and the Self-Employed Women’s Association gave recipients up to 300 rupees (~4.50 USD) per month for up to 17 months, with smaller payments given to parents on behalf of children. These payments were intentionally sized below the level required for basic needs, so as not to make them a “substitute for employment.” The study found improved nutrition, improved school attendance, increases in assets, and - unlike the NIT trials - increases in hours worked.

Namibia Basic Income Grant Pilot Project

Sample size: 930
Eligibility: All residents of two villages under 60 years old
Period: 2007-2009

This study by the Basic Income Grant Coalition (a local network of churches and social service organizations) gave recipients in two villages 100 Namibian dollars (~7.00 USD) per month for roughly two years, with equally sized payments given to parents on behalf of children. While it found a number of positive outcomes, it was conducted without a control group.


3. Recent government cash transfer programs in developing countries

As our website summarizes, a large body of research has established that cash transfers in general work to reduce poverty. And while basic income has not been fully implemented as government policy in any country, certain other types of cash transfer programs have already taken root around the world. The World Bank reports that over 50% of countries have incorporated cash transfers into their social safety nets, with the three largest programs in Brazil, Mexico, and China. While none of these programs delivers a true basic income, they may be seen as relevant examples of cash transfers at a large scale.

Program

Recipients

Universal

Basic

Long-term

RCT

Bolsa Familia (Brazil)

~58M

 *

PROSPERA (Mexico)

~32M

 *

✔ **

 **

Dibao (China)

~78M

✗ ***

✗ ****

*

Unclear what proportion of basic needs are covered by transfers. Benefit amounts vary and may lift some families above national poverty thresholds, but average transfer per individual is below global poverty line.

**

Program itself is long-term, but RCT covered only short term (first 3 years)

***

While means-testing has been lax, program still attempts to target poor, and does not reach everyone in given area

****

Benefit amounts vary widely by area; average per individual is well below global poverty line

Brazil and Mexico have the two largest conditional cash transfer programs. Both target poor families living below the poverty line, and require as conditions the enrollment of children in school and the acceptance of vaccination and pre-natal care.

Bolsa Familia (Brazil)

Current coverage: ~58M individuals (~29% of population)
Amounts: Average of ~182 reals (57 USD) per family per month
Eligibility: Families in poverty whose mothers accept medical care surrounding births and whose children have been vaccinated and attend school

Brazil’s cash transfer program began in 1995 as a set of regional pilots, expanding to national scope in 2001 and merging under the Bolsa Familia banner in 2003. Brazil did not launch the program with any impact measurement elements, but subsequent studies have found a significant impact in poverty reduction with no effect on work hours.

In theory, Bolsa Famila is the first phase of a commitment to implement a universal basic income over time (which Brazil’s president signed into law in 2004). No additional plans have been announced for this transition. (Brazil’s current testing of UBI is restricted to a single small village, where a local organization is sponsoring basic incomes for certain families.)

PROSPERA (Mexico)

Current coverage: ~32M individuals (~26% of population)
Amounts: Average of ~1000 pesos (55 USD) per family per month
Eligibility: Families in poverty who visit health clinics and whose children attend school

Initially called “Progresa,” Mexico’s cash transfer program started with 300 thousand rural households in 1997. It moved to national scale in 2012 as “Oportunidades,” then was renamed “Prospera” in 2014 as part of an ongoing push to couple the transfers with complementary government services. The impact of the program has been examined by a number of studies, including an RCT over its first three years. Findings include an increase in consumption and school enrollment, a decrease in child malnourishment, and no effect on time spent working.

China, meanwhile, has an even larger cash transfer program than Brazil or Mexico (though it covers a smaller fraction of its population). The Chinese program is unconditional, with income level the only criterion for eligibility.

Dibao (China)

Current coverage: ~75M individuals (~6% of population)
Amounts: Large variation by county: 40 to 250 yuan (6 to 38 USD) per individual per month (rural program)
Eligibility: Households and individuals in poverty

China’s cash transfer program began as a series of rural experiments in the 1990s, and expanded nationwide 2007. It has remained a decentralized program, with separate arms for rural and urban areas and high reliance on local administration to set and distribute benefits. Research has found negligible effect on poverty reduction, along with targeting difficulties in rural areas that cause around 70% of transfers to go to non-poor households. (The urban arm appears to be more successfully targeted.) Commentators have also raised questions about the wide variance in benefits by region and elements of the program that could discourage work.

Many other programs, with both unconditional and conditional models, have continued to emerge across the developing world. A growing body of research continues to test their effects - one notable aggregate result so far is no apparent effect on working hours.

In the developed world, conditional programs like Brazil’s and Mexico’s have inspired additional government experimentation. For example, in the US a “Family Rewards” program inspired by Oportunidades has experimented with a variety of payouts to guide poor families education- and health-related decisions. Separate pilots have taken place in New York and Memphis and the Bronx; the latter was planned as an RCT, but full results do not seem to have been released.


4. Upcoming basic income pilots

As interest in basic income grows, a few governments and organizations have recently announced pilots to test aspects of the idea in particular regions.

Program

Recipients

Universal

Basic

Long-term

RCT

Finland Pilot

5,000-10,000

✗ *

Utrecht Pilot

250

Oakland (YC) Pilot

30-50+ **

Details not available

*

Based on latest “partial” plan presented

**

Estimate for first phase only

Finland Pilot

Sample size: 5,000-10,000
Eligibility:
 Diffuse spread across nation, plus 10-30% saturated municipalities
Launch date: TBD 2017
Duration: 2 years

The Finnish basic income pilot is the largest currently being contemplated by a government, and the only one planned on a national scale. It will be administered by Kela, the Finnish Social Insurance Institution that runs the country’s current benefit programs.

The pilot was originally planned to experiment with multiple models for a national program, including “full” basic income (replacing nearly all current social service transfers), “partial” basic income (replacing only some), a negative income tax mechanism, and cash bonuses for community service. The latest research report suggests this scope will be scaled down to only the “partial” basic income model, roughly €550-750/month.

The plan for the pilot is still being developed, and has not yet received the government’s final go-ahead. The end of the report breaks down the pros and cons of various models, including discussion of work incentives, administrative streamlining, poverty gaps, economic feasibility, research feasibility, and level of public support.

Utrecht, Netherlands Pilot (with other Dutch cities)

Sample size: 250 (roughly half with basic income)
Eligibility:
 Households currently within the city’s welfare system
Launch date: January 2017
Duration: 2 years

On a much smaller scale, Utrecht (the fourth-largest city in the Netherlands) will be piloting basic income as part of a spectrum of alternatives to the current welfare regime. The pilot will be conducted by the city government in partnership with Utrecht University.

The study has multiple test groups, half of which will receive a basic income (with or without community service bonuses), while the other half remains on existing welfare benefits (with or without a relaxation of current work requirements). Four other Dutch cities (Wageningen, Tilburg, Groningen and Nijmegen) plan to conduct similar pilots; it is unclear if these will join together into a larger study.

Oakland, USA Pilot (Y-Combinator)

Sample size: 30-50 (with potential to grow)
Eligibility: TBD
Launch date: TBD 2016
Duration: 1 year (with potential to grow)

Y-Combinator will be investigating basic income as its second large research project (after Open AI). It recently hired a full-time director for the project, and has announced plans to launch a small-scale pilot in Oakland, California to test research methods and inform plans for a larger study.

A recent interview suggests YC will spend roughly $1.5M to give a small group $1,500-2,000 per month in unconditional income for a year, with potential to expand the experiment over the subsequent five years. So far no additional detail has been released.

Other regions like Ontario have announced their intentions to pilot a basic income, but have not yet announced more details on their plans.

In addition to these explicit plans to test basic income, active debates about it have begun in a number of countries and regions. In Switzerland, a referendum that would have mandated a country-wide basic income was recently defeated by a wide margin (though supporters noted that 77% of those polled favored testing UBI at a local level). In FranceNew ZealandScotland, NamibiaGermany, India [1, 2, 3], Vermont, and Washington, D.C., political bodies have expressed interest in exploring basic income further.


5. Other cash transfer examples and anecdotes

In addition to these formal studies and programs, a number of other cash transfer examples are often considered anecdotally relevant to the basic income debate.

Program

Recipients

Universal

Basic

Long-term

RCT

Alaska Permanent Fund Dividend

644,000

*

Eastern Band of Cherokees Casino Dividend

15,000

 *

Government Patronage in Gulf States

Uncertain

Details not available

Mein Grundeinkommen

49

London “Rough Sleepers”

13

*

Benefit size varies with profits of underlying fund.

Some of these programs are “natural experiments”: patterns of money distribution with some characteristics of basic income, but without that stated intention or defined study conditions. For example:

Alaska Permanent Fund Dividend 

Alaska has a state government program to distribute proceeds from state oil revenues as a "dividend" to eligible Alaskan residents. The annual amount has varied from $331 to $2,072. No major study has been conducted on the effects of the dividend (partly due to the difficulty of creating a control group). One Alaskan economist has assessed the roadblocks to research, and hypothesized likely impacts on job creation, poverty reduction, and increased migration to the state.

Eastern Band of Cherokees Casino Dividend

A Native American tribe in North Carolina has distributed casino profits equally among its 8,000 members since 1996, totaling ~$9,000/year per recipient in 2006. An epidemiologist focusing on mental health issues analyzed youths living with these recipients compared to those in neighboring areas and found the additional cash led to a decrease in behavioral problems like substance abuse, school absence, and minor crimes.

Government patronage in Gulf States

States that harness most of their wealth from natural resources, then distribute it to their citizens, are sometimes presented as possible analogues to basic income. While countries like Saudi Arabia do confer large, non-means-tested benefits to their citizens, these often come in forms other than cash (such as salaried government positions and fuel subsidies). Cash bonuses on top of these benefits have tended to be unpredictable. We are not aware of any research on the anti-poverty impact of this government patronage; scholars have instead examined whether such close ties to the government harm civil society.

Smaller-scale case studies have also come up in discussions of basic income. For example:

Mein Grundeinkommen ("My Basic Income”)

Started by German entrepreneur Michael Bohmeyer, this organization provides one-year, 1,000 euros/month basic incomes to individuals, and has begun to spread to the US. Its stated goal is to raise awareness about basic income, and it is not the subject of a study. So far the project has financed 49 one-year basic incomes through crowdfunding.

London “Rough Sleepers” Program

In 2009, a local charity gave 13 homeless people the equivalent of $4,500, and told thm to spend it how they liked. 11 of the recipients ended up off the streets, which inspired positive press coverage.