Labor supply response to an unconditional basic income : comparison between different case studies

(1)

Labor supply response to an unconditional basic income:

Comparison between different case studies

Student: Niels Heining

Student number: 10220429

Supervisor: A. Kamm (MSc)

Date: July 11, 2014

Specialization: Economy

(2)

Abstract

In this paper, different surveys that investigated the labor supply response to an unconditional basic income are reviewed. An unconditional basic income is an income issued by a political community to all its citizens, without means test and work requirement. Opponents argue that a basic income will have negative impacts on the labor market. Negative income tax experiments in North America show that there was an insignificant negative work response in developed countries. Surveys with conditional cash transfer programs in Latin America found no significant effects on labor supply. Positive effects on labor supply caused by unconditional basic income are found in Namibia, Uganda, Kenya and India. These results suppose that there is a different work response on unconditional basic income between different economic regions. Unconditional basic income could be more preferable in developing countries than in developed countries, but further research is necessary.

(3)

In the wonderful world of Adam Smith, as a result of price mechanisms, markets will move to equilibrium where supply equals demand. This theory is also being used for labor markets. People who are willing to work, are supplying labor. Firms that attract employees have a demand for labor and instead of price mechanisms, wages will move this market to equilibrium. But, as Myles (1988) pointed out, the labor market could not be seen as a commodity market. There are some fundamental differences: first, non-strategic demographic processes. Secondly, people are not created, nor destroyed by the expected value on the market and last society has to deal with those unable to sell labor. When the society is organized in a way that labor was produced for sale in a market, this would lead to nonfunctional markets and even the destruction of society (Myles, 1988). Hence, labor markets need welfare states; society is in need of a mechanism that reduces inefficiencies that labor markets create. The problem is that we have to find a welfare state organized in such a way that it does not destroy labor markets and creates a fair and efficient economic society. In this paper I will discuss a proposal that is potentially able to meet these requirements.

In Switzerland a group of people raised enough signatures for a referendum on an unconditional basic income. When the proposal is accepted, every adult Swiss citizen will receive 2500 Swiss Francs per month from the state, unconditionally. The aim of the proposal is to provide a financial safety net for all the citizens and to increase individual freedom. Switzerland could be the first country in the world where every citizen will receive such an unconditional grant.

The idea of an unconditional basic income (UBI) is that every citizen of a country or region receives a monthly income, independent whether somebody is working or not. There are two important issues that must be solved, when UBI is implemented. The first issue concerning UBI is that there must be found a way to finance a basic income. The second point of discussion is the effect on labor supply (Atkinson, 1996; Van Parijs, 2004). This second question will take a central place in this paper. When there is a significant negative effect on labor supply, UBI is not economically feasible.

Different papers have been written about the effect on labor supply. One example is the Alaska permanent fund, where money, earned from the natural resources in the state, is collected and of which parts of these earnings are distributed among Alaska residents. Another example is a program called Bolsa Família that the Brazilian government started in 2004. The program supports poor families with direct cash transfers and it also gives free education to children who are not able to go to school (Soares, 2010). A third example was an experiment, conducted in the 1970s in Dauphin, Canada. In this village the government gave every resident a minimum cash benefit (Forget, 2011). In this paper I will compare these examples among others and I will look if there are any differences in results between regions. The central question of this paper is: what is the effect on labor supply caused by unconditional basic income?

(6)

2. General literature review

In scientific literature and media different ideas are named basic income. In this section, I will state the definition of UBI that will be used in this paper. First, I will answer what unconditional basic income is and what it is not. Negative income tax (NIT) is often directly compared to UBI, although they are not the same. The third paragraph of this section covers the advantages of UBI, followed by a paragraph about the disadvantages.

2.1 Definition unconditional basic income

Van Parijs (2004, p.8) gives a definition of a basic income that I will use throughout this text: “A basic

income is an income paid by a political community to all its members on an individual basis, without means test or work requirement”. There are different aspects in this definition that need some

further explanation.

First of all, it is an income that is preferably paid in cash, not in food stamps, housing grants or other kinds. One of the main drives behind UBI is that people increase their freedom of choice (Van Parijs, 1997). When the government supplies a basic income in kind instead of cash, this enhancement in freedom of choice is partly removed. With a payment in kind it is not possible to save or invest and to consume in any period you want. The second feature of the definition says that a political community must pay the basic income. This does not mean that the government of a nation-state must pay the grant, but that a political institution that controls some publicly controllable resources pays the basic income (Van Parijs, 2004). The third feature states that UBI is paid to all its members on an individual basis. This means that the grant is also paid to children, non-permanent residents, pensioners and inmates. A requirement for receiving a grant could be that you must be a legal resident and that you live in a political community for at least a minimum timescale. ‘On an individual basis’ means that a basic income is not paid to households, but to every individual citizen. Van Parijs (2004) states that for the individualistic nature of the proposal, the amount of basic income must not be dependent of the household to which the individual belongs.

The last section of the definition states that a basic income must be without means test or work requirement. ‘Without means test’ means that the level of dividend is paid irrespective of income level. This is fundamentally different compared to existing social insurance systems. In the current situation, benefits to households are paid ex post; it is paid on basis of a prior assessment. A basic income is paid ex ante, before any assessment is made (Van Parijs, 2004). There is no income test involved, nor are their any other means tests taken into account when determining the amount of basic income. A basic income does not contain a work requirement. This is one of the fundamental features that evoke resistance from offenders of the proposal (Elster, 1986; White, 1997). It is possible to be voluntary unemployed and still receive basic income grant. I will discuss these arguments in the next sections.

(7)

2.2 A comparison between UBI and NIT

The negative income tax proposal (NIT) is similar to a UBI, but not the same. This proposal gains popularity by the work of Friedman (1962) and Tobin, Pechman and Mieszkowsy (1967). NIT complements the income of households living below a minimum subsistence level. Tobin argues that it is a way of fighting poverty without losing too much working incentives (1967). The proposal uses a flat tax rate, but this tax could be negative when someone’s income is below a break-even level. For example, the break-even level is one thousand dollars per month and the flat tax rate is 30 percent. When someone in this case earns two thousand dollars a month, he has to pay 30 percent tax over the thousand dollars he earns above the break-even level. That would be 300 dollars. When an individual earns 1000 dollars he does not have to pay any taxes. When someone is unemployed and has no income at all, such a person would receive a NIT. With no income this person has an income that is thousand dollars below the break-even level. These 1000 dollars are again taxed at a rate of 30 percent, but in this case it is a negative tax. This person receives 300 dollars from the state. When in another case, UBI is set at the level of 300 dollars a month, the way poverty is reduced is the same. Everybody, working or not and independent of individual situation has a minimum income of three hundred dollars. But in the situation of a NIT, the benefit someone receives from the state reduces with income. This would also happen in a conventional means-tested system, but with a NIT this happens at a slower rate. The marginal income could still be very low when an unemployed individual accepts a low-paid job (Van Parijs, 2004). This is a big difference with a UBI scheme, because in that case the benefit from the state does not reduce with income and work incentives are higher compared to NIT. With UBI, the benefits are paid to all residents in a community, where NIT schemes only benefit the residents that have an income below a break-even point. These differences could be seen in figure 1 and 2. Figure 1 shows a NIT, with E as the break-even income and G the minimum guaranteed income. The differences between N and M are the taxes. Figure 2 shows a UBI; with g as basic income level. In both situations, net income never fall below g or G, but with NIT you lose of part of your benefit when gross income increases, while with UBI the benefit remains the same, but taxes increase with gross income.

Van Parijs (2004) states that the main difference UBI

and NIT is that NIT is paid ex post, like a means-tested system, where UBI is paid ex ante. Van Parijs (2004) argues that a system paying the benefits ex ante is preferable to a benefit system paying ex

Figure 2. Universal basic income, N: gross income, M: net income, g: guaranteed basic income. (Tondani, 2009, p.4)

Figure 1. Negative income tax, N: gross income, M: net income, G: guaranteed minimum income. (Tondani, 2009, p.4)

(8)

post. With an ex post system you still have the negative impacts from the unemployment trap. With

the unemployment trap Atkinson (1996) means that people have an incentive to stay unemployed instead of participate in the labor market. There are two reasons for the occurrence of the unemployment trap. The first reason is the uncertainty argument. Because you lose part of your benefit when you accept a job, it could feel safer to keep the benefit from the government. The other reason is that the marginal wage could be very low. This reason arises, because you lose part of your benefit when you accept a job. The second reason takes place in a less severe manner with NIT than with a means-tested system, but is still a problem, while they are both eliminated with a UBI. The first reason does not play a role for UBI and NIT. With a negative income tax there are also more administration costs involved compared to UBI. UBI is simpler to administer, because you do not have to calculate the benefit for every individual. With a NIT, you are still in need of this information, but the proposal is still simpler and covers therefore less administration costs than a means-tested social insurance system (Van Parijs, 2004).

Tondani (2009) mentions that there are also some ethical and normative differences. The two schemes differ in the freedom they reflect. NIT reflects a right-libertarian freedom, which views freedom from a negative point of view. Freedom exists when another individual or group does not thwart an individual. Therefore, in this view the state must be as small as possible to create self-ownership for individuals. In this view redistribution must be as small as possible, because they distort the free will. A NIT scheme is seen as a redistribution system that underlies these principles as much as possible without the risk of free riding (Tondani, 2009). UBI underlines egalitarian principles. These reflect positive freedom, which emphasizes the role of opportunities. An individual must have as much opportunities as possible and obstacles on this aim, have to be removed. A basic income could be a vehicle to provide as much opportunities as possible (Van Parijs, 1997).

2.3 Advantages of a UBI

There are several reasons why a basic income would be advantageous. The first reason is that it could alleviate poverty (Atkinson, 1996). It could totally eliminate poverty when a basic income is set at a level that the monthly or annual grant is above or on subsistence level. Especially in countries where there is a lack of a social insurance programs this first argument is important. In most European countries this argument is less valid, because there already exist a social insurance system that prevents for extreme poverty. In developing countries this is often one of the main arguments of proponents of a UBI. The U.S. NIT experiments were also executed, because of the will to reduce poverty (Burtless, 1986).

A basic income above minimum income level will also improve the independence of individuals. All adults would receive the same basic income, regardless of their private circumstances. The grant is not dependent of marital status or the income level of their partner. This could be especially beneficial for women (Atkinson, 1996; Pateman, 2004). Generally, there is a relatively larger population of women who are not involved in paid work. In this situation they are dependent of their partner’s income. Pateman (2004) argues that a basic income could improve the autonomy and self-government of individuals and that this is the basis for democracy. Therefore Pateman (2004) argues that a basic income is a democratic right for every citizen.

(9)

The third aspect Atkinson (1996) named in favor of a basic income is that it is not dependent of employment status. With current social insurance systems there are special payments paid to those who are unemployed. With a UBI scheme these special payments would disappear. With the special payments paid in the current social insurance program the unemployment trap appears. When the unemployed found a low-paid job, they lose their payment from the state. It could happen that a means test is in effect a 100 percent tax on workers earnings. For every dollar he earns, his benefit from the state could decrease by one dollar (Tobin et al., 1967). So their marginal benefit from that job could be low and they won’t accept a low-paid job (Van Parijs, 2004). Van Parijs mentioned also another aspect of the unemployment trap. Namely, for an unemployed person accepting a job depends not only on the difference between net income, but also on the uncertainty that could arise when someone accepts a job. Staying unemployed could mean that there is a certain flow of income. With administrative time-laps involved there could be a gap in income flow. Risk-averse people may prefer the constant income flow provided by the government. A basic income could solve both problems caused by the unemployment trap, because a constant income is secured. Risk-averse people will therefore sooner accept a job and the marginal wage is higher than in current social benefit systems.

The fourth argument Atkinson (1996) mentions the reduction in administration costs for government and taxpayers. Current means-tested social insurance systems are complicated and costly to monitor. A UBI would be much simpler and easier to monitor. This could eliminate some administrative costs for the government and would reduce the differentiation in benefit amounts. For The Netherlands, Allers (1994) found that the administration and compliance costs of the Dutch income tax system in the 1990’s was around 6 percent of the total revenue raised with these taxes. A lot of money could be saved when the system is less complex.

A fifth argument in favor of a basic income is that it enhances the flexibility of the labor market. This will in turn reduce unemployment through three dynamics. First of all, taking for granted that some people will reduce their labor supply, it gives unemployed people the chance to enter the labor market. The people who reduce their labor supply could still be working, but maybe fewer hours or become voluntary unemployed. This opens opportunities for people being involuntary unemployed (Marx and Peeters, 2007). Secondly, a UBI could improve incentives for entrepreneurial activity. In the case when people are receiving a basic income there is less risk involved with setting up a business, because they are always in favor of a guaranteed income (Euzéby, 1987). More entrepreneurial activity could lead to more economic growth, which results in increased labor demand. This risk-reduction mechanism could also lead to more innovation and a dynamic business landscape. Thirdly, a logical step to take when a UBI is implemented, is the abolishment of some labor market institutions (Marx and Peeters, 2007). One such institution that could be removed is a minimum wage policy. With UBI it is not necessary to set a minimum wage, because people are receiving an income that is high enough to live at subsistence level. Without a minimum wage there could be a re-emergence of jobs that are out of the market in the current situation. These are the so-called low-productivity jobs that were not worth a minimum wage. For low-educated and low-skilled workers the removal of a minimum wage could be an opportunity to enter the labor market.

(10)

Opponents of UBI have several reasons to argue that UBI is unrealistic. The first argument that is given is that UBI is not affordable. Basic income for every citizen is not efficient and too expensive. Secondly, there are some ethical points of discussion. Working citizens must pay for those who are voluntary unemployed. The last argument I will discuss is the possibility of labor market distortions.

2.4.1 How to finance UBI?

Opponents claim that UBI is too expensive and therefore not affordable. The government needs a way to finance a cash grant for every citizen in the case of UBI. Van Parijs (2004) comments that the cost of a UBI scheme depends on the height of the grant. When the grant must be at a level that the basic income is high enough to fulfill the basic needs of individuals, then it is difficult to finance a UBI. Harvey (2006) calculated that the cost for the U.S. of a UBI that pays every citizen above eighteen years old a grant equal to the official poverty rate would be $1.69 trillion. De Mooij (2006) calculated the net costs of a UBI at 50 percent of the current social minimum for those aged above eighteen years old and found that this will cost 45 billion euro. In the literature three options to finance a UBI are mentioned. First a basic income could be paid by the elimination of the current social insurance system. Secondly a tax raise could finance a UBI and finally, revenues from natural resources could be used to finance a basic income.

The first thing a state can do to finance UBI, is to replace current social insurance systems. The current welfare systems, especially in some European countries, are expensive (De Mooij, 2006). UBI could replace these systems, when it is set at a level sufficient to fulfill basic needs (Van Parijs, 2004). However, a UBI program also provides grants to people having an income above the poverty income level. Only replacing current social assistance programs are not enough to finance a UBI set at the social minimum level. Though, a UBI scheme covers less administration costs than a means-tested social insurance system (Atkinson, 1996).

A second way to finance UBI is to raise taxes. De Mooij (2006) calculated that for implementing UBI in The Netherlands, a 53.15 percent tax rate is necessary. A progressive tax rate could offset the benefit from UBI for middle- and high-incomes. A difference with current social minimum income systems and UBI is that with UBI also higher incomes receive a benefit. This benefit for higher incomes is unnecessary and a progressive tax rate could let the higher incomes pay for their own benefit. UBI is meant to serve the low-income households. These households benefit from the UBI, but do not pay it back through a high tax rate (Van Parijs, 2004). How much the tax rate must be, depends on the height of the basic income. It’s therefore difficult to say if a UBI is affordable with a tax raise, this must be assessed case by case. Atkinson (1996) states that a flat tax rate of 40 percent would be enough to finance an adequate basic income, by which he mean a negative income tax proposal that guarantees an income at the poverty level. Clark (2005) estimated that a basic income in the U.S., where citizens under eighteen receive $3,500 annually and adults receive an annual grant of $9,359 would need a flat tax rate of 35.2 percent.

The third way to (partly) finance UBI is to use the revenues from natural resources. In Alaska and Iran basic income schemes financed by natural resources already exist. However, it’s hard to finance a substantial UBI out of the revenues from natural resources alone (White, 1997). In combination with tax revenue and the substitution of the current social insurance system, it could be enough to finance a substantial UBI. Natural resources could be seen as assets belonging to all the citizens of a state. Therefore, it could be argued, the revenues flowing from these natural resources should

(11)

belong to all citizens (Van Parijs, 1997). Resource-rich countries like the oil states and some African countries, could eliminate poverty and increase opportunities for their citizens by using these resources to finance UBI.

2.4.2 Is UBI ethically justified?

Van Parijs (1997) states that every person has the right to real freedom. According to Van Parijs (1997) real freedom exists of three key elements. First, there has to be a clear structure of rights that guarantees the security of the citizens. Secondly, this structure has to be organized in a way that every person has self-ownership. Thirdly, this structure has to be defined in a way, that every person has the greatest possible opportunities to do whatever he or she wants. Van Parijs argues that a basic income must be an ethical right, because it can sustain real freedom (1997). With a basic income, every citizen has a given amount of money that gives this person the freedom to do whatever he or she wants, without being restrained to earn an income. The real freedom of a person increases when the opportunity sets of an individual increase. These sets increase when an additional combination of work-leisure is added and no other combination is removed.

Opponents of UBI are saying that UBI is morally undesirable, because of a free-riding problem. They see it as unfair that productive, tax-paying citizens have to pay for citizens who are capable of working, but just not willing and that these people could choose to live from their UBI. White (1997) calls this problem the exploitation objection. By which she means that citizens choosing to live from their basic income, exploit the productive tax-paying citizens.

Van Parijs (1997) counters the exploitation objection with a few arguments. The first is what he calls the external assets argument. Ideally, UBI is paid out of some external assets, where no one is especially entitled. When a country has enough natural resources, a government should collect these assets and share the revenue with their citizens. Natural resources have to be taxed at 100 percent and the proceeds have to be equally distributed among the citizens in the form of a basic income. Van Parijs (1997) states that this is fair, because no one has the exclusive right to fully own these resources. Van der Veen (1998) says this is fair for two reasons. First, it meets the criterion of envy-freedom, because no one is better off than another with the initial endowments and it is efficient in a way that no mutual gains from trading one’s initial bundle are possible.

When the benefits from the natural resources are high enough to finance a substantial UBI, this would give all the citizens equal opportunities. Van Parijs proposes a UBI, which could provide every citizen enough money to facilitate in his or her basic needs substantially (1997). When there are enough natural resources to finance a substantial UBI, the exploitation objection is solved, because individuals who are not willing to work do not exploit productive employees. The problem is that there are not many countries that have enough natural resources to give all their residents a substantial UBI (White, 1997).

Van Parijs argues that jobs could also constitute a form of external wealth that could be used to finance a substantial basic income (1997). In the Walrasian model of the economy there are no frustrated buyers and sellers. For the labor market this would mean that insiders, people who are employed, wouldn’t get a higher real wage than outsiders are asking. In reality, the presumptions underlying the Walrasian model do not hold for two reasons. First, in the real world there may often be a positive relation between worker productivity and real wage. High wages motivate workers to

(12)

work harder, what increases profitability of firms. This takes away the incentive for firms to hire unemployed workers, who are asking smaller wages. Secondly, in reality there exist turnover costs related to the hiring and firing of workers. Therefore insiders are in the position to claim a higher wage than the market-clearing price would suggest. These two reasons lead to the existence of involuntary unemployment and lead to the existence of unequal opportunities between insiders and outsiders. Van Parijs (1997) argues that this inequality is the result of some bad luck of the outsiders and the extra wage insiders receive employment rents. Wages are a form of brute luck and this generated wealth should be taxed and distributed along those disadvantaged by bad luck (Van Parijs, 1997).

White (1997) concludes it is ethically right to implement UBI out of natural resources, but this would be a relatively small UBI. He argues that the exploitation objection only is solved when a UBI is financed by natural resources alone. However, for a substantial UBI jobs must be seen as economic benefits of social cooperation as well, which is ethically unjustified (White, 1997).

2.4.3 Negative effects on the labor market

Marx and Peeters (2007) mention a few positive effects of UBI on labor supply. Remarkably, opponents predict exactly the opposite. Between 1968 and 1980 there were four negative income tax experiments in the United States and one in Canada. The results of these experiments are often used in the debate about UBI. Proponents of UBI say that the experiments showed that UBI has not a very large effect on labor supply, at least smaller than expected (Hum and Simpson, 1993). Opponents use the results to say that UBI has a large negative effect on labor supply and that UBI is therefore not desirable (Anderson and Block, 1993). The main question in this discussion is how large the negative work response can be, without disturbing economic stability. A criterion for this could be that the decrease in labor supply must not threaten the financial viability of the project (Widerquist, 2005). Widerquist states that there is no evidence (with the North American experiments) this is indeed the case, but the result fell in a range between no disincentives at all and a system that is not affordable. Therefore opponents and proponents of UBI can both use the results as an argument for their case (2005).

Another problem for the labor market with UBI could be a low-wage trap (Myles, 1988). UBI could eliminate the unemployment trap, but a low-wage trap appears. Myles (1988) argues UBI encourages expansion of low-skill industries and crowd-out high skilled industries. In most proposals with UBI, minimum wage policies will disappear. For entrepreneurs it will be profitable to invest in low-skilled industries, industries that were not able to survive with minimum wage. Workers who are willing to work for lower wage could be used instead of high-skilled, high-waged workers. Therefore, UBI could be a system to facilitate lazy capitalists. A consequence could be that there is less room for high-skilled workers. UBI could increase the flexibility at the lower end of the wage distribution, while crowding out the high end of the wage distribution (Myles, 1988).

There is a lot of discussion about the effect of unearned income on economic activity and especially labor supply. Lotteries could generate unearned income, which is comparable to the case with a UBI. Imbens, Rubin and Sacerdote (2001) estimated the effect of an unearned income on labor earnings. They found that for lottery winners the marginal propensity to earn out is -11 percent. This effect is the average effect of lottery winners that won between zero and $100,000 and means that these people earned 11 percent less income from other activities, since they have won a lottery (Imbens et

(13)

al., 2001). Marx and Peeters (2008) also investigated the effect on labor supply for winners of the Belgian Win for Life lottery. They found no significant effect of the unearned income on labor supply. However, they found that couples, two partners working, worked fewer hours after the unearned income in 8 percent of the cases. Singles didn’t work less after they won the lottery and couples, one partner working, worked fewer hours in 9 percent of the cases. Because of the small sample, these statistics are not significant (Marx and Peeters, 2008).

(14)

This part will contain several case studies with proposals similar to UBI. Cases in the U.S. and Canada, Latin America, regions where natural resources are used as a basic income and developing countries are reviewed. None of these case studies present a nationwide UBI as Van Parijs (2004) defined it, but all these cases have some characteristics of UBI. I will give a brief description of the different cases and look at the results with regard to labor supply responses. In the conclusion and discussion I will summarize these results and give some possible explanations for the difference results.

3.1 Experiments in the U.S. and Canada (1968-1980)

Between 1968 and 1980 there were four negative income tax experiments in the United States and one in Canada. A summary of the basic principles of these experiments is given in table 1. The experiments conducted a NIT variant, what is not the same as UBI, as I previously mentioned. Though, the results are often used in the UBI debate and the effects on the labor market could be comparable to those of UBI, because in both proposals citizens receive an income unconditionally paid by the government. Labor supply results in both cases should at least be comparable. There are a few things that are problematic in the statistical analysis of the data. First, households involved are nonrandom. With a NIT scheme, it is only beneficial to participate, if your income is below the break-even level. Secondly, households were not forced to stay in the experiment and many participants left the experiment before maturation date. Although the proposal of the five experiments were all a bit different, all the experiments investigated the effect on labor supply in hours worked (Hum and Simpson, 1993). Different guaranteed incomes and tax rates were tested in the experiments and the results presented here, are weighted averages. There are differences in the results found by the difference researches. This is due to the use of different econometric models. The models make use of different corrections for biases that occur in statistical analysis.

Greenberg and Halsy (1983) have critique on the method used to investigate the work response results. They say that households involved in the experiment had incentives to understate their work effort, so they will have lower taxable income and receive a higher NIT. The method used in all four U.S. experiments was to interview people about their hours worked, but participants had incentives to misreport these hours worked, delay reporting the beginning of a new job or fail to report entire jobs (Greenberg and Halsy, 1983). These misreporting opportunities could lead to overstating the effect on labor supply.

The experiments in North America investigated high-developed countries. The main reason why these NIT experiments took place was to improve the welfare state (Euzeby, 1987). The proposal could eliminate poverty and therefore the experiments focused on poor households. Although, the results in this section are from developed countries and therefore could the experiments be used for the effect from UBI on work effort for people that could fulfill their basic needs without a basic income.

(15)

Table 1, Summary of the NIT experiments in the U.S. and Canada. (Widerquist, 2005, p. 53)

3.1.1 New Jersey

The first NIT experiment started in 1968 in New Jersey and Pennsylvania and is called The New Jersey Graduated Work Incentive Experiment (NJ). The project was set up to investigate the consequences of NIT or guaranteed annual income on labor supply (Watts, 1969). The project involves only households in urban industrial areas with at least one male between 18 and 58 years old and not permanently studying or disabled to work and the household must contain at least two members. Households that were involved had an income not exceeding 150% of the poverty income, so they are in poor or near-poor categories (Watts, 1969). Because this experiment contains only one subpopulation it was necessary to investigate other subpopulations as well to create a complete picture of the effects of NIT on labor supply.

In the New Jersey experiment 1216 households were involved. At the end of the program, only 983 households were still receiving a negative income tax, the fallout rate was 19 percent (Widerquist, 2005). Two different methods are used to investigate the effect on labor supply. The first method uses ANOVA, which analyses significant different work responses, between the control group and the treatment group. The other approach is a structural model, which looks if there is an experimental response in terms of conventional income and substitution effect on labor supply (Hum & Simpson, 1993). Three different researches have investigated the New Jersey experiment and found different results, but they found all a decrease in labor supply for husbands and for wives. Keeley (1981) found that labor supply in annually hours worked for men decreased by 116, which is 7 percent of the total hours worked and for women with 75 hours (33 percent). Robins (1985) used a

(16)

regression model and found that husbands decreased their labor supply by 34 hours (2 percent) and women with 56 hours (25 percent). Finally, Burtless (1986) used a statistical model and found that husbands worked 21 hours (1 percent) less on average and wives 56 hours (25 percent). There are some big differences between the three researches, due to the different statistical methods they use. On average, the results show a decrease in labor supply for husbands of 3.3 percent and for wives a decrease of 27.7 percent (Hum and Simpson, 1993).

3.1.2 Rural

The second experiment took place in Iowa and North Carolina from 1970 till 1972 and was called the Rural Income Maintenance Experiment (RIME). The difference with the New Jersey experiment is that this experiment focused on the rural sector Only farmers and those living in towns with a population smaller than 2,500 were involved (Bawden and Harrar, 1977). The goal of the experiment was to investigate the labor supply response for the rural population, but also school performance, expenditure patterns, job search behavior, farm production, financial decisions, family structure, nutrition, geographical mobility, psychological well-being and political involvement were investigated (Bawden and Harrar, 1977). I will only focus on the results in labor supply response. Researches expected a different result in work response between the urban areas of New Jersey and the rural areas in Iowa and North Carolina, because of different alternative employment opportunities and the proportion of the self-employed population (Bawden and Harrar, 1977). Households were randomly selected, but they must have an income less than 150% of the poverty line. Three different guaranteed incomes were tested: 50, 75 and 100 percent of the poverty line and three different tax rates: 30, 50 and 70 percent.

The initial sample size in the rural experiment was 809 households. At the end of the experiment 729 households were still involved (Widerquist, 2005). Keeley (1981), Robins (1985) and Burtless (1986) investigated the rural case. They found quite similar results in the rural experiment compared to the New Jersey experiment. Keeley (1981) found a decrease in labor supply for husband of 9 percent and for wives of 29 percent. The effect found in the rural experiment is more than in the New Jersey experiment for husbands, while the effect for wives is less in the rural experiment. The results of Robins (1985) show a decrease in labor supply of 3 percent for husbands and a decrease of 28 percent for wives. The labor supply of rural husbands and wives are more sensitive to NIT than the response of urban citizens. Finally, Burtless (1986) found also a decrease in labor supply for husbands of 3 percent and a decrease for wives of 28 percent. These results are the same as the results of Robins (1985) in the rural case. The results also show a bigger effect in labor supply for the rural case compared to the urban experiment. The results show that people in urban and rural areas, both have a negative work response when they receive an annual guaranteed income. The results also show that this effect is larger for rural citizens than for urban citizens, but this difference is not significant (Hum and Simpson, 1993).

3.1.3 Seattle/Denver

The biggest of the five experiments took place in Seattle and Denver, the Seattle/Denver Income-Maintenance Experiments (SIME/DIME). The experiment started in 1970 in Seattle, but due to an unemployment problem caused by changes in the major aerospace industry, the experiment was extended to Denver (Spiegelman and Yaeger, 1980). The intended duration of the experiments was six years, but researches obtain approval to extend the experiment for twenty years with a few

(17)

households. Finally, in 1980 the experiment stopped. The households still participating in the experiment thought they were going to receive a guaranteed income for another ten years. The experiment included all kind of families; there was no selection on household composition. 4800 Households were involved in SIME/DIME, more than in the other experiments combined. The difference with the previous experiments was that this experiment was more complete and more complex tax systems were used (Spiegelman and Yaeger, 1980). Households that were involved had an income below $11,000 for single-parent families and below $13,000 for two-parent families (Widerquist, 2005). The goal of the experiment was to test different consequences of NIT, but researches were also interested in the interaction between NIT and a program of subsidized job training (Spiegelman and Yaeger, 1980).

In the previous experiments, a linear tax model was used with constant tax rates. In the SIME/DIME the projects used a two-parameter tax function:

T(Y) = − ,

Where is the initial tax rate, R is the rate of decline of the average tax rate and Y is income (Kurz and Spiegelman, 1971). This means there is a linear declining tax rate in this experiment, the higher your taxable income, the lower the tax rate. Following Kurz and Spiegelman (1971), it will improve work incentives, because of diminishing marginal tax rates. In combination with the job training program of the experiment, they expected that the effects on labor supply should be less than in the other experiments.

For the Seattle/Denver experiment Keeley (1981) found that there was a negative work response for husbands of 8 percent and for wives of 21%. With the method Keeley (1981) used, the effect on labor supply for wives is less than in the previous experiments. For husbands there is no big difference between the experiments. Because in the SIME/DIME single parent families were also included Keeley (1981) investigated the work response for single female heads. In Seattle/Denver he found a 15 percent decrease in labor supply for single female heads. In the study of Robins (1985) the results were quit the same. A 7 percent decrease in labor supply for husbands, a decrease of 21 percent for wives and for single female heads the decrease was 16 percent. The method Burtless (1986) used, gives some different results. For husbands he found the same decrease as Keeley, 8 percent, but for wives and single female heads the declines were smaller. For wives he found a decrease of 17 percent and for single female heads a decrease of only 9 percent. The SIME/DIME was the only experiment, where the results were significant. We see a decline in labor supply in all the other cases as well, but these declines were insignificant (Hum and Simpson, 1993). Greenberg and Halsey (1983) also calculated the work response in the SIME/DIME using a different method that utilizes data from employers instead of data from interviews. They found that husbands decrease working effort with 8.3 percent, wives with 21.9 percent and female family heads with 20.1 percent. In the SIME/DIME Greenberg and Halsey (1983) did not found an overestimation of the work response, the results they found, even show a higher effect on labor supply.

Because the program in Seattle and Denver was the most generous one, we would expect that the work response was the largest here. The payments made to the families were the greatest and payments were made to higher income families than in the other experiments (Burtless, 1986). Results in table 2 show that this is not the case. This is maybe due to differences in the income

(18)

distribution of the four experiments. Though, because of the larger sample the negative work responses in the SIME/DIME were the only declines that were significant (Burtless, 1986).

3.1.4 Gary

The fourth US NIT experiment has taken place in Gary, Indiana. In 1971, 1799 households started with the experiment. In 1974, only 967 households were still involved. This experiment focused on black households, primarily single female-headed (Widerquist, 2005). The Gary experiment involved also some higher income participants, maximum income of 240 percent of the poverty line. Gary was selected, because the black, female-headed families were not involved in the other experiments (Kelly and Singer, 1971). Important is the consideration of nonmarket labor, because of the large fraction of single-female headed families. The focus of Gary was on the economic responses of NIT on labor supply, consumption patterns and investments in human capital (Kelly and Singer, 1971). In this experiment a guaranteed income of 70 percent of the poverty income and a flat tax rate of 40 percent was used.

For the Gary experiment, Keeley (1981) found a negative work response of 5 percent for husbands, 3 percent for wives and for single female heads a decrease of 28 percent. The big difference between the decline for wives and single female heads is remarkable. An explanation for this difference could be that the single female heads could spend more time on taking care of the children and other domestic tasks. Robins (1985) found some different results in the Gary case. For husbands the decline in labor supply he found was only 2 percent, for wives 20 percent and for single female heads 10 percent. Here there is also a big difference between wives and single female heads, but the other way around. Work response for wives is much greater than for female heads in Robins’ research. Burtless (1986) found similar results as Keeley (1981) for husbands and single female heads. The negative work response for husbands he found was 7 percent. For wives he found the only positive work response. Wives in the Gary experiment increased their labor supply by 5 percent. Single female heads decreased their labor supply by 30 percent. It’s hard to say why the research of Robins gives such different results for the Gary experiment. Greenberg, Moffitt and Friedmann (1981) made an analysis of the underestimation bias due to the incentive of participants to understate their work effort. They found an underreporting error in the data, especially for wives and female family heads. When this bias is taken in account in the analysis, no negative work response is found in the Gary experiment (Greenberg, Moffitt and Friedman, 1981).

3.1.5 Mincome

The only Canadian NIT experiment started in Winnipeg and Dauphin, Manitoba in 1975 and ended in 1978. The experiment started and ended with 1300 families (Widerquist, 2005). The experiment was set up, because the concern of poverty (Forget, 2008). In the Mincome experiment families were selected in Winnipeg, rural communities near Dauphin and some other small rural communities. Annual income of a family of four may not exceed $13,000. Families selected in Winnipeg were randomly selected and dispersed with each other, so it was possible to test different guaranteed income levels and tax rates. The main goal of the experiment was to test for work response and therefore those disabled to work were excluded (Forget, 2008). In Dauphin the proposal was a bit different, because every citizen of the population of 10,000 were eligible to participate in a guaranteed income program. This was an aspect that was not earlier investigated in the U.S. The idea behind this purpose was that a less artificial situation and more similar with the case of a UBI

(19)

would be implemented (Forget, 2008). Officially, the experiment ended in 1979, but for two years earlier no research had been done. Financial problems and ebbing away of political problems made the results were told to archive and had been collected for later statistical analysis (Forget, 2008). Hum and Simpson (1993) investigated the Mincome case for work response. The response they found in labor supply was less severe than in all the U.S. experiments. The decrease in labor supply for husbands was 1 percent. For wives the negative response was 3 percent and for single female heads 7 percent. The results show that the disincentive effects were minimal; the outcomes were not significant. Beside the weak labor supply effects found in Mincome, there were some interesting side effects, especially in Dauphin, the saturation site. Forget (2008) found that fertility rates for women below the age of 25 significantly decreased. Subjects were less likely to be hospitalized, for physical threats and mental health care. Also a decline in accidents and injuries during the experiment was found (Forget, 2008).

3.1.6 All North American experiments

Robins (1985) and Burtless (1986) combined the findings of the four U.S. experiments. They used weighted averages of the data from the experiments. The weightings to each experiment were selected by sample size. Robins (1985) found an average decrease in labor supply for husbands of 5 percent, for wives of 21 percent and for single female heads of 13 percent. The results of Burtless (1986) are quite similar: a decline for husbands of 7 percent, for wives of 17 percent and for single female heads also 17 percent.

There is an overall decline in labor supply, what was expected by the theory. However, the results were not significant and the decline was less than opponents had feared, but was more than opponents had hoped (Burtless, 1986). Compared to no experimental studies, the experimental data shows a small work response. The theory says that when NIT is implemented, there should be an income and substitution effect, both decreasing the work effort (Burtless, 1986). The substitution effect occurs, because a NIT will lead to higher tax rates, reducing net wage. This could cause a decrease in labor supply. The second effect is the income effect, due to an increase in family income, even when working effort is decreased. Burtless (1986) stated that the substitution effect is greatly overstated in the findings of the experiments and that the income effect is greatly understated. Widerquist (2005) concludes that these experiments show that there is no evidence that NIT would cause a reduction in the labor participation rate and there is no evidence that the supply response will increase the cost of the program in a way that it would be unaffordable. Widerquist (2005) states the experiments showed a guarantee level of 150 percent of the poverty level would be financially feasible.

However, all five experiments did not take longer than six years. The long-run effects would be a greater reduction in working effort due to the income effect. The participants knew that the payments would stop in a couple of years, so they were limited in quitting their job. Therefore, the U.S. experiments were an indicator of what would happen when a guaranteed annual income is implemented, but did not give a definitive answer to the question. More investigation is needed, to investigate long-term effects.

(20)

Table 2, Nonstructural labor-supply response estimates from the five NIT experiments (in annual

hours worked).

(Hum and Simpson, 1993, p. 279)

3.2 Latin America

A few Latin America countries have implemented social welfare systems that support the poorest households. Conditional cash transfer (CCT) programs are taken place at the moment in Brazil (Bolsa Família), Mexico (Progresa or Oportunidades), Colombia (Familias en Acción), Chile (Subsidio Unitario Familiar), Nicaragua (Red de Protección Social), Argentina (Jefes de Hogar) and Ecuador (Bono de Desarrollo Humano) (Hall, 2006). The programs are invented as major social policy instruments to reduce poverty. The intention of the programs is that the cash payments could be used as a vehicle to enhance education and health care opportunities. Therefore, the focus of the programs is on families with children, which could improve the long-term human capital and could break the inter-generational transmission of poverty (Hall, 2006). The programs are proposed in such a way that they could prevent long-term poverty, instead of poverty in times of crisis, which is the aim of many conventional welfare programs. The benefits are given in cash, rather than in kind,

(21)

food stamps or vouchers, what is more cost-effective, flexible and does not distort secondary markets.

In this section I will discuss the proposals and the results of the different programs. The programs are conditional, so they are not the same as UBI. Although, the unconditionally was a basic principle of UBI as mentioned in Van Parijs (2004) and this paper, the CCT programs could be a useful indicator for the work response of people receiving an unearned income. The programs are conditional, because payments are only made to selected families below a certain income level. What make these programs comparable to UBI, is that there is no work requirement. This feature makes that we can compare the effect on labor supply with that of UBI (Widerquist, 2013). The conditions that select the families are different in the various countries, but they all aim to improve the living standard of the poorest families. UBI would by definition be paid to every citizen in a certain community. Therefore, results of CCT programs could be misleading, because high-income citizens are excluded and this group could have a different work response than low-income citizens.

3.2.1 Brazil (Bolsa Família)

In Brazil, the CCT program is called Bolsa Família and it supplements families with at least one child below the age of seventeen with an income per capita less than R$140 (US$63) per month. Extremely poor households that earn less than R$70 (US$32) are also involved in the program independent of composition, for them the program is unconditional (Soares, Ribas and Osório, 2010). Former Brazilian president Fernando Henrique Cardoso (1995-2002) started with increase social assistance spending by expanding social assistance programs such as Bolsa Escola (Hall, 2006). The current president Lula (2003-2010) further increased social assistant spending. Bolsa Família started in 2003 with the merger of four social assisting programs. Bolsa Escola has the goal to improve education opportunities, Fome Zero and Bolsa Alimentaçãu to improve food securtity and Vale Gás to help poor households to buy gas for cooking (Soares et al., 2010). By the end of 2006, 11 million households were receiving benefits from Bolsa Família, making it the largest social assistance program in the world (Hall, 2006).

The results of Bolsa Família show a reduction in poverty of 12 percent measured by the poverty gap and a reduction of 19 percent measured by the poverty severity index (Soares et al., 2010). Results of Bolsa Família are based on propensity score matching (PSM) techniques, a quasi-experimental method that assumes that households have a certain set of features that could be matched with households in a control group that have similar characteristics (Attanasio, Meghir and Vera-Hernandez, 2004). Opponents of CCT programs expect that labor supply will decrease with such a system. The first results of Bolsa Família show the opposite. Soares et al. (2010) found an increase in labor supply for men of 2.6 percent and for women of 6.9 percent. The results could have something to do with child labor supply, but this need further investigation. One of the main goals of Bolsa Família is to increase education participation for children, what would likely decrease child labor supply (Hall, 2006). Parent’s labor supply could be a substitute or a complement for child labor, but this relationship is not investigated yet.

3.2.2 Mexico (Progresa/Oportunidades)

The CCT program in Mexico was established in 1997 and was first called Progresa and changed name in 2002 to Oportunidades. The dual goal of the program consists of direct reducing poverty and

(22)

breaking with the intergenerational transmission of poverty. This is done through improving education, nutrition and healthcare for poor families (Fernald, Gertler and Neufeld, 2008). In 2004, four million families were covered by the program with a budget of $2.2 billion and by 2007, five million families were involved and the budget rose to $3.7 billion (Fernald et al., 2008). Low-income families were means tested selected on basis of socio-economic status. Benefits are given in two ways. First, families receive conditional payments each month on basis of family members obtaining preventive medical care and healthy nutrition. It aims at improving nutrition, both quantitative and qualitative. The second way is through educational scholarships and is conditional to children in a way that they have to intent at least 85 percent of the lessons and not repeating a grade more than two times (Fernald et al., 2007).

There is no condition on working effort in the program, thus an effect in work response only depends on a possible income effect. Cash transfers are 20 percent of pre-program consumption. A negative work response could be the result of spending time of fulfilling the requirements to participate in the program (Skoufias and Di Maro, 2008). Empirical research shows that there was not a significant labor supply response for adults, though there was a significant reduction in poverty. The effect on labor supply is not exceeding 1 percent for men and women (Skoufias and Di Maro, 2008). In the study the researchers used panel data of households living in treatment areas and control households living in comparable villages, not covered by Oportunidades.

3.2.3 Colombia (Familias en Acción)

The CCT program in Colombia, Familias en Acción, is based on Progresa and is implemented in 2000 (Attanasio, Battistin, Fitzsimons and Vera-Hernandez, 2005). The program has the goal to improve education rates and nutrition. In the specific areas that were chosen, poor families with children below the age of seventeen were eligible for the benefits. The chosen areas were towns that have less than 100,000 citizens and have sufficient educational and healthcare centers and a bank that administer the program (Attanasio et al., 2005). The program focuses on the rural part of Colombia with good infrastructure. Poor families with children below the age of five receive a basic nutritional subsidy, when mothers go to growth and development check-ups. Families with children between six to seventeen years old receive a grant when these children attend school classes in at least 80 percent of the classes (Attanasio et al., 2005).

Attanasio et al. (2004) make use of panel data to make a comparison for labor supply between treatment and control groups. They make use of the same method as Bolsa Família, the propensity score matching technique. Attanasio et al. (2004) found no big discrepancies between the treatment and the control group, concluding that Familia en Acción has no influence on labor supply for men and women. Although the program had no influence on labor supply of adults, it decreased the labor supply of children (Attanasio et al., 2004).

3.2.4 Chile (Subsidio Unitario Familiar)

The CCT program implemented in Chile in 2002 is called Subsidio Unitario Familiar and is set up in the war against poverty. It is different than the programs in Brazil, Mexico and Colombia, because it serves only the extremely poor (Galasso, 2006). The program has a two-pronged strategy. At the demand side, extremely poor households receive direct cash payments and intensive support through social workers for two years and could get additional payments for another three years. This

(23)

side must reduce direct poverty. The direct cash payment is independent of household composition and there is no conditional aspect. The second component aims at increasing the coordination among different programs. This must lead to better long-run opportunities for poor households (Galasso, 2006). The program tries to cover all the extremely poor households, which are 225,000 families.

The data that are used come from 2003 and 2004. The program started in 2002, thus only short-term results are presented. Galasso (2006) found some positive results of the program on labor supply. Participation rates in employment programs of the government increased significantly. However, the results are not followed by an increase in labor supply (Galasso, 2006). A positive work response of the program could be a long-term process that could not be seen yet. Galasso (2006) found a positive and significant labor supply response in rural areas.

3.2.5 Nicaragua (Red de Protección Social)

In Nicaragua the CCT program is called Red de Protección Social and targets poor households in rural areas. Rural households were chosen, because 75 percent of the poor in Nicaragua live in rural areas. There were two core components in the program, food security and education (Maluccio and Flores, 2005). Every eligible households receive a monthly grant, the food security transfer and families with children between seven and thirteen also receive a school supplies transfer and a school attendance transfer. The program is modeled after Progresa and is conditional, because households are monitored to ensure that their children go to education and visit preventive healthcare institutions (Maluccio and Flores, 2005). Red de Protección Social has three specific objectives. First, the program issues income to increase food expenditures. Second, dropout rates in the first four years of school should be reduced. The third objective is to improve the health status of children below the age of five (Maluccio and Flores, 2005). The program started in 2000 with phase 1 and lasted for three years. After evaluation of phase 1, positive findings led to another phase for three more years. Maluccio and Flores (2005) found that there was no effect on labor participation. No difference between the treatment and the control group in the number of people involved in any kind of work was found. Maluccio and Flores (2005) found an insignificant decrease in labor supply in hours worked per week. Women involved in the program worked in 2002 three hours less than in 2000, while women from the control group worked one hour less. Men involved in the program worked on average 3.3 hours more than in 2000, but men that were not involved worked 8.9 hours more on average. This extra work incentive could be due to an income effect. For women, an explanation for the small difference could be that this is consistent with the extra responsibilities associated with the program (Maluccio and Flores, 2005).

3.2.6 Ecuador (Bono de Desarollo Humano)

Of all the Latin American social assistance programs, Ecuador’s is the most comparable to UBI. Bono de Desarollo Humano (BDH) origins in its current form in 2003 and gives a cash payment of $15 per household per month, which is seven percent of monthly expenditures for these households. The payment is the same for every eligible household and does not have any conditions attach to it (Edmonds and Schady, 2009). Contrary to the other cash transfer programs, BDH does not contain a direct stimulus for education participation or food and health security. However, it is not the same as UBI, because not every citizen is involved in the program, only the poorest twenty percent.

(24)

Edmonds and Schady (2009) did not found a significant adult work response caused by the cash transfers. However, they found a significant decrease in child labor supply, which is. The luxury axiom states that child labor supply could decrease with an increase in exogenous income. Edmonds and Schady (2009) found that the luxury axiom holds in this case. They found a 40 percent decline in child labor supply, although the cash transfer is less than 20 percent of foregone income. The effect is the largest in families were children were not able to go to school without the benefits.

3.2.7 All Latin American social assistance programs

In all the six Latin American social welfare programs, no significant negative effect on working effort correlated to the payments was found. In Brazil, Soares et al. (2010) found even a positive work response. The results found by these case studies could be an indicator for the work response of UBI. However, these programs are making payments to the poor and not to middle- and high-income families. Middle and high-income households could have more incentives to decrease their working effort in the case of UBI. The Ecuador case is the most similar to UBI, but also in Ecuador only poor families are involved. In the other cases, also other requirements were set to improve education participation, nutrition and healthcare. Most of the research is done about five years after implementing the programs and long-term results are not known yet. In the long run there could be larger effects on work incentives than in the short and middle run.

Table 3, Labor supply response estimates from the CCT programs in Latin America

Country Work response men Work response women Significant?

Brazil (Soares et al.,

2010) 2.6% 6.9% Yes

Mexico (Skoufias and

Di Maro, 2008) ± 0% ± 0% No

Colombia (Attanasio,

2004) ± 0% ± 0% No

Chile (Galasso, 2006) ± 0% ± 0% Significant positive

increase in rural areas

Nicaragua (Maluccio

and Flores, 2005) - 5.6 hours - 2 hours No

Ecuador (Edmonds and

Schady, 2009) ± 0% ± 0% No

3.3 Basic income based on natural resources

There are a few countries or states where a UBI financed by the revenue from natural resources is implemented. Revenue from natural resources is one way UBI could be financed by the government. In this section I will investigate what kind of program is implemented in Alaska and Iran and look if there is evidence of an effect on labor supply. In Alaska and Iran, oil revenues are used as a benefit to all their citizens. In Iraq and Libya there is a discussion about implementing a basic income. West (2011) states that it is possible to pay every adult Iraq citizen a grant of $220 in October 2012. However, this has not happened, but West’s statement that a dividend program out of oil revenues could eliminate poverty in two years give some attention to a UBI program in Iraq (2011). In Libya oil

(25)

revenues are already used in a national fund. The Qadaffi family used this Libyan Investment Authority, but a new political community could invest the returns from this fund to finance a basic income. Which could in turn reduce the impacts of the resource course (Bremner, 2012).

3.3.1 Alaska

For more than thirty years, citizens of Alaska have received an annual grant from the Alaska Permanent Fund. The cash payment from the fund accounts in 2002 for six percent of the total household income of the 600,000 residents (Goldsmith, 2002). The program started in 1977, the year that the largest oil field of North America was discovered in Alaska. The government was aware that the field would be depleted in the future and therefore they started the fund to set aside a share of the revenues for future generations. A second reason was to keep some of the oil revenues away from politicians (Goldsmith, 2002). The intention was to put 25 percent of the oil revenues into the fund, but in reality ten percent is invested in the fund. At the end of 2002, the fund covers $23.6 billion. The fund is used as investment capital and the payments do not depend on the economic performance of Alaska, but on the return on investment (Goldsmith, 2002). In 1981 the first payments were made to Alaskan residents, when the Alaska Permanent Fund Dividend program started. Half of the five-year average realized earnings of the fund were distributed along all the residents intended to stay in Alaska, including children. The largest dividend was $2,096 paid out in 2008 (APFC, 2014). The program is completely unconditional. Every citizen is eligible to the payments, regardless of circumstances.

There is no experimental data that give information about labor supply response in Alaska. Though Goldsmith (2002) argues that the effect on labor supply is small. There are two reasons for this statement. First, residents see the payment as a gift, rather than a part of their regular income. Therefore, Alaska residents do not change behavior caused by the benefits. Secondly, labor supply response is difficult to investigate, caused by migration effects. The migration effect is moderated, because of the one-year residency requirement. It is possible that retirees and young adults stay in Alaska because of the fund, what could offset a negative labor supply response (Goldsmith, 2002).

3.3.2 Iran

In December 2010 Iran implemented the most generous UBI in the world. Before, Iran had a system with price subsidies. These subsidies were cut, what lead to a raise in fuel prices and other commodities. Iranian residents receive an unconditional grant equivalent to $40-45 a month (Tabatabai, 2012). The payments are paid from natural resources, mainly oil. The payments are made irrespective of income and age. The benefits make up to fifteen percent of Iran’s national income and are therefore relatively larger than the Alaska Permanent Fund Dividend payments (Tabatabai, 2012). The monthly grant substitutes a large price increase for consumer goods, following the cutting of subsidies. The price increase caused a decrease in consumption of goods and services in the range of 20 to 30 percent (Tabatabai, 2012). Therefore, the continuation of the program is far from safe. The universalities are maybe too generous and the economic situation did not improve. Iran is thinking of making the program means-tested or to ask high-income households to not participate in the program.

Because the Iranian basic income program is recently implemented, no research is done yet about the effect on labor supply. It is an interesting case, because a relatively generous UBI is implemented.

Labor supply response to an unconditional basic income : comparison between different case studies

Labor supply response to an unconditional basic income:

Comparison between different case studies

Table of contents