Public interest in Basic Income, or BI (known as Universal Basic Income in the UK), has increased significantly during the last few years. In addition to academic writings, there are new BI initiatives, intense debates on BI in political elections, and real-life BI experiments initiated around the world.
Although Basic Income’s success critically depends on how the model is implemented, debates on the subject tend to be abstract. Few advocates have explicitly considered questions of implementation. A more detailed analysis – as well as valuable lessons for implementation – can be found in BI field experiments in countries such as Finland.
Finland put basic income into practice
The Finnish government launched an experiment on BI in 2017. The main idea of the experiment is to find out how a BI affects the labour market behaviour of individuals. The tested model has fewer conditions and requirements for receiving basic social benefits than the current system, and it increases the monetary incentives for working.
The Finnish experiment follows the idea of a randomised controlled trial (RCT) where the causal effect of a policy is estimated by comparing two similar population groups. The only difference between the two groups is that one receives the BI and the other continues living under the current tax-benefit system.
The Finnish experiment is the result of a government-initiated research project and an intense preparation process. While the main objective of the experiment is the study of the policy outcomes, the preparation work also provides us with valuable information about developing a feasible model of a BI. During the design process, we learned many important lessons on conducting social experiments in general.
Here we will limit our discussion to four questions. We believe that they serve the often neglected, administrative analysis of BI. Different answers to our questions lead to different redistributive outcomes: some households end up winning, some end up losing. Therefore, the questions are political in nature.
1. Which population groups are eligible for basic income?
In Finland, all registered residents are insured for, for example, unemployment, disability, sickness and retirement. If we followed this principle, our BI model would cover a very large proportion of the population.
When thinking about the coverage, the designer must keep the main goal of a BI reform clearly in mind. Is the goal to increase the income security of certain, particularly vulnerable population groups (e.g. the homeless or the poor, immigrants, or those who are dependent on several social benefits at the same time)? Or is the main idea to replace (either totally or partially) the existing income transfer system for the whole population?
2. What happens to the existing social benefits?
Distributing a basic income to all beneficiaries on top of their current benefits would not be economically sustainable, so a BI would naturally replace some of the existing social benefits. Thus, the central question is: which benefit systems are we going to replace?
In the model tested in Finland, BI replaces the minimum level of benefits that the beneficiaries are entitled to by status (e.g., studying, unemployment, or sickness). Additional parts of basic benefits, earnings-related components, compensation for housing costs, and last resort social assistance remain intact to maintain the current level of social security.
3. What should the amount of basic income be?
The goal to replace existing benefits with a BI leads us to think about its level. If the level were set according to the lowest primary benefit (e.g., the study grant of €250 a month), we would still need all the other benefits to maintain the beneficiaries’ level of income.
Setting the level higher, for example, equal to the guarantee pension (€775 a month), makes studying and taking care of children at home economically more attractive than today. The designer should ask, whether this is the objective of a BI reform or not.
Problems of replacing the existing schemes lead us back to our first question: If we cannot define only one meaningful level of BI, should we then think about a model where the level is different for different population groups? On the other hand, this would increase the administrative burden of a BI system, because the beneficiaries’ changes of status would require following. As a result, we would lose the simplicity that BI promises.
4. How should we adjust income taxes with basic income?
An economically feasible BI system would also entail adjustments in income taxation. The obvious objective would be to keep the level of disposable income of middle-income groups more or less unchanged. In the Finnish basic income experiment, there was no need to find a solution to taxation, because the government excluded a tax reform from the experimental model.
Before the experiment, we analysed several cost-neutral BI models by running tax-benefit simulations. We learned that in a scheme where the tax rate on personal income is flat, the biggest losses would come to part-time workers who receive unemployment benefits (income-adjusted in-work benefits). This approach to find an economically feasible BI seems to produce outcomes that are contradictory to what is often expected from a BI policy.
Radical vs. incremental reform
Based on our simulations we conclude that, at least in the Finnish context, an introduction of a BI as such does not lead to any major improvements when it comes to the costs for the public economy, work incentives, or income distribution. The comparative advantage/disadvantage of a BI depends on the details of the model: At whom is the BI targeted? What is its level? What are the conditions to get it? How are additional benefits and taxes adjusted to the BI?
We could also choose a politically easier way without a system-wide BI reform. We could change the levels of the current basic security benefits, coordinate them more uniformly – possibly merge some of them – and adjust taxation to achieve the intended outcomes of income distribution and work incentives.
From models to experiments
Above, we discussed the choices in designing a BI scheme without any systematic consideration of their behavioural effects. Even if the income levels did not change as a result of the BI model, the behavioural outcomes might change the picture. Definitions and simulations do not tell how an unconditional BI affects the recipients’ behaviour. Does it activate people in job seeking? How much does a BI increase activities outside the labour force, such as studying, taking care of children at home, or participating in voluntary work? When the designers have defined the model, the next step is to start experimenting with it.