Ville-Veikko Pulkka is a visiting postgraduate scholar at the Institute for Policy Research (IPR). He has previously worked with the Social Insurance Institution of Finland, Kela, on their recommendations for the Finnish basic income experiment.
The long wait for basic income advocates and social policy scholars is finally over, following this morning’s release of the first results of the Finnish basic income experiment (2017–2018). The preliminary results, based on the first year’s register data and a survey that was carried out in the autumn of 2018, will be followed by a more in-depth analysis in March, and a final report due in spring 2020.
In summary, the data reveals that the amount of days in employment for the test group (2000 randomly selected unemployed who received unconditional basic income of €560 a month*), did not differ in a statistically significant manner from the control group (173, 000 unemployed in November 2016, who did not receive basic income).
However, descriptive analysis of the survey** results indicate that the test group experienced better subjective well-being than the control group, and whilst firm conclusions cannot be drawn from this preliminary analysis, it is likely that the results will be welcomed by many basic income advocates. For instance, the recipients of basic income experienced fewer problems related to health and stress, and were more confident in their own future. Moreover, the results suggest that receiving basic income did not make people more passive, quashing the most fundamental moral hazard argument against basic income.
It is possible though, that some basic income recipients were more passive while others became more active, which would have caused a neutral effect. Either way, the neutral employment effect combined with the positive survey results could increase the social legitimacy of basic income in Finland.
But what can we actually learn from the results?
The Finnish experiment highlights that social scientific experiments are extremely prone to methodological issues resulting from a lack of political commitment. In the Finnish case, the lack of political commitment materialised as an inadequate budget and an overly strict timeline imposed by the government. It is evident that the experiment did not provide enough representative results, and the tax treatment of the test group also raises the question of the validity of the results for further basic income analyses.
Additionally, reckless interpretations of the results by media, policymakers, and the public are a plausible risk and can have far-reaching consequences on the future of policymaking. To mitigate the risk of uninformed interpretations, it is essential to point out the major limitations of the research design, and their implications for interpreting the results.
Narrow study population does not provide representative results
By definition, basic income is a universal benefit paid by a political community to all its members. The level of universalism may vary in different proposals, but in most cases the entire working age population is eligible to receive the benefit. In other words, when providing representative results from an experiment, the study population should represent its potential recipients. This highlights the most visible complication of the Finnish basic income experiment, which is undoubtedly its narrow test group, consisting of only 25–58-year-old long-term unemployed and unemployed individuals with a short working history.
If the primary focus of a basic income experiment is to study employment effects – as was the case in the Finnish experiment – it is reasonable to sample socio-demographic groups that can be expected to respond to changes in social benefits. From this perspective, focusing on the unemployed was a justified decision by the Finnish government. However, as the sample consisted of only the long-term unemployed and unemployed with short working history, it still excluded almost half of the unemployed – individuals eligible for earnings-related unemployment benefits.
The effects of basic income on labour market behaviour is one of the most fundamental questions, and should be studied with scientifically solid experiments. It is reasonable to argue that the effects on labour supply may even determine whether basic income can de facto become a politically feasible option in the future. The Finnish basic income experiment did not make the long-term unemployed passive, but the case could have been different, for instance, among precarious workers. To address the labour supply question in a comprehensive enough manner, the low-income working population should have been included in the study population, at the very least.
Tax treatment complicates the interpretation of the results
Implementing a revenue-neutral basic income does not necessarily change people’s net taxation (i.e. taxation when the benefit is considered). Yet, implementing a revenue-neutral basic income implies unavoidably higher marginal tax rates.
Excluding tax reform in the experiment has serious ramifications for interpreting the results. First, if the test model was implemented at the state level, the estimated budget deficit would be as high as €11 billion. Given Finland’s state budget of €55.3 in 2019, it is somewhat easy to argue that the test model was at least economically unrealistic, if not unfeasible. Second, and more importantly, since applying the current progressive taxation in the experiment substantially improved monetary incentives to participate in the labour market, the employment effect could be interpreted as a result of this.
Contrary to deep-seated beliefs, and due to the relatively high marginal tax rates that are required to fund a realistic revenue-neutral basic income, it is difficult, if not impossible, to increase monetary incentives to participate in the labour market without diluting the current level of social benefits. Abandoning means-testing does not automatically make work always pay if complementary benefits are still needed to guarantee an adequate income. In advanced welfare states such as Finland, replacing housing benefits and earnings-related benefits would require a higher basic income, and thus, an unrealistically higher marginal tax rate. As an illustration, according to microsimulations carried out in Finland, a level of €1500 a month would already require a flat rate tax of 79%.
Determining whether the changes in people’s behaviour are a result of monetary incentives, or unconditionality and the removal of means-testing, may have implications for not only future basic income discussions, but also the principles of any future social benefit reform. If monetary incentives are increasingly emphasised in discussions concerning behavioural effects of social benefits, the target may be achieved by weakening social benefits and tightening sanctions. However, if the changes are primarily interpreted as a result of reduced bureaucracy, the meaningfulness of conditions and obligations attached to unemployment benefits might be revised and gradually eased.
Interestingly, considerable monetary incentives did not increase basic income recipients’ participation in the labour market. On the other hand, the neutral employment effect is not particularly surprising given the persistent difficulties for the long-term unemployed to find work. Nonetheless, the unrealistic tax model raises the question of whether the employment effect would have been negative without the substantial improvement of incentives. To put it differently, it could be maintained that the unrealistic incentives neutralised the negative effects of unconditionality.
The only reliable manner to study the effects of unconditionality and the removal of means testing would have been to test a revenue-neutral model – either alone or alongside the actual model that substantially improved monetary incentives. The survey results offer indirect information on the bureaucracy effect, but only a register-based evaluation can provide data on the causal effects.
Extending the experiment is essential
The preliminary results of the Finnish basic income experiment provide long-awaited empirical information on the potential effects of basic income. The results are an intriguing adjunct to the theoretically-oriented basic income discussion, but the limitations of the research prevents us from drawing unambiguous or exhaustive conclusions about the actual effects of basic income.
The only scientifically justified political decision would be to extend the research experiment immediately after the next parliamentary elections held in April. Studying a representative population and introducing a revenue-neutral tax model are essential if we are genuinely interested in evidence-based decision-making on basic income. In an ideal situation, the extension would also include testing different levels and tax models.
Since studying basic income’s employment effects were the centre-right government’s primary reason to launch the experiment, the Finnish trial focused excessively on this dimension. This unavoidably neglected studying the potential social and health effects in a comprehensive manner. As the survey results suggest, it is necessary to further explore the social and health effects of basic income if new experiments are to be launched in the coming years. If the meaningfulness of basic income is merely assessed from an employment point of view, the burden of proof, is unreasonable.
Randomised controlled trials can provide valuable information for reforming social benefit systems if robust political commitment is secured from the start. In the case of basic income, it is, however, likely that even the most comprehensive research setting cannot reveal the absolute truth about the long-term effects of basic income. No matter how sophisticated data researchers can offer, politicians are still needed to do their part.
*€560 a month corresponds to the net level of basic social security benefits and was supplemented with housing allowance and social assistance in the experiment. Given the relatively low net level, approximately one third of basic social security benefit recipients are dependent on social assistance which creates major bureaucracy traps resulting from means testing.
**It is necessary to note that response rates for the survey were rather low (31.35% among the treatment group & 20.29% among the control group) and no baseline survey was carried out.