The preliminary results of the Finnish Universal Basic Income (UBI) experiment (2017-2018) have been released. So, what does the preliminary evidence from the experiment show? The unwelcome answer is: not much – at least not much that we didn’t already know based on previous studies.

In the famous thought experiment by the physicist Erwin Schrödinger, the cat in the box is simultaneously dead and alive. Similarly, the Finnish UBI experiment is simultaneously a success and a failure, depending on one’s background assumptions and interpretive slant. In behavioural psychology, the concept of “framing” explains how the same situation can be explained in wildly different ways (sometimes even diametrically opposed), depending upon one’s frame.

As an illustration, here are two contrasting headlines about the same experiment:

1. “Universal income study finds money for nothing won't make us work less.” - New Scientist, February 8, 2019.

2. “Universal basic income trial in Finland fails to help unemployed people back to work.” – The Independent, February 8, 2019.

Which one is correct? Or are they both right? (Spoiler alert: yes.)

Despite some false reporting in the global media, the Finnish UBI experiment ran its uninterrupted course for the two years (from January 2017 to December 2018) and has now concluded. The primary aim of the Finnish UBI experiment, as stated by the centre-right government, was to test whether UBI could be used to incentivize employment. The target group consisted of 2,000 long-term unemployed people, between the ages of 25 and 58, randomly selected from all over Finland. Participation was compulsory. The target group were given an unconditional monthly payment of €560, no strings attached, with the ability to top up their income from any other sources.

We will have to wait until 2020 for the final report, but the government has released some preliminary results based on a) employment data from the first year of the experiment and b) a phone survey conducted towards the end of the pilot. According to the government’s own report, the preliminary results suggest that “self-perceived wellbeing improved” and “during the first year [there were] no effects on employment.” If we combine these findings with comparable data from studies around the world, it is likely that UBI would not significantly reduce work incentives. Nor would it significantly improve them. The glass is half full and half empty, at least in the short run. The results also suggest UBI is likely to boost psychological wellbeing and health indicators of poor people—but again the size of the effect is subject to uncertainty; and it is unclear whether the boost is permanent.

Furthermore, it is important to realize that the Finnish experiment suffered from several methodological problems from the very beginning:

• Observing the long-term effects, both economic and psychological, will take years, even decades.

• The experiment excluded various key demographics – e.g. part-time workers, entrepreneurs, young people, and disabled people – who might benefit from the in-built flexibility of the UBI system.

• The experiment did not include a regional saturation study, so we cannot say anything about the “network” effect of a whole community being involved with UBI.

• The income tax system was not changed to accommodate for the UBI system.

• Many UBI-recipients were still eligible to apply for various conditional unemployment and other benefits; and most of them did. This means that most UBI recipients were still subject to similar conditions as the people in the control group (which goes against the notion that UBI automatically liberates people from the constraints of conditional bureaucratic controls).

That’s a lot of problems with the Finnish UBI experiment. But, at least from the narrow perspective of employment prospects and work incentives, and based on all the accumulated evidence, it seems that UBI stands roughly shoulder to shoulder with workfare programs and conditional unemployment benefits – neither better nor worse. Similar results were found in the Canadian and United States experiments half a century ago. And from the point of view of intangible benefits – such as psychological wellbeing, freedom, and dignity – UBI seems to perform better than its competitors. Therefore, given that there is no evidence of a looming incentive catastrophe – such as an epidemic of laziness – it seems to me that the experimental evidence allays many of the common fears about UBI.

However, the current fascination with randomized control trials (RCTs) and evidence-based policy (EBP) is a double-edged sword. Ideally, it can be used to stress test controversial policies. In practice, however, it contributes to a misunderstanding of both politics and science, because it obscures a) the biases and heuristics that drive actual government policy and b) the inherent uncertainty in the scientific data. Furthermore, a small-scale experiment can only give a glimpse of the behavioural responses that can be expected in a wider-scale implementation. An excessive use of experimental data may therefore lend a thin layer of scientific credibility to technocratic policy making – a credibility that it desperately craves but rarely deserves.

It is a little bit ironic that UBI is being tested by such a technocratic tool as an RCT experiment, since UBI supporters often express a healthy scepticism about the capacities of technocratic governance. UBI would constitute a paradigm shift that transfers powers away from the meddlesome, convoluted bureaucracies that characterize the existing benefit system. It could hopefully give poor people an increased sense of freedom, security, and dignity. But UBI is not a perfect system, and it comes with many potential dangers. Firstly, it needs to be set on a sound fiscal footing and constrained by institutional checks and balances. Secondly, it should be implemented as part of a broader range of liberalisation reforms in labour markets, entrepreneurship, and technologies. (For a broader discussion of the design parameters of an ideal UBI system, see my research paper for the Adam Smith Institute.)

Overall, these kinds of experiments are a Rorschach test for the reader: one is liable to see what one wants to see. The lack of an employment effect, for example, can equally be given a positive or negative spin. The power of narrative shapes the political discourse as analysts from all sides are liable to interpret the same incomplete and inchoate data as definitive proof of their own biased predisposition. Drawing overconfident and unwarranted conclusions from the partial and incoherent data makes for exciting headlines for zealous newspaper editors, but it does not constitute good science nor good governance. In order to avoid diluting the scientific value of these experiments, it will be necessary to educate both the public and the politicians about the dangers of over interpreting experimental data; and to show them that the welfare of the poor is not amenable to scientific micromanagement. UBI will not kill us – but technocratic hubris might.