Why Have an Experiment at All?

Abstract

This chapter considers whether it is, after all, worthwhile (both strategically and scientifically) to have a Universal Basic Income (UBI) experiment, given that earlier chapters have shown so many difficulties experiments have in addressing the most important issues in the public discussion of UBI.

This book’s goal is to examine the many potential pitfalls of UBI experiments, so that people learn more from the experiments we’re doing. The book is not about whether experiments are after all a good idea. I’m largely neutral on this question, but given the many shortcomings I’ve pointed out, I feel obliged to consider it. You can approach it both scientifically and strategically.

Strictly speaking, science cannot tell you whether to do anything or not. That depends on your values. But I can think of at least three ways to approach this question from a scientific perspective.

First, can science settle the disagreement? It can’t, but cases where experiments can settle disagreements are rare. The distance from experiments and other research methods to anything like a bottom line is a common methodological problem across the social sciences.¹ To expect it on an issue like this is to expect more than most social science can deliver.

Second, do UBI experiments add to our understanding of this policy? Certainly, what they can do is limited and tentative, and to get people to truly understand the contribution they do make, researchers will have to point out how limited and tentative their contribution is, but doing so risks giving people the impression that they aren’t very valuable at all.

Perhaps the most compelling reason to use the experimental method for an issue like this is, because “all the available methods of studying politics are pretty bad.”² Given the limitations of the four other methods mentioned in Chap. 17, it’s plausible that field experiments can make a valuable contribution. Enlightening the discussion with improved evidence requires open-minded self-reflection on the limits of what each method contributes to our understanding, which will still be limited even as evidence gradually improves it.³

Experiments aren’t great, but neither are micro- or macro-economic simulation models. There are lot a of unknowns about this largely untried policy (UBI). An experiment—used in combination with other also-limited methods—is a way for social scientists to fill in a few of those gaps, while a lot of unknowns remain. If we think UBI experiments—or any other social science method—can do more than that, we have unrealistic expectations.

There is no strong, scientific downside to conducting an experiment. It’s not prohibitively expensive or dangerous to the subjects. Most of the past experimental evidence available on UBI is very specific to the time and place where it was gathered. If one polity conducts a UBI experiment, it can learn something about how UBI works relative to existing alternatives in that context. If many different polities experiment with UBI, we can hopefully piece that information together into a slightly better shared understanding of UBI’s effects in more general terms.

Third, one can ask whether there is a scientific need to conduct an experiment. Would it be irresponsible for policymakers to seriously consider this policy without testing it first, the way we learned that it was irresponsible to allow the sale of Thalidomide to pregnant women without adequate testing? Here the answer is clearly no. While there is no strong, scientific downside to conducting an experiment, there is no strong, compelling downside to introducing UBI without further experimentation. Most major policy changes are simply rolled out without advanced experimentation. And this roll-out can begin modestly and increased gradually, while policymakers fix problems as they come up.

UBI is certainly compatible with this kind of process. Some level of it is sustainable; some level isn’t. For UBI to be unsustainable would require not just some reduction in work effort, but a massive labor-market withdrawal that made essential industries unprofitable in ways that could not be counteracted either by automation or by enticing workers back to the labor force with better wages and working conditions. This process does not seem likely even with a substantial UBI. And if it seemed to be moving that way, we could simply reduce the UBI to a more modest level. I suspect the bigger problem with UBI would be the political difficulty of raising it to a level that is high enough rather than cutting it back if it is unaffordably high.

The strategic question is very different: will good scientific research help demonstrate the efficacy of UBI and attract support? Perhaps, but experiments have a lot of risks for UBI supporters. Even if experiments are good science and find promising results, Van Parijs and Vanderborght warn of the possibility of “damaging backlash analogous to the one that followed the North American experiments.”⁴ Others worry about a double standard: why is UBI subject to so much testing when most social policy is rolled out with little or no advanced experimentation?

Although UBI supporters may be rational to desire the immediate introduction of UBI, that is still an uphill battle. At the rate the UBI movement has grown over the last few years, that could change, but at the moment, UBI remains an outside long shot, and experiments are a strategic attempt to build the movement further. I’ve argued that the Namibian and Indian experiments played an important role in sparking the current UBI movement.⁵ Whether the 5–10 experiments getting underway will push the movement further remains to be seen. They provide the opportunity for UBI supporters to show they’re interested in evidence-based reasoning and are willing to subject their idea to testing and revision if necessary.

Evaluating experiments as a political action requires comparing them to other strategies to promote UBI. In this sense, UBI experiments come off very well because, for the most part, they are not coming at the expense of the other things supporters are doing to promote it. If you’re a major donor to Y Combinator, the Economic Security Project, or GiveDirectly, this strategic question might be important for you. If you’re anyone else, you can look at the experiments as a bonus. UBI supporters are free to go on with just as much activism as before. As long as the experiments have even a minor contribution to the UBI movement, supporters can consider them a publicity windfall.

Although the risk that experiments will backfire exists, not all experiments have backfired, and past experience provides lessons on how to resist backlash this time. I don’t think either researchers or UBI supporters are capable of controlling the reaction to experimental findings to prevent negative spin. And they are not immune to doing their own spin. But I do think they’re better prepared to handle it fairly than researchers or BIG supporters were in the 1970s.

And we should not look at the 1970s experiments as negative on the whole. The media response at the time was negative, but the NIT movement was already in serious decline before the major negative media discussion got under way. The mere fact that government conducted these experiments has given BIG credibility ever since. And the popular understanding of the 1970s experiments has greatly improved in the last 10–15 years. Even if the experiments had a net negative effect on the BIG discussion at the time, perhaps, by now, they have had a net positive impact on the current UBI movement.

Finally, the question of whether we should have UBI experiments is moot. We are having them now. We are having them not because of a careful consideration of strategic or scientific perspectives on why to have an experiment, but because of the complex political process discussed in Chap. 9. The question is not whether to conduct an experiment, but how to make the best of the experiments being conducted now.