Social Experiments 101 Continued

Michael Lewis. Professor, Hunter College.

Several years ago, I wrote this piece about UBI experiments, more precisely unconditional cash experiments. It focused on an experiment that had been run in Finland. There have been many more experiments, some of them called “pilots,” since then. So, this is a good time to take stock. In this piece as well as a series of forthcoming ones, that’s exactly what I’ll do.

We’ve gotten to the point in the proliferation of unconditional cash experiments where some are wondering aloud whether we’ve experimented enough. At a recent UBI conference, I heard a panelist bluntly ask how many more experiments we need. They continued claiming, “We know unconditional cash works; why must we keep showing this over and over again?” I wonder what we’ll gain from conducting more unconditional cash experiments too. But my wondering about this isn’t based on the claim that we know what works. Before I dig into this point, I need to say a few things about why we might want to conduct unconditional cash experiments.

One reason for conducting such experiments is that they are ways of getting much needed cash to trial participants who were lucky enough to be randomly assigned to receive unconditional cash. Being randomly assigned to receive, say, $1,000 a month for 2 or 3 years is a huge boost to many of those taking part in these trials.

A second reason for conducting these experiments is that they may draw or sustain attention to UBI or related policies. I first started writing about UBI about a quarter of a century ago. There was much less attention paid to the policy at that time, and I suspect unconditional cash experiments, starting with Castro and West’s work in Stockton, CA, have something to do with this.

A third reason for conducting unconditional cash experiments is that it gives statisticians/social scientists interested in UBI (or related policies) a way to enter the terrain. Admittedly, it seems a bit yucky to say that one reason to conduct experiments is that it gives work to researchers. But if we think there’s a role for academics to play in the UBI discussion and some of them are best positioned to play that role by running experiments, I suppose there’s something to be said for this third reason.

So far, the reasons I’ve proposed have nothing to do with experiments helping us gain knowledge that’s relevant to whether we should enact a national level UBI or how, if enacted, it should be implemented. These reasons, having nothing to do with acquiring knowledge, are what philosophers might call “non-epistemic” ones. Reasons having to do with gaining knowledge relevant to adoption of a national level UBI are epistemic ones. Now I can state why I wonder what we’ll gain from conducting more unconditional cash experiments; I’m not convinced that conducting more experiments will have much of an epistemic payoff. That is, I’m not convinced we’ll learn much from them that will help with enactment/implementation of a national level UBI. Some of my forthcoming pieces will focus on why I feel this way.

I suspect, though, that my doubts about how much we’ll learn from more unconditional cash experiments will have little impact on whether more of them are conducted. So, these doubts notwithstanding, it may be worth thinking about how such experiments might be conducted in a way that will enhance our knowledge. Some of my forthcoming pieces will also address this question. Before doing either of these things, though, I first must get something out of the way called the is/ought fallacy.

A distinction called the “is-ought” distinction comes from philosopher David Hume; he told us that you can’t deduce a claim about what ought to be the case from a claim about what is the case. This distinction is related to unconditional cash experiments. To see how, suppose an experiment finds that an unconditional cash grant resulted in a reduction in labor supply or, God forbid, an increase in recreational drug use. Some might be tempted to conclude from this that unconditional cash is illegitimate or unjustified. The underlying argument here would be something like the following:

• Premise: the unconditional cash grant reduced labor supply or increased recreational drug use
• Conclusion: therefore, an unconditional cash grant shouldn’t be implemented

The problem, Hume told us, is that it’s fallacious to derive an ought to claim solely from an is claim. This is the is-ought fallacy. If we want to draw the conclusion above, we’d need to add a premise such as the one added in the following:

• Premise: the unconditional cash grant reduced labor supply or increased recreational drug use
• Premise: if a policy reduces labor supply or increases recreational drug use, that policy shouldn’t be implemented
• Conclusion: therefore, the unconditional cash grant shouldn’t be implemented

When we state the argument this way, it should be clear that even if the first premise is true, the second one is debatable. If a policy reduces labor supply or increases recreational drug use and even if we assume these are bad outcomes (a questionable assumption in itself), it may have other good outcomes that outweigh these bad ones. Anyone conducting an unconditional cash experiment or consuming findings from one should keep the is-ought fallacy in mind: just because a study showed that unconditional cash led to something many people don’t like, that doesn’t necessarily mean unconditional cash shouldn’t be implemented. Having gotten this is-ought business out of the way, my next piece will begin explaining why I have doubts about the epistemic payoff of more unconditional cash experiments.