17 Comments
User's avatar
J.K. Lund's avatar

Thanks for sharing this Maxwell , disappointing, thought not entirely surprising to see the results of UBI. This is one reason I have backed away from a full endorsement of UBI. We need more data.

I am curious if you have any good studies of this kind for a negative income tax?

Expand full comment
Maxwell Tabarrok's avatar

There was a famous negative income tax RCT in the 70s I think. I've heard people mention it in econ classes but don't know too many details.

Expand full comment
J.K. Lund's avatar

I will have to look for it. Thank you.

Expand full comment
mathew's avatar

Very interesting results. Seems to confirm the pessimistic view of the welfare state.

And the many unintended consequences that can happen when trying to help people

Expand full comment
JD Free's avatar

Whatever the limitations of trials like these, they are still SOME evidence, whereas the people arguing against their implications have none.

Expand full comment
Kartik's avatar

Well CDLZ (Cengiz et al) performs a very meticulous review of state-wide minimum wage data and concludes that they have no employment effect. It is _some_ evidence

Expand full comment
Thomas L. Hutcheson's avatar

On #4 a question would be did total wages rise? Did the treatment successfully transfer income at the expense of employment. It's the magnitude of the elasticity that is important, not just the sign.

Expand full comment
Simas Kucinskas's avatar

Thank you for the great writeup.

That said, I think we need to be careful not to over-extrapolate from single studies. Even very careful experiments can get things wrong, for various reasons.

Also, I feel it would makes sense to update more on the three "no-effect" studies than on the minimum-wage experiment. First, most things fail, so our prior should probably be these interventions won't work, and if we see that confirmed, that seems pretty definitive (https://inexactscience.substack.com/p/most-ideas-fail-and-thats-fine). Second, there's the argument of Abadie (2020, AER) that non-rejections of null hypotheses are generally more informative.

Expand full comment
Kartik's avatar

You link to "the average of empirical findings" on minimum wage - but the graph you linked is from Doucouliagos and Stanley, which account for bias and "corroborate that Card and Krueger’s findings were nevertheless correct".

https://www.ctdol.state.ct.us/lweab/Doucougliagos%20&%20Stanley%20Publication%20Selection%20Bias%20in%20Min%20Wage%20Research-A%20Metaregression%20Analysis.pdf

Expand full comment
Peter Defeel's avatar

> at a sufficiently high minimum wage, the probability of hiring goes down,

Of course it does, I doubt anybody thinks that paying $1M a year for a service worker would lead to anything else than mass unemployment for service workers. What would be more useful is an economic analysis that worked out what that „sufficiently high level“ actually is. That would be proper theory.

Expand full comment
Maxwell Tabarrok's avatar

Well, the minimum wages they set were 2, 3, or $4 dollars an hour so I think they are relevant

Expand full comment
Michael A Alexander's avatar

I don't get how people are hired at these rates in the first place. Who is willing to work for $2/hour (or less it would seem in the control group).

Expand full comment
Kartik's avatar

Cengiz et al suggest "between 37% and 59% of the median wage" as having no employment effect

Expand full comment
Dmitrii Zelenskii's avatar

These results are an epitome of "no way, Jose". Something wasn't controlled for, something went wrong.

Expand full comment
ConnGator's avatar

Are you saying this only because the results are (somewhat) counter-intuitive? I am not saying these results have to be true, but lots of counter-intuitive things indeed are true.

Expand full comment
Dmitrii Zelenskii's avatar

I'm saying this because the results are not just counter-intuitive in general (which is, indeed, allowed), but go out of their way to be counter-intuitive in details. Like, "no effect for medical debt cancellation" can have a dozen of explanations, but "detrimental results for mental health"? Null results for things somewhat correlated with income after UBI is one thing, but losing improvements in food insecurity? "Substitution of higher productivity workers for lower productivity workers" is also a bit suspicious, imo, although less so.

The usual disclaimer: I am not an economist, not by a long shot.

Expand full comment
ConnGator's avatar

I agree that the productivity substitution claim seems backward, but maybe it is just poorly written. Perhaps Maxwell could clarify these items.

Expand full comment