An interesting working paper was published this month by economists Rik Chakraborti (Christopher Newport University) and Gavin Roberts (Weber State University), “How Price-Gouging Regulation Undermined COVID-19 Mitigation: Evidence of Unintended Consequences.”
These price controls created shortages, which, according to economic theory, would have been more severe in the 42 states that already had price-gouging laws on the books or (inexplicably for an economist) rushed to legislate them after Covid hit. The federal Defense Production Act, invoked by Donald Trump, added more biting price controls on pandemic-related supplies (such as personal protection equipment) but is not considered in the Chakraborti-Roberts paper.
The authors used a database of cellphone-tracked mobility to calculate “average exposure of smartphones to each other within commercial venues.” Comparing states with and without price-gouging laws between January 22 and May 3, 2020, the econometric study confirmed that these laws were associated with more physical visits to commercial venues (especially from individuals in the lowest income quartile), as people were frantically looking for sanitizer and other goods in shortage. This increased shopping is likely to have increased contacts and infections. After controlling for state population density (which can have a compounding effect on infection), lockdown orders, and other factors, the econometric estimates suggest that price-gouging laws explain at least 25% of the early-April, first-wave Covid deaths in states with such laws.
We’ll have to see if these results are confirmed by other studies but they make economic sense. The regulatory welfare state may not be as nice as we thought, at least in its consequences. As for intentions, an old saying in many languages suggests that the road to hell is paved with them.
READER COMMENTS
Jon Murphy
Mar 19 2021 at 12:27pm
I have to admit: I’m a little skeptical of their findings. 25% is a huge proportion. Coupled with testing issues and the fact we still don’t know how this thing spreads (eg see the latest CDC guidelines which are a contradictory mess), teasing out any causal effects from the shortages are going to be tough.
Gavin Roberts
Mar 19 2021 at 12:44pm
I was pretty shocked by the 25% as well but I think it is possible given the exponential nature of contagion. My view is the strongest evidence is the theoretical part as emphasized by Pierre here. We are currently working on putting together a dataset with county-level data and hope that will allow for a more granular and robust econometric analysis.
Jon Murphy
Mar 19 2021 at 2:27pm
Awesome. I agree with the theory and I’m excited to see how the paper goes.
Chris
Mar 20 2021 at 11:21am
I don’t honestly know how you’d account for the lack of testing at that time, combined with our continued issues of really not fully understanding when spread occurs.
also, in a quick event like we had at the beginning, a jump in price wouldn’t have had an immediate impact on supply. It takes time to ramp up supply, and manufacturers already were because of the huge demand. Therefore we wouldn’t have seen increased the availability of supplies for weeks at best. What it would have done is limit supplies to the poorest, who also happen to be those in jobs and living conditions with the highest COVID risks. Are you accounting for that in any way? It seems like you should.
also, just a nitpick but a theory isn’t evidence. You establish theory through evidence. It undermines your argument when you make it sound like you are trying to support a specific theory because that’s not how the scientific method is supposed to work.
Jon Murphy
Mar 20 2021 at 11:52am
Chris-
As a point of fact and theory, supply is only half of the equation. Whether or not supply could rapidly ramp up does not alone explain shortages. A higher price would also reduce quantity demanded.
Not a nitpick because this is a significant point: one does not establish theory through evidence. One tests theory with evidence, but one cannot establish theory through evidence. You need theory to interpret evidence. Evidence without theory is unintelligible.
As an aside, the theory-evidence connection is one of the major empirical issues I have with the high-school scientific method story. It implies that all one need to is stare hard enough at data and Truth emerges. It ignores the countless judgement calls the scientist must make in his analysis, not the least of which is which theory to use to analyze the data. In other words, the “scientific method” is a lot like the perfect competition model in economics insofar as there is no science in the scientific method. Just like there is no competition in the perfectly competitive model.
Pierre Lemieux
Mar 19 2021 at 2:03pm
Jon: I am not totally persuaded by the magnitude of their result either. Their method is interesting but their time frame is short and their less significant results for toilet paper is strange. But it doesn’t disprove price theory and it does show the way for future research.
Jon Murphy
Mar 19 2021 at 2:28pm
Agreed, Pierre. I think the paper, at first blush, is important. My fear is teasing out causality, but the theory is solid.
Rik Chakraborti
Mar 22 2021 at 10:58am
Thank you for your interest in our research, and your valuable feedback, Pierre, and Jon!
Pierre, I’m guessing your reference to the weaker toilet paper result is about our other paper published at the Journal of Private Enterprise, because the CGO working paper referenced here — related to increased retail store visits and resulting COVID-19 transmissions — doesn’t really deal with product-based online searches at all. The weak toilet paper result you note in that earlier paper is simply because we had a much smaller sample to work with at that point: Feb 15 – March 25. Chances are the full effects hadn’t manifested by then. But since we started the work in April, that range was all we had access to. However, in a follow up paper (https://dx.doi.org/10.2139/ssrn.3672300), currently under review, we are looking at the differences in online searches driven by activation of preexisting price-gouging laws versus activation of new ones introduced in response to the pandemic. Conveniently, the sample is longer (Jan 22 – May 3rd), and we do find strong results for toilet paper as well.
Jon Murphy
Mar 22 2021 at 1:26pm
Hi Rik-
I saw the Journal of Private Enterprise article (I have an article in the same issue). I love the concept you guys are going with. My concerns have nothing to do with the theory; just the quality of data. I’m glad to hear the follow-up paper is generating more robust results. Your work is tangential to some of the work I am doing with Roger Koppl on expert failure during the pandemic
Rik Chakraborti
Mar 22 2021 at 5:29pm
Thanks for your response, Jon! I’ll definitely give your JoPE article a read, and check out your work on expert failure during the pandemic (sounds fascinating!). On a separate note, I loved your crisp note clarifying the significance of theory in response to Chris’s comment above. Very clear, and on point.
Pierre Lemieux
Mar 23 2021 at 9:56am
Rik: You’re right about TP. I had confused your two articles.
JFA
Mar 20 2021 at 8:42am
As the author is reading the comments, I thought I’d add some requests. Some visualizations of the raw data would be quite useful, particularly trends overtime. A grid that displays each state with a vertical line at the emergency declaration date would be nice. What do changes in social contacts look like in the no-gouging states vs. gouging states? The no-law states (mostly empty plains states) seem quite different for the others; it might be a good idea to either compare each state to a similar state or to do a synthetic control group. You’ll probably also want to do a placebo event study for the no-law states (probably using some mid-March date) and compare the effect sizes to the anti-gouging-law states. The average difference of those event study estimates should be similar to your coefficient on Active, correct?
Gavin Roberts
Mar 21 2021 at 11:35pm
Thanks for these suggestions. We will apply some type of matching (maybe synthetic control) in a revised version with county-level data. Is One “placebo” we tried recently is applying the DiD to outdoor spaces like parks and we don’t find an effect on social contacts there. Not perfect but we thought it was a good check.
JFA
Mar 22 2021 at 6:48am
I look forward to seeing the revised version. I’d say if you could add anything (even just in an online appendix), it’d be nice to see more visualizations of the raw data. That’s something I think is missing from a lot of social science research. Good luck.
Rik Chakraborti
Mar 22 2021 at 11:07am
The synthetic control group approach is an excellent suggestion, thank you! With regards to a visual depiction, we do take a crack at it. I’m pasting Figure 4 from the appendix here. This event-study figure compares the difference between normalized search trends from control and treatment groups through the entire range of the sample. But your proposed alternative about presenting comparisons of the raw data may be more convincing. We’ll definitely keep that in mind.
Pierre Lemieux
Mar 22 2021 at 2:26pm
Rik: Thanks for your comment. (Idem to Gavin.) For some reason that only the god of the electrons knows, we seem to be incapable of posting images in the comments. But our readers can click the link to your article in my post above.
Comments are closed.