Mini Post #6: Touch of Evil
Do people buy fewer sandwiches if the sandwich salesman is a pedophile?
I’ve decided to try something new: once a week, to write a shorter, narrower post focusing on one specific paper. Last week, I wrote about how movies praising the Ku Klux Klan made the Klan more popular. This is the tab with all the previous (and current) posts.
Onto the actual post: do people buy fewer sandwiches if the guy selling them is a pedophile?
Most people are generally morally repulsed by pedophiles, rightfully so. How far does that dislike for child molesters go? For example, if the celebrity spokesman of a company was outed as a pedophile, would consumers stop shopping there?
This is an interesting question, one that the paper “The Role of Repugnance in Markets: How the Jared Fogle Scandal Affected Patronage of Subway” (2023) by John Cawley, Julia Eddelbuettel, Scott Cunningham, Matthew Eisenberg, Alan Mathios & Rosemary Avery aims to answer.
Firstly, some background knowledge: who is Jared Fogle? Fogle was a man who, sometime in the 2000s, rose to prominence by claiming he had lost 90kg (or 200 pounds for the Americans) by sticking to a Subway-heavy diet. This story, which proved compelling, resulted in Fogle becoming a high-profile Subway spokesman, starring in several (extremely poorly aged) commercials. In 2015, his house was searched by authorities due to suspicions of posession of child pornography, and months later he struck a deal with prosecutors to plea guilty to several offenses, including at least two charges of sex with minors, in exchange for a reduced sentence of 12 years. Some of Fogle’s crimes, such as the child pornography, were obtained with the participation of people aligned with his anti-obesity foundation.
Why would Jared Fogle’s scandal be damaging to Subway? Because he was the company’s spokesman, “pitchman”, and main public-facing figure. The economic theory correlate would be the concept of repugnance: the idea that certain economic transactions be disallowed, or frowned upon, not by any direct economic concerns (including material harms such as externalities) but rather, due to moral disapproval of the supplier or the products. The banning of horse meat, for instance, is one clear example of repugnance: while eating meat from cows, pigs, goats, and rabbits is allowed, the breeding and slaughtering of horses under similar conditions is not, not because of health concenrs, but because people do not approve of it.
If repugnance was binding in the Jared Fogle case, then Subway would have lost market share or revenue from this; if, on the contrary, it hadn’t, then the company would have been separated from the heinous acts of its spokesperson. It should be noted that Fogle was deeply enmeshed in the Subway corporation’s public image over the entirety of the 2000s up until his plea bargain, and that he was a well-known public figure whose arrest and conviction were front page national news.
The first complication to test whether Subway was affected is that the corporation is privately owned, and thus does not trade in the stock market or distribute earnings reports. To examine the impact of “Jaredgate” on Subway sales, the authors use the Simmons National Consumer Survey, which tracks patronage at specific brands over time. This survey is aimed at marketing agencies, and thus intends to have top-notch data on demographic coverage, as well as accurately creating a sample based on spending patterns and not total-population representation, i.e., purposefully overrepresents higher income individuals. The data provided by Simmons NCS covers the 18 months before and after the initial raid on Fogle’s house (July 2015).
But who could be compared to Subway? There’s no other high-profile sandwich purveyor of a similar scope that was not affected by its main spokesperson being a child molester. Rather, what the authors do is look at other fast food chains, and construct a synthetic control for Subway. A synthetic control is, more or less, a fake control group to do statistical analysis with, if there is no single decent control group but a large number of half decent ones. The methodology is quite straightforward: one constructs an average from before the event in question to create as similar as possible a trend to the variable of interest’s, such that one can do an event study where the performance of the selected synthetic controls (as in. 25% McDonald’s, 25% Burger King, 30% Wendy’s, etc) is used as a reasonable counterfactual for after the even. That is, if the “Synthetic Subway” constructed from the other fast food chains had a better performance, then it would be reasonable to assume that the difference is caused by the Jared Fogle scandal; contrarily, if there isn’t, then the working hypothesis that Subway was punished should not be considered.
A battery of statistical tests, particularly ones concerning estimates for other periods and whether the consumer demographics match Subway’s, found that Synthetic Subway was a workable comparison group to real Subway. The two are compared in an event study methodology, i.e. compared differences between the two Subwaysin each of their sales over time after the Jared pedophilia scandal, and dropping certain parts of the sample (three quarters of the sample before, half before, all the odd months, and all the even months) to test for robustness of the results. They also estimate the impact by gender, since women are disproportionately likely to be victims of sexual abuse and thus might have a stronger threshhold for repugnance than men. Lastly, a similar distinction is made for those with and without minor children, for the same reason: parents may be more upset than non-parents.
The results are simple: across all specifications and for all groups of consumers, there is no evidence that Subway Jared being a pedophile affected consumer demand for Subway sandwiches, not even for women or for parents of minor children. Subway corporation also did not increase its ad spending or engage in any relevant campaign to try to assuage consumers. Thus, it can be said that the repugnance theory may have its limits, or that it may be limited directly to concerns over the trasnaction itself. This is surprising becase other studies, such as of oil companies during the BP oil spill, or car manufacturers during the VW emissions scandal, did find negative effects. Similarly, sexual abuse allegations hurt both the Bostonian Catholic church and the Michael Jackson estate. And restaurants are indeed flexible to disclosure of information, such as health ratings by government agencies.
To finish the post, some links
The paper in question
Nobel Laureate Alvin Roth’s 2007 paper outlining repugnance as a concept
A paper exploring synthetic controls in depth
A similar paper exploring the impact of a sex abuse scandal on Catholic affiliation
A previous blog post about repugnance and the movie NOPE
> This is surprising becase other studies, such as of oil companies during the BP oil spill, or car manufacturers during the VW emissions scandal, did find negative effects. Similarly, sexual abuse allegations hurt both the Bostonian Catholic church and the Michael Jackson estate.
This doesn’t seem too surprising? In the first two cases, the scandal is about the main product of the companies, so consumers would feel that the actual goods are tainted. With the Catholic church and the Jackson estate, the scandals are central to the entity themselves, and neither entity has fully recanted. In all of those cases, the core issue is still there. But with Subway, Jared was ultimately and obviously an advertising gimmick. A well-used advertising gimmick, but one that Subway quickly cut ties with. No one going to Subway afterward thought that their sub was funding pedophilia or made by pedophiles.
How a company reacts really makes an impact. I despise Jared, but Subway's quick cutting of ties with him made me not see them any differently. They didn't know. Compare that with how I reacted to Spotify after they enriched Joe Rogan, or Twitter with its new owner. Even Facebook after the 2016 election, Rohyinga genocide & Zuckerberg's infamous Holocaust deniers, "there are things different people get wrong". I cancelled my FB account.
As gmt says, how a company or organization reacts to repugnant acts is what really determines if they are punished in the marketplace.