Perverse Outcomes and Perverse Incentives
“When a measure becomes a target, it ceases to be a good measure.”
(Commonly referred to as Goddhart’s Law, this actually comes from a paper published in 1997 by anthropologist Marilyn Strathern, who generalized the law beyond statistics and control to the topic of evaluation more broadly.)
What’s a perverse incentive?
The classic example a perverse incentive is when someone does something that undoes the point of the incentive. The well-known example is the Cobra Effect.
In the days of the Indian Raj during the late 19th century, the British government, concerned about the number of venomous cobras roaming freely in Delhi, decided to offer a bounty for every dead venomous. Bring in a dead cobra, get a reward for your efforts. This was a huge success as large numbers of snakes were killed and turned in for their cash reward. Naturally, and perversely, enterprising Delhiites began breeding cobras for the income. When the government finally figured this out, they scrapped the reward program figuring that this just didn’t work out well. The part-time cobra breeders who now had a lot of dangerous mouths to feed, chose to set the now-worthless snakes free. As a result, the cobra population further increased in the city. The apparent solution for the problem made the situation even worse, becoming known as the Cobra effect, which describes an incentive to motivate people to a particular end but causes them to do exactly the opposite. Nobody wanted vast numbers of cobras in the city, but that’s what they got. [Siebert] {Footnote: With subsequent research into Delhi, cobras, and incentives, it’s not entirely clear that this tale is real—it might well be apocryphal. But the story is so well known that the moniker—Cobra effect—has stuck around for at least 100 years. So take this story with a grain of salt, but keep the name in mind.}
But incentives arise in many places and in many disguises. Consider potholes in the street outside your house. The pothole is a problem. It’s going to cause real issues for you and any driver on the road that drives over it.
You have an obvious incentive to get the pothole filled in and repaired. That makes perfect sense. However, if you look at the problem a bit differently, you might see an opportunity where others see only problems. You could set up a company to fix potholes and get paid to fix the problem. That’s a perverse interpretation of the pothole problem; you’ve just created a perverse incentive.
But for any perverse interpretation, there are often even more contrary ways to think about things. What could make even more money would be to lobby your local government so that they don’t repair the pothole, and then open a shop near the pothole that will repair cars suspensions after the car hits the crater in the road, breaking suspensions or cracking their car’s alignment out of whack. Alternatively, you could go into towing services. In either case, you have a repeated source of income.
And of course, anyone who’s trying to fix the pothole becomes a threat to your bottom line and becomes a threat to your livelihood that rests on a perverse basis.
There are perverse incentives leading to perverse outcomes all around us.
Examples of perverse incentives abound! For instance, in education, standardized testing regimes can create perverse incentives for teachers and schools to "teach to the test." That strategy yields great scores for the school but can easily lead to an outcome of a narrow focus on test-taking strategies rather than a deep and valuable education.
Similarly, a company might pollute less if they are fined for exceeding specified pollution limits. But if the fines are lower than the cost of complying with regulations, they might simply choose to pay the fines instead of investing in cleaner technologies and processes, preferring to see the fines and penalties merely as the “cost of doing business.” This is a perverse outcome as a consequence of trying to get companies to “do the right thing.”
Setting up regulatory fines and penalties is tricky—people and companies have strong incentives to find ways to work around any such costs. Often people and companies are willing to spend many more hours to find workarounds than the regulators have invested in designing the incentivizing regulations.
Here we’re going to look at several examples of perverse incentives and perverse outcomes, drawing Rules of Thumb for each. Let’s start with perhaps the best-known example of a perverse outcome and what we can learn.
The Perverse Incentives Of Trying to Suppress Information
In 2003, the famous singer and actor Barbra Streisand sued Kenneth Adelman and Pictopia.com for posting a photograph of her home online. [SCLA] Before the lawsuit had been filed, only 6 people had downloaded the file, two of them Streisand's attorneys. The annoying photo was one of thousands taken along the California coastline as part of a project to document the long-term changes to the coast due to climate change. It was just one among the many taken by the passing plane; relatively few people knew of that photo’s connection to Streisand.
As you might expect, the lawsuit drew attention to the photograph, leading to more than 420,000 people visiting the site. [Rogers] This came to be known as the Streisand Effect, which happens when an attempt to hide, remove, or suppress specific pieces of information leads instead to a much greater amount of attention to the material. [Masnick] The perverse consequence is that the material you want private turns into something that’s widely known.
And now, naturally, the image of Streisand’s house is essentially everywhere, including here.
The Streisand Estate, Malibu. N34 00.65. W118 47.24. Image 3850. Mon Sep 23 13:30:47 2002. Copyright (C) 2002 Kenneth & Gabrielle Adelman, California Coastal Records Project, www.californiacoastline.org (See: Wikimedia page for copyright details.)
People just don’t learn. A decade later, the Right to be Forgotten law became a European Union (EU) regulation that allows people to request that search engines remove certain results from searches on their name. This EU right originated from a 2014 ruling by the European Court of Justice (ECJ) in the case of Google Spain SL, Google Inc v. Agencia Española de Protección de Datos, for their client NAME-REDACTED (but you can find it easily). While the ruling did, in fact, agree that individuals have a right to privacy, and that Google had to remove the offending results, the person filing the original complaint managed to achieve completely the opposite effect and becoming extremely well-known in the process, another victim of the Streisand Effect. [BBC]
Rule of Thumb: Be very, very careful when trying to suppress information. There are many (many more than here) examples of individuals and governments trying to make a photo or news article go away. They often find that the process of attempting to suppress actually highlights the information.
Perverse Incentives When Trying to Eliminate Varmints
Rats: After reading about the Cobra Effect, it might come as a surprise to learn that French colonial officials in 1902 Hanoi, Vietnam, did something very similar by putting a bounty on rat tails. One rat tail, one payout. This led to the Great Hanoi Rat Massacre (Vietnamese: Đại thảm sát chuột ở Hà Nội; chữ Nôm; French: Massacre des rats de Hanoï). After beginning a general program to kill off rats, the Government-General of French Indochina hired professional Vietnamese rat-catchers to go down into the sewers to hunt the rats, and be paid for each rat that they had eliminated. While they reported killing up to 15,041 rats in one day, it quickly became obvious that this wasn’t nearly enough. The pro rat-catchers were making insufficient progress, and so a rat-tail bounty program began that paid a reward of 1¢ for each rat killed. [Venn] {Footnote: This is quite possibly the best documented perverse incentive story ever! Michael Vann wrote his PhD thesis on this story, including data he discovered in Hanoi in a folder labelled "Destruction of Hazardous Animals: Rats" concerning rat control in the city. The archived file was from the French government of Indochina detailing the number of rats that were killed on each day and the amount of money that the French had awarded to the rat hunters. They documented numbers and prices paid.}
However, it didn’t take long before the local authorities started noticing rats in Hanoi with no tails. The Vietnamese rat catchers would capture rats, sever their tails, then release them back into the sewers so that they could produce more rats and thereby assure future income.
What’s more, there were also reports that some locals were deliberately smuggling in rats from outside Hanoi into the city. The Cobra effect reached its peak when French health inspectors discovered rat farming operations on the outskirts of the city, operations that were breeding rats solely for their tails. [Trương]
Pigs: In 2007 the Army had a problem with pigs. There were just too many feral pigs tearing up the land and causing endless problems for the Army at Fort Benning, Georgia (now named Fort Moore). The Army offered hunters a $40 bounty for every pig tail they turned in. [NBC] Over the course of the 2007–2008 program, they killed hundreds of feral pigs—but the feral pig population in the area increased dramatically, quite the perverse and opposite of the intended effect.
While there were some reports that a few people bought pigs' tails from nearby meat processors then resold the tails to the Army at the higher bounty price, a careful study of the situation showed that both the pigs' fertility rate and offspring survival rates increased under the hunting scheme. Completely counterintuitive, this unexpected outcome was due primarily to the improved pig nutrition because the hunters used high quality feed as bait to attract animals to hunting sites. What’s more, hunters were more likely to target large males to shoot as trophies, while ignoring females and small juveniles as targets. It turns out that removing mature males from the population has a negligible impact on population growth, as remaining mature males can each consort happily with many breeding sows, producing a bumper crop of new pigs on an accelerated schedule. [Ditchkoff]
It's pretty clear that creating eradication efforts based on bounties of animal tails is a pretty bad idea, despite many attempts. But the efforts to use bounties-on-tails as an eradication mechanism is a long and proud, if hapless, tradition.
Squirrels: In the west, ground squirrels were widely considered deplorable pests—evil consumers of grain and destroyer of crops. In multiple attempts, bounties were offered for squirrel tails and massive campaigns mounted to destroy as many squirrels as possible.
In 1887, Bozeman, Montana offered a bounty, managing to kill around 275,000 of them. [Bozeman] While that helped crop yields later that year, squirrels showed a robust compensatory reproduction effect. [Wiki] When a fraction of the total population of a very fecund animal is destroyed, those left behind compensate with a variety of responses—the remaining animals have less competition for food, there’s more food available overall, females have faster reproduction cycles, and they have larger broods each time.
A bit after Bozeman’s attempt at squirrel population control, California launched many attempts to do the same with the same mechanism: bounties for tails, with more or less the same result. In 1913, San Luis Obispo paid 5 cents/tail [SLO], and a few years later a state-wide effort was launched to control the ground squirrel population. The last week of April in 1918 was proclaimed “Squirrel Week” by the governor of California, William Stephens, who directed everyone, including school children, “to do their utmost to relieve the country of all ground squirrel pests.” Hundreds of thousands of squirrels were killed that week, to virtually no effect—it was a mere temporary dip in the overall numbers. [Barry] If anything, a year later, there were more squirrels on the ground than before. A truly perverse outcome.
Newspaper ad in California advertising “Kill the Squirrels” week, including the injunction for small children to handle poison responsibly. While hundreds of thousands of squirrels were eliminated, the record does not record the number of children who were injured in the process. Note that the squirrels in the lower left and lower right corners are wearing the despised Pickelhaube helmet of the German army, with which the US was at war at this time.
In some sense, perverse incentives are any incentives for which people can find another way to achieve the outcome. Given how very clever people are—and especially if the incentive can be gained at a lower cost—it’s not a surprise that crafting incentives that do not turn perverse is really tough.
And unintentional outcomes abound as well. You wouldn’t think that someone seeking privacy would suddenly cause world attention to focus on them—but such is the price of attention in the internet age. Things go viral, especially when celebrities or surprisingly controversial regulations (e.g., right-to-be-forgotten), are involved.
In future posts we’ll be discussing how to avoid such disasters. You won’t be surprised to learn that it’s mostly a matter of paying attention to the right things. The trick is knowing what the right things are.
Stay tuned.
======
[Barry] University of California, Agriculture and Natural Resources. “A Brief History of Ground Squirrel Control Efforts in California” https://ucanr.edu/sites/BayAreaRangeland/files/288938.pdf
[BBC] “EU court backs 'right to be forgotten' in Google case.”BBC News (13 May 2014) https://www.bbc.com/news/world-europe-27388289
[Ditchkoff] Ditchkoff, Stephen S.; Holtfreter, Robert W.; Williams, Brian L. "Effectiveness of a bounty program for reducing wild pig densities". Wildlife Society Bulletin. 41 (3): 548–555. (September 2017) doi:10.1002/wsb.787
[NBC] "Fort Benning puts a bounty on boars". NBC News. Associated Press. 1 March 2008.
[Masnick] Masnick, Mike. “Since when is it illegal to just mention a trademark online?” TechDirt (Jan 5, 2005) https://www.techdirt.com/2005/01/05/since-when-is-it-illegal-to-just-mention-a-trademark-online/
[Rogers] Rogers, Paul "Photo of Streisand home becomes an Internet hit". San Jose Mercury News (June 24, 2003). https://www.californiacoastline.org/news/sjmerc5.html
[SCLA] Barbara Streisand v. Kenneth Adelman Et. Al., Cal.Super. (Superior Court of California, County of Los Angeles May 20, 2003)No. SC077257 https://www.plainsite.org/dockets/2zd3hv14c/superior-court-of-california-county-of-los-angeles/barbara-streisand-vs-kenneth-adelman-et-al/
[Siebert] Siebert, Horst (2001). Der Kobra-Effekt. Wie man Irrwege der Wirtschaftspolitik vermeidet (“How to avoid wrong paths in economic policy” in German; Munich: Deutsche Verlags-Anstalt. ISBN 3421055629.
[SLO] “An Ordinance” San Luis Obispo Tribune (Weekly), Volume XLIV, Number 85, 10 June 1913.
[Strathern] Strathern, Marilyn (1997). "'Improving ratings': audit in the British University system.” European Review. John Wiley & Sons. 5 (3): 305–321.
[Trương] Linh Trương "Bài học từ cuộc 'thảm sát' chuột ở Hà Nội đầu thế kỷ 20” (“Lessons from the 'massacre' of rats in Hanoi in the early 20th century.”) (in Vietnamese) VNExpress (17 June 2017) https://vnexpress.net/bai-hoc-tu-cuoc-tham-sat-chuot-o-ha-noi-dau-the-ky-20-3600268.html
[Vann] Vann, Michael G. (2003). "Of Rats, Rice, and Race: The Great Hanoi Rat Massacre, an Episode in French Colonial History". French Colonial History. 4: 191–203. https://muse.jhu.edu/pub/26/article/42110
[Wiki] https://en.wikipedia.org/wiki/Reproductive_compensation