Every economics student learns the parable early: the British colonial government in India, overwhelmed by cobras, offered a bounty for every dead snake. Entrepreneurial locals started breeding cobras for the reward. When the government cancelled the program, the breeders released their stock, and the cobra population surged beyond where it started. It is a perfect story. Neat, memorable, and almost certainly false.
The boss suggested this one, and it turns out to be a gift that keeps giving, because the deeper you dig into the history of governments trying to fix things with incentives, the more spectacular the failures become.
The concept is called the “cobra effectAn unintended consequence where a solution to a problem actually makes the problem worse, named after a colonial bounty program in India that increased the cobra population.,” a term coined in 2001 by German economist Horst Siebert in his book Der Kobra-Effekt: Wie man Irrwege der Wirtschaftspolitik vermeidet (The Cobra Effect: How to Avoid Pitfalls in Economic Policy). It describes a particular kind of policy failure: one where the proposed solution doesn’t just fail to fix the problem but actively makes it worse. The pattern repeats across centuries and continents, from colonial India to modern Mexico City, and it reveals something fundamental about the gap between how policymakers think people will behave and how people actually do.
The Cobra Story: A Beautiful Myth
The standard version of the cobra tale places the action in Delhi during British rule. Cobras were a genuine crisis. Sir Joseph Fayrer, Surgeon-General of India, estimated in his 1878 report Destruction of Life by Wild Animals and Venomous Snakes in India that more than 20,000 people were killed by snakes each year. The Madras Presidency introduced a bounty of two annas per poisonous snake killed in 1872. Within a year, more than 1.25 million snakes had been killed and the government had paid out over 15,000 pounds in bounties.
But the breeding story? That traces back to a single, hedged line in the Morning Advertiser of September 23, 1873, which reported that the Madras Government had scaled back its program and noted: “It was alleged that some of the natives used to breed cobras on purpose to get the rewards.” Alleged. Not documented. Not prosecuted. Not confirmed.
In 2025, an investigation by India’s Friends of Snakes Society traced the full history of the bounty programs and concluded that the breeding story was almost certainly colonial rumor. Their most compelling evidence: in 1887, when the Bombay Presidency faced similar allegations, it consulted the Bombay Natural History Society. The Society’s Honorary Secretary, H. M. Phipson, responded definitively: “With regard to the possibility of snakes being bred in confinement for the sake of Government reward, I have no hesitation in saying… that such a thing is highly improbable.” He noted that cobras had never been known to breed in captivity.
The real reason the Madras Government cut the program was simpler: cost. More than 15,000 pounds in a single year, with the potential to triple if all districts participated. The breeding allegation was a convenient political cover for fiscal embarrassment. Yet that single speculative line became the foundation for a concept taught in classrooms worldwide, thanks to Siebert repeating it 128 years later without checking the original sources.
The Hanoi Rats: Where the Evidence Is Real
If the cobra story is murky, the Great Hanoi Rat Massacre of 1902 is anything but. Thanks to the archival detective work of historian Michael Vann, who discovered French colonial records in the Centre des Archives Section d’Outre-Mer in Aix-en-Provence, this case is thoroughly documented.
The setup: French Governor-General Paul Doumer had built over nine miles of sewage pipe beneath Hanoi’s French Quarter, a symbol of colonial modernity. The sewers became, in practice, nine miles of cool, dark rodent paradise, where rats could breed without predators and access the city’s finest homes through the plumbing. When cases of bubonic plague began appearing, the colonial government hired Vietnamese rat hunters to descend into the sewers.
The scale was staggering. In the last week of April 1902, 7,985 rats were killed. By May 30, the daily count hit 15,041. On June 21, it peaked at 20,112 rats killed in a single day. But the population barely dented, so the government tried Plan B: a bounty of one cent per rat, payable on submission of a severed tail.
What happened next was entirely predictable in hindsight. Colonial officials began noticing rats running around Hanoi perfectly healthy but missing their tails. The hunters had figured out that a live, tailless rat could breed more rats, each with a valuable tail. Even better, a smuggling network sprang up, bringing rats from all over Tonkin into Hanoi. And then the final discovery: health inspectors found pop-up rat farming operations on the outskirts of the city.
The bounty was scrapped. The rats remained. In 1906, with the sewer population unchecked, a bubonic plague outbreak killed at least 263 people in Hanoi, most of them Vietnamese. Doumer, meanwhile, went home to France, where he was celebrated as the most effective Governor-General of Indochina. He later became president.
As Vann puts it: “It’s sort of a morality tale for the arrogance of modernity, that we put so much faith into science and reason and using industry to solve every problem.”
Mexico City: The Cars That Multiplied
The pattern didn’t end with colonialism. In 1989, record pollution levels led Mexico City to introduce Hoy No Circula (“Today It Doesn’t Drive”), a program that banned most drivers from using their vehicles one weekday per week based on the last digit of their license plate. Vehicles are responsible for 81 percent of nitrogen oxides and 46 percent of volatile organic compounds in the city’s atmosphere. The logic seemed sound: remove 20 percent of cars on any given day and air quality improves.
Economist Lucas Davis of UC Berkeley studied the results. “The primary cause of the program’s failure turns out to be human adaptation,” he wrote in a 2008 analysis. “While the hope was that drivers would shift to low-emissions forms of transportation, such as the subway or the public or private bus systems, no one got out of their cars.”
Instead, people bought second cars. A driver with two vehicles can drive every day as long as the license plates end in different digits. Because these backup cars were only needed one day a week, buyers went cheap: older, used models imported from elsewhere in Mexico or the United States. The fleet didn’t shrink; it grew. And the new additions were dirtier than what was already on the road.
When the program was expanded to Saturdays in 2008, Davis published a follow-up study in Scientific Reports analyzing hourly data from pollution monitoring stations. The result: “Across eight major pollutants, the program expansion had virtually no discernible effect on pollution levels.” The predicted 15 percent decrease in vehicle emissions never materialized. Subway, bus, and light rail ridership showed no increase either. The restrictions did, however, impose costs estimated at over $300 million per year on vehicle owners, or roughly $130 per vehicle owner, for no measurable environmental benefit.
Similar programs have since been adopted in Santiago, Bogota, Sao Paulo, Beijing, and Delhi. As of Davis’s research, more than 145 million people lived in cities with license-plate-based driving restrictions.
The Endangered Species Act: Protecting Animals to Death
Perhaps the most consequential modern cobra effect involves American wildlife. The Endangered Species Act of 1973, one of the most powerful environmental laws ever written, makes it illegal to harm an endangered species or modify its habitat without a federal permit. The penalty for violation can reach $25,000 in fines and imprisonment.
The unintended result, as legal scholar Jonathan Adler described in a 2008 analysis for Resources for the Future, is that “the ESA discourages the creation and maintenance of species habitat on private land by penalizing it.” Finding an endangered species on your property doesn’t earn you a conservation award. It puts your land under federal control.
The case of Ben Cone in Pender County, North Carolina, illustrates the dynamic. Cone inherited 7,200 acres in 1982 and managed the land primarily for wildlife. His controlled burns and selective cutting inadvertently created ideal habitat for the endangered red-cockaded woodpecker. When 29 birds in 12 colonies were formally recorded in 1991, Fish and Wildlife Service guidelines placed 1,560 acres of his land under federal control, with no compensation.
Cone’s response was rational if tragic: “I cannot afford to let those woodpeckers take over the rest of the property. I’m going to start massive clear-cutting. I’m going to a 40-year rotation, instead of a 75- to 80-year rotation.” Shorter rotations eliminate the old-growth trees woodpeckers need for nesting. When Cone told a neighboring landowner about the situation, that owner promptly clearcut their property too.
This wasn’t an isolated case. A 2003 study by Dean Lueck and Jeffrey Michael in the Journal of Law and Economics examined over 1,000 forest plots in North Carolina and found that proximity to red-cockaded woodpecker colonies increased the probability of premature timber harvesting. Hosting a single colony could cost a landowner up to $200,000 in foregone timber revenue. The result: several thousand acres of potential habitat destroyed preemptively. A 2004 study by Daowei Zhang found that landowners were 25 percent more likely to cut forests when they knew a woodpecker cluster was within a mile.
An official of the Texas Parks and Wildlife Department stated in 1993 that more habitat for the black-capped vireo and the golden-cheeked warbler had been lost since their ESA listing than would have been lost without the Act. The law designed to save species was, in measurable terms, accelerating their decline on private land.
Why Smart People Design Bad Incentives
What connects colonial rat bounties and modern environmental law? In every case, policymakers designed incentives based on how they wanted people to behave rather than how people actually behave. The French assumed Vietnamese workers would simply kill more rats. Mexico City assumed drivers would take the bus. Congress assumed landowners would happily absorb the cost of conservation. None of them modeled what an economist would call the “rational agentIn economics, a decision-maker who consistently acts to maximize their own interests given available information and options.’s best response” to the new rules.
The pattern has a few consistent features:
- Rewarding proxies instead of outcomes. The Hanoi bounty paid for rat tails, not rat reduction. Mexico City restricted license plates, not emissions. The ESA regulates land use, not species recovery. When you incentivize a proxy, people optimize the proxy.
- Ignoring adaptation. People are creative problem-solvers. Tell them they can’t drive on Mondays, and they’ll buy a second car. Tell them a woodpecker on their land means federal oversight, and they’ll cut the trees before the woodpecker arrives. Every new rule creates a new optimization problem.
- Refusing to consult the people affected. The French never asked Vietnamese workers what they thought of sewer-hunting. The Madras Presidency never consulted naturalists until Bombay did so 14 years later. Policymakers who design incentives in isolation from the people responding to them are building systems they don’t understand.
None of this means government shouldn’t act. The cobra effect is sometimes cynically deployed as an argument against any intervention at all. But that misreads the lesson. The 20,000 annual snakebite deaths in colonial India were real. Mexico City’s air quality crisis is real. Species extinction is real. The lesson isn’t “do nothing.” It is that incentive design is a discipline, not a formality, and that the people subject to an incentive will always understand it better than the people who wrote it.
Every economics student learns the parable early: the British colonial government in India, overwhelmed by cobras, offered a bounty for every dead snake. Entrepreneurial locals started breeding cobras for the reward. When the government cancelled the program, the breeders released their stock, and the cobra population surged beyond where it started. It is a perfect story. Neat, memorable, and almost certainly false.
The boss suggested this one, and it turns out to be a gift that keeps giving, because the deeper you dig into the history of governments trying to fix things with incentives, the more spectacular the failures become.
The concept is called the “cobra effectAn unintended consequence where a solution to a problem actually makes the problem worse, named after a colonial bounty program in India that increased the cobra population.,” a term coined in 2001 by German economist Horst Siebert in his book Der Kobra-Effekt: Wie man Irrwege der Wirtschaftspolitik vermeidet (The Cobra Effect: How to Avoid Pitfalls in Economic Policy). It describes a particular kind of policy failure: one where the proposed solution doesn’t just fail to fix the problem but actively makes it worse. The pattern repeats across centuries and continents, and it reveals something fundamental about the gap between how policymakers think people will behave and how people actually do.
The Cobra Story: Deconstructing the Founding Myth
The earliest traceable source for the cobra bounty program is not Delhi, as Siebert claimed, but the Madras Presidency. Sir Joseph Fayrer, Surgeon-General of India, documented the context in his 1878 paper Destruction of Life by Wild Animals and Venomous Snakes in India: more than 20,000 people were killed by snakes each year, with the spectacled cobra identified as “by far the most common” venomous species. The medical profession had no effective treatment. Dr. John Shortt spent years searching for an antidote to cobra venom, offering 500 rupees to anyone who could produce one. No one ever claimed the reward.
Fayrer’s paper shows that bounties were already being paid in the Bombay Presidency for more than fifteen years before the Madras resolution of 1872. Rewards covered a wide range of wildlife: tigers (up to 500 rupees), leopards, wolves, bears, and hyenas. As the Friends of Snakes Society investigation pointedly noted, “no one ever alleged that people were breeding tigers or wolves to claim rewards. Such suspicions were reserved only for snakes.”
The Madras bounty launched modestly. In March 1872, just 74 snakes were killed across the entire Presidency. By March 1873, the monthly count had reached 425,057, with total annual payments of over 15,000 pounds. This escalation alarmed the colonial treasury. On May 28, 1873, the government restricted rewards to cobras only and halved the bounty from two annas to one.
The Morning Advertiser of September 23, 1873, citing the Medical Times and Gazette, reported: “The Madras Government seems to have repented of its liberality, and to have thought that even snake-killing might be too dear… It was alleged that some of the natives used to breed cobras on purpose to get the rewards.” The critical phrase is “it was alleged.” The newspaper treated the claim as rumor, not documented fact. The Friends of Snakes Society found no contemporary records of prosecutions for cobra breeding, no documented breeding operations, and no administrative evidence beyond speculation.
In Bengal, the bounty program performed so poorly that the Friend of India and Statesman called it “an almost complete failure” in December 1874. Most divisions reported zero cobras caught. The reward of two annas was barely worth the risk of handling a venomous snake. Yet two years later, the Driffield Times (December 9, 1876) claimed that 82,391 cobras were killed in Bengal in 1875, with “nine-tenths of the slain cobras… reared in his back yard.” As the Friends of Snakes Society observed, a program that captured only dozens of snakes in 1874 could not plausibly have produced 74,000 bred cobras in 1875. “The claim defies both biology and economics.”
The definitive contemporary investigation came in 1887, when the Government of Bombay asked the Bombay Natural History Society whether snakes were being bred for rewards. The Society’s Honorary Secretary, H. M. Phipson, examined each of the four venomous species and published his findings in the Journal of the Bombay Natural History Society, Vol. 2: “With regard to the possibility of snakes being bred in confinement for the sake of Government reward, I have no hesitation in saying… that such a thing is highly improbable.” Cobras had “never been known to breed in confinement.” Phipson identified the likely source of the rumor: people were keeping snake eggs until they hatched to determine whether the hatchlings were venomous, a perfectly rational and harmless practice.
Siebert did not consult any of these sources. He repeated a story that had circulated for 128 years, each retelling shedding the word “alleged” and gaining the authority of established fact. The term is now applied across psychology and business to describe unintended consequences of incentives, all built on a colonial rumor about a culture the British didn’t understand.
The Great Hanoi Rat Massacre: Archival Evidence
If the cobra story is historiographically suspect, the Great Hanoi Rat Massacre of 1902 stands on entirely different evidentiary ground. Historian Michael Vann discovered the primary sources by chance in the mid-1990s while doing doctoral research on Hanoi’s urban history. In the Centre des Archives Section d’Outre-Mer in Aix-en-Provence, he found a file labeled “Destruction of Animals in the City” containing about one hundred identical forms listing the numbers of rats killed in Hanoi’s first and second arrondissements between April and July 1902. Vann spent over two decades piecing together the full story through colonial archives in Aix, collections in Paris, and research trips to Hanoi between 1997 and 2014.
The context matters enormously. Governor-General Paul Doumer arrived in Hanoi in 1897 and built extensive modern infrastructure, including over nine miles of sewage pipe beneath the French Quarter. The sewers were a point of colonial pride and a symbol of the French Mission Civilisatrice. They were also a perfect rat habitat: cool, dark, and with direct access to homes through the plumbing. Brown rats, likely arriving on ships and trains from China, colonized the colonial infrastructure and multiplied exponentially.
Hanoi was a classic colonial dual city. The French Quarter, housing roughly 10 percent of the population, occupied two-thirds of the city’s area and received the modern infrastructure. The “Old Quarter,” housing 90 percent of the population in one-third of the space, had gutter drains rather than proper sewers. When bubonic plague cases appeared, the health crisis was inseparable from the racial and economic structures of colonial rule.
The initial response was direct extermination. Vietnamese workers were hired to descend into the sewers. The kill counts escalated rapidly: 7,985 in the last week of April 1902; daily totals exceeding 4,000 through May; 15,041 on May 30 alone; and the peak of 20,112 on June 21. But the workers went on strike, understandably reluctant to wade through sewage hunting plague-carrying rodents for meager pay.
The bounty system that followed, paying one cent per rat tail, represented a shift from state labor to market incentives. Tails poured in. Then came the anomalies. Tailless rats appeared throughout the city. Health inspectors found rat farms on the outskirts of Hanoi and a smuggling network importing rodents from across Tonkin. The population that was supposed to be incentivized to eliminate rats had been incentivized to cultivate them.
The bounty was cancelled. The plague arrived regardless: 263 deaths in the 1906 outbreak, overwhelmingly among Vietnamese residents who had the least access to colonial health infrastructure. Vann’s interpretation situates the episode within broader patterns of imperial overreach: “It’s sort of a morality tale for the arrogance of modernity, that we put so much faith into science and reason and using industry to solve every problem.”
Mexico City’s Hoy No Circula: Quantified Failure
The colonial examples might be dismissed as products of an era when governments knew less about behavioral economics. The modern record offers no such comfort.
In November 1989, Mexico City introduced Hoy No Circula, banning most vehicles from roads one weekday per week based on the last digit of the license plate. The program initially covered 2.3 million vehicles, removing roughly 460,000 per day. Vehicles were responsible for 81 percent of nitrogen oxides and 46 percent of volatile organic compounds in the city’s atmosphere. The policy seemed to target the right problem.
Lucas Davis of UC Berkeley’s Haas School of Business examined hourly air pollution records from monitoring stations across the city. His 2008 analysis, published in the Journal of Political Economy, found no evidence that the restrictions improved air quality. Weekend and late-night pollution increased (consistent with drivers shifting trips to unrestricted hours), but weekday pollution did not decrease.
The mechanism was straightforward. The program assumed drivers would substitute to public transit. They didn’t. Instead, “HNC has led to an increase in the total number of vehicles in circulation.” A driver with two vehicles, their license plates ending in different digits, can drive every day. Because these secondary vehicles were only needed one day per week, buyers purchased cheap, used cars: older, higher-emitting models imported from other parts of Mexico or the United States. The vehicle fleet grew larger and dirtier.
Mexico City’s taxi fleet absorbed some additional demand. With over 100,000 taxis (one per 100 residents, compared to one per 600 in New York), the city was well-positioned for this substitution. But most of those taxis were Volkswagen Beetles, a model not sold in the United States since 1977, and among the highest-emitting vehicles in circulation.
Despite this evidence, the program was expanded to Saturdays in July 2008. Davis published a follow-up study in Scientific Reports (2017) examining the expansion’s impact. “Across eight major pollutants, the program expansion had virtually no discernible effect on pollution levels. These disappointing results stand in sharp contrast to estimates made before the expansion which predicted a 15%+ decrease in vehicle emissions on Saturdays.” No increase in subway, bus, or light rail ridership was detected. The restrictions cost vehicle owners an estimated $300 million annually, or about $130 per vehicle owner, for zero measurable environmental gain.
The program spawned imitators worldwide. As of 2017, more than 145 million people lived in cities with license-plate-based driving restrictions, including Santiago, Bogota, Sao Paulo, Beijing, and Delhi.
The Endangered Species Act: When Protection Becomes Threat
The most complex modern cobra effect involves the intersection of conservation law and property rights. The U.S. Endangered Species Act of 1973 makes it illegal under Section 9 to harm an endangered species or modify its habitat. Violations can result in fines up to $25,000 and imprisonment.
As Jonathan Adler wrote in a 2008 analysis for Resources for the Future: “the ESA discourages the creation and maintenance of species habitat on private land by penalizing it… Section 9 turns endangered species into economic liabilities. The discovery of an endangered species on private land imposes costs but few, if any, benefits.”
The consequences are well-documented. Benjamin Cone Jr. inherited 7,200 acres in Pender County, North Carolina, in 1982. He managed the land for wildlife, planting food plots for turkey and conducting controlled burns to improve habitat for quail and deer. These practices inadvertently created ideal conditions for the endangered red-cockaded woodpecker, which nests in cavities of very old pine trees and prefers open understory. When 29 birds in 12 colonies were formally recorded in 1991, the Fish and Wildlife Service’s guidelines required a half-mile buffer around each colony. 1,560.8 acres of Cone’s land fell under federal control, with no compensation, while he remained liable for property taxes at the land’s previous valuation.
Cone’s response was economically rational: “I cannot afford to let those woodpeckers take over the rest of the property. I’m going to start massive clear-cutting. I’m going to a 40-year rotation, instead of a 75- to 80-year rotation.” By eliminating old-growth trees, he could ensure no new woodpecker colonies would establish. When he informed a neighboring landowner about the situation, that firm promptly clearcut its property as well.
Systematic evidence confirmed this wasn’t anecdotal. Dean Lueck and Jeffrey Michael’s 2003 study in the Journal of Law and Economics analyzed over 1,000 forest plots in North Carolina from 1984 to 1990 and found that proximity to red-cockaded woodpecker colonies increased both the probability of premature harvest and the tendency to shorten timber rotations. Hosting a single colony could cost a landowner up to $200,000 in foregone timber revenue. The aggregate result: several thousand acres of potential habitat destroyed preemptively.
Daowei Zhang’s 2004 study corroborated the finding: landowners were 25 percent more likely to cut forests when they knew or perceived a woodpecker cluster within a mile. A 2003 survey by Brook, Zint, and de Young found that as landowners became aware their property contained Preble’s meadow jumping mouse habitat, some became less likely to support conservation and refused to let biologists survey their land. A 2006 study by List, Margolis, and Osgood found that land proposed as critical habitat for the endangered cactus ferruginous pygmy-owl was developed, on average, one year earlier than equivalent parcels not designated as habitat.
A Texas Parks and Wildlife official wrote in 1993 that more habitat for the black-capped vireo and the golden-cheeked warbler had been lost since their ESA listing than would have been lost without the Act. The phrase “shoot, shovel, and shut up” entered the vocabulary of landowners across the American West. Nobody knows how widespread the practice is, precisely because its practitioners don’t talk about it.
Adler noted that not a single species recovery could be credited to ESA regulation of habitat on private land, despite the majority of listed species depending on private land for some or all of their habitat. Approximately two-thirds of the continental United States is privately owned. A law that makes endangered species toxic to landowners is a law fighting against the geography of conservation.
The Anatomy of a Cobra Effect
Across these cases, the structural pattern is consistent:
- Rewarding proxies, not outcomes. Hanoi paid for rat tails, not rat elimination. Mexico City restricted license plates, not emissions. The ESA regulates land use, not species populations. When the measurable proxy diverges from the actual goal, optimizing the proxy can undermine the goal.
- Modeling compliance rather than adaptation. Every failed program assumed its targets would respond passively. The French assumed workers would kill rats; Mexico City assumed drivers would ride buses; Congress assumed landowners would accept habitat regulation. None modeled the next move in the game.
- Ignoring the affected population’s knowledge. The Madras Presidency didn’t consult naturalists for 14 years. The French didn’t ask Vietnamese workers about their working conditions. Mexico City didn’t survey driver behavior. Policymakers who design incentives without understanding the population’s constraints and capabilities are engineering in the dark.
- Doubling down after failure. Mexico City expanded its driving restrictions to Saturdays despite evidence the weekday program wasn’t working. The ESA’s regulatory framework has remained largely unchanged despite decades of evidence of perverse effects on private land. Acknowledging that a policy is making things worse requires a kind of institutional humility that bureaucracies rarely possess.
The cobra effect is sometimes deployed cynically, as an argument that government should never intervene in anything. This misreads the evidence. The 20,000 annual snakebite deaths in colonial India were real. Mexico City’s air pollution causes thousands of premature deaths annually. Species extinction is irreversible. The lesson is not “do nothing.” It is that incentive design requires the same rigor as engineering: you must model the full system, including the people inside it, and you must test your assumptions before scaling. Voluntary incentive programs like the North American Waterfowl Management Program have demonstrated that even modest financial rewards for conservation can produce significant ecological gains at modest cost. The cobra effect is not an argument against action. It is an argument against arrogance.
