What is the cobra effect, and why do good policies backfire?

The cobra effect is a label for a perverse incentive: a reward that produces the opposite of its intended result. In the usual telling, a colonial government in Delhi paid a bounty for dead cobras to reduce their number. People responded by breeding cobras to collect the money, and the policy made the problem worse. The term is now used for any scheme in which the reward rewards something other than the real goal. [1] [2]

According to the standard account, the bounty at first worked, and cobra numbers fell. As wild cobras became harder to find, entrepreneurial residents began raising them at home, because killing a captive snake was easier than hunting a wild one. Officials noticed the gap between the few cobras seen in the city and the many dead ones being redeemed, found the breeding sites and cancelled the bounty. The breeders then released their now worthless snakes, leaving more cobras than before. [3] [2]

The historical record is doubtful. One commentary calls the tale probably apocryphal and notes that the historian Michael Vann suggests calling it the rat effect instead, after a comparable bounty scheme in Hanoi that is better documented. [4]

A 2025 investigation by the Friends of Snakes Society, as summarized on Wikipedia, went further. It reported finding no contemporary records of cobra breeding or prosecutions in British India, and traced the story to an 1873 newspaper article that said only that breeding was alleged. It also cited an 1887 Bombay Natural History Society inquiry that called confined breeding highly improbable. The investigation said a Madras bounty was cut back in 1873 because of cost, not fraud, and recommended retiring the term. Other sources in the results simply repeat the story as fact, so the results show a split between retellings and this skeptical research. The skeptical findings are the only archival evidence offered in these results. [5] [3]

The common structure is that an incentive targets something measurable, people find the easiest way to produce it, and that route undermines the goal. Fieldmarks stresses that the cause is not dishonesty but rational people responding to incentives as designed rather than as intended. [1]

The Center for Global Development describes this as a principal-agent problem: the party paying cannot fully control how the implementer delivers. Paying for an intermediate step, such as snakes killed, invites gaming, while paying closer to the real objective, such as fewer snakebite deaths, leaves room for cheaper and better methods. It adds that paying for results is no cure-all, because funders must still judge whether the paid actions deliver value. Its other examples include empty food-aid huts in Somalia and inflated export figures used to earn tax credits. [4]

Rules other than bounties can behave the same way. Mexico City's driving ban by license plate led some residents to buy second cars with different plate numbers, and others to use taxis, which on average polluted more than the typical car. [2]