Administrators in colonial-era Delhi, the story goes, had a problem. The growing capital city bordered agricultural land, bringing more people into occasionally deadly contact with a subcontinental rogues gallery that included cobras and banded kraits.
The administrators’ novel solution was to give the private sector an incentive to supply a public good: they offered a bounty for dead snakes. Supply did indeed rise to meet demand in response, but not quite in the way they had hoped: local entrepreneurs reacted to the bounty by farming snakes.
Civil servants realized they were running a costly but socially useless snake-farm subsidy program and ended the scheme. That must have seemed like a sensible response to a scheme that wasn’t working as intended. But the farmers, with their urban snake farms now unprofitable, released their stocks into the wild. The administrators’ well-intentioned plan may have made Delhi safer at first but much less safe in the long run.
Like most good parables, the story of the so-called cobra effect is probably apocryphal; the historian Michael Vann suggests using “the rat effect” instead in an article about a similar incentive scheme gone awry in Hanoi.
The term’s historical veracity notwithstanding, however, cobra effects are a useful way to think of problems that crop up a lot in international development: whether it’s empty “rice huts” that don’t house anyone set up by entrepreneurial Somalis to increase food aid to camps whose population is estimated by aerial surveys, or companies lying about the value of their exports to earn tax credits.
Economists characterize these as “principal-agent problems”: how does the party who pays (the principle) maximize the chance the implementer (the agent) does her job well—or at all? Our CGD colleagues have helped to develop and promote the concepts of Cash on Delivery (COD) Aid and Payment by Results (PbR) that, if applied properly, help mitigate principal-agent problems by paying implementers for results rather than prescribing or supplying inputs and hoping that we choose the right ones, in the right combination, at the right time, that are appropriate for the local context.
But didn’t our figurative colonial administrators pay for outcomes? Not quite. They rewarded people based on a measure that (it turned out) could be increased or decreased without any change in snakes on the ground, creating opportunities for gaming.
Linking payments to intermediate steps or inputs, whether snakes killed or shelters counted, risks creating incentives for partners to goose metrics rather than deliver the output we really care about. The cobra effect, then, was gaming with pernicious consequences, causing harm beyond merely absorbing budgets and other scarce resources.
If we designed a similar scheme today, we might pay local administrators based on reductions in the number of snake bite deaths, providing them with an incentive to award bounties for captured animals rather than farmed ones and to experiment with more effective and cheaper approaches such as education campaigns, encouraging people to report snakes for capture by professionals,or rolling out innovative, cost-effective medical protocols to reduce mortality.
Generally, the further we go along the chain of causality, the closer we come to the outcome on which we should contract. Cobra effects are more likely to arise when a funder rewards an intermediate step (snakes captured) rather than the underlying objective (fewer snake bites).
Studying cobra effects reminds us that payment by results is no panacea: funders still need to think hard about whether the actions they pay for will actually deliver valuable outcomes. Gaming has negative connotations, but entrepreneurs in Somalia or Hanoi show us that incentives work, even in resource-poor environments.
The lesson is that the onus lies squarely on funders to minimize waste and maximize impact by contracting on the right outcomes—whether it’s food aid delivered to those in need instead of just more food aid, actual exports rather than inflated statistics, or fewer deaths rather than more snakes in the bag.
CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise.
CGD is a nonpartisan, independent organization and does not take institutional positions.