The experiment is described in the paper, The Demand for Bad Policy when Voters Underappreciate Equilibrium Effects, which has been submitted for review and future publication.
“Our evidence suggests that unfamiliar policy options can be a challenge for voters when these policies contain hidden costs or benefits that will accrue once behavior changes,” says Dal Bó.
A tax cut, to use a simple example, has an immediate benefit – more money in a voter’s pocket – but may later result in a reduction in services such as street repair, an indirect consequence resulting in more potholes.
Voters have difficulty understanding that a change in policy will eventually change the behavior of other people, which in turn affects the outcome of a given policy.
Consider, for example, the choice between a tax on carbon and regulations mandating more efficient automobiles, says Dal Bó.
Tougher miles-per-gallon standards have an immediate benefit – more efficient cars – but in the longer run it encourages people to drive more, and thus add more carbon to the atmosphere. Economists call that a rebound effect, which voters tend to discount.
A tax on carbon has the opposite effect: It results in immediate costs to the voter, but in the long run has a social benefit – a lowering of carbon emissions, argues Dal Bó.
Not everyone would agree with experts who believe that passing a carbon tax is ultimately beneficial. Judgments of the value of a given policy are subjective, and experts can be wrong, he acknowledges. So, to perform their experiment, Dal Bó and co-authors Pedro Dal Bó of Brown University and Eric Eyster of the London School of Economics, needed to find a simplified setting in which there was no doubt about the value of a given policy.
The researchers recruited 768 students from UC Berkeley and Brown University. They were divided into small groups to play a classic game-theory exercise called the Prisoner’s Dilemma. It works like this.
Two prisoners, who can’t see or speak to each other, are told to either betray the other prisoner by testifying that he committed the crime, or to cooperate with the other prisoner by remaining silent. If they betray each other, both serve two years. If one betrays his partner and the other stays silent, the betrayer goes free and the other serves three years. If they both stay silent they each do one year in prison.
The safest bet for both is to betray the other. Since no one in the experiment is going to jail, the players get a payoff of cash, the amount keyed to the choices they’ve made at the end of each round.
After playing a number of times, the students were presented with “the Harmony Game,” an invention of the researchers. The game is similar to the Prisoner’s Dilemma, but taxes are imposed on both choices: to cooperate and to defect. However, the trick lies in the fact that the tax on cooperating is lower than that on defecting – as with a carbon tax that hits the more infrequent driver less hard than the frequent one.
As a result, the Harmony Game leads players to cooperate – the opposite of what happened in the Prisoner’s Dilemma.
Which has the better payoff? If players had cooperated in the Prisoner’s Dilemma, that game would have a higher payoff than the Harmony Game, no matter what strategy the players adopted. But since players generally don’t cooperate in the Prisoner’s Dilemma, and they do in the Harmony Game, the payoffs in the Harmony Game are, on average, significantly higher.
Nevertheless, when the students were asked to vote on which game to play, a majority chose the Prisoner’s Dilemma. Since playing the Prisoner’s Dilemma results in smaller payoffs, it’s fair to call it a bad policy, and the fact that most subjects decide to play it shows that, under certain circumstances, voters demand bad policies, say the researchers.
Voting for the Prisoner’s Dilemma is only logical if the voter underestimates how differently the other players will behave in the two games, says Dal Bó. A subject who believes that other players will cooperate as frequently in the Prisoners Dilemma as in the Harmony Game (which they won’t) will prefer to play the Prisoner’s Dilemma.
The players have all the information they need to make a rational choice, and if they were the rational people of the economic models, they would. The results, Dal Bó says, “mean that because we are not always the rational actor postulated in theory, societies with access to democratic means of reaching decisions may fail to resolve social dilemmas like the one captured by the Prisoners’ dilemma. This is concerning since some of the most challenging problems of our time, such as global warming, have the structure of social dilemmas.”
Why do voters often opt for bad policies? Is it because the media and the schools haven’t done a good job educating them, or is it the fault of politicians who pander to short term electoral considerations? A novel experiment by Professor Ernesto Dal Bó, the Phillips Girgich Professor of Business at UC Berkeley’s Haas School of Business, and his co-authors pinpoints another cause.