Electoral Issue #2: Why We Demand Bad Policy
Under-appreciating equilibrium effects can result in voters demanding sub-optimal policy.
Thank you for reading our work! Nominal News is an email newsletter that focuses on the application of economic research on current issues. Subscribe for free to stay-up-to-date with Nominal News directly in your inbox:
If you would like to support us further with reaching our 1,000 subscriber goal by year end, please consider sharing and liking this article!
With the US presidential elections coming up in November, several issues have come to the forefront. Over the next couple of months, we will focus on what economic research has to say about these issues. We will not be focusing on or assessing specific policy proposals, but rather on what the economic literature can contribute to our understanding of the issues. Although the inspiration for these articles is the US election, these issues are broad enough to impact many countries.
In democracies, it is assumed that voters ‘optimally’ choose policies that result in the highest welfare to them personally. Often, however, this does not materialize, as it appears that voters may be choosing sub-optimal policies. This ‘political failure’ has been studied by social scientists, including economists, and various reasons for non-optimal choices have been proposed such as bad institutions, special interests or poor quality politicians.
But what if the voters are partially responsible for the choice of bad policies due to systematic errors in evaluating policies? This is the question Dal Bo, Dal Bo and Eyster (2018) (shortened to DDY) set out to answer.
Inferring ‘Bad Policy’ Choices
It won’t come as a surprise that it is difficult to demonstrate that voters choose bad policies based on real-world policy data. Government policies do not exist in a vacuum and even determining if a particular policy choice by a voter is ‘wrong’ is difficult. DDY, thus, decided to run a controlled laboratory experiment to determine whether individuals, in a simpler, controlled setting, are able to consistently choose the correct policy.
The Experiment
DDY set up a ‘game’ played between two people. In this two-player game, each player has to choose one of two actions. Once both players choose an action, each player receives a payoff based on the action-pair selected by the players. This is referred to as a 2x2 game (two players, each has two actions, resulting in 4 payoffs).
The game is visually represented below:
Thus, if Player 1 selects option A, and player 2 selects option B, player 1 will receive $3, and player 2 will receive $11. In this example above, known as the Prisoner’s Dilemma, it is easy to determine that the ‘only’ equilibrium is for each player to select B. Why?
If Player 2 selects A, then Player 1 will prefer to play B, as the payoff is higher for Player 1 than if they played A. If Player 2 selects B, Player 1 will likewise prefer B. Choosing B is the ‘strictly dominant’ strategy for Player 1, as it dominates choosing A for every decision of Player 2.
Since the game is symmetric (both players face an identical situation), both players will choose to play B in equilibrium, resulting in an equilibrium payoff of $5 for each. Additionally, to get acquainted with the mechanics of the game, players repeated this game several times.
The Change
Suppose you’re a player in this game. Suppose DDY now offers you the following choice – do you want to change the game such that everyone who plays A will pay a $1 tax, and everyone who plays B, will pay a $4 tax. The new game looks as follows:
It is important to emphasize again that the player can see the entire payoff matrix, before the player chooses to switch games. If we look at the changes to the payoffs in the games, we will notice that all the payoffs in the second game (‘Harmony Game’) are lower than in the first game – the Prisoner’s Dilemma. But, the optimal choice within this game has also changed. Instead of choosing B, it is now optimal to choose A. Both players would optimally choose A. The payoffs in the Harmony Game when both players choose A is actually greater than the payoff in the Prisoner’s Dilemma if both choose B. This occurs because the tax impacted the payoff pairs differently.
So you should choose to pay the tax, because your payoff will be higher. The tax in this example acts as a ‘direct cost’. But the tax changes the game. The indirect benefit comes from the equilibrium effect – that is both players will actually act differently under the new payoffs, resulting in both players being better off.
Results
You might not find it surprising, but DDY found that a majority of participants in these games chose not to ‘pay’ the tax, and preferred to play the Prisoner’s Dilemma, reducing their overall payoff (i.e. ‘welfare’). Clearly, participants focus on the direct cost of the tax much more than the indirect effect of a changing equilibrium outcome. It is worth emphasizing that the optimal and only correct strategy is to always pick the Harmony Game – every player should ultimately prefer the Harmony Game.
You may wonder – what if we switched which game was played first? Suppose the Harmony Game is played first, and then a ‘tax cut’ allows the player to switch to the Prisoner’s Dilemma game. The player should prefer to not switch. But DDY found that a majority of players did prefer to switch and play the Prisoner’s Dilemma resulting in lower payoffs, than if they stuck to the Harmony Game.
Why the Bad Policy Preference
To establish what the cause of the incorrect choice was, DDY experimented with people’s beliefs. That is, DDY asked participants, prior to them making a choice about whether to switch the game, what they think other participants would pick as their actions in the game. For example, a participant may have been asked what they think will be the probability that their partner player picks decision A (the correct choice for the Harmony Game) in either the Prisoner’s Dilemma or Harmony Game.
Individual’s beliefs were wildly off. In theory, in the Harmony Game, one should assume that other will pick decision A 100% of the time, while in the Prisoner’s Dilemma, decision A will never be picked, giving a “belief difference” between the two games for decision A of 100 percentage points (100% – 0%). Participants, however, believed the difference would, on average, be 35 percentage points (as an example to explain this difference – a person may have assumed in the Harmony Game, decision A will be chosen with 60% probability and in the Prisoner’s Dilemma, decision A will be chosen 25% of the time). Since DDY could actually observe player’s decisions in the two games, the actual difference between games was 76 percentage points. That is, decision A in the Harmony Game was selected 76 percentage points more often than decision A in the Prisoner’s Dilemma. Additionally, DDY found that participants that thought the belief difference was smaller were more likely to choose to play the Prisoner’s Dilemma game, which has the lower payoffs in equilibrium.
Participants significantly underestimated what their fellow participants would do. Moreover, DDY also conducted an experiment whereby some participants were shown actual results from the games prior to stating their beliefs about fellow participants. The participants that were shown statistics for games where the difference was large (i.e. for example, a participant was told in the Harmony Game, decision A was chosen 83% of the time, while in the Prisoner’ Dilemma, decision A was selected 17% of the time) were more likely to have a larger ‘belief difference’ (note – the correct ‘belief difference’ should be 100 percentage points). This means that actual data does influence the beliefs participants had about equilibrium behavior.
Relevance and Solutions
A natural question to ask is whether this experiment has any relevance to real-world policies. DDY believes it does, in part, because the games are quite simple. First, the game was played by college students, who are typically assumed to be more ‘sophisticated’ and able to figure out the game. Second, the games had ‘perfect’ information – everybody knew the exact payoffs from each game. Real world policies, on the other hand, are much less clear and their outcomes cannot be perfectly pinned down.Thus, if in such a simple game, correct ‘policy’ choices cannot be made, in real-life scenarios, it might be even more challenging.
One glimmer of hope is that DDY did find that participants learned to take into account the equilibrium effects after repeated exposure to both games. Thus, observing the policy in action can lead to people voting for it.
There are many real-life situations, some of which we have described, where individuals underappreciated the equilibrium effects. A recent example was the proposed congestion pricing in New York City – a direct tax that had large indirect benefits for all parties involved. Other examples are the child tax credit or the student loan forgiveness for low income individuals – both are direct costs that have large indirect benefits to society. Lastly, a common example is the Pigouvian tax (a tax levied on actions that have negative externalities) which is often mentioned in the case of carbon taxes. Although benefits accrue to everyone since people and firms would reduce pollution by making different decisions (equilibrium effect benefit), the direct cost (additional tax) dominates the perception of this policy.
Unfortunately, politicians may not be of much help here. As DDY state in their paper:
.. a vast literature in economics and political science – both theoretical and empirical – has considered politicians as reflecting, more than shaping, the positions of voters.
Interesting Reads from the Week
Rate Cut: In a bit of a surprise, the Federal Reserve cut interest rates by 50 basis points to 4.75%-5.00%. The Federal Reserve believes that the balance of risks is now tilting towards the weakening labor market, rather than any additional inflationary spikes. Moreover, the Federal Reserve governors (i.e. the individuals that vote on interest rate policy) see at least another 50 basis points of cuts this calendar year.
Article: Having discussed early childhood investment from a US perspective here at Nominal News,
discusses research on the topic of child investment in developing countries and how light-touch programs may have significant payoffs.Research Paper: A new study by Simon Buchler and Elena Lutz looked at the impacts of upzoning in Zurich. Results showed that very large upzonings (30%+ increases to buildable area) resulted in 9% increase in housing in 10 years. The effect on rents, however, remained ambiguous.
The Electoral Issues Series:
Electoral Issue #1: Investing in the Future – Child Tax Credit
Electoral Issue #2: Why We Demand Bad Policy
Electoral Issue #3: Immigration
There is nothing wrong with the structural analysis, but I am much more (un) impressed by incorrect beliefs by both politicians and the publics about what "payoffs" are. I think his rather than equilibrium dynamics is the reason we do not have merit based immigration, low deficits, and taxation of net emissions of CO2.
Ideally, and to some extent it happens, we want voters to favor things that are for the common good even if they are a bit inconvenient for them personally just as we want legislators to vote for the common good even if its unpopular. The problem is that voters are often wrong about what is in the common good and everyone would be better off if they just correctly favored what was in their individual good.