Prof. Bryan Caplan
Week 11: Behavioral Political Economy
What Is Behavioral Political Economy?
The key intuition behind “behavioral economics”: The behavior of
flesh-and-blood people is very different from the behavior of rational
actors. So if we want to understand the
world, we need to spend more time studying human psychology, and less time
analyzing irrelevant models.
The “behavioral” revolution has made big inroads in almost every area
of applied economics – except the economics of politics.
Even if I’m wrong to think that people are especially irrational in
politics, this is a puzzling oversight. And
if I’m right, it’s perverse!
This week we’re going to try to correct this neglect of behavioral
We’ll begin by putting irrational voters into standard public choice
models to see what happens.
Then we’ll look at some “advanced prototype” models that use specific
assumptions about voter irrationality to explain otherwise puzzling facts about
Irrationality in the Median Voter Model
How does irrationality affect policy in a simple median voter framework? Let's consider two simple cases:
Case 1: Voters are identical in all respects - including identical
near-neoclassical demand-for-irrationality curves.
Case 2: Voters are identical except that they have different
near-neoclassical demand-for-irrationality curves.
Case 1: Voters all want to
maximize social income, but also want to believe that protectionism works.
On one graph, we can show the unbiased and biased estimates of the
wealth-maximizing level of protection.
On the second graph, we can contrast the optimal and the winning
Case 2: Voters all want to maximize social income, but the median voter
wants to believe that protectionism works.
On one graph, we can show the unbiased estimate and the biased median
estimate of the wealth-maximizing level of protection.
On the second graph, we can contrast the optimal and the winning
Public choice economists have typically seen protectionism as a product
of special interests taking advantage of the public's rational ignorance.
Big puzzle for this theory: Protectionism is popular!
The median voter model with voter irrationality can easily explain the
People hold rationally irrational beliefs about trade policy, as the
SAEE and many other data sources confirm.
Politicians offer protectionist policies to get their votes.
The real puzzle: Why isn't policy far more protectionist than it is?
The Inefficiency of Political Irrationality
Economists' efficiency calculations must count the consumption value of
irrationality as a benefit. However,
this hardly implies an efficient outcome.
Why? Voters enjoy the full
benefit of their own irrationality, but pay only an infinitesimal fraction of
the cost. Each voter subconsciously
thinks "My irrationality makes no perceptible difference on policy, so I
might as well believe whatever makes me feel best."
If enough voters rely on systematically biased beliefs to decide how to
vote, disastrous policies may be adopted.
Ex: With enough protectionist voters, protectionist policies may
Rational irrationality, like expressive voting, even allows for
“inefficient unanimity.” Ex: Suppose
voters are trying to ascertain whether their nation will be able to defeat a
hated national enemy.
Each voter is willing to pay up to $100 in order to believe that
"One patriot can lick twenty foreigners, so victory is assured." If they hold this belief, they vote Yes.
But if a majority votes Yes, and war is actually declared, the country
will be thrown into a bloody conflict that costs each voter an average of
Each person believes in his country's invincibility so long as .
Since everyone is identical by assumption, it follows that as long as , 100% of all voters vote for war, even though the net per-capita
social benefit of war is -$99,900!
How is this possible? There are
massive externalities of irrationality.
Just as all polluters can be better off if everyone pollutes less, all
voters can be better off if everyone consumes less irrationality.
The Interaction Between Voter Motivation and Cognition
When they analyze politics, economists have two standard
These two assumptions imply four logical possibilities. How does each play out in a median voter
Unselfish motivation and rational cognition
imply unanimous voter support for efficient policy; selfish motivation and
rational cognition imply less favorably results.
Most critics of public choice blame its
pessimistic conclusions on the assumption of selfish motivation. They have a point, but Wittman’s work
suggests that they over-state their case.
Selfish rational voting leads to policies at most mildly less efficient
than unselfish rational voting.
What happens if unselfish motivation is combined
with irrational beliefs? Very bad
things. Choosing the optimal policy
given wildly erroneous assumptions normally leads to very bad policies.
In fact, selfish motivation probably partially mitigates the harm of irrational beliefs. Why?
Heterogeneous interests reduce the support for so-called “socially
Ex: Gas price controls
In sum, then, a plausible welfare ordering looks
Unselfish motivation, rational cognition
Selfish motivation, rational cognition
Selfish motivation, irrational cognition
Unselfish motivation, irrational cognition
Neither selfish motivation nor rational
cognition hold up empirically, suggesting that we are in the worst quadrant.
The Supply Side of Politics
What does rational irrationality say about
politicians? Politicians have strong
incentives to be rational about how to
Their incentive to rationally assess the effects of policies are much less
clear. If voters rate politicians on the
sole basis of agreement with their policy views, politicians have no incentive to rationally assess
If voters can detect sincerity, politicians have
a negative incentive to rationally
assess policies’ effects!
However, if voters also reward politicians for
their results, politicians get some electoral
benefit from second-guessing their constituents. But does this benefit outweigh its costs?
Added difficulty: Do voters at least have
rational beliefs about which politicians affect which outcomes? My work in progress says they don’t.
Politicians are an extremely select group. But they’re selected for persuasively telling
voters what they want to hear, not their high-quality analysis of public
Do advisors help? Again, it all depends on how voters reward
politicians. If they reward politicians
who agree with their policy views, politicians want advisors to figure out the
most compelling way to tell voters what they want to hear – not what works.
If you were a politician, how would you sell
more immigration to the American public?
Drug legalization? A free market
in human markets? If you were an
advisor, how would you sell them to a politician?
Irrationality and Slack
Economists usually blame political failure on
“agency problems,” not voter irrationality.
My claim: Democratic agency problems are largely
the result of the principals’
negligence. If voters were rational, these
problems would have been largely solved before they started.
As we’ve discussed, rational voters have
powerful tools to discipline politicians, even in the face of severe ignorance
and high monitoring costs:
Use Beckerian punishment strategies.
Reward/blame the top.
When in doubt, say no.
Give the Miracle of Aggregation a hand: Vote only
on what you know.
If these strategies seem overly intellectual,
note that voters seem to understand them well enough to use them on special
occasions – most notably to punish offensive remarks.
The existence of a big bureaucratic pyramid does
not fundamentally change anything. A
competent politician’s most basic order to his underlings is: “Do what I would
have done if I knew all the details.” So
if a subordinate does something voters don’t like, his superior either…
wanted him to do it,
managed him incompetently, or
is a bad judge of character.
This doesn’t mean that agency problems don’t
exist, just that you need voter irrationality to explain big, persistent agency
Ex #1: Rational voters would not accept
“buck-passing” or “plausible deniability.”
They would roll their eyes if a president tried to blame an underling
for e.g. torture. (“I’m shocked, simply
However, irrational voters might be willing to
buy lame excuses – ensuring a steady supply of deception.
Ex #2: Many models of political failure require
the assumption that politicians can’t be paid for performance. So why don’t voters just pay them for
Perhaps voters prefer to see politicians as
selfless servants of the public good, so they see no need for better
incentives. Given public choice
economists efforts to disabuse the world of “politics as romance,” it is hard
to deny that this idea is widespread.
Answering Wittman, II
To my mind, rational irrationality is the second key pillar of a
thoughtful answer to Wittman.
Yes, public choice arguments frequently require "extreme voter
stupidity," as Wittman charges. But
so what? Voters - even smart ones - become extremely stupid (“irrational”)
when they deliberate on political/economic questions.
Voter irrationality is both:
Plausible in theory
Easy to detect empirically on a large scale
Key asymmetry between politics and markets: Incentives for
rationality. In markets, ignorant actors
do their best with what they know. In
politics, they scarcely try.
Rational irrationality helps explain why politicians cater to voters'
prejudices rather than trying to "educate" them. Voters like candidates who share their
confusions, not pedants who lecture them.
Can rational irrationality breathe new life into old political
Pork barrel politics
Political advertising and special interests
Wittman has engaged my criticism in a three-round exchange – the first
two rounds in Econ Journal Watch, and
the last round in Critical Review. While Wittman has not officially changed his
position, I think he has lowered his standard of “rationality” so much that
almost anything would be consistent with it.
He has even written a paper (“Utility When There is a Preference for
Beliefs”) explicitly assuming that people’s beliefs become less rational as the
incentive for rationality falls.
Why Isn’t Democracy Worse?
Before people study public opinion, they often wonder:
“Why is policy so bad?” After studying
public opinion, they often wonder: “Why isn’t policy much worse?”
Answer #1: The median voter is more educated
than the median citizen – and the more educated are less irrational.
How convincing is the Australian
Answer #2: Voters reward politicians for both
policies and results, so politicians deliberately water down the voters’ worst
Intriguing implication: Voters will dislike
politicians because they seem either dishonest or incompetent.
Answer #3: Voters’ inept monitoring strategies
give politicians and bureaucrats a lot of slack – and on balance they use their
slack in the public interest!
Are government economists the real “special
interest” behind free trade agreements?
Cognitive psychologists have found that people frequently estimate
probabilities based upon how easy it is
to think of examples. Psychologists
call this the “availability heuristic.”
This often leads to systematically biased estimates. Psychologists call this "availability
This bias is normally demonstrated in simple experiments. How does it play out in the real world?
One fascinating answer (Kuran and Sunstein): The interaction between
availability bias and the media leads to a never-ending series of mass
hysterias, or "availability cascades.”
The cycle of hysteria:
The media gives massive coverage to shocking but rare events in order
to get good ratings.
The public watches. Watching
makes it easier for the public to think of examples of the events the media
One effect: The public begins to think the problems are quantitatively
serious, so it gets easier to sell the public similar stories.
Other effect: Politicians begin trying to solve the "problem"
to win votes.
It is easy to combine this with my rational irrationality
approach. Mass hysterias provoke strong
political responses but weak personal responses because the price of
irrationality is lower in the former case.
The Idea Trap
Standard finding in growth econ: The convergence
hypothesis fails. Poor countries do not
on average catch-up to rich countries.
However, poor countries do catch-up if they have good policies. (Sachs and
Warner) Convergence fails because poor
countries persistently stick with bad policies.
Remember my finding that income growth “makes people think like
economists”? If we assume that this
finding generalizes across countries and over time, a simple model that I call
the “idea trap” can explain all these facts.
The model: A country has three attributes:
growth, policy, and ideas. Each
attribute can be good, mediocre, or bad.
First two “laws of motion” for countries are
Good ideas cause good policy (say, through a
median voter mechanism).
Good policy causes good growth (near-tautology).
The last “law of motion” is non-obvious:
“Negative feedback”: Bad growth could lead to
good ideas, and vice versa, through a learning/hubris mechanism. In this case, there is a unique equilibrium
where growth, policy, and ideas are all mediocre.
“Positive feedback”: Good growth could lead to good ideas, and vice versa. In this case, there are three equilibria – one where all variables are good, one where
they’re all mediocre, and one where they’re all bad.
The model with positive feedback fits the
stylized facts. But it seems
counter-intuitive. Don’t countries learn
from their mistakes?
My answer: On average, NO. Disaster usually leads to more disaster, not
reform. So when disaster does lead to reform, we should interpret
it as a positive shock, not an inevitable result of events.
The Great Depression
Allende, Pinochet, and Chile
The Crisis of 2008?
Government Growth and Crisis
Government as a percentage of GDP has grown
tremendously since 1900. Other measures
of the size of government, economic freedom, etc. reinforce this conclusion.
Why has this happened? If you approve of these changes, you’ll
probably just say that government is a normal good, or that government grew as
awareness of market failure grew, or as the majority overcame the plutocrats’
conspiracy to keep them down.
If you don’t approve of these changes, however,
you might turn to behavioral political economy for an explanation.
Simplest model: Voters have become increasingly
irrational over time. But is this
More popular explanation: Higgs’ “ratchet model”
of government growth from Crisis and
Leviathan. You don’t get big
government gradually, or because people wisely see its advantages. Instead, a “crisis” (war or depression) hits,
and the frightened population looks to government for salvation. By the time the crisis recedes, people take
big government for granted. It might
shrink below its peak level, but it never goes back to where it started.
What are Higgs’ underlying assumptions about
voter cognition? They seem similar to
those in my idea trap paper – voters are especially
irrational during a crisis. Then he
adds on something like status quo bias to explain why things don’t go back to
normal later on.
Tyler Cowen’s alternative: Government growth is
driven by technology. Voters were always
about as irrational (or rational) as they are now. There has always been a “latent demand” for
the welfare state. But it wasn’t
feasible to create one with the technology of 1850. (Furthermore, wasteful experiments are deadly
when income is near subsistence).
Cowen also criticizes Higgs’ model by pointing
to countries like Sweden that stayed out of both world wars still got big
Healy and Malhotra’s “Myopic Voters and Natural
Disaster Spending” (APSR 2009) provides an provocative new take on how
democracies respond to crises: Voters reward “cure” spending, but not
“prevention” spending, even though prevention is much more cost-effective. If you were a politician, how would you
respond to these incentives?