Notes for a talk on Economic Research and Politics
Public Choice Economics
The application of Economic Modeling techniques to Politics
What do economists assume about people in the marketplace?
People are motivated by self-interest
People are rational
People rationally pursue their self interests
And the sum of all these interactions – the market
Homo-economicus “economic man”.
Economic theory, applied to people in the marketplace, is probably the most powerful set of tools that social scientists have developed so far. Theoretically elegant, empirically robust, true over time and societies, and widely agreed upon.
The origins of public choice economics
by the early 1940's, Keynesian economics had swept the science of economics
Economists understood the micro-economy, and the Macro-economy
And computers had reduced all problems to one of data
We could truly “manage” the economy as never before
Economics was now like engineering – the theory was set, it was all application
But the problem was politicians were not cooperating
Why – they were slow?
or just uneducated? Or maybe they were evil?
Public choice arose out of the fact that now that economists knew everything, they had to explain why everybody was ignoring them.
What was the problem with Politicians?
Nothing, the problem was with economists – we were assuming that people, when making collective decisions, were different from people making private decisions. We assumed
Benevolent Despots
Rational, informed voters
Public Spirited Bureaucrats
What if they were not? What if, in making collective decisions, voters, politicians, and bureaucrats were actually...
Voters, Politicians, and Bureaucrats are motivated by self-interest
Voters, Politicians, and Bureaucrats are rational
Voters, Politicians, and Bureaucrats rationally pursue their self interests
And the sum of all these interactions – Politics
Homo-economicus “economic man”, applies to collective decisions as well.
Public choice economics is, therefore, the application of economic reasoning, assumptions and techniques to politics
three questions about Politics that Public choice (claims) that it answers
Part I Why are voters “irrational”?
Part II Where did Thaksin come from? (politics and voting rules)
Part III Does voting produce consistent, rational results?
Part IV Can Voting rules change who wins elections?
Topic I: Voting Behavior
Why Vote?
The Rational Hypothesis
Probability of Success
Probability of Being Correct
Probability of Influencing the Election
E(Vv = PcPmVp – (Ci + Ct)) where 0 < Pc,Pm < 1, dPc/dI > 0 and Pm = Pm(V) and Vv = Value of Voting, Pc probability of being correct, Pm probability of it mattering, Vp = Value of your preferences, Ci = cost of information, Ct cost of time, V = total voters, I = Information about candidates
When is it rational to vote?
The Social Good Hypothesis
“if everybody didn’t vote…..”
Mandatory voting
Problem of level of preferences…
The Socialization Explanation
You vote because your parents did. It is important
Expressive Voting
Vote to express yourself
Voting as an emotional act
Voting pays off, emotionally not rationally
Problems of explaining voting behavior
Problem of rationality
Problem of irrationality
Chances of effecting an election
Assume 2n +1 voters (no ties)
Probability of a tie = 1/(nπ)1/2 (4p – 4p2)n
Note that 1/(nπ)1/2 < 1, and gets smaller as n grows
Note that (4p – 4p2)n < 1, as P is between 1 and 0, and thus this decreases as n grows
A close supreme court ruling (9 voters, evenly split) gives a 30% or so chance of being the deciding vote
A non-close supreme court ruling (p=66%) is a 1 % chance of being the deciding vote
A close national election (n = 50,000,000 or so) gives a 1.1*10-21
Imaging winning a 1/100,000,000 lotery, 11 times in a row………..
Problems of voting
Information gap
Knowledge gap
Attenuated costs of mistakes
Community Knowledge vs. Individual Knowledge
Problem of preference Intensity
Problems of Mechanics
How to vote?
Best system: Majority voting
Super-majority voting, Unanimous votes
Voting systems matter
First past the post voting vs. party list voting
Negative advertising
works when positions are fixed, set amount of seats....
Voting as a zero sum, vs. negative or positive sum, game...
When is voting zero sum?
Part II: The median voter theorem – where did Taksin come from?
Medium Voter Model – Numeric Example
Eleven Voters, arranged on a High spending – Low spending axis (Left – Right)
Eleven voters deciding on spending levels for a public parkVoter / A / B / C / D / E / F / G / H / I / J / K
Desired Bud / 20 / 17 / 13 / 12 / 11 / 9 / 8 / 8 / 5 / 5 / 4
The Model:
All voters are Fully Informed and Rational
All voters will in fact Vote, for the candidate who is closest to their preferred position
The policy that gets the most votes will win
Candidates can choose any position they want to run on.
Weak Medium voter theorem: Any policy that the medium voter favors will defeat any other policy that it is competing with:
Example – if current spending is 4, and somebody proposes to spend 8 instead, 8 will defeat 4. Voters A,B,C,D,E,F,G,H will all voter for 8, only I,J,K will vote for 4
Strong Medium voter theorem: The policy preferred by the medium voter will eventually win.
Example – we have reached a spending level of 8. A new candidate will arise, proposing to instead spend 9. A,B,C,D,E,F will support it, G,H,I,J,K will oppose it. Spending is increased from 8 to 9, by 6 votes to 5.
The medium vote theory and political parties.
Each voter, rather then directly voting for a position, votes for a candidate who promises to spend at a certain level. Candidates (in this example) are also voters.
Party list systems; you can be a representative if you pass a certain threshold of votes, for our example that means 10%.
In the above, you need at least two voters to vote for your party. If only one voter votes for a party, the party does not reach the minimum threshold, and that voter is not represented at all.
The above model predicts a rightwing party consisting of I,J,K, and another party consisting of G,H, and maybe F. It predicts a leftwing party of A,B and maybe C. Another party will exist on the left, consisting of C,D or D,E, or possibly E,F (it is hard to predict)
Parties will always consist of at least two voters. Parties will never be larger then 3 voters. The party structure will be unstable.
First Past the Post Voting
In first past the post voting, whoever gets the most votes, wins.
This will lead to a different party structure; one with larger parties. Imagine the parties we had from the previous example: A,B,C and D,E,F and G,H and I,J,K. In this situation, a coalition of either ABC and DEF (majority), or DEF and GH (controlling minority), will dominate.
Now, the only candidate who wins is the one who gets the most votes. If ABC and DEF each voter for their own candidate (say B and E), GH and IJK can merge, and win the election. Candidate I could get 5 votes, while B and E each only received 3. Voter I wins, even though he only received 5 of eleven votes. But now, parties ABC and DEF have an incentive to combine, because if they do so, they will have 6 votes, and win. But DEFGH has an incentive to attract F, so that they have a majority.
Both parties are now large, and competing for the median voter.
Eleven voters deciding on spending levels for a public parkVoter / A / B / C / D / E / F / G / H / I / J / K
Desired Bud / 23 / 17 / 13 / 12 / 11 / 9 / 8 / 8 / 5 / 5 / 4
Extensions of the model
Alienation. Not all voters will vote. They will only vote if a candidate adopts a policy within 2 of what they believe. So for example, if a candidate proposes spending of 7, GHIJ would all support it (depending on other proposals) but K would not, it is to far from his desired position so he sits the election out.
Party Lists……..
In this example, voter A will never participate in elections, under either party lists or first past the post. Any proposal close enough to his position (spending of 21 or more) has no appeal to anybody else. B and C would thus form a party (2 votes to spend 15), and DEFGH could form a super party (5 votes to spend 10). IJK would also form a party, 3 votes for spending of 4 or 5.
But would this be stable? If somebody proposes to spend instead 6, IJK would vote for it, and maybe G and H. Other examples could be made: the net effect is the outlier voters become increasingly irrelevant. They form fringe parties (K and IJ) or are unrepresented (A). B will never be in the same party as anybody other then C, his views again are to extreme.
First past the Post……
A winning coalition does exist, based on 6. A candidate who proposes 6 will attract voters GHIJK. This would defeat a party unified around 11, which would attract voters CDEF. If this center/left party switched to a position of 10, they would win if they could attract GH, but note now that C withdraws from politics. In effect, GH are the new median voters….
Eleven voters deciding on spending levels for a public parkVoter / A / B / C / D / E / F / G / H / I / J / K
Desired Spending / 23 / 17 / 13 / 12 / 11 / 9 / 8 / 8 / 5 / 5 / 4
Primaries
Up to now, the assumption has been that parties form effortlessly, and candidates just “appear” without any transaction costs. In reality, transaction costs do exist.
Closed Primaries……
Parties need to select candidates, again lets assume everybody votes. Imagine a Rightwing party consisting of GHIJK, and a Leftwing party consisting of ABCD, and EFG are independents. In the republican primary, I is the median voter, so a candidate who proposed I would win the primary. In the leftwing primary, C or D is the median voter, (for this example, say that D won).
Now, in the General Election, the Left supports D, the right supports I. The median voter F, and A (if he votes) all vote for D. The position that wins the Rightwing primary looses the General Election. The position that won the Leftwing primary also won the general election.
Open Primaries…….
In an open primary, anybody can vote in the primary. This leads to strategic behavior, and different dynamics. In the Rightwing primary, if F (the independent) can vote in the primary, the median voter changes from I to either I or H. If it changes to H, then in the general election the rightwing party wins.
Primaries and voter intensity….
Assume again voters only vote within two of their desired position. If H and D win the primaries, H will now lose the general election. His position, 8, while it appeals to F, has lost him IJK, so he has FGH. That only ties CDE.
Part III: The problem of vote Cycling (Condorcet)
The median voter model assumes single peaked preferences along a single dimension
This works for many allocative economic decisions, such as roads or bridges
But non-economic decisions (or distributive ones) often can’t be defined this way
For example, imagine the U.S. after a terrorist attack. Policy makers (and by extension, the voters they represent) would fall into one of three categories
Hawks: It is their fault, so lets make them pay….
Conquer them, or if you can’t, Nuke them, or if you can’t, appease them.
Doves: It is our fault, we need to understand them and show them we love them
Appease them, or conquer them, but don’t Nuke them…..
Isolationists: It is their fault, we must punish them, but who wants to run an Arab country?
Nuke them, or appease them, or conquer them.
We have three voters, H,D and I
We have three policies, N, C, A
Vote Cycling in the Presence of Dual Peaked PreferencesPolicy/Voter / Hawks / Doves / Isolationists / Policy voted for
Nuke them / 2 / 3 / 1 / N will lose to C
Conquer them / 1 / 2 / 3 / C will lose to A
Appease them / 3 / 1 / 2 / A will lose to N
Our preferences for
H is C > N > A
D is A > C > N
I is N > A > C
Thus, a lower number is our preferred policy, higher numbers represent less preferred policies.
Q: In a pairwise vote, which policy will dominate?
Graphically, we can represent this in the graph to the Right. Note that our Isolationist has two policies that he prefers to conquest, either don’t get involved, or Nuke ‘em.
In a situation like this, the voting result will depend very much on the skill and power of an agenda setter.
The Problem of Condorcet Vote Cycling III: Multi-dimensional voting, “Bundled Votes”
What happens when we extend our analysis into two dimensions.
Imaging our indifference curve as being around an Ideal Point. In this analysis, each voter has a single ideal point (single peaked) and any deviation from that point leads to a lower level of utility. Thus, we can draw an indifference curve around the ideal point for each voter.