Probabilistic Voting Model Flashcards
In reality, sometime voters can best elect, but not affect policies. What is this known as
Political principle agent problem (agent i.e parties who act on behalf of principles i.e voters, act in self interest)
Why may voters not exclusively consider candidates policy when voting for candidates (2)
(What non-policy factors do they consider) (2)
Vote on ideology/sympathy.
May also not have/be able to process the full information
Main criticism of MVT
Unrealistic to believe individuals base votes on just policy - Non-policy factors like sympathy exist irl!, likeability, looks!
Similarity between MVT and PVM
Both predict policy convergence
Probabilistic voting model main characteristic
Introduces voters’ subjective perceptions on top of their preferences for policy
Mechanism of MVT vs PVM
MVT - median voter determines policy outcome
PVM - swing voter determines policy outcome
PVM setup:
2 candidates A and B: each opportunistic
Candidates simultaneously but independently choose policy which consists of 2 issues e.g gov size and foreign policy
What can we immediately notice from this that differs from MVT
Multi-dimensional unlike MVT which only considers one policy
3 voter groups, poor, middle, rich
Yp<Ym<Yr
Proportion of population is thus
ap, am, ar
Every voter i can differ in ideology
Voting behaviour depends on: (3)
Policy
Individual ideology towards a candidate σij
Candidate average popularity before election (the shock) δ
Voter i votes for B if…
Uj (Xb,Yb) + σij + δ > Uj(Xa,Ya)
(utility from candidate B’s 2 policies) + individual ideology + shock
has to be > utility from the other candidate A’s 2 policies
Ideology σij (ideology of a voter i in group j.
Key assumption
B) what does σij mean if
>0
<0
=0
Assume ideology follows uniform distribution centered on neutral ideology i.e where = 0
b)
σiJ > 0 means voter i is ideologically closer to B.
σiJ < 0 means voter i is ideologically closer to A.
σiJ = 0 means that voter i is neutral.
Scandals δ can reduce one and increase others. Assumption for popularity
B)
δ>0
δ<0
Uniform distribition centred on neutral ideology
δ=0
B)
δ>0 = shock favours B
δ<0 = shock favours A
Timeline: step 1 - electoral campaign: candidates announce simultaneously
(Xa, Ya) and (Xb, Yb)
Step 2: shock δ may occur
Step 3: election takes place, voters vote
Step 4: winner implements policy
Swing voter
The voter who after considering each policy platform, and average popularity (the shock) is indifferent between the 2 candidates
Swing voter expression
σj > Uj (Xa,Ya) - Uj (Xb, Yb) - δ
Why is swing voter the most importanst
Because a small change in policy platform is sufficient to gain her vote. Easiest to change opinion
Issue candidates face
Do not know average popularity is known, i.e do not know who the swing voter is (only know probability of shock)
Candidates strategies: they have to set policies before average popularity is known.
So how do they maximise probability of being elected?
Consider scandals
Who will vote for candidate A
People left of swing voter.
Area of the voters for A
When does A win election
When Πa > 0.5
Rearrange to find δ (scandal threshold below which A wins)
First result we get:
A wins if not hit by a scandal. But they do not know the value of the threshold, so maximises probability of winning
So A maximises probability of winning:
A’s maximisation problem
B) what to note about the weightings between groups i.e which groups are given more relevance (2)
More relevance given to larger group (higher aj i.e proportion) and to the less ideological group (higher ΩJ density)
(Since more likely contain swing voters)
What does B do?
Since also unaware of value of shock threshold, Performs same maximisation process to max probability to win, mirroring A
Thus set same policy platform! Xa, Ya = Xb, Yb
Policy convergence!
So PVM on why candidates please swing voter
Less ideological i.e less care about sympathy, looks etc (no loyalty towards a certain candidate) and more sensitive to policy of candidates, so thus easier to change their vote.
