Partisan Politicians Flashcards
So far we assumed politicians are opportunistic (only care about winning) to extract exogenous (salary) or endogenous rents (so do not care about policy)
However they may have strong ideological/policy preferences. What is this known as
Partisan preferences
We could still see policy convergence if partisan politicians
When
If politicians have to make binding commitments
If not binding (lack of enforceability), policy divergence
We use citizen-candidate model again for partisan politicians
(extension to political agency; last topic (i.e whether endogenous rents could be extracted under a) MVT or b) PVT)
Candidates care about 2 things. What?
Winning
And implementing their preferred policy
Setup for model: median voter model
Citizen-voter i has income Yi and utility (as usual)
Ui = Ci + ln(g)
Individual budget constraint of i is Ci = (1-t)Yi
Gov budget constraint
g = tYbar
Politicians propose policy (a level of government spending)
But which level of government spending do voters prefer. (Same as working for checking endogenous rents under MVT)
Find indirect utility of voter i
Vi(g) = Yi - Yi/Ybar g + ln(g)
Found by subbing in individual BC and rearranged gov BC (t=g/Ybar) into utility function
Then differentiating we get
g*i = Ybar/Yi
Same as last time!
Now this bit is different.
We will now assume no rents (r=R=0) Why?
Since no rents, only motivation to win is thus partisan i.e to implement preferred policy!
Candidates A and B have to propose FULLY BINDING (enforceability exist) in their policy platforms: ga and gb (we will see non-binding scenario later)
As candidates are also citizens. Assume Ya<Ym<Yb
Indirectly, what would their g distribution look like?
ga>gm>g*b
As A is poorest, wants more government spending! So A is left-wing, B is right wing
Partisan politicians:
Candidates set their proposed policy to maximised what?
Expected utility
(NOT PROBABILITY OF WINNING SINCE PARTISAN)
So they will set policy to maximise their expected utility
What does this look like for A; expression (pg12)
B) what is important to see here
E[Ua(ga)] = pa Ua(ga) + (1-pa)Ua(gb)
If A wins If B wins
B) as partisans, A cares about which policy is implemented, if they win Ua(ga) and also if they lose Ub(gb)
Symmetrically, what would B’s expression be
Eb[Ub(gb)] = pa Ub(ga) + (1-pa) Ub(gb)
If A wins If B wins
Key: Now what is Pa dependent on?
B) and so what 3 possible results do we get
Depends on median voter utility!
B)
If median voter gets more utility from gb than ga, Pa = 0!
If median voter gets more utility from ga than gb, Pa=1
If median voter indifferent between ga and gb, Pa=1/2!
2 opposing forces we see for candidates optimisation
Centrifugal force - A wants to increase Ua(ga)
This would be further away from median voter bliss point.
Centripetal force - A also wants to increase Pa
This would move closer to median voter bliss point
As a result of these centrifugal and centripetal forces; what is the policy outcome given our assumption we made of BINDING policies
B) which force has dominated
ga = gb = g*m
They converge to median voter bliss point! (Since every time they move closer to M, their probability goes to 1! (FC 11) So keep going till convergence.
B) centripetal force (to maximise Pa) has dominated, since both have moved closer to median voter bliss point
That was when Ya<Ym<Yb.
What about when Ya<Yb<Ym?
Still see convergence, but to g*b.
Reasoning behind this: assume start at gm.
gm<gb<ga
They both benefit from moving further away from median, and towards their bliss points to gain higher utilities (centrifugal), but A needs to stop at gb, cannot move closer to ga without losing election for sure
3 scenarios summary:
Ya<Ym<Yb
Ya<Yb<Ym
Ym<Ya<Yb
A) convergence to g*m (as shown)
B) convergence to g*b (as shown)
C) convergence to g*a (same intuition as prev)
Now assume not binding policies (maybe lack of enforceability)
(Recall no enforceability caused problems when endogenous rents existed, but in this topic we’ve assumed r=R=0! So will be different….
What do we assume now
Promises are cheap talk, and voters are aware of this
As mentioned, with no binding commitments we see policy divergence.
How?
B) how do voters respond
With no enforceability to set to g*m, so can maximise its own utility!
For example if A won, maximise Ua(ga) by setting at g*a!
If B won, set g*b, so policy divergence!
B) Voters know this, that candidates won’t do g*m, and thus vote for policy which appeal most to median voter! E.g
ga if Um(ga) > Um(g*b)
Why is CC model good (2)
Predicts policy divergence which is realistic
Shows how shifts in power/policy arise due to changes in voters or changes in candidates (e.g the 3 cases of Ya<Ym<Yb etc) ; in standard MVT/PVM, equilibrium only shifts with median/swing voter.
So far we assumed exogenous candidates i.e A and B with Ya and Yb.
Now allow for endogenous;
Each citizen i decides whether to run, at cost c and preferred policy gi
Election is held amongst only those who choose to run.
Assume no binding commitment so winners implement preferred policy. If no one runs status quo gbar)
Suppose median voter M runs for election.
When would this happen?
If Um(gm) - Um(gbar) >= c
If profitable to run
(Benefit from not going with status quo)
What is important to note if M runs
No one else will run, since M’s policy is condorcet winner (wins in pairwise vote)2
When would 2 candidates (A and B) run
If median voter is indifferent Um(ga) = Um(gb)
A must find it profitable to run given B runs.
1/2 [Ua(ga) - Ua(gb) >= c
(Symmetrical for B)
Basically the same intuition as if only median voter ran: If Um(gm) - Um(gbar) >= c , but this time 1/2 since 2 candidates, and also paired against gb not gm.
Why does a third candidate not enter e.g some one with preference between A and B
Assume median voter M runs, and you have ideology close to right of M.
So M is the better vote. However, voting for M instead of B, reduces chance of B winning, so A would win!
(Our vote isn’t enough to make M win, but can make B lose to A! Since we would be substituting B vote for M!)
Thus, M is not willing to incur cost of running.
Does identity of elected candidates matter for policy
MVT and PVT - no, only matters about median/swing voter
CC model - yes! (E.g if Ya<Ym<Yb vs Ya<Yb<Ym)
Does gender matter for policy then?
Study findings
Rule: 1/3 of GP heads must be women
Found different preferences per gender, women want more health/welfare expenditure
Men: roads, irrigation, education
So women as policy makers shape policy towards their preferences
