Lobbying and Corruption Dr James Tremewan - - PowerPoint PPT Presentation

Lobbying and Corruption

Dr James Tremewan (james.tremewan@univie.ac.at) Common Agency Model

Is lobbying a good thing?

• Two relevant aspects of efficiency.
• 1. Is expenditure on lobbying wasteful? Not clear:
• Campaign contributions are just transfers, so no efficiency loss.

Money spent on lobbying goes to restaurants, wages etc. so can boost economy.

• But 15,000-30,000 presumably well-educated lobbyists in

Brussels could be employed in the search for malaria cure or cold fusion...

• 2. Is the efficient policy chosen?
• In the Tullock model with a buyer, the efficient outcome was

when the buyer won (no dead-weight loss from monopoly).

• The buyer only won a proportion of the time because of lottery,
• r randomized strategies in perfectly discriminating auctions.
• Today we will look at some theory and an experiment which

focusses on the question of whether or not the efficient policy is chosen.

2/19

Common Agency Models

• Basic idea:
• A politician must choose between a number of different policies.
• Firms can bid for as many policies as they like.
• The politician chooses the policy with the highest total in bids.
• Firms pay only bid for the policy that is chosen.
• In contract theory terminology, the lobbyists are ”principals”

and the politician the ”common agent”.

• Main differences with Tullock model:
• Several firms may benifit from the same policy.
• Firm’s payment is conditional on the politician’s action.
• For now we assume that contributions are just transfers from the

firms to the politician (either campaign contributions or implicit/explicit bribes), so no efficiency loss from lobbying expenditure, only from the ”wrong” policy being chosen.

3/19

Common Agency Models

• One agent; m principals denoted M = {1, ..., m}.
• The agent chooses an alternative from a finite set S.
• Each principal j ∈ M chooses contribution schedule

tj = {tj

1, ..., tj s} for any or all of the alternatives s ∈ S (all

tj

s ≥ 0).

• The agent chooses s to maximize πP =

j∈M tj s.

• Each principal j ∈ M derives utility G j

s from s so receives net

payoff πj = G j

s − tj s.

• Game has two stages:
• Principals choose contributions simultaneously and without

communication.

• Agent observes principals’ contribution schedules and selects

alternative.

• (so we are interested in subgame-perfect Nash equilibria.)

4/19

Example

Two firms, A and B. e.g. project to build and operate toll-road: I-A does both; II-A builds, B operates; III-B does both. Alternative I II III Payoffs G A 10 6 G B 6 8

• Many SPNE, some efficient (i.e. II chosen), some not.
• Here ”efficiency” means the sum of payoffs of the firms and the
• politician. Because contributions are just transfers from a firm

to the politician, they are irrelevant for efficiency. All that matters is which alternative is chosen.

• A SPNE should specify a complete contingent strategy for the

politician, i.e. what they would do after every possible pair of contribution schedules. We will be lazy here and just specify which option is chosen, but remember that for a SPNE the politician must always maximise

j∈M tj s after any deviation.

5/19

Example, continued

Alternative I II III Payoffs G A 10 6 G B 6 8

• Equilibrium 1 (inefficient):
• tA = {8, 2, 0}; tB = {0, 4, 8}; I chosen.
• πA = 2, πB = 0, πP = 8.
• Note that even though 8 is offered for both I and III, if III was

chosen this would not be an equilibrium as A would have an incentive to increase their bid for I to 8 + ǫ so it would be chosen.

6/19

Example, continued

Alternative I II III Payoffs G A 10 6 G B 6 8

• Equilibrium 2 (efficient):
• tA = {4, 3, 0}; tB = {0, 4, 2}; II chosen.
• πA = 3, πB = 2, πP = 7.

7/19

Coalition-Proof Equilibria

• SPNE only considers unilateral deviations by firms.
• Coalition-proofness considers assumes that firms can

communicate and jointly deviate if it makes them better off.

• An equilibria is coalition proof if there is no ”self-enforcing” joint

deviation that makes a subset of the firms better off.

• ”Self-enforcing” means that no subset of the deviating coalition

has a further profitable deviation.

8/19

Example, continued

Alternative I II III Payoffs G A 10 6 G B 6 8

• Equilibrium 1:
• tA = {8, 2, 0}; tB = {0, 4, 8}; I chosen.
• πA = 2, πB = 0, πP = 8.
• Equilibrium 2:
• tA = {4, 3, 0}; tB = {0, 4, 2}; II chosen.
• πA = 3, πB = 2, πP = 7.
• Equilibrium 1 is not coalition-proof because A and B can jointly

deviate to contribution schedule in Equilibrium 2: both firms are better off and this deviation is self-enforcing (as any subset of deviating coalition is an individual, and Equilibrium 2 is a Nash equilibrium).

9/19

Truthful Equilibria

• Out of many possible equilibria, which should we predict?
• Bernheim and Whinston1 suggest truthful equilibria.
• Truthful equilibria: The contribution schedule tj is said to be

truthful if it can be written as tj

s = max(0, G j s − uj) for all s ∈ S,

where uj is a constant. A truthful equilibrium is an equilibrium in which all principals offer truthful contribution schedules.

• It can be shown that truthful equilibria:
• exist in any common agency game.
• choose an alternative that is efficient.
• Why should these equilibria be more likely?
• They require only simple strategies (may be ”focal”).
• They are coalition-proof (and all coalition-proof equilibria are

equivalent to a truthful equilibrium).

1Menu Auctions, Resource Allocation, and Economic Influence; B.

Douglas Bernheim and Michael D. Whinston, Quarterly Journal of Economics (1986).

10/19

Natural Equilibria

• Kirchsteiger and Prat2 suggest an alternative class of equilibria

(and run an experiment to see what is chosen in the lab).

• Natural equilibria: The contribution schedule tj is said to be

natural if tj

s = 0 for all s ∈ S except, at most, one. A natural

equilibrium is an equilibrium in which all firms offer natural contribution schedules.

• It can be shown that natural equilibria:
• exist in any common agency game,
• do not necessarily choose an alternative that is efficient,
• are not necessarily coalition-proof.
• Why should these equilibria be more likely?
• Even simpler than truthful equilibria.

2Inefficient equilibria in lobbying; Georg Kirchsteiger and Andreas Prat,

Journal of Public Economics (2001)

11/19

Digression

• Why do we care whether one type of equilibrium is more

common than another?

• If truthful equilibria tend to be chosen, then outcomes will be

socially efficient (assuming lobbying takes the form of transfers, etc.). This means that as long as all interested parties are able to make unlimited cash payments to politicians, then the

• utcome will be the socially efficient one, and there is no

immediate reason to regulate lobbying.

• If natural equilibria (or other outcomes) tend to arise, then

ineffient outcomes may occur and regulation of lobbying may be desirable.

12/19

Experiment: common agency game

Alternative I II III Payoffs G A 17 11 G B 7 12

• Subjects play three times in one role, then three times in the
• ther.
• Payoffs in Dutch guilders. Minimum increment was 0.05 (rather

than continuous as assumed before), so there are a number of truthful and natural equilibria, but they differ from the following by at most 0.05.

• Politician was ”automated” (not played by participant) and just

chose alternative with highest total contributions, or randomized in case of tie.

• For these reasons there are a number of truthful and natural

equilibria, but they differ from the following by at most 0.05.

13/19

Experiment: common agency game

Alternative I II III Payoffs G A 17 11 G B 7 12 Natural eq. tA 12 (I chosen) tB 11.95 Truthful eq. tA 10.95 5 (II chosen) tB 6 10.95

14/19

Experiments: Results (Chosen alternative)

• Alternative II hardly ever chosen (no support for truthful

equilibrium).

• Alternative I is chosen most of the time (support for natural

equilibrium).

15/19

Experiments: Results (Contributions)

16/19

Experiments: Results (Contributions)

• Average contribution schedules do not resemble those of either

type of equilibrium:

• All contributions too low for truthful equilibrium.
• Contributions to preferred outcomes are too low, and

contributions to II too high for natural equilibria.

• Contribution schedules within 5 cents of equilibrium:
• Neither class of equilibrium predicts contributions well, but

natural better, and improving with learning.

17/19

Experiments: Aggressive Bids

• Note that if
• tA

I − tA II > 7, A has ensured II can never be chosen, or

• tB

III − tB II > 11, B has ensured II can never be chosen.

• Let ˆ

ti

II be the highest previous observed contribution by the

• ther role. If
• tA

I − tA II > ˆ

tB

II , A can expect II will not be chosen, or

• tB

III − tB II > ˆ

tA

II , B can expect II will not be chosen.

• 94% of A’s and 83% of B’s schedules were such that it would be

unlikely II would ever be chosen.

18/19

Experiment: Conclusions

• The truthful equilibrium was a poor predictor of both outcomes

and schedules. The authors speculate that this is because of risk

• f miscoordination: if A plays the truthful schedule and B plays

the natural schedule, A gets zero; but if A plays the natural schedule, A gets 5 for sure if B plays either eq schedule, and A loses only 1 compared to the the efficient truthful eq payoffs.

• The natural equilibrium predicts outcomes well, but is not a

good predictor of individual decisions.

• My interpretation: subjects go for the selfish option (as

predicted by the natural equilibrium), but contribute a little to II to fool themselves into thinking they are trying for the fair

• ption too, even though their contributions are low enough to

make this outcome unlikely.

• Main conclusion: The truthful equilibrium was seldom played,

and lobbying led to inefficient outcomes.

19/19