Decentralized Bribery

57
Decentralized Bribery Sergey V. Popov Moscow, December 14, 2012 Centre for Institutional Studies Higher School of Economics

Transcript of Decentralized Bribery

Decentralized Bribery

Sergey V. Popov

Moscow, December 14, 2012

Centre for Institutional StudiesHigher School of Economics

Introduction

Bribery

Bribes

I Likhoimstvo = taking bribes for doing bad stuff.I Mzdoimstvo = taking bribes for doing stuff you’re supposed to do for free.

Problems

I First one — obvious, ex-post verifiable.I Second one — a transfer.I This paper: there are welfare implications for the second type of bribe.

Introduction

Bribery

Bribes

I Likhoimstvo = taking bribes for doing bad stuff.I Mzdoimstvo = taking bribes for doing stuff you’re supposed to do for free.

Problems

I First one — obvious,

ex-post verifiable.I Second one — a transfer.I This paper: there are welfare implications for the second type of bribe.

Introduction

Bribery

Bribes

I Likhoimstvo = taking bribes for doing bad stuff.I Mzdoimstvo = taking bribes for doing stuff you’re supposed to do for free.

Problems

I First one — obvious, ex-post verifiable.

I Second one — a transfer.I This paper: there are welfare implications for the second type of bribe.

Introduction

Bribery

Bribes

I Likhoimstvo = taking bribes for doing bad stuff.I Mzdoimstvo = taking bribes for doing stuff you’re supposed to do for free.

Problems

I First one — obvious, ex-post verifiable.I Second one —

a transfer.I This paper: there are welfare implications for the second type of bribe.

Introduction

Bribery

Bribes

I Likhoimstvo = taking bribes for doing bad stuff.I Mzdoimstvo = taking bribes for doing stuff you’re supposed to do for free.

Problems

I First one — obvious, ex-post verifiable.I Second one — a transfer.

I This paper: there are welfare implications for the second type of bribe.

Introduction

Bribery

Bribes

I Likhoimstvo = taking bribes for doing bad stuff.I Mzdoimstvo = taking bribes for doing stuff you’re supposed to do for free.

Problems

I First one — obvious, ex-post verifiable.I Second one — a transfer.I This paper: there are welfare implications for the second type of bribe.

Introduction

Literature

Empirical Literature

I Exposure to corruption ⇒ less investment, slower growth.I Corrupt economies are heavily regulated.I Putin is blamed for economic development, but not for corruption.

Theoretical Literature

I Stealing from governmental coffers is bad for development.I Rent-seeking is vacuous.I Bribes can improve the allocation.

Introduction

Literature

Empirical Literature

I Exposure to corruption ⇒ less investment, slower growth.I Corrupt economies are heavily regulated.I Putin is blamed for economic development, but not for corruption.

Theoretical Literature

I Stealing from governmental coffers is bad for development.I Rent-seeking is vacuous.I Bribes can improve the allocation.

Introduction

This Paper

Questions

I How does the “transfer bribe” affect the capital market?I Can it be beneficial?I Can it harm?

Answers

I It can make the society worse off by scaring small businesses away.I It might not be a good idea to decentralize potential bribery.

Introduction

This Paper

Questions

I How does the “transfer bribe” affect the capital market?I Can it be beneficial?I Can it harm?

Answers

I It can make the society worse off by scaring small businesses away.

I It might not be a good idea to decentralize potential bribery.

Introduction

This Paper

Questions

I How does the “transfer bribe” affect the capital market?I Can it be beneficial?I Can it harm?

Answers

I It can make the society worse off by scaring small businesses away.I It might not be a good idea to decentralize potential bribery.

Model

Fundamentals

Agents

I Agents consume a single good.I Agents have roles: investor or inspector.

Roles

I investor gets a random project, need to invest K, after investmentobserves return R, needs to pass an inspection.

I inspector asks for a bribe, if not paid does not pass the project.

Model

Fundamentals

Agents

I Agents consume a single good.I Agents have roles: investor or inspector.

Roles

I investor gets a random project, need to invest K, after investmentobserves return R, needs to pass an inspection.

I inspector asks for a bribe, if not paid does not pass the project.

Model

Fundamentals

Agents

I Agents consume a single good.I Agents have roles: investor or inspector.

Roles

I investor gets a random project, need to invest K, after investmentobserves return R, needs to pass an inspection.

I inspector asks for a bribe, if not paid does not pass the project.

Model

InvestorsI Investor observes K — project size.I Expects to pay a bribe s.I Investor chooses whether to start up a project:

I After investment, project return R is observed.I If s > RK, investor can decide to not pay the bribe and walk away.I Expected return is E[RK − s∗]+ −K.

I Will start up the project if E[R− s∗/K]+ > 1.

ResultWhen K > K∗(s∗), investor participates.

Model

InvestorsI Investor observes K — project size.I Expects to pay a bribe s.I Investor chooses whether to start up a project:

I After investment, project return R is observed.I If s > RK, investor can decide to not pay the bribe and walk away.I Expected return is E[RK − s∗]+ −K.

I Will start up the project if E[R− s∗/K]+ > 1.

ResultWhen K > K∗(s∗), investor participates.

Model

InvestorsI Investor observes K — project size.I Expects to pay a bribe s.I Investor chooses whether to start up a project:

I After investment, project return R is observed.I If s > RK, investor can decide to not pay the bribe and walk away.I Expected return is E[RK − s∗]+ −K.

I Will start up the project if E[R− s∗/K]+ > 1.

ResultWhen K > K∗(s∗), investor participates.

Model

InspectorsI Inspector know neither K nor R of the project.I Inspector decides on the bribe size s, believing in K∗.I The inspector’s problem is:

maxssP (RK > s).

I When K is trivialmax

ss P (R >

s

K)︸ ︷︷ ︸

1−FR(s/K)

.

I The solution iss/K =

1− FR(s/K)

fR(s/K).

Model

InspectorsI Inspector know neither K nor R of the project.I Inspector decides on the bribe size s, believing in K∗.I The inspector’s problem is:

maxssP (RK > s).

I When K is trivialmax

ss P (R >

s

K)︸ ︷︷ ︸

1−FR(s/K)

.

I The solution iss/K =

1− FR(s/K)

fR(s/K).

Model

InspectorsI Inspector know neither K nor R of the project.I Inspector decides on the bribe size s, believing in K∗.I The inspector’s problem is:

maxssP (RK > s).

I When K is trivialmax

ss P (R >

s

K)︸ ︷︷ ︸

1−FR(s/K)

.

I The solution iss/K =

1− FR(s/K)

fR(s/K).

Model

InspectorsI Inspector know neither K nor R of the project.I Inspector decides on the bribe size s, believing in K∗.I The inspector’s problem is:

maxssP (RK > s).

I When K is trivialmax

ss P (R >

s

K)︸ ︷︷ ︸

1−FR(s/K)

.

I The solution iss/K =

1− FR(s/K)

fR(s/K).

Model

InspectorsI In general, the solution is

s∗ =

∫ +∞0 (1− FR(s∗/K)) fK(K)dK∫ +∞0

1/KfR(s∗/K)fK(K)dK=EK [P (R > s∗/K)]

EK [1/KfR(s∗/K)].

I R ∼ Exp(α), and two levels of investment size (KH and KL) produce

s

Only H types

Both H andL types

Model

InspectorsI In general, the solution is

s∗ =

∫ +∞0 (1− FR(s∗/K)) fK(K)dK∫ +∞0

1/KfR(s∗/K)fK(K)dK=EK [P (R > s∗/K)]

EK [1/KfR(s∗/K)].

I R ∼ Exp(α), and two levels of investment size (KH and KL) produce

s

Only H types

Both H andL types

Model

InspectorsI In general, the solution is

s∗ =

∫ +∞0 (1− FR(s∗/K)) fK(K)dK∫ +∞0

1/KfR(s∗/K)fK(K)dK=EK [P (R > s∗/K)]

EK [1/KfR(s∗/K)].

I R ∼ Exp(α), and two levels of investment size (KH and KL) produce

s

Only H types

Both H andL types

Model

InspectorsI In general, the solution is

s∗ =

∫ +∞0 (1− FR(s∗/K)) fK(K)dK∫ +∞0

1/KfR(s∗/K)fK(K)dK=EK [P (R > s∗/K)]

EK [1/KfR(s∗/K)].

I R ∼ Exp(α), and two levels of investment size (KH and KL) produce

s

Only H types

Both H andL types

Model

InspectorsI In general, the solution is

s∗ =

∫ +∞0 (1− FR(s∗/K)) fK(K)dK∫ +∞0

1/KfR(s∗/K)fK(K)dK=EK [P (R > s∗/K)]

EK [1/KfR(s∗/K)].

I R ∼ Exp(α), and two levels of investment size (KH and KL) produce

s

Only H types

Both H andL types

Model

EquilibriumI The equilibrium is (K∗, s∗) such that:

I all projects bigger than K∗ are implemented;I the bribe size is s∗;I both are optimal decisions subject to rational beliefs.

I Equilibrium existsI autarky: no projects are implementedI restricted: a subset of projects is implementedI abundance: all projects are implemented

Model

EquilibriumI The equilibrium is (K∗, s∗) such that:

I all projects bigger than K∗ are implemented;I the bribe size is s∗;I both are optimal decisions subject to rational beliefs.

I Equilibrium existsI autarky: no projects are implementedI restricted: a subset of projects is implementedI abundance: all projects are implemented

Predictions

Capital MarketI The expected return of a project of size K is [R− s/K]+ − 1.I The total profit is [RK − s]+ −K.I The derivative of that with respect to K is

Average return︷ ︸︸ ︷(∫ +∞

sK

(R− s

K

)dFR − 1

)+K

Extra chance of no cancelling︷ ︸︸ ︷(∫ +∞

sK

( s

K2

)dFR

).

I Tobin’s marginal Q is bigger than 1.I Data in a corrupt economy would suggest increasing the scale of

investment...I but increase in scale will only increase the bribe size.

Predictions

Capital MarketI The expected return of a project of size K is [R− s/K]+ − 1.I The total profit is [RK − s]+ −K.I The derivative of that with respect to K is

Average return︷ ︸︸ ︷(∫ +∞

sK

(R− s

K

)dFR − 1

)+K

Extra chance of no cancelling︷ ︸︸ ︷(∫ +∞

sK

( s

K2

)dFR

).

I Tobin’s marginal Q is bigger than 1.I Data in a corrupt economy would suggest increasing the scale of

investment...

I but increase in scale will only increase the bribe size.

Predictions

Capital MarketI The expected return of a project of size K is [R− s/K]+ − 1.I The total profit is [RK − s]+ −K.I The derivative of that with respect to K is

Average return︷ ︸︸ ︷(∫ +∞

sK

(R− s

K

)dFR − 1

)+K

Extra chance of no cancelling︷ ︸︸ ︷(∫ +∞

sK

( s

K2

)dFR

).

I Tobin’s marginal Q is bigger than 1.I Data in a corrupt economy would suggest increasing the scale of

investment...I but increase in scale will only increase the bribe size.

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

Inspector’schoice

KLα

s s∗1

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

Inspector’schoice

KLα

s s∗1

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

Inspector’schoice

KLα

s

s∗1

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

Inspector’schoice

KLα

s s∗1

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

Inspector’schoice

KLα

s s∗1

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

KLα

Inspector’schoice

ss∗2

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

KLα

Inspector’schoice

s

s∗2

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

KLα

Inspector’schoice

ss∗2

Eqm bribe

Predictions

The SquanderingI Consider a situation where both restricted and abundance equilibria exist.

s

s 45◦ line

KHα

KLα

Inspector’schoice

ss∗2

Eqm bribe

Predictions

The SquanderingI What if there is a signal about the investment size?..

I Say, with probability q the signal is correct (H when investor is of type Hand I if investor is of type I).

I Then inspectors will believe their signals if both types of firms start up...

Predictions

The SquanderingI What if there is a signal about the investment size?..I Say, with probability q the signal is correct (H when investor is of type H

and I if investor is of type I).

I Then inspectors will believe their signals if both types of firms start up...

Predictions

The SquanderingI What if there is a signal about the investment size?..I Say, with probability q the signal is correct (H when investor is of type H

and I if investor is of type I).I Then inspectors will believe their signals if both types of firms start up...

Predictions

Imperfect ObservationI What if there is a signal about the investment size?..

s

s 45◦ line

KHα

KLα

s

H signal

L signal

Bad eqm bribe

Predictions

Imperfect ObservationI What if there is a signal about the investment size?..

s

s 45◦ line

KHα

KLα

s

H signal

L signal

Bad eqm bribe

Predictions

Imperfect ObservationI What if there is a signal about the investment size?..

s

s 45◦ line

KHα

KLα

s

H signal

L signal

Bad eqm bribe

Discussion and Conclusion

ExtensionsI Private information about returns

I If investors have a signal about R before investment, almost the same story.I If inspectors have a signal about each project, need to be able to convince

the investor that he needs to pay higher bribe.

I Honest inspectorsI Improve the participation constraint, might invite small businesses.I Will let go the “big fish”

I Complaining to superiorsI Lowers the participation constraint.I Does not have to lower the bribe.I Total welfare should increase...

I ...

Discussion and Conclusion

ExtensionsI Private information about returns

I If investors have a signal about R before investment, almost the same story.I If inspectors have a signal about each project, need to be able to convince

the investor that he needs to pay higher bribe.I Honest inspectors

I Improve the participation constraint, might invite small businesses.I Will let go the “big fish”

I Complaining to superiorsI Lowers the participation constraint.I Does not have to lower the bribe.I Total welfare should increase...

I ...

Discussion and Conclusion

ExtensionsI Private information about returns

I If investors have a signal about R before investment, almost the same story.I If inspectors have a signal about each project, need to be able to convince

the investor that he needs to pay higher bribe.I Honest inspectors

I Improve the participation constraint, might invite small businesses.I Will let go the “big fish”

I Complaining to superiorsI Lowers the participation constraint.I Does not have to lower the bribe.I Total welfare should increase...

I ...

Discussion and Conclusion

ExtensionsI Private information about returns

I If investors have a signal about R before investment, almost the same story.I If inspectors have a signal about each project, need to be able to convince

the investor that he needs to pay higher bribe.I Honest inspectors

I Improve the participation constraint, might invite small businesses.I Will let go the “big fish”

I Complaining to superiorsI Lowers the participation constraint.I Does not have to lower the bribe.I Total welfare should increase...

I ...

Discussion and Conclusion

ConclusionI Transfer bribery is not welfare-neutral.

I It might improve welfare when outside wage is low.I It might put the economy in a bad equilibrium, where not all projects start

up.

I Decentralization creates an inefficiency: bribe-takers cannot coordinate toswitch to a better equilibrium.

I Recovery rate higher ⇒ bribe can go down (does not have to!)I There are some nontrivial implications for the size and efficiency of

government.I Not rent-seeking.I No strategic complementarities.

Discussion and Conclusion

ConclusionI Transfer bribery is not welfare-neutral.

I It might improve welfare when outside wage is low.I It might put the economy in a bad equilibrium, where not all projects start

up.

I Decentralization creates an inefficiency: bribe-takers cannot coordinate toswitch to a better equilibrium.

I Recovery rate higher ⇒ bribe can go down (does not have to!)I There are some nontrivial implications for the size and efficiency of

government.I Not rent-seeking.I No strategic complementarities.

Discussion and Conclusion

ConclusionI Transfer bribery is not welfare-neutral.

I It might improve welfare when outside wage is low.I It might put the economy in a bad equilibrium, where not all projects start

up.

I Decentralization creates an inefficiency: bribe-takers cannot coordinate toswitch to a better equilibrium.

I Recovery rate higher ⇒ bribe can go down (does not have to!)

I There are some nontrivial implications for the size and efficiency ofgovernment.

I Not rent-seeking.I No strategic complementarities.

Discussion and Conclusion

ConclusionI Transfer bribery is not welfare-neutral.

I It might improve welfare when outside wage is low.I It might put the economy in a bad equilibrium, where not all projects start

up.

I Decentralization creates an inefficiency: bribe-takers cannot coordinate toswitch to a better equilibrium.

I Recovery rate higher ⇒ bribe can go down (does not have to!)I There are some nontrivial implications for the size and efficiency of

government.

I Not rent-seeking.I No strategic complementarities.

Discussion and Conclusion

ConclusionI Transfer bribery is not welfare-neutral.

I It might improve welfare when outside wage is low.I It might put the economy in a bad equilibrium, where not all projects start

up.

I Decentralization creates an inefficiency: bribe-takers cannot coordinate toswitch to a better equilibrium.

I Recovery rate higher ⇒ bribe can go down (does not have to!)I There are some nontrivial implications for the size and efficiency of

government.I Not rent-seeking.

I No strategic complementarities.

Discussion and Conclusion

ConclusionI Transfer bribery is not welfare-neutral.

I It might improve welfare when outside wage is low.I It might put the economy in a bad equilibrium, where not all projects start

up.

I Decentralization creates an inefficiency: bribe-takers cannot coordinate toswitch to a better equilibrium.

I Recovery rate higher ⇒ bribe can go down (does not have to!)I There are some nontrivial implications for the size and efficiency of

government.I Not rent-seeking.I No strategic complementarities.