Lobbying and Corruption Dr James Tremewan - - PowerPoint PPT Presentation

Lobbying and Corruption

Dr James Tremewan (james.tremewan@univie.ac.at) Gender and Corruption

Introduction

Gender and Corruption

• Women are often perceived as less susceptable to corruption

than men.

• In Mexico City traffic officers all women, in attempt to reduce

bribe demands.

• Similar scheme in Lima

(http://www.americasquarterly.org/node/2802)

2/33

Introduction 3/33

Introduction

Gender and Corruption: cross-country regressions

• Greater female influence is associated with lower levels of

corruption:

• “Are women really the fairer sex? Corruption and women in

government”, Dollar et al (2001):

• Proportion of women in parliament negatively correlated with a

number of corruption measures.

• Around 90 countries and three years (1985, 1990, 1995).
• Correlation remains after controlling for a number of variables

(e.g. civil liberties, legal origin).

• “Gender and Corruption”, Swamy et al. (2001):
• A variety of corruption measures are negatively correlated with

proportion of women in parliament, in government ministerial positions, and the labour force in general.

• Correlation survives similar controls as Dollar et al (2001).

4/33

Introduction

Gender and Corruption: cross-country regressions

• But will increasing gender-equity reduce corruption? Correlation

is not causation!:

• Reverse causality: in corrupt countries, male-dominated

networks may provide greater barrier to female participation.

• Omitted variable: both corruption and gender-inequity may be

caused by third variable.

• “Fairer sex or fairer system? Gender and corruption revisited”,

Sung (2003):

• Similar exercise to the two earlier papers.
• Finds that relationship between gender-equity and corruption

disappears after controlling for “liberal democracy” variable.

• “In sum, liberal democratic insttitutions and spirit increase

female participation in government and restrain systemic corruption, but the latter two factors are not causally related.”

5/33

Introduction

Gender and Corruption: WVS (63 countries)

6/33

Introduction

Gender and Corruption: survey data

• Swamy et al (2001) analyze data from the World Values Survey.
• Women say they are less tolerant of corrupt behaviour than

men.

• Significant difference remains after controlling for a number of

variables.

• (Less likely to be a gender difference in developing countries).
• Criticisms:
• Men and women may have different experiences of corruption

depending on social role:

• Higher exposure to corruption in daily life may increase

tolerance.

• Greater victimization by corruption may decrease tolerance.
• Attitudes towards corruption may not fully determine propensity

to act corruptly (maybe gender difference in relationship between attitudes and behaviours?).

• No incentive to tell the truth: women may have different need

to maintain social or self-image when answering survey.

7/33

Introduction

Gender and Corruption: experiments

• Experiments can overcome problems with empirical and survey

data.

• Causality can be identified by placing males and females in

identical environments.

• Equalizes opportunity to act corruptly.
• Can study behaviour with real consequences rather than just

attitudes.

• Financial incentives encourage genuine responses.

8/33

Introduction

Gender and Corruption: experiments

• We will look at three experiments related to gender and

corruption.

• “An Experiment on Corruption and Gender”, Rivas (2012).
• Finds lower rates of corrupt behaviour among females in

Barcelona.

• “Gender, Culture, and Corruption: Insights from an

Experimental Analysis”, Alatas et al (2009).

• Finds lower rates of corrupt behaviour among females in

Australia, but not India, Indonesia, or Singapore.

• “Deterrence versus intrinsic motivation: Experimental evidence
• n the determinants of corruptibility”, Schulze and Frank

(2003).

• Examines whether lower rates of corrupt behaviour by females

can be explained by greater risk-aversion rather than different attitudes towards corruption.

9/33

Introduction

Rivas (2012)

• Same game as Abbink (2004) (see previous set of slides).
• Neutral framing.
• Firm can offer bribe to Public Official.
• Public Official can accept or reject bribe, then decides whether

to select policy A or B:

• A is socially efficient.
• B give the Firm a higher payoff, but creates costs for all other

subjects in the experiment.

• If a corrupt transaction takes place, there is a small probability

the Firm and Public Official are caught and lose all their earnings.

• Four treatments: Male Firm and Male PO; Male Firm and

Female PO; Female Firm and Male PO; Female Firm and Male PO (subjects know the gender of their partner).

10/33

Introduction 11/33

Introduction 12/33

Introduction 13/33

Introduction 14/33

Introduction 15/33

Introduction 16/33

Introduction

Conclusions

• Females are less likely to offer bribes, and offer lower bribes on

average.

• Females are less likely to accept bribes even after controlling for

the size of the bribes.

• This effect is stronger for Female/Female pairs.
• Females are less likely to reciprocate a bribe and choose the

socially inefficient option even after controlling for the size of the bribes..

• Social costs are significantly lower (average earnings higher) for

Female/Female pairs than for Male/Male pairs.

17/33

Introduction

Alatas et al (2009)

• Three player game:
• Firm chooses whether or not to offer a bribe.
• If a bribe is offered, Official decides whether to accept or reject.
• If a bribe is offered and accepted, it damages a Citizen, who can

choose whether or not to punish the Firm and Official (at some cost to themselves).

• Subgame perfect equilibrium: Bribe is offered and accepted and

the Citizen does not Punish.

• The experiment was run in Australia, India, Indonesia, and

Singapore using Corruption framing (also run in Australia with neutral framing).

18/33

Introduction 19/33

Introduction 20/33

Introduction 21/33

Introduction

Conclusions

• Females were less likely to offer or accept bribes, and more likely

to punish corruption, but only in Australia.

• In Australia, the gender difference almost disappears with

neutral framing, suggesting that the gender effect is due to differences in attitudes towards corruption, rather than simply differences in general social preferences or risk aversion.

• Strong gender differences in corruption experiments found only

in developed wester countries where (petty) corruption is not such a big issue (Waithima (2010) uses a similar design in Kenya and also finds no gender effect).

22/33

Introduction

Shulze and Frank (2003)

• Experiment run before a screening of a film, organised by a

students’ film club.

• Experiment 1: Subjects told to imagine that a 200 German Mark

bank note (about e100) has fallen down a drain-pipe and cannot be retrieved without a plumber. The subject must select

• n behalf of the film club amongst plumbers, who may be willing

to pay the subject in exchange for getting chosen.

• The cost to the film club for each plumber, and the “kickback”

each plumber offers is shown on the next slide.

23/33

Introduction 24/33

Introduction

Shulze and Frank (2003)

• Subjects had to fill out a form stating their decision (and some
• ther information).
• About half the subjects were told they would receive an extra 40

Marks regardless of their decision (to test for the effect of an increase in wage on corrupt behaviour: see first set of slides from this course).

• One subject was randomly chosen and paid (confidentially)

according to their decision.

• Experiment 2: The same as Experiment 1, but now there is a

possibility of getting caught and earning nothing (losing also the extra 40 Marks for those subjects for whom it was available): see next slide for choices.

25/33

Introduction 26/33

Introduction 27/33

Introduction 28/33

Introduction

Shulze and Frank (2003): Conclusions

• There is no gender difference in the risk-free context.
• When there was a risk of detection and punishment, females

were less corrupt.

• Greater risk aversion apears to play a role in gender differences

in corrupt behaviour.

29/33

Introduction

Gender and Corruption: Conclusions

• Females have been found less likely to act corruptly in laboratory

experiments.

• Strong gender differences have been found mostly in Western

countries.

• Greater risk aversion apears to play a role in gender differences

in corrupt behaviour, but there is also evidence that females sometimes have less favourable attitudes towards corruption.

• There has been no study finding men to be less corrupt, so

increasing the proportion of women in bureacracy and government can be recommended as having, at worst, no increase corruption (also for pure gender equity reasons).

30/33

Introduction

Intrinsic vs Extrinsic Motivation

• Extrinsic motivation: e.g. financial incentives (piece-rates,

bonuses, fines), imprisonment.

• Intrinsic motivation: e.g. altruism, work ethic, social norms.
• Traditionally economics only considered extrinsic motivation, but

there is increasing interest in the possibility that using extrinsic motivation can backfire if it “crowds out” intrinsic motivation, e.g.:

• A Fine is a Price - Gneezy and Rustichini (2000): introducing

fines for picking up children late from school increased lateness as it put a price on the behaviour which parents calculated they were willing to pay.

• Crowding out in blood donation: Was Titmuss Right? -

Mellstrom and Johanesson (2008) - introducing payments for blood donations can decrease giving as people no longer get satisfaction from making altruistic choice.

31/33

Introduction

Back to Shulze and Frank (2003)

• “Monitoring” and the possibility of punishment increased the

average bribe! (see next slide for distribution)

• Fewer people chose the honest choice (zero bribe).
• More people chose the profit maximising choices (maximum

bribe in risk-free treatment; bribes of 80 and 112 in risky treatment.

• Risk of being caught deterred some people, but crowded out

intrinsic motivation to be honest among others: net effect was to increase average bribe.

• Economics students were more corrupt in risk-free treatment,

but no different in risky treatment (no intrinsic motivation to crowd out?)

32/33

Introduction 33/33