Lobbying and Corruption Dr James Tremewan - - PowerPoint PPT Presentation

Lobbying and Corruption

Dr James Tremewan (james.tremewan@univie.ac.at) An Embezzlement Game

Introduction

Embezzlement

• Embezzlement is the act of dishonestly withholding assets for

the purpose of conversion (theft) of such assets by one or more individuals to whom such assets have been entrusted, to be held and/or used for other purposes.1

• Embezzle may be “grand” or “petty” corruption:
• Grand corruption: e.g. presidents stealing public money.

Top 10

• Petty corruption: public officials, teachers, hospital staff stealing
• ffice supplies, medicines etc.

1https://en.wikipedia.org/wiki/Embezzlement

2/22

Introduction

Today’s Class

• Transparency, wages, and the separation of powers: An

experimental analysis of corruption. O. Azfar and Wi. R. J. Nelson (2007).

• Introduced the embezzlement game.
• Roles framed as President, Attorney General, and Citizens.
• Different treatments to study impact of different institutional

factors on level of corruption.

• Uses US undergraduate students as subjects.
• Corruption in public service delivery: An experimental analysis.

Barr et al. (2009).

• Uses the same game (with slight differences) to study similar

institutional factors.

• Game framed either neutrally, or as health-care scenario.
• Subjects are Ethiopian nursing students.
• Looks at whether behaviour by nurses may be driven by

professional norms or adapting to a corrupt environment.

3/22

Azfar and Nelson (2007)

Azfar and Nelson (2007): Game

• Election of president:
• Candidates randomly selected (3 in first round, 2 in later rounds

to compete with incumbent).

• Each gives 15 second election speech.
• One is elected by popular vote.
• Selection of attorney general:
• Either selected by president, OR
• Elected at same time as president (if same person elected for

both positions, runner-up in attorney general election gets position).

• Distribution of tiles:
• President secretely rolls dice and receives that number of

valuable tiles.

• Valuable tiles mixed with worthless tiles to add to 10, 14, or 22.
• President chooses six tiles to place in bag to be distributed to

voters.

4/22

Azfar and Nelson (2007)

Azfar and Nelson (2007): Game, continued

• Monitoring:
• The attorney general can choose up to four tiles to flip over and

see if they are valuable.

• The first two tiles can be flipped for free, the third costs $5,

and the fourth $10 (deducted from attorney general’s wages).

• Valuable tiles that are not exposed are worth $15 to the

president, exposed tiles are worth nothing.

• Each voter takes a tile from the bag: valuable tiles are worth

$30 to the voter.

• Wages:
• President: $30 in Low Wage treatment, $60 in High Wage.
• Attorney general: $20 in Low Wage treatment, $40 in High

Wage.

• Subjects played two 6-round games.
• Game played face-to-face (reduction in experimental control, but

increase in realism?).

5/22

Azfar and Nelson (2007)

Probability of exposure (Transparency)

• The impact of more effective monitoring institutions:
• Standardized accounting systems.
• Regular declaration of assets by public officials.
• In presence of “unexplained riches”, burden of proof transferred

to accused (Hong Kong, Singapore).

• Effectiveness of monitoring institutions is manipulated in this

experiment by varying the total number of tiles the “president” receives:

• Treatments with either 10, 14, or 22 tiles in total.
• The more tiles there are, the less likely an embezzled tile will be

found.

6/22

Azfar and Nelson (2007)

Separation of powers

• In most countries, attorney general is selected by executive

branch of government, so has weak incentives to investigate corruption by the executive.

• Alternatives may be better:
• Attorney general is directly elected in 44 US states.
• External oversight of police department (Hong Kong’s

“Independent Commission Against Corruption”) and customs (Singapore’s “Corrupt Practices Investigation Bureau).

7/22

Azfar and Nelson (2007)

Results: re-election of the executive

• Presidents exposed as corrupt less likely to be re-elected (5% if

corrupt; 32% if not corrupt).

• Higher revenues improve chances of re-election (rolling a 6

rather than a 1 improves probability of re-election by 35 p.p.).

• Re-election probability correlated more with total voter revenue

rather than individual revenue (subjects not informed about this, but guess from facial expressions?).

8/22

Azfar and Nelson (2007)

Results: vigilance of the attorney general

• Elected attorney generals flip more tiles.
• No impact of (High vs Low) wages.
• Fewer tiles flipped in 6th and 12th rounds (lower/zero

probability of participation in another election).

• Size of president’s roll increases vigilance (secret, but facial

expression?).

9/22

Azfar and Nelson (2007)

Results: executive corruption

10/22

Azfar and Nelson (2007)

Results: executive corruption

• Higher wages reduce corruption.
• Decreasing effectiveness of monitoring increases corruption.
• Method of selecting attorney general has no effect despite

greater vigilence (maybe requires more learning).

• Higher revenue (more valuable tiles) increases corruption.
• More corruption in 6th and 12th rounds (lower/zero probability
• f participation in another election).

11/22

Barr et al. (2009)

Barr et al. (2009): Game

• “Health worker” (equivalent of president) randomly selected

(subjects have equal probability in 1st round or if exposed as corrupt, otherwise incumbent has 50% chance of remaining).

• Selection of “Monitor” (equivalent of attorney general):
• Either randomly selected, OR
• Candidates randomly selected (2 in first round, 1 in later rounds

to compete with incumbent), then elected by majority of five remaining subjects (called “Community members” - equivalent

• f voters).
• Distribution of tiles:
• Health worker secretely rolls dice and receives that number of

valuable tiles.

• Valuable tiles mixed with worthless tiles to add to 10 or 18.
• Health worker chooses six tiles to place in bag to be distributed

to Community members.

12/22

Barr et al. (2009)

Game, continued

• Monitoring:
• The monitor can choose up to four tiles to flip over and see if

they are valuable.

• Each tile costs 5 Birr (currency in Ethiopia) to flip (deducted

from monitor’s wages)..

• Valuable tiles not exposed are worth 40 Birr to the president. If
• ne valuable tile is exposed, all retained tiles are worthless.
• Each voter takes a tile from the bag: valuable tiles are worth 60

Birr to the voter.

• Wages:
• Health worker: 20 Birr in Low Wage treatment, 60 Birr in High

Wage.

• Monitor: 60 Birr (in all treatments).
• Subjects played two 6-round games.
• Sessions were framed as above (and told tiles represented

bandages, drugs, etc.), OR framed neutrally (e.g. “Player P”).

13/22

Barr et al. (2009)

Experimental context (Ethiopian health sector)

• Health care professionals often work alone or in teams of two or

three in remote areas.

• Underpaid and poorly resourced.
• Often no monitoring, or monitoring by poorly resourced and

unmotivated local public officials.

• Interviews with Ethiopeian health workers suggest they

frequently expropriate drugs and other consumables (and charge extra fees and don’t show up to work).

• Some evidence that people enter sector in anticipation of these

“perks”.

• In these respects, Ethiopia is typical of many developing

countries.

14/22

Barr et al. (2009)

Subjects: Ethiopian nursing students

15/22

Barr et al. (2009) 16/22

Barr et al. (2009)

Results: embezzlement

• Embezzlement was significantly reduced by:
• Having an elected monitor.
• Higher monitoring effectiveness (i.e. 18 rather than 10 tiles).
• Higher health worker wages.
• Framing has no effect on average, but depends on age:
• An 18 year old keeps half a tile less when experiment is framed.
• A 38 year old keeps two tiles more when experiment is framed.

17/22

Barr et al. (2009) 18/22

Barr et al. (2009)

Results: monitoring

• Monitoring effort was significantly increased by having an elected

monitor (no other simple treatment effects were significant).

• Framing has no effect on average, but depends again on age and

years training:

• Younger subjects and subjects with fewer years training turn

more tiles when experiment is framed.

• Older subjects and subjects with more years training turn fewer

when experiment is framed.

• (No gender effects here or in embezzlement behaviour).

19/22

Barr et al. (2009)

Results: monitoring

Monitors who turn more tiles receive significantly more votes.

20/22

Conclusions

Conclusions

• More evidence that higher wages for public officials reduce

corruption (see first set of slides).

• More evidence that empowering victims of corruption to take

action is effective (see Transparency and Corruption slides).

• More effective monitoring can reduce corruption (what about

crowding out of intrinsic motivation?).

• Framing experiments can be useful when interested in a

particular context. The results of the second experiment regarding framing effects from both health worker and monitor roles suggest that:

• Younger nursing students may be particularly idealistic about

reducing or not engaging in corrupt behaviour in the workplace, BUT

• People adjust to the (corrupt) social norms of the health sector

environment.

21/22

Back 22/22