Online Appendix

Appendix A: Additional Results

Figure A1: Corruption Behavior of the Firms. Results for each round.

  1. Extensive Margin
/
  1. Intensive Margin

Notes: Created by authors using experimental data. Panel A refers to the percentage of individuals acting as firms who offer a bribe to a public official. Panel B includes only individuals who offer a bribe. The figures of 5% and 25% monitoring refer to the probability of getting caught.

Figure A2: Corruption Behavior by Public Officials. Results for each round.

  1. Extensive Margin
/
  1. Decision to grant permit

Notes: Created by authors using experimental data. Panel A refers to the percentage of individuals acting as public officials who accept a bribe from a firm. Panel B refers to the percentage of public officials who decide to grant a permit. The figures of 5% and 25% monitoring refer to the probability of getting caught.

Figure A3: Punishment Points by Treatment Group

Notes: Created by authors using experimental data. Vertical axis shows the average units used as punishment by leaders in all rounds. The figures of 5% and 25% monitoring refer to the probability of getting caught. Leader refers to a public goods game with centralized punishment, while counter-punishment refers to a public goods game with centralized punishment and counter-punishment of leaders allowed. Error bars refer to 95 percent confidence intervals.

Figure A4: Earnings by Treatment Group

Notes: Created by authors using experimental data. Vertical axis shows the average earnings in all rounds. The figures of 5% and 25% monitoring refer to the probability of getting caught. Cooperation refers to the standard public goods game. Leader refers to a public goods game with centralized punishment, and counter-punishment refers to a public goods game with centralized punishment and counter-punishment of leaders allowed. Error bars refer to 95 percent confidence intervals.

Table A1: Descriptive Statistics of Covariates

Covariates / Corruption Treatment / N / Mean / Standard Deviation
Control / 36 / 0.64 / 0.49
Sex / Low Monitoring / 64 / 0.66 / 0.48
High Monitoring / 64 / 0.67 / 0.47
Control / 36 / 18.94 / 1.77
Age / Low Monitoring / 64 / 18.8 / 1.85
High Monitoring / 64 / 18.61 / 1.95
Control / 30 / 21.66 / 13.31
Income / Low Monitoring / 56 / 21.1 / 16.36
High Monitoring / 52 / 17.81 / 14.31
Control / 30 / 2.25 / 1.05
Father’s education / Low Monitoring / 64 / 2.55 / 1.27
High Monitoring / 64 / 2.27 / 1.14
Control / 30 / 2.3 / 0.86
Mother’s education / Low Monitoring / 64 / 2.34 / 1.16
High Monitoring / 64 / 2.25 / 1.19

Notes: Created by authors using experimental data. The covariates used are the sex of each participant, age in years at the date of the experiment in years, average household monthly income in thousands of Mexican pesos, and parents' education. Education has a value of 0 for primary or less, 1 for junior high school, 2 for high school, 3 for college, 4 for master's degree, and 5 for doctoral degree. Income data are restricted to exclude 26 observations with extreme values over 100,000 Mexican pesos.

Table A2: Regression Results for Corruption

Pr(Offer Bribe) / Avg Bribe / Pr(Accept Bribe) / Pr(Grant Permit)
[1] / [2] / [3] / [4]
High Monitoring / -0.175 / 0.188 / -0.137 / 0.012
[0.067] / [0.049] / [0.035] / [0.085]
N / 320 / 132 / 320 / 320

Notes: Each column shows a regression using robust standard errors clustered by session. Probit regressions are shown in columns 1, 3, and 4 with marginal effects at the mean of the covariates. Tobit regression is shown in column 2 with marginal effects at the mean of the covariates and restricted to individuals who offered a bribe.The coefficient is percentage of maximum bribe allowed. All regressions include the following controls: round, age and age squared, dummy of female, educational levels of father and mother (0 for primary or less, 1 for junior high school, 2 for high school, 3 for college, 4 for master's degree, and 5 for doctoral degree.)

Table A3: Summary of Results for Cooperation by Treatment Status

Cooperation Treatment / Corruption Treatment / N / Mean / Standard Deviation
Control / 20 / 14.26 / 5.03
Standard PG / Low Monitoring / 24 / 10.06 / 5.05
High Monitoring / 24 / 11.71 / 4.75
Control / 12 / 13.58 / 3.48
Leader PG / Low Monitoring / 20 / 13.05 / 3.38
High Monitoring / 20 / 15.93 / 3.51
Control / 4 / 15.43 / 4.93
Counter-Punishment PG / Low Monitoring / 20 / 10.67 / 3.98
High Monitoring / 20 / 13.23 / 3.91

Notes: Created by authors using experimental data. Cooperation Treatment refers to the type of game played by participants, either a standard public goods game (Standard PG), a public goods game with centralized punishment and without counter-punishment (Leader PG), or a public goods game with centralized punishment and counter-punishment (Counter-Punishment PG). Corruption Treatment refers to the type of corruption game played in the first stage, either Low Monitoring, High Monitoring, or as a Control with no exposure to corruption. N is the number of participants in each scenario. Mean represents average contribution to the public good (in ECU) for all rounds of each scenario. Standard deviation represents the standard deviation of contribution to the public good for all rounds of each scenario.

Table A4: Regression Results for Cooperation

[1] / [2] / [3] / [4]
Low Monitoring / -2.628** / -2.457** / -3.543***
[0.776] / [0.708] / [0.532]
High Monitoring / -0.862 / -0.733 / -2.078
[1.272] / [1.163] / [1.733]
Leader / 1.567** / 1.643** / -1.268
[0.445] / [0.434] / [0.899]
Counter-punishment / -1.117** / -0.748* / 0.297
[0.375] / [0.313] / [0.976]
Low * Leader / 3.285***
[0.450]
High * Leader / 4.477**
[1.537]
Low * Counter-punishment / -0.756
[0.691]
High * Counter-punishment / -1.095
[1.868]
N / 820

Notes: Each column shows a regression using robust standard errors clustered by session, an OLS regression of the cooperation levels for five rounds in the second stage of the experiment. Low and High Monitoring are dummy variables for groups with their respective levels of monitoring in the corruption game in the first stage. Leader and Counter-punishment represent dummy variables for the treatments in the second stage of the experiment, with Leader including centralized punishment, and Counter-punishment including centralized punishment with the option of counter-punishment of the leader. The last four variables are the interactions for the two treatments in the game. Covariates for the four models arethe respective round, participants’ sex, age and age squared, and parents' level of education. Omitted variables in the regression arethe control group for the corruption game and the standard public goods game in the second stage.

Table A5: Regression Results for Punishment

[1] / [2] / [3] / [4]
Low Monitoring / -0.196 / -0.193 / -0.472
[0.491] / [0.479] / [0.668]
High Monitoring / -0.275 / -0.222 / -0.523
[0.456] / [0.459] / [0.465]
Counter-punishment / -0.515** / -0.505** / -1.447***
[0.113] / [0.120] / [0.285]
Low * Counter-punishment / 1.173*
[0.521]
High * Counter-punishment / 1.197*
[0.540]
N / 96 / 96 / 96 / 96

Notes: Each column shows a regression using robust standard errors clustered by session, an OLS regression of the leaders’ punishment levels for four rounds in the second stage of the experiment. Low and High Monitoring are dummy variables for groups with their respective levels of monitoring in the corruption game in the first stage. Counter-punishment is a dummy variable for the treatment in the second stage of the experiment, with centralized punishment and the option of counter-punishment of the leader. The last two variables are the interactions for the two treatments in the game. Covariates for the four models arethe respective round, participants’ sex, age and age squared, and parents' educational level. Omitted variables in the regression arethe control group for the corruption game and centralized punishment in a public goods game for the second stage.

Appendix B: Instructions

Instructions

General instruction for the corruption game (the experimenter read the instructions to all participants)

Welcome, you’re going to be part of a decision-making experiment. The amount you’re going to receive depends on the decisions you choose, so please follow carefully the next instructions. Each one of you will be paid in private and in cash at the end of the experiment. To all participants, besides the gains you earn by playing, you’ll receive a show-up fee of 20 pesos, just for having assisted today.

All over the experiment you’ll gain experimental currency units (ECU), which will serve to be converted in pesos at the end of the experiment. Therefore, the more experimental units you gain, the more money you'll have in the end. Communication is not allowed during the experiment between the participants, until the end of the session. If anybody has a question, please raise your hand and one member of the staff will attend you in private. Those participants that break the rules will be automatically disqualified. The session consist of two different stages, so you must be careful for each session's instructions.

At the beginning of the experiment, each one of you will receive two envelopes: one contains all the material needed to complete the experiment (yellow envelope), and the other one will be used to deliver your decisions and to receive the final results at the end of each period (white envelope). Be careful to keep safe both envelopes. Both envelopes have been marked with your personal ID number that you'll use at the end of the session to claim your payoff.

Stage 1 (read out loud by the experimenter to all participants)

Each stage of the experiment will be played by rounds. The first stage will be played by couples, formed randomly on each round, so no participant will know the identity of his partner, and no couple will play more than once on this stage.

The stage starts by considering a situation between a firm and a public official and both represent a couple for the game. The firm is interested on installing a new industrial factory that causes a damage for the environment and to the public. Additionally, the public official is in charge to give the no-pollution certificates to all new factories, therefore, he must decide to grant (or not) to this new factory. Before the public official have taken his decision, the firm can offer him a bribe to encourage him to give it the certificate.

There are two important points to remark: first, if a firm offers a bribe and the public official accepts the bribe, the latter is not forced to grant the certificate to the firm, that is, the public official can take the bribe and do not deliver the certificate. Second, if the public official accepts the bribe, there is a chance that the transfer is discovered and both players would be punished.

Now you can open the yellow envelope where it's specified the role you're going to play for the rest of this stage and read the next instructions.

Decisions step by step (individual sheet of instructions contained in each yellow envelope)

Step 1. The firm must decide whether to offer or not a bribe to the public official. If it decides to offer a bribe then the firm pays 2 ECU and the game continues in step 2. If the firm decides not to offer a bribe then the game continues in step 4.

Step 2. If the firm decides to offer a bribe, then it must choose the amount to offer, that could go from 1 to 9 (integers) ECU. Next, the period continues at step 3.

Step 3. The public official must decide to accept or rejects the bribe from the firm.

  • If the public official accepts the bribe, the amount of the bribe is reduced from the firm's account. This amount is multiplied by three and it's added to the public official's account. Right away, a random number between 1 and 20 will be selected by the experimenter, where:
  • If the number chosen is 1 then the couple of players will be disqualified and both lose automatically their earnings of the period,
  • If the number chosen is 2, 3, ..., 20, then the game continues in step 4,
  • If the public official rejects the bribe then the accounts of both players remain equal (although the firm already paid the 2 ECU for offering the bribe). The game continues in step 4.

Step 4. The public official must decide whether to grant or not the certificate to the firm.

  • If the public official don't grant the certificate to the firm, then theaccounts of both players increase by36 ECU.
  • If the public official grants the certificate to the firm, the firm's account increases by 56 ECU, while the public official's account increases by 30 ECU.

After step 4 the round is over. You should introduce your answer sheet into the white envelope and pass it to the experimenters when needed. Next, the experimenters will capture the answers from all participants, write the outcome for everyone and deliver you the profit from the round. After this, a new round will start and you'll be matched to a new anonymous partner.

Payment for the corruption game

(Instructions read out loud by the experimenter after each player has read the last sheet individually)

At the end of this stage, one round will be randomly selected for all players and everybody will be paid according to that round's outcome. The exchange rate on this stage is 0.40 pesos by each ECU. So, if in the selected round you gain 100 ECU you'll receive 40 pesos. You have one minute to respond in each round.

Decision table and some examples (read and presented by the experimenter to all participants)

To collect the information on each round, the players on the firm's role must write in the selected answer sheet if they offer (or not) a bribe, and if they offer it, specify the bribe's amount (The firm's answer sheet is presented to all players).

For those in the public official's role, you should fill in the next table (the corresponding table is shown). For each possible bribe you must specify whether to accept or not the bribe offered and whether to grant or not the certificate to the firm. For each decision you choose you'll see the corresponding profit at the end of the row.

In each round the experimenters will match the decisions for each couple of firms and public officials and will establish the outcome of the round. In order to check that all instructions are clear we'll see some examples: (we show an example of the firm's table and another one of the public official).

Example 1: we suppose the firm decides to not offer a bribe; so, a public official choose to not grant the certificate in this case. So, the final outcome is that both, the firm and the public official get 36 ECU each, which equals 14.4 pesos.

Example 2: we assume the firm offers a bribe of 5 ECU. The public official choose to not accept a bribe of 5 ECU, but decided to grant the certificate for the firm. In the end, the payoff for the firm is 54 ECU while that of the public official equals 30 ECU.

Example 3: we assume now that the firm decides to offer a bribe of 5 ECU. The public official choose to accept a bribe of 5 ECU. However, the random number selected determines that the illegal transfer is discovered and both players lose their profits.

Example 4: finally, we assume the firm offers a 6 ECU bribe. The public official choose to accept the bribe (and the transfer is not discovered). Additionally, the public official decided to not grant the certificate to the firm, which means that the firm's payoff equals 28 ECU and the public official's payoff is 54 ECU.

Now you can start answering the first round of the game.

Treatment 1: public good game (instruction read out loud to all participants)

Now we'll randomly form groups of 4 members where no one will know the identity of their colleagues. Each group will play together for the rest of the experiment. Each member of your group played the same game as you played before, so each group is randomly formed by firms and public officials that decided to offer (accept) or not a bribe.

On this part of the experiment each player has an endowment of 20 ECU. You can either assign an amount of the endowment to your private account or invest a part of it into an investment project. Each ECU not invested in the project will be automatically saved in your private account. Each ECU saved in your private account will be added to your round's payoff, and each ECU invested in the project will be summed with those of your partner’s investments, multiplied by two and divided equally among all members of your group.

Earnings by your private account (20 ECU minus your investment to the project)

+ Earnings by your investment in the project (0.5 x Sum of all investments in your group)

= Your round's total earnings

Therefore, your earnings by round correspond to the sum of the earnings by investing on the common project plus the amount saved on your private account.

As we stated before, the identity of your partners will be anonymous and you will play together for the rest of the experiment. You must write your investment decision to the project in the sheets contained on the yellow envelope where is marked your group and your member number (you should write your answer according to your group's member number). So, the steps to follow are these:

  1. You must write in the answer sheet your contribution to the project (the amount not invested will be saved in your private account), and you should pass your decision to the experimenters inside the white envelope.
  2. After this, the experimenters will capture all participants decisions,
  3. Then, your answer sheet will be given back to you with the information of your total earnings of the round and the investments made by your group's partners.
  4. Finally, the round ends and a new round will start.

Next we present to you some examples to illustrate the exercise: