Gender, education and reciprocal generosity: Gender, education and reciprocal generosity: Evidence from 1,500 experiment subjects   Pablo Brañas-Garza, Universidad de Granada (Spain) Juan C. Cárdenas, Universidad de los Andes (Colombia) Máximo Rossi, Universidad de la República (Uruguay)

Motivation Previous papers have explore gender (“generous women”), education, … But they have very few data… and their econometric analysis use to be poor and too simple: for instance non parametric tests. and…. They have evidence for students only. This paper provides evidence about gender, education using an enormous dataset.

Unique dataset Data from experiments conducted under identical conditions (and monetary payoffs) in: Bogotá, Buenos Aires, Caracas, Lima, Montevideo and San José. 3,107 experimental subjects. Ordinary people recruited like in a survey. Controls by education, socio-economic level, income, gender and age. Economic incentives. The complete experiment was conducted by professionals .

Design Activity 1 (TRUST GAME): half of the sample as Player 1 and the remaining as Player 2. Activity 2 (VCM), Activity 3 ( RISK GAMES) , Activity 4 (RISK POOLING)  We focus on second movers in Activity 1 (receivers the SMUG)

Strategy Method Trust Game 1st movers SENT a % of the pie (endowment): 0, 25%, 50%, 75% and 100%. The amount of money is tripled 2nd mover receives the money (3 times % sent) 2nd mover returns 0%, 25%, 50%, 75%, 100%

Classical measures ARi (Average return) is the player 2 mean behavior, that is, the [∑inreturn_i/n]; for n=1,…,5. σi (Sensibility or variance) how generosity varies across the periods.

New measures PEi (Price effect) when generosity in the last period (with larger pie) is larger, that is, return5>return1. SPEi (Strict price effect)when player 2 never decreases across all the periods, that is, when return5 ≥ return4 ≥ return3 ≥ return2 ≥ return1 NRRi (Non-reciprocal generosity) when player 2 shows positive rates of generosity even when he does receive nothing, that is, if return1>0.

Behavioral measures Players 1 investment SURPLUS Players 2 return free ride sharing exact Players 2 return

I1 [Free-ride] number of times individual return is less than original player 1 investment I2 [Exact] number of times that player 2 returns player 1’s investment (then player 2 keeps the entire surplus for herself). I3 [Sharing] number of times that player 2 shares the surplus with player 1.

Variables Player’s expectations about the other players’ behavior. e(money) Gender (Female=1). Race (White=1; Black=1; Indig=1) Age (Age and Age2) and, Years of Education Married or Single. Relative wealth Financial exclusion Heads of households as Chief=1. ** Subjects are asked: "Imagine a 10-step ladder in which the poorest house- holds in your city are on step 1 and the wealthiest are on step 10. Where would you position your own household?"

Variables (cont.) As regards race, our dataset was composed of white (56.4%), indigenous (2.1%), mixed race (39.17%) and black (2.3%). Our reference group was always mixed race. For instance, there are differences in education among cities. The average value in Bogotá is 11.2 years, in Buenos Aires 13.4, in Caracas 12.9, in Lima 12.1, in Montevideo 10.9 and in San José it is 9.9 years. The complete distribution of marital status is as follows: unmarried couples (15.9%), married (28.9%), widowed (3.5%), divorced (11.1%) and single (40.4%).

results

Types: summary Expectations are crucial: people who are more optimistic are more disposed to share; those who are less optimistic are more disposed to free-ride Men are likely to share the surplus and women are inclined to return the investment (but keep the surplus) Education enhances pro-social behavior Responsibility makes males less generous

Monetary return: Summary As Andreoni & Vesterlund (2001) maintain, gender bias appears –in favor of more generous men- for larger pie sizes (6,000, 9,000 and 12,000), that is when monetary costs are proportionally cheaper. Expectations are crucial: individuals who are more optimistic return more money. Education plays a determinant role in reciprocal behavior: better education makes subjects return very little (much) when they receive very little (much) of the pie. In addition, better education prompts people to punish anti-social behaviors. The fact of being single seems to be relevant but the effect for women is different from that for men in that females are more generous and males more selfish. Responsibility has an effect for males: men who support a family are less generous.

Replication Why previous research is wrong?

Replication (cont.) We repeated the entire analysis but limited the sample to subjects with at least 12 years of education, that is to individuals who had completed at least the first year of undergraduate studies. Secondly, we restricted the subject pool by age in that we specifically explored subjects only during their university years (17<years<25).

Replication (cont.’) The replication for the reduced sample yields a surprising result, which is that women are not more disposed to free-ride. Men are more willing to share the surplus but women tend to return the complete investment to player 1. This result is confirmed in the second part of the analysis shown in table A1. When we explored the return ratio for each scenario (0x3, 6,000x3 and 12,000x3) we found that gender bias vanishes! Only a weak effect is found for the intermediate case.

Replication (cont.’’) Using the sample of 18-24 year-old subjects (N=374) we made 250 random draws of sub-samples of 100 individuals.

Discussion Levitt and List (2007) Our subjects were recruited in the streets. heterogeneity among participants: they would find difficulties in guessing their partner’s behavior (1st and 2nd order beliefs) given that all of them are “real” strangers. we did not use computerized rooms (neither PC’s), we have monitors explaining the complete procedures (and assisting people with very low education level), we use paper & pencil, etc.

Main Results The role of beliefs in reciprocal behavior. We consistently found that optimism is positively correlated to sharing and negatively to free- riding. Therefore optimism enhances reciprocity. We showed that women are more selfish than men when it comes to sharing a surplus, but fairer when paying back an initial investment.

Main Results II We found that selfish and generous behavior in females and males is motivated by different reasons. Individual level of education enhances reciprocal behavior. Better- educated subjects are kinder to people who were kind first. In addition, we found that better-educated subjects punish selfish behavior more. There was no gender bias in these findings.