Socio-economic level, neighborhood segregation and determinants of reciprocity: evidence using representative artefactual data from Latin American cities

ABSTRACT

We study whether urban segregation is linked to social capital and, in particular, whether socio-economic level, measured by neighborhood of provenance, is detrimental to the formation of reciprocity. We employ representative data for six Latin American cities, an underrepresented region in terms of experimental research. Our main findings provide robust evidence that individuals with higher socio-economic level increasingly reward larger levels of trust in comparison to lower socio-economic level participants.

KEYWORDS:

• Social capital
• reciprocity
• socio-economic level
• Latin America

JEL CLASSIFICATION:

• C90
• D01
• O12

Supplemental data

Supplemental data for this article can be accessed here.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. What defines socio-economic status in Latin America is a rather difficult question that researchers have tried to answer for a long time. While the specifics depend on each city, overall, it tends to be determined by a combination of race, income, family historical background and education (Angotti 2013).

2. Contrary to the conventional wisdom, there is no evidence to support the hypotheses of a negative effect of religion on social capital (Brañas‐Garza, Rossi, and Zaclicever 2009).

3. Trust constitutes one of the essential pillars for social interaction development and effective economic functioning (Tonkiss 2009; Yamagishi 2011). Higher levels of trust have been associated with greater financial development, more efficient judicial structures, lower corruption, and, in general, greater economic success (e.g. Fukuyama 1995; Knack and Keefer 1997). Given its importance for society, multiple efforts have aimed at measuring and exploring what motivates trust using mostly experimental approaches.

4. Some researchers argue that the amount sent may also reflect the first player’s altruism towards the second player, independent of the first player’s trusting (e.g. Cox, Sadiraj, and Sadiraj 2008; Ashraf, Bohnet, and Piankov 2006).

5. Interestingly, no effect was found when analyzing income in absolute terms.

6. When the authors exclude the laboratory sample, and only keep external representative sample, holding a job no longer shows a significant link with reciprocity.

7. The data are available at https://www.iadb.org/en/about-us/departments/res and come from a broader research effort funded by the Inter-American Development Bank. This section draws from previous research that also describes both this same experimental design as well as the data collection process (Cardenas et al. 2013, and Chong, Rios-Salas, and Nopo 2020). This is done so that this article can remain self-explanatory.

8. Additional information on the canvassing process, the identification of the neighborhoods of each city and other related aspects of the city grid are provided in Candelo and Polania-Reyes (2008).

9. Admittedly, our findings may be viewed more about the community exposure rather than individual characteristics. We are grateful to an anonymous referee for this comment.

10. Each session held three different experimental games applied in the following order: a Trust Game, a Voluntary Contribution Game and a Risk Game (Chong, Rios-Salas, and Nopo 2020).

11. People on average send half of the initial endowment to player 2. The net return from player 2 is typically positive and between 10% to 20% of what was sent initially (Carpenter, Harrison, and List 2005; Ashraf, Camerer, and Loewenstein 2005).

12. We run an ordered Probit using neighborhood socio-economic level classification as the dependent variable and household expenditures as the independent variable, while controlling for socio-economic covariates and city of residence. We obtain statistically significant marginal effects for each of the values of the expenditure quarter and monotonic predictions across quarters. Similarly, simple correlation between the two categorical variables is 0.4 and is statistically significant at conventional levels.

13. Not all participants reported monthly household expenditures, which results in a significant decrease in our sample size Summary statistics of these and all the variables reported in this paper are presented in.

14. We also take advantage of additional data, including information from the other experimental exercises pursued during the sessions. Please, see Footnote 10.

15. In we provide additional graphical reference using the mean estimated return ratio when testing our most comprehensive specification Column 8 from which gives an approximation of the aggregate return function for each socio-economic level. We show a stepper slope for high socio-economic level individuals and slightly lower intercept. Using ordinary least squares, we estimate the slope of the return strategy of players 2, which Ciriolo (2007) and Altmann et al. (2008) use as an alternative measure of reciprocity. We find that high and middle socio-economic levels yield a positive link and is statistically significant with respect to the steepness of the slope, providing additional evidence that socio-economic level association with reciprocity depends on the trust provided. Using household expenditure as socio-economic level construction results similar results.

16. As shown in returned units in absolute values do increase along higher levels of trust.

17. We include well-accepted social capital controls including: (i) Participation in a social organization, (ii) Decision making in a social organization; (iii) Hours spent in a social organization; (iv) Pro-social index based on the agreement degree with five pro-social affirmations; (v) Voting in the last election.

18. This is shown in.

19. As shows, the coefficient of our education variable is not statistically significant at conventional levels.