Natural Disasters and Social Preferences: The Effect of Tsunami-Memories on Cheating in Sri Lanka

Abstract

This study analyses how past tsunami-memories affect cheating in Sri Lanka. Subjects are assigned to a treatment (control) group in which they watch a video about the calamity before (after) participating in a trust game. Cheating is elicited by asking trustors how much they need to receive not to feel cheated and trustees how much they need to return not to make the trustor feel cheated. Finally, participants report whether the video mostly reminded them about solidarity, looting or the calamity. Trustors show lower cheating standards and trustees more often satisfy the trustors’ cheating notion if they mostly recall solidarity.

Acknowledgement

I thank Francesco Amodio, Maria Bigoni, Stefania Bortolotti, Jeffrey V. Butler, Colin Camerer, Vincenzo Carrieri, Marco Casari, Alessandra Cassar, Marcello D’Amato, Giacomo Degli Antoni, Christopher J. Flinn, Giulia Fuochi, Pauline Grosjean, Fabio Landini, Eliana La Ferrara, Martin A. Leroch, Ethan Ligon, Karen Macours, Ugo Merlone, Matteo Migheli, Dilip Mookherjee, Stefan Napel, Ernesto Rezk, Robert Sugden, Sigrid Suetens, Alessandro Tarozzi, Bertil Tungodden, Mark Willinger and the participants to the 2015 Meeting on Experimental and Behavioral Social Sciences (Toulouse), the GRASS VIII Meeting (Torino), the 2014 Summer School in Development Economics (Ascea), the 54th Meeting of the Italian Economists Society (Trento), the International Herbert A. Simon Society Workshop (Torino), the 11th Young Economists’ Workshop On Social Economy (Bologna) and the 2013 Symposium on Economic Experiments in Developing Countries (Bergen) for useful comments and suggestions. I gratefully acknowledge Angelico, C. Pagano, E. Agostino and N. Kurera for the precious support to the entire data-collection process. Etimos Foundation, Etimos Lanka and AMF are fully acknowledged for the financial and logistic aid. Data and codes will be provided by the author upon request.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. A series of experimental results suggest that emotions influence reciprocity (Kirchsteiger, Rigotti, & Rustichini, 2006), altruism (Mellers, Haselhuhn, Tetlock, Silva, & Isen, 2010), trust (Dunn & Schweitzer, 2005; Kausel & Connolly 2014) and fairness (Bonini et al., 2011). From a psychological point of view, recent experimental studies (Halali, Bereby-Meyer, & Meiran, 2014) identify the theoretical basis of these findings in the association of emotions with the automatic and impulsive decision system which, by mutually interacting with the reflective decision system, shapes social behaviour (Posten, Ockenfels, & Mussweiler, 2014; Strack & Deutsch, 2004). For an overview on the dual-process theory see Evans (2008).

2. After data-cleaning the sample size reduced to 386 observations because four participants failed to complete the entire interview process.

3. As confirmed by the AMF staff, the selected sample is not likely to be affected by post-tsunami migration since soon after the calamity damaged (and, indirectly, non-damaged) individuals received incentives to stay in the form of i) incoming flows of recovery aid and ii) concession of micro-loans at favourable conditions. In fact, AMF’s loan-portfolio suffered severe losses because of the insolvency of tsunami-affected individuals. However, soon after the calamity it was recapitalised in an effective way as shown by Becchetti and Castriota (2010, 2011)). Empirical evidence on the absence of out-migration from damaged areas is also provided by Paul (2005) who found that victims of the 2004 tornado in Bangladesh did not migrate since they received more disaster relief (in monetary terms) than the damage they incurred.

4. There are no effects of the treatment/control assignment and the consequent memory activation on behaviour in the lottery and the risky investment game. Results are omitted for reasons of space and consistency with the main focus of this paper but are available upon request.

5. The wording was kept neutral in all games in order to avoid frame effects. For instance, the game was never presented as a ‘trust game’, but rather denominated ‘TG’. Roles were phrased as ‘player 1’ and ‘player 2’ respectively for TR and TE.

6. First order beliefs (FOBs) for TR and TE – that is the amount expected from the other player – were also elicited through money incentivised questions (see Supplementary Material A.2).

7. The video is available at the following link: https://drive.google.com/file/d/0Bx_sVRhc-2PaUTVwNlhOcmI3YlE/ edit?usp = sharing. An English version of the script is reported in the Supplementary Material C.

8. Eliciting subjects’ mood after a movie is not a novel methodology in experimental psychology. Andrade (2005) and Andrade and Ariely (2009) exposed participants to a five minute clip followed by a task in which they were asked to describe a personal experience related to the movie; participants in their setting were assigned to angry or happy affect-inducing treatments. Closely related to the current study is the work by Shahrabani et al. (2009). The authors analyse the role of recalled emotions about the 2006 Israel-Lebanon war on perceived risks and manipulate the former using a short videoclip that recalled the war. Västfjall et al. (2008) find that the affect elicited by reminding Swedish undergraduates of the 2004 tsunami disaster negatively influences their judgments of wellbeing, future optimistic thinking and risk perceptions. Importantly, they assume that recalling tsunami triggers negative feelings; this, however, may not always be the case if victims’ social preferences are emotionally affected by experiences of solidarity under the form of recovery aid (Becchetti et al., 2017).

9. The variable Avg Pr(Return ≥ TE cheat) has been built by generating an indicator equal to one when TE returned in each strategy choice no less than the expected TR’s cheating threshold (that is variable Pr(Return ≥ TE cheat)); then the indicator has been averaged over all 11 TE’s choices so to have the TE’s mean (strategic) propensity to play as a ‘reliable agent’ or, adopting the definition by Butler et al. (2016a), as ‘non-intentionally cheaters’.

Additional information

Funding

This work was supported by the ETIMOS Foundation.