Economics of Culture and Institutions

April 11, 2015
Alberto Bisin of New York University and Paola Giuliano of the University of California, Los Angeles, Organizers

Oded Galor, Brown University and NBER, and Ömer Özak, Southern Methodist University

The Agricultural Origins of Time Preference (NBER Working Paper No. 20438)

Galor and Özak explore the origins of the distribution of time preference across regions. They advance the hypothesis, and establish empirically that geographical variations in the natural return to agricultural investment have had a persistent effect on the distribution of time preference across societies. In particular, exploiting a natural experiment associated with the expansion of suitable crops for cultivation in the course of the Columbian Exchange, the research establishes that pre-industrial agro-climatic characteristics that were conducive to higher return to agricultural investment triggered selection and learning processes that had a persistent positive effect on the prevalence of long-term orientation in the contemporary era.

Daron Acemoglu, MIT and NBER, and Matthew Jackson, Stanford University

Social Norms and the Enforcement of Laws (NBER Working Paper No. 20369)

Acemoglu and Jackson examine the interplay between social norms and the enforcement of laws. Agents choose a behavior (e.g., tax evasion, production of low-quality products, corruption, harassing behavior, substance abuse, etc.) and then are randomly matched with another agent. There are complementarities in behaviors so that an agent's payoff decreases with the mismatch between her behavior and her partner's, and from overall negative externalities created by the behavior of others. A law is an upper bound (cap) on behavior and a law-breaker, when detected, pays a fine and has her behavior forced down to the level of the law. Equilibrium law-breaking depends on social norms because detection relies, at least in part, on whistleblowing. Law-abiding agents have an incentive to whistle-blow on a law-breaking partner because this reduces the mismatch with their partners' behaviors as well as the negative externalities. When laws are in conflict with norms and many agents are breaking the law, each agent anticipates little whistle-blowing and is more likely to also break the law. Tighter laws (banning more behaviors) have counteracting effects, reducing behavior among law-abiding individuals but inducing more law-breaking. Greater fines for law-breaking and better public enforcement reduce the number of law-breakers and behavior among law-abiding agents, but increase levels of law breaking among law-breakers (who effectively optimize their behavior conditional upon matching with law-breakers). Within a dynamic version of the model, the researchers show that laws that are in strong conflict with prevailing social norms may backfire, while gradual tightening of laws can be more effective in influencing social norms and behavior.

Daniel Chen, ETH Zurich, and Susan Yeh, George Mason University

How Do Rights Revolutions Occur? Free Speech and the First Amendment

Does law shape values? Chen and Yeh test a model of law and norms that predicts when law has backlash or expressive effects. Several Supreme Court decisions coincide with the onset of sexual liberation. Since 1958, Democrat judges have been 10% more likely than Republicans to favor progressive free speech standards. Using the random assignment of U.S. federal judges setting geographically-local precedent, the researchers estimate that progressive free speech standards liberalized sexual attitudes and behaviors and had secondary effects on sex-related crimes and diseases. The authors distinguish expressive from deterrence effects by assigning 1,345 data-entry workers to transcribe news reports about free speech decisions. Exposure to progressive decisions liberalized sexual attitudes and shifted norm perceptions but not self-reported behavior. These results verify the model and help explain a reversal of free speech law's effects from backlash to expressive during the sexual revolution.

Yihui Pan, University of Utah; Stephan Siegel, University of Washington; and Tracy Yue Wang, University of Minnesota

The Cultural Origin of Preferences: CEO Cultural Heritage and Corporate Investment

Does culture shape risk preferences? In this study, Pan, Siegel, and Wang exploit variation in cultural heritage across CEOs of public U.S. companies and document a significant association between CEOs' culturally transmitted risk preferences and corporate investment decisions. A one standard deviation increase in a CEO's uncertainty avoidance, a proxy for risk preference, is associated with a 17% reduction in acquisitions and a 7% reduction in capital expenditures (Capx). The researchers findings are robust to controlling for economic and institutional differences as well as genetic differences across countries of origin and do not depend on first-generation immigrant CEOs. CEOs' risk preferences seem to have a causal influence on acquisitions, while the association between CEO risk preferences and Capx is largely explained by firm-CEO matching. The results provide novel evidence of important social transmission of risk preferences, their effect on corporate investment policies, and the interplay of the culturally transmitted preferences of CEOs, corporate boards, and other top executives.

Eugenio Proto and Andis Sofianos, Warwick University, and Aldo Rustichini, University of Minnesota

Higher Intelligence Groups Have Higher Cooperation Rates in the Repeated Prisoner's Dilemma

Systematic differences in cooperative behavior in social groups, or cities or neighborhoods, might emerge as outcome of average or typical personality characteristics of the groups, like conscientiousness or intelligence. Proto, Sofianos, and Rustichini provide a systematic study of the link between typical intelligence in a group and cooperative behavior of the members in an experiment where two groups of subjects with different levels of intelligence, but otherwise similar, play a repeated prisoner's dilemma. The two groups were created in the laboratory with careful choice of the subjects in the two pools on the basis of an IQ test (Raven Advanced matrices) administered in a separate session two days before the main treatment. The initial cooperation rates are similar, but it increases in the groups with higher intelligence to reach almost full cooperation, while declining in the groups with lower intelligence. The difference is produced by the accumulation of small but persistent differences in the response to past cooperation of the partner. In higher intelligence subjects, cooperation after the initial stages is immediate and becomes the default mode, defection instead requires a longer decision time. In lower intelligence groups this difference is absent, both decisions take similar time, longer than for the higher IQ group. Since cooperation rates are similar in the initial stages, cooperation of higher intelligence subjects is learned, rather than automatic. That higher intelligence groups are not unconditional cooperators is also demonstrated by our additional finding that in a treatment with lower continuation probability there is no difference between different in behavior between the two groups.

Giovanni Mastrobuoni, University of Essex, and Daniele Terlizzese, Bank of Italy

Rehabilitating Rehabilitation: Prison Conditions and Recidivism

Mastrobuoni and Terlizzese use quasi-random variation in the fraction of time served in the Italian "open-cell prison" of Bollate to estimate the effect of rehabilitation efforts on recidivism. The researchers deal with the endogeneity of rehabilitation assignments by focusing on those sources of variability in the length of exposure to rehabilitation efforts that are plausibly unrelated to recidivism. The authors' most stringent test restricts the analysis to inmates who are displaced to Bollate due to overcrowding in nearby prisons, controlling for measures of observed (based on a revealed preference argument) and unobserved potential selection. Spending one more year at the rehabilitating prison (and one less year at an ordinary one) reduces recidivism by around 10 percentage points. For the group of displaced inmates, which is shown to be negatively selected, the effects of rehabilitation efforts on recidivism are larger. While the researchers find evidence that over time Bollate inmates become more likely to work outside the prison, more than a single mechanism seems to underlie these effects.

Jeffrey Butler, Einaudi Institute for Economics and Finance; Pierluigi Conzo, University of Turin; and Martin A. Leroch, University of Mainz

Social Identity and Punishment

Third party — or, bystander — punishment is crucial for sustaining cooperative behavior. Through laboratory experiments, Butler, Conzo, and Leroch investigate the interaction between group identification and a bystander's punishment preferences by inducing minimal groups and giving a bystander the opportunity to levy a fixed amount of punishment on the perpetrator of an unfair act towards a defenseless victim. The researchers elicit the bystander's valuation for punishment in four cases: when the perpetrator, the victim, both or neither are members of the bystander's group. For predictions, they construct two separate frameworks differing by whether the primary effect of group identity is to create an empathetic bond between in-group members or to affect the weights placed on others' money earnings in a distributional social preferences model. The two frameworks yield starkly different ordinal predictions about the bystander's value for punishment across two cases: i) when the perpetrator and victim are both members of the bystander's group; ii) when only the victim is an in-group member. The empathetic bond framework predicts that punishment will be more highly valued in the latter case, while the distributional preferences framework suggests the opposite. The authors' data support the predictions of the first. Finally, they conduct control sessions where groups are not induced and find that bystanders tend to treat others as in-group members unless specifically divided into distinct groups.