CULTURAL VALUES AND BEHAVIOR IN DICTATOR, ULTIMATUM, AND TRUST GAMES: AN EXPERIMENTAL STUDY
Sun-Ki Chai, Dolgorsuren Dorj and Katerina Sherstyuk
ABSTRACT
Culture is a central concept broadly studied in social anthropology and sociology. It has been gaining increasing attention in economics, appearing in research on labor market discrimination, identity, gender, and social preferences. Most experimental economics research on culture studies cross-national or cross-ethnic differences in economic behavior. In contrast, we explain laboratory behavior using two cultural dimensions adopted from a prominent general cultural framework in contemporary social anthropology: group commitment and grid control. Groupness measures the extent to which individual identity is incorporated into group or collective identity; gridness measures the extent to which social and political prescriptions intrinsically influence individual behavior. Grid-group characteristics are measured for each individual using selected items from the World Values Survey. We hypothesize that these attributes allow us to systematically predict behavior in a way that discriminates among multiple forms of social preferences using a simple, parsimonious deductive model. The theoretical predictions are further tested in the economics laboratory by applying them to the dictator, ultimatum, and trust games. We find that these predictions are confirmed overall for most experimental games, although the strength of empirical support varies across games. We conclude that grid-group cultural theory is a viable predictor of peopleās economic behavior, then discuss potential limitations of the current approach and ways to improve it.
Keywords: Laboratory experiment; two-person games; survey; culture; grid-group typology; social preferences
JEL classification: C7; C91; Z1
1. INTRODUCTION
Culture is a central concept broadly studied in social anthropology and sociology. There is a large political science and sociology literature on culture and economic development, dating from Max Weber (1904), with more recent work in fields such as modernization theory (e.g. Banfield, 1958; Bellah, 1957) and social capital theory (Putnam, Leonardi, & Nanetti, 1994). Culture has been linked to economic growth (Hofstede & Bond, 1988) and other economic outcomes such as national savings and income redistribution (Guiso, Sapienza, & Zingales, 2006). It has been gaining increasing attention in economics in relation to research on identity (Akerlof & Kranton, 2005; Eckel & Grossman, 2005), gender (Croson & Gneezy, 2009), and social preferences (Andreoni, Castillo, & Petrie, 2003; Charness & Rabin, 2002; Falk & Fischbacher, 2006).
Most experimental economics research on culture focuses on cross-national or cross-ethnic differences in economic behavior (Buchan & Croson, 1999; Chuah, Hoffmann, Jones, & Williams, 2009; Fershtman & Gneezy, 2001; Henrich et al., 2001; Roth, Prasnikar, Okuno-Fujiwara, & Zamir, 1991; Stoddard & Leibbrandt, 2014). These studies reveal clear behavioral differences across ethnic groups, and have greatly enriched the conventional rational choice model by encompassing a richer set of preference assumptions. However, they are based on an inductive approach in which specific preference assumptions are adopted in order to explain specific empirical findings. Thus, they await integration via a priori general theoretical frameworks that seek to specify a broad range of possible human preference configurations applicable to behavior under a wide range of environmental states. Such a theoretical underpinning is crucial in generalizing findings about behavior under one set of conditions to predict behavior under a wider set of conditions. Without a general framework, empirical results will be difficult to cumulate into general findings about how cultural differences affect behavior.
The main purpose of this chapter is to use a prominent general cultural framework in contemporary social anthropology, grid and group, to predict laboratory behavior in the dictator, ultimatum, and trust games, and to test these predictions. Groupness measures the extent to which individual identity is incorporated into group or collective identity; gridness measures the extent to which social and political prescriptions intrinsically influence individual behavior. Thus, one objective is to show that the grid-group framework, despite its origins in comparative ethnography, is adaptable to an experimental setting and indeed provides a parsimonious framework for generating testable behavioral predictions across a variety of experimental games. Another is to test the predictions of the grid-group framework on a number of simple games widely employed by experimental economists.
We first briefly discuss some prominent cultural frameworks that have been applied to economic experiments, and then introduce Mary Douglasā grid-group model from social and cultural anthropology. We argue that the model provides an alternative general but parsimonious cultural framework, one that is widely used across multiple social science fields and has already been applied successfully to a diverse set of social phenomena using a diverse range of methodologies. We describe a process by which the grid and group dimensions can be operationalized to a laboratory setting, and how hypotheses for a variety of standard games are generated from the theoretical logic of model. Next, we report the result of experiments involving dictator, ultimatum, and trust games, showing, for most games, confirmation for the predictions made by the theory. Finally, we briefly discuss the limitations of the current approach, and how it could be improved in future work.
2. GENERAL CULTURAL FRAMEWORKS AND ECONOMICS
While general frameworks for representation of cultural differences have not yet been developed independently within economics, a number of cultural frameworks1 have been borrowed and adapted from other social science disciplines in some studies of economic behavior. Although culture is a macro property, many scholars have adapted cross-cultural psychology frameworks that measure the manifestation of culture at the individual level using attitudinal surveys, hence conceptualizing it as the distribution of such attitudes across a group, organization, or society.2
Perhaps, the most widely utilized cultural model for empirical study of economic behavior can be found in cross-cultural management studies, particularly in the work of Geert Hofstede. The work originally identified four major cultural value dimensions for business organizations: power distance, individualism vs collectivism, masculinity vs femininity, and uncertainty avoidance (1980). Later versions of Hofstedeās work added the additional dimensions of long-term orientation aka Confucian dynamism (1991) and indulgence vs restraint (Hofstede, Hofstede, & Minkov, 2010). These cultural dimensions were generally measured at the individual level, then aggregated across organizations, then eventually across entire cultures. For instance, uncertainty avoidance is measured via questions pertaining to tension at work, competition between employees, and the qualities of good managers; long-term orientation by the importance placed on thrift and respect for tradition; and individualism by the importance of family life, physical working conditions, and adventure (Hofstede, 2001). In recent years, number of researchers have sought to build upon and improve Hofstedeās framework by positing alternative and more extensive sets of conceptual dimensions, the most notable the work of Shalom Schwartz (Schwartz, 1992, 1994; Schwartz et al., 2012)3, and the GLOBE project (Chhokar, Brodbeck, & House, 2007; House, Hanges, Javidan, Dorfman, & Gupta, 2004).
Another type of cultural model is due to Inglehart (1997) who suggests the distinction across societal cultures along two main dimensions of value orientations: ātraditional versus secular-rational orientations towards authorityā, and āsurvival versus self-expression values.ā These dimensions are measured using the data from the World Values Survey (WVS) from more than 60 countries, allowing to study cross-cultural differences among the countries, and are an extension of his earlier work on the rise of āpost-materialisticā values in Western European politics among cohorts who came of age after World War II (Inglehart, 1977, 1990). Post-materialism was designed to explain the lower priority younger voters placed on growth and their relative willingness to extend participation outside the boundaries of established party politics. In later work, Inglehart and Baker (2000) cluster countries based on their mean scores along the two dimensions, partitioning the globe into six regions, and examine the relationship between the two dimensions and economic development.
Both Inglehartās value orientations and Hofstedeās cultural dimensions measures have been applied by a number of experimental economics studies to explain differences in behavior across geographic regions or countries. Oosterbeek, Sloof, and Van De Kuilen (2004) perform a meta-analysis of several ultimatum game studi...