The impact of social value orientation and risk attitudes on trust and reciprocity

Abstract

Prior experimental studies provide evidence that the levels of trust and reciprocity are highly susceptible to individuals’ preferences towards payoffs, prior experience, capacity to learn more about personal characteristics of each other and social distance. The objective of this study is to examine whether social value orientation as developed by Griesinger and Livingston [Griesinger, D. W., & Livingston Jr., J. W. (1973). Toward a model of interpersonal motivation in experimental games. Behavioral Science, 18, 173–188] and Liebrand [Liebrand, W. B. G. (1984). The effect of social motives, communication and group size on behavior in an n-person multi-stage mixed-motive game. European Journal of Social Psychology, 14, 239–264] and risk preferences can help to account for the variability of trust and trustworthiness. We use the Berg, Dickhaut and McCabe [Berg, J., Dickhaut, J., & McCabe, K. (1995). Trust, reciprocity, and social history. Games and Economic Behavior, 10, 122–142] investment game to generate indices of trust and reciprocity. Prior to their participation in the investment game, all subjects participated in two other games. One is used to measure their social value orientation (a measure of other regarding behavior) and the second to measure risk attitudes. These variables are introduced as treatments in the analysis of the trust and reciprocity data. In addition to these preference related variables, gender is introduced to capture any differences between men and women which may not be encompassed by value orientation and risk attitudes. The statistical analysis indicates that the social value orientation measure significantly accounts for variation in trust and reciprocity. As well, the level of trust exhibited by an investor significantly affects the reciprocity of the responders and this measure of trust interacts with social value orientation. Individuals who are highly pro-social reciprocate more as the sender’s trust increases, while those who are highly pro-self reciprocate less as the sender’s trust increases. For this sample of participants, the gender variable does not capture any differences in the behavior of men and women that is not already reflected by the differences captured by their value orientations. Risk attitudes do not significantly account for variation in trusting behavior, except for the case where individuals have neither strongly pro-social nor pro-self social value orientations. In this case, more risk-seeking individuals are more trusting.

Introduction

Arrow (1974) stresses the ubiquity of trust (as a transaction cost depressant) in almost every economic transaction. He finds that higher rates of investment and growth are positively associated with higher levels of trust. More recently, Dincer and Uslaner (2007) show that a 10% increase in trust leads to a 0.5% increase in the growth rate of per capita income. For more than a decade, researchers have been examining trust and reciprocity issues in a behavioral game context. The most common artifact for this work is the trust (investment) game in a one-shot setting (see for example, Berg et al., 1995, Cox and Deck, 2006, Croson and Buchan, 1999).

Many experimental studies (see for example, Cox and Deck, 2006, McCabe et al., 2003) have shown that measures of trust and trustworthiness are highly variable and are associated with various factors such as individual preferences towards payoffs, prior experience, a player’s capacity to learn more about personal characteristics of each other and social distance. In these experiments, various contextual factors associated with the structure of the experiments are manipulated to partially explain the variance in the levels of trust and trustworthiness.

In addition to the above factors, social preferences may also influence both trusting and reciprocal behavior. According to Croson and Gneezy (2004), social preference is modeled in the economic literature in a number of ways, including altruism (Andreoni, 1989, Becker, 1974), envy (Mui, 1995), inequality-aversion (Bolton, 1991, Bolton and Ockenfels, 2000) and reciprocity (Charness and Rabin, 2002, Rabin, 1993). Cox (2004) employs a three-game (or triadic) design to decompose trust from altruism and reciprocity from altruism or inequality-aversion. We use a measure of other-regarding preferences which social psychologists describe as capturing a range of preferences from competition through altruism (see, for example, the development of a measure of social value orientation by Griesinger and Livingston (1973), and Liebrand (1984)) in an attempt to account for variation in trust and trustworthiness in a one-shot Berg et al. (1995) investment game. While this measure has been used to study voluntary contributions to public goods, we have not seen it applied to the trust game.

Eckel and Wilson (2004b) study trust as a risky activity. They note (p. 448) that “While many researchers appear to accept this relationship between risk and trust, the relationship between the risk orientation of the trustor and trusting behavior has not been explicitly studied.” Ashraf, Bohnet, and Piankovn (2003) and Eckel and Wilson (2004b) conduct experiments that include behavioral risk measures and report no statistical relationship between behavioral risk measures and the decision to trust.¹ We use the results of a series of lotteries to construct a continuous measure of risk attitude, similar to that used by Fellner and Maciejovsky (2007), for the participants in our investment game, and use this risk measure to augment our measure of other-regarding preferences to understand the variation in our trust data.

The statistical analysis in this paper indicates that the social value orientation measure significantly accounts for variation in trust and reciprocity. Furthermore, risk attitudes as we have measured them do not significantly explain variation in trusting behavior using the aggregated data, which is consistent with the findings of Eckel and Wilson (2004b), who use a different measure of risk attitudes than what we use. However, we do find a significant positive relationship between risk attitudes and trust for participants whose social value orientations are neither strongly pro-social nor pro-self.² This disaggregation of the social value orientation measure permits us to identify a notable relationship between risk attitude and trust which has not been identified in the literature.