Wage claim detracts reciprocity in labor relations: experimental study of gift exchange games

The tendency of individuals to conflate what is fair with their self-interests is known as self-serving bias, which is an important factor in the bargaining impasse (e.g., Babcock and Loewenstein 1997).

For example, Klaas (1989) argued that after filing a grievance associated with a feeling of inequality, an employee might unconsciously attempt to justify this action by making cognitive adjustments that result in a heightened sense of inequity.

Such experiments using a cover story, confederates, and deception are common in psychology to ensure that the subjects behave as they would in a real-world situation, that is, enhancing external validity. Meanwhile, experimental economists avoid deceiving subjects since deception will reduce the subjects’ ability to make informed choices about game rules and incentives. See, for example, Ariely and Norton (2007) or Madsen and Stenheim (2015) for a comparison of the experimental methods of psychologists and economists.

Roth (1995) noted that face-to-face interaction might have many confounding and uncontrolled effects.

See Crawford (1998) and Farrell and Rabin (1996) for a detailed review.

While most of these studies have focused on cheap talk as a means to facilitate coordination or improve efficiency in social dilemmas, some researchers have focused on the shortcomings of cheap talk, such as being a tool for deception (e.g., Wilson and Sell 1997; Croson et al. 2003). We will argue later about what this shortcoming of cheap talk implies in our experiment.

In Gächter and Falk (2002), the firm’s wage offer is an integer, taking values from 20 to 120.

See Gächter and Fehr (2002) for a survey.

Instructions for the IT and practice problems are shown in Appendix A in the supplementary materials.

Preference for mathematics is measured on a 5-point scale, with 1 meaning very favorable and 5 very unfavorable. For monthly income, we use six dummy variables. Each dummy variable is 1 if the income is less than 9,999 yen, between 10,000 yen and 29,999 yen, 30,000 yen and 49,999 yen, 70,000 yen and 89,999 yen, and more than 90,000 yen.

This result is consistent with the results of the dictator game experiment in Yamamori et al. (2008), in which dictators reduce their offers to greedy recipients.

One may wonder why we do not consider the intention-based reciprocity models (e.g., Charness and Rabin 2002; Dufwenberg and Kirchsteiger 2004; Falk and Fischbacher 2006; Rabin 1993) rather than the models of distributional preferences (e.g., Andreoni and Miller 2002; Bolton and Ockenfels 2000), which rely only on the outcomes, such as Fehr and Schmidt’s model. The intention-based reciprocity models can also explain the worker’s reciprocal behavior in a gift exchange game. However, our estimation’s aim is neither to find the best-fit model to explain our data nor to elicit the preference of subjects (which should be defined on a broader domain, including the other parts of the game form, i.e., context-dependence). To achieve these goals, we need more elaborate experiments designed for them (See Weibull (2004) for an argument for the difficulty of eliciting preference). Instead, the purpose is to examine whether workers’ ranking on the payoff distributions was biased toward more self-interest or self-serving, which cannot be verified by only observing descriptive statistics of the effort schedules. Capturing the impact of workers’ voices on the parameters of the inequality aversion model is a simple way to achieve this goal.

Our estimation follows the method described in Chapter 16 of Moffatt (2016).

For example, see Breitmoser and Tan (2013) for more sophisticated models on reference-dependent social preferences.

This does not mean that the use of deceptive messages will be eliminated in repeated interactions. For example, Wilson and Sell (1997) showed, in their public goods experiment with a pre-announcement that repeats multiple times in a group of the same six people, that the subjects rarely contribute at the levels they announce.

Some experimental studies on gift exchange games indicate that repeated interactions improve efficiency relative to the one-shot situation through the so-called “repeated game effect” (Falk et al. 1999; Gächter and Falk 2002). We do not focus on this repeated game effect to address inefficiency but rather due to its ability to make messages credible.

Note that reservation wages and effort schedules are no longer proxies for their preferences on the outcomes of the underlying game in the repeated situation since they would be influenced by strategic considerations.