«IZA DP No. 4262 PAPER Gift Exchange and Workers’ Fairness Concerns: When Equality Is Unfair Johannes Abeler DISCUSSION Steffen Altmann Sebastian ...»
The strong diﬀerence in eﬀort choices suggests that the degree to which gift exchange can mitigate contract enforcement problems depends on the payment mode that is used. Wage equality hampers eﬃciency, and we hypothesized above that this might be due to horizontal fairness concerns. However, performance diﬀerences might also 60%
be driven by diﬀering monetary incentives across treatments. To rule this out, we take a closer look at principals’ wage setting and the resulting monetary incentives for the agents.
Figure 3 plots the average wage per eﬀort level in the two treatments. For both treatments we take the wage paid by the principal for each individual eﬀort decision and calculate averages for a given eﬀort level. The graph exhibits the upward sloping eﬀort-wage relation of many gift-exchange experiments. For example, an agent in the equal wage treatment who exerts an eﬀort of 1 receives on average a wage of 6.3 while an agent exerting an eﬀort of 10 receives an average wage of 30.3. In the individual wage treatment, the corresponding wages are 1.7 and 39.5.8 The eﬀort-wage relation indicates that gift exchange indeed occurs between principals and agents. In both treatments, higher eﬀort levels are reciprocated with higher wages.
Since principals in the EWT have to pay the same wage to both agents, an interesting question concerns how they choose this wage when confronted with a low and a high eﬀort. To answer this question, we assume that the wage-eﬀort relation from the IWT reﬂects the “true” wage-setting preferences of principals because wage choices are not constrained in this treatment. We regress wages on eﬀort in the IWT and calculate predicted wages for all possible levels of eﬀort. We then calculate the diﬀerences between actual wages paid in the EWT and these predicted wages. This analysis shows that the actual wage in the EWT is very close to the average between the predicted wage for the higher and lower eﬀort. This means that principals in the EWT weight the higher and lower eﬀort about equally when deciding on the wage payment.
Result 2: Principals reward higher eﬀort levels with higher wages in both treatments.
Reciprocal behavior of principals generates monetary incentives for the agents. In order to calculate the monetary incentives entailed in principals’ wage decisions, one has to take into account agents’ cost of eﬀort exertion (see Table 1). Qualitatively, this does not change the picture of the eﬀort-wage relation: higher eﬀort levels seem to lead not only to higher wages, but also to higher proﬁts for the agents. To check this in more detail, we estimate an OLS-model where we regress the agent’s proﬁt per period πAi on his eﬀort level ei and a constant. To account for potential diﬀerences between treatments we include a treatment dummy IW T, and an interaction term of the treatment dummy and the agent’s eﬀort. IW T equals 1 for the individual wage treatment and 0 for the equal wage treatment. Reported robust standard errors are adjusted for clustering within matching groups. Estimation results are shown in Column 1 of Table 3. The coeﬃcients indicate that the eﬀort-proﬁt relation is indeed positive in both treatments. On average, an additional unit of eﬀort increases the agent’s proﬁt under equal wages by 1.031 points. This coeﬃcient is weakly signiﬁcant.
In the individual wage treatment the eﬀort-proﬁt relation is slightly steeper: an eﬀort increase of 1 leads to an increase in agent’s proﬁt of 1.804 points (1.031 + 0.773). The diﬀerence between treatments, however, is not signiﬁcant.
Table 3: Proﬁt regressions. Robust standard errors are adjusted for clustered matching groups and are given in parentheses. For each ﬁrm, one observation per period is included in the analysis. The dummy “IWT” is equal to 1 for the individual wage treatment. Signiﬁcance at the 10%, 5% and 1% level is denoted by *, **, and ***, respectively.
Column 2 of Table 3). The results indicate that the co-worker’s eﬀort choice has a substantial impact on an agent’s proﬁt under wage equality while it has a negligible inﬂuence if individual wages are paid. An increase in agent j’s eﬀort increases agent i’s proﬁt in a given period by 2.774 points in the EWT, while the (insigniﬁcant) inﬂuence in the IWT is −0.404 (= 2.774 − 3.178). However, it is still individually proﬁtable for the agents to exert high eﬀorts in the EWT. An additional unit of (own) eﬀort increases the agent’s proﬁt by 0.854 points.9 Our ﬁndings concerning agents’ monetary One could object that subjects in the experiment did not have access to the analyses we just presented, because these are “ex-post” examinations while subjects only observed behavior and outcomes of their previous groups. We therefore calculate the proﬁt-maximizing eﬀort level for each agent in each period based on the information this subject actually has. If we assume that agents choose the eﬀort level that was on average the most proﬁtable of all eﬀort levels they have observed so far, the calculations show that agents in the EWT could have increased their eﬀorts and proﬁts considerably incentives can thus be summarized as follows.
Result 3: The wages paid by principals imply similar monetary incentives in both treatments. A higher eﬀort level leads to a higher proﬁt in both treatments.
3.3 The Importance of Equity In light of the previous result, the strong diﬀerences in actual eﬀorts and especially the low eﬀort levels under equal wages are remarkable and stress the signiﬁcance of non-pecuniary motivations for agents’ performance. Agents under equal wages predominantly choose low eﬀorts, thereby foregoing considerable proﬁts. Apparently, equal wages are not reconcilable with agents’ horizontal fairness considerations. On the other hand, agents under individual wages provide very high eﬀort levels. Thus, aggregate behavior is consistent with the predictions of equity-concerned agents. We therefore focus our analysis of non-monetary motivations on the question whether individual behavior is also in line with a concern for the fulﬁllment of the norm of equity.
3.3.1 Agents’ Reactions to Norm Violations
We ﬁrst analyze how agents react to a violation of the norm of equity. Equity theory argues that agents experience distress from inequity and take action to reduce it— which in our setup means to increase or decrease work eﬀort. The direction of the eﬀort adjustment should depend on the type of norm violation. An equity-concerned agent who works more but does not receive a higher payoﬀ than his co-worker faces a disadvantageous norm violation. To restore equity, he can only decrease his eﬀort.
Analogously, his co-worker who exerts a lower eﬀort and earns a higher proﬁt faces an advantageous norm violation and should increase his eﬀort.10 even by using only their limited information. In the last period, the average proﬁt-maximizing eﬀort level exceeds the average actual level in that period by 61%. By contrast, the average actual eﬀort levels of subjects in the IWT are very close to the proﬁt-maximizing levels.
More precisely, an advantageous norm violation comprises all cases when eﬀorts are equal but proﬁt is higher, or when eﬀort is lower but proﬁt is not. A disadvantageous norm violation occurs if eﬀorts are equal but proﬁt is lower, or if eﬀort is higher but proﬁt is not.
Table 4 shows how often agents decrease, increase or do not change their eﬀort from period t to t + 1 after they experienced no, an advantageous or a disadvantageous norm violation in period t. The top panel of Table 4 reports data for the equal wage treatment. When the norm is fulﬁlled, most agents keep their eﬀort constant (54%) and slightly more agents increase their eﬀort than decrease it. After experiencing an advantageous norm violation, agents tend to increase their eﬀort (44%) and only few reduce it (12%). The opposite is true after a disadvantageous norm violation: the majority of agents decrease their eﬀort (53%) and only few increase their eﬀort in the following period (14%). In line with equity theory these numbers suggest that agents change their eﬀort provision in the direction that makes a violation less likely to occur in the next period.
Behavior in the individual wage treatment (bottom panel) is very similar to behavior in the EWT for the cases of no violation and disadvantageous violations. When the norm is not violated agents mostly keep their eﬀort unchanged. After a disadvantageous norm violation eﬀorts are decreased rather than increased, as in the EWT. The only diﬀerence between treatments is observed when agents experience an advantageous norm violation: agents in the IWT tend to decrease their eﬀort while the EWT agents tend to increase it in this case.11 We checked the robustness of the reaction patterns in several ways. For example, it could be that The pattern of individual reactions to norm violations indicates that agents care about equity; we therefore check next how often norm violations occur in the two treatments. We expected to see more norm violations in the EWT than in the IWT, because the equal wage institution forces principals to set wages that are not in line with the norm of equity whenever agents exert diﬀerent eﬀorts. This is indeed what we observe. While the norm is violated in 87% of all cases (460 out of 528) in the EWT, the ﬁgure for the IWT is only 15% of all cases (80 out of 528). Thus, even if individual reactions in a given situation are similar, agents in the EWT are far more often exposed to norm violations than agents in the IWT. Principals in the IWT seem to understand quite well that agents care about equity and use the possibility to set diﬀerent wages in a sophisticated way. If eﬀorts diﬀer, they reward the more hardworking agent with a higher wage in 90% of these cases. If agents exert the same eﬀort, principals pay equal wages in 90% of the cases.
Result 4: Agents mostly react to disadvantageous violations of the norm of equity by reducing their eﬀort and by increasing it after an advantageous norm violation. The norm of equity is far more often violated in the equal wage treatment.
So far we have seen that agents’ reactions are largely in line with the hypotheses of equity theory and that treatments diﬀer with respect to the frequency of equity-norm violations. Yet, this is not suﬃcient to explain the treatment eﬀect, since a norm violation is always advantageous for one agent and at the same time disadvantageous for the other one. If both agents adjust their eﬀort in a similar way but in opposite directions the adjustments will cancel out. However, previous evidence suggests that reactions to a disadvantageous norm violation are stronger than reactions to an advantageous one (e.g., Loewenstein et al. 1989, Mowday 1991, Th¨ni and G¨chter 2008).
o a If this is the case, norm violations could explain the downward trend in the EWT and agents react diﬀerently to norm violations if they are paid very high or low absolute wages. However, performing the analysis only for agents receiving a wage out of the top or bottom quartile of the ex-post wage distribution does not alter the result. An implicit assumption of our analysis is that the gift-exchange relation is generally intact between principal and agent, i.e., that agents exert a non-minimal eﬀort and that principals pay a positive wage. The results do not change if one restricts the analysis to these cases. Also if one deﬁnes gift exchange as requiring the agent’s proﬁt to be positive, i.e. wi c(ei ) instead of wi 0, the results are very similar.
the treatment diﬀerence in eﬀort provision.
Figure 4 shows the average magnitude of changes in eﬀort provision from period t to period t + 1 after an agent experienced no norm violation, a disadvantageous or an advantageous norm violation in period t. The width of the bars corresponds to the number of observations in the respective category (cf. last column of Table 4).
When the equity-norm is not violated agents tend to keep their eﬀort constant or even slightly increase it. After a disadvantageous norm violation, agents in the EWT react strongly. They decrease their eﬀort by 1.30. Their co-worker, experiencing an advantageous norm violation, increases his eﬀort but not as strongly. He raises his eﬀort by only 0.75. The diﬀerence is statistically signiﬁcant (Wilcoxon test of the absolute values: p = 0.01). In the IWT, both groups of agents experiencing a norm violation decrease their eﬀort. The strength of reactions indicate that agents suﬀer more from a disadvantageous norm violation than from an advantageous one. This results in an overall decrease of eﬀorts after a norm violation.
Result 5: Agents’ reactions to a violation of the norm of equity are asymmetric: the negative reaction of the disadvantaged agents is stronger than the reaction of the advantaged agents. This asymmetry in agents’ reactions leads to an overall negative time trend in eﬀorts for the EWT and in the strong treatment diﬀerence in eﬀort.
The analysis above suggests that agents care about equity and experience the equal wage scheme as unfair. Interestingly, even the principals consider the equal wage scheme as less fair. In the post-experimental questionnaire, principals were presented three hypothetical game situations that included eﬀort choices, wage choices, and the resulting payoﬀs for all players. They were asked whether they considered the resulting allocation as just. One of the three situations reﬂected their own average behavior in the experiment.12 The principals did not know that they were facing their own past decisions when answering the question. 63% of the principals in the IWT considered their own decisions fair while only 38% of the principals in the EWT shared this view This situation was constructed as follows: We calculated the average eﬀort of the higher-eﬀort and of the lower-eﬀort providers that the principals actually faced during the experiment. We then took the average of the wages the principals paid to the two groups. Finally, we calculated hypothetical payoﬀs for all three “average” players by considering the costs of the average eﬀorts.