TX Antenna

contributions. The sequential treatment LEAD delays the first occurrence of the event, that at least one individual does not strictly improve, on average, by one round. The behavioral type of the leader is again crucial in this context.

8.5.4 Further Analysis of the LEAD Treatment

In this section we want to take a closer look behind the mechanisms of effec-tive leadership and investigate why some groups in LEAD are more successful than others. We begin by analyzing the followers’ reactions and then proceed by looking at the leaders’ behavior.

8.5.4.1 Follower behavior

The followers’ reaction is decisive for successful leading-by-example. Only if they respond adequately, it will be ensured that leadership (i) is successful in terms of the final endowments for the group members, and additionally, (ii) that the leader does not fare worse either.

We consider the reactions to a leader’s contribution both relative to a sub-ject’s own endowment and in absolute terms. Focusing on absolute contri-butions first, Table 8.7 (left) indicates that for every Taler a leader gives in a round, followers give on average 0.822 Taler to the public good. Additionally, follower type heterogeneity leads to different reactions. Although all types re-act positively to higher contributions of the leader, this rere-action is particularly pronounced for followers that were classified as conditional cooperators and altruists, as the coefficients in Table8.7 reveal. Selfish follower types match a Taler given by the leader by only 0.56 Taler, which is much less compared to the amount given by conditional cooperators and altruists. A similar pattern emerges for relative contributions (see Table8.7(right)).

8.5 results 141

Rel. leader contribution (RLC) 0.835^∗∗∗

(0.027)

Note: OLS regression with period controls. Cluster-robust standard errors (on the subject-level) are in parentheses.

∗p<0.1,^∗∗p<0.05,^∗∗∗ p<0.01.

Table 8.7: Regression Table - Matching of leader contribution (absolute and relative)

This high rate of matching a leader’s contribution further illustrates why the behavior of the leader matters a lot, since even selfish types react positively on the amount spent by the leader. Our result is summarized in Observation 4:

Observation 4. In general, followers react positively to a leader’s contribution, matching it to a large degree, both in absolute and in relative terms. The type of the follower plays a major role as selfish followers exhibit a much smaller reaction.

8.5.4.2 Leader behavior

We have already pointed out that leadership is beneficial. In the next step we will go more into detail to explain the mechanisms behind successful leader-ship. Overall, a high early contribution by the leader in round 1 of a period has a large impact on final earnings in round 7 of the respective period.

Figure 8.5 depicts the relationship between a leader’s first contributions and final wealth. We see a pattern of increasing final wealth of a group when the leader contributes much in the beginning. However, we can observe that there is a large heterogeneity between leader types for the 20 Taler bracket.

There are many selfish leaders that contribute their entire endowment in the beginning. But they are less successful than conditional cooperator leaders who do the same.

8.5 results 142

Figure 8.5: Leader’s first contribution and final group earnings

Figure 8.6: Leader’s contribution pattern over rounds by type

Figure8.6indicates why this is the case. We see that while conditional coop-erators have the lowest average initial leader contribution of all types, they are persistent in contributing. That makes CC types successful. By contrast, we discover signs of strategic leadership by selfish types. They start with relatively high contributions in the first two rounds, revising them sharply afterwards. What came to our mind was a strategy that the New York Times (Gleick1986) described as a “tranquilizer strategy” for the repeated prisoner’s dilemma: to lull the opponent for a few moves and then try to exploit him.

We find evidence for this claim in our data as selfish types contribute more in the beginning when they are leaders. In round 1 of LEAD, 61.54% of the leaders, which we classified as “selfish”, contributed their whole endowment of 20 Taler. However, only 21.74% of the selfish followers also contributed this amount. In treatment NOLEAD by comparison, 25.00% of the selfish types contributed 20 Taler in round 1. A behavior that can probably be explained by a false consensus effect, is the one shown by those subjects classified as

al-8.5 results 143

truists. When they are leaders, they start with very high relative contributions in the beginning, but display a faster decline than the CC types afterwards.

Endowment (end of round 7)

(1) (2) (3)

Is leader -26.540^∗∗ -25.703^∗∗ 24.404 (12.829) (12.884) (27.012) Leader first contr. 7.566^∗∗∗ 8.351^∗∗∗

(Leader1st) (1.332) (1.433)

Leader1st X Is Leader -3.567

(2.539) Constant 80.073^∗∗∗ -15.345 -26.520^∗

(11.066) (13.489) (15.160)

Period controls YES YES YES

Observations 192 192 192

Subjects 48 48 48

R²overall 0.029 0.194 0.196

Note: Random effects regression with period controls. Cluster-robust standard errors (on the subject-level) are in parentheses.

∗p<0.1,^∗∗p<0.05,^∗∗∗ p<0.01.

Table 8.8: Regression Table - Final endowment (only LEAD)

In columns (2) to (3) of Table 8.8we can see that a leader’s first contribution has a large effect on final wealth. Column (2) indicates that for each Taler the leader gives in round 1, the final endowment of every group member (including the leader) increases by around 7.6 Taler, on average.

In general, being the leader is not necessarily good for the own payoffs as column (1) of Table 8.8 reveals. Leaders face some kind of curse as their earnings are lower than those of other group members.

Figure 8.7: Average endowment by treatment + leader/no-leader

8.5 results 144

Figure 8.7 illustrates this circumstance graphically. However, leaders do not fare worse than average participants in NOLEAD in the end. This, too, is depicted in Figure 8.7. The leader himself in turn, does not lose money by contributing more in round 1, but he profits less from each Taler compared to the other group members as it can be seen by the interaction effect in column (3) of8.8. We can thus summarize our last finding as follows:

Observation 5. The leader type has a significant effect on final earnings of the group