Inverse Stackelberg public goods game with multiple hierarchies under global and local information structures (Q467489)

The aim of the paper under review is to solve the pure strategy equilibria of the game with some typical hierarchical structures by theoretical analysis and find the fewest hierarchies under the local information structure with which full cooperation can be reached. A Stackelberg game is a game involving hierarchical decision making among leaders and followers. An inverse Stackelberg game considers the situation that the leaders' strategies are a mapping of the followers' action. The author of the paper analyzes the inverse Stackelberg public goods game by formulating the multiple hierarchies for players, adapts the constructive and ``guess-prove'' ways to find equilibria. Let \(L_1\) be the set of players at the top level. They announce their actions \(A_1\) to the players from \(L_2,\dots,L_h\), the players from \(L_2\) at the second level announce their actions \(A_2\) to \(L_3,\dots,L_h\) and so on. The players \(L_h\) at the bottom choose actions \(A_h^* = f_h(A_1, \dots, A_{h-1})\) such that \(A_h^*\) is the Nash equilibrium of \(L_h\) if the actions profile of \(L_1, \dots, L_{h-1}\) is \(A_1, \dots, A_{h-1}\), respectively. From \(A_h^*\), there is a sequential decision \(A_{h-1}^*\) etc. with \(A_i^*\) \((i=1, \dots, h)\) that fit the inverse Stackelberg equilibrium of the game. Section 1 introduces the background and literature reviews with a summary of the paper. Section 2 devotes to the formulation of the setup of the inverse Stackelberg game and its equilibrium in a hierarchical structure. The main results of the paper are given in Section 3. Theorem 3.1 shows that the strict TFT strategy is better for the leader than the moderate player, where the leader takes the strict TFT as his/her strategy, the leader will act as C if and only if there are all \(n\) cooperators among followers. This may help the leader to announce the strategy or make rules in practice. Theorem 3.2 shows that the strategy r-TFT seems too strange to be adapted by the leaders since it will increase the payoff for the follower when he defects. However, in order to induce the followers to cooperate, more temptation from the leaders seems necessary. The author claims that no previous work has noticed and pointed this out. It is certainly not new among game theorists. Theorem 3.3 points out that the leaders cannot influence all the followers, they can only influence part of the followers to ensure that the temptation is large enough so that the followers will cooperate. Theorem 3.4 takes the local information into consideration by analyzing the followers who are only responsible to the nearest leaders or the followers who are responsible to all the leaders with a summary in Table 1. All the technical proofs are given in Section 4. The paper concludes in Section 5.