Gradualism and Irreversibility (Q4789808)

The authors study the following model of a two-person non-zero-sum dynamic game \(\Gamma\): There are two players \(i=1,2\). In each period \(t=0,1,\dots\), both players \(i=1,2\) simultaneously choose an action \(c_{i,t} \in \mathbb{R}_{+}\), measuring \(i\)'s level of cooperation. The one step payoffs in period \(t\) to players 1 and 2 are \(\pi(c_{1,t},c_{2,t})\) and \(\pi(c_{2,t},c_{1,t})\), respectively, where \(\pi(x,y)\) is a continuous function strictly decreasing in \(x\) and strictly increasing in \(y\), and satisfying additionally the following: there is a \(c^*\) such that the function \(w(c) = \pi(c,c)\) is strictly increasing in \(c\) for all \(0\leq c \leq c^*\). Payoffs in game \(\Gamma\) are discounted sums of one step ones over infinite horizon with a discount factor \(\delta, \; 0 < \delta < 1\). It is assumed that the choices of actions by the players are restricted by the condition: \(c_{i,t}\geq c_{i,t-1}\) for \(i=1,2\) and \(t=1,2\dots\). The paper characterizes pure-strategy subgame-perfect Nash equilibria in game \(\Gamma\), described in the form of a sequence of nodecreasing actions \(\{\widehat{c}_{1,t},\widehat{c}_{2,t}\}_{t=1}^\infty\). They are widely discussed in two cases, symmetric (with \(\widehat{c}_{1,t}=\widehat{c}_{2,t}\) for all \(t\)) and nonsymmetric. Similar results are given for a ''sequential'' modification of game \(\Gamma\) where at even periods only player 1 can choose his action, while player 2 only at odd periods. The paper also contains an application of the obtained results to a model with adjustment costs, described in the form of dynamic voluntary contribution games.