Consider a game with $N$ players, each indexed by $i=1,...,N$. Every player $i$ has to choose a $J\times 1$ vector of actions $a_i\equiv (a_{i,1},...,a_{i,J})$ where each $a_{i,j}$ can be zero or one. For each player $i$, let $a_{-i}$ denote the actions of the other players.

The payoff of each player $i$ is $$ u_i(a_i, a_{-i}; \theta)+v_i(a_i; \delta) $$ where $u_i$ and $v_i$ are some parametric functions of the parameters $\theta$ and $\delta$. Moreover $v_i$ is monotone decreasing in $\sum_{j=1}^J a_{i,j}$.

A pure strategy Nash equilibrium (PSNE) of the game is $a^*\equiv (a_1^*,...,a_N^*)$ solving $$ a_i^*\in \arg\max_{a_i\in \{0,1\}^J} u_i(a_i, a^*_{-i}; \theta)+v_i(a_i; \delta) \quad \forall i=1,...,N $$

Question: Suppose that I'm able to show that a PSNE exists for $\theta=\theta_0$ and $\delta=\delta_0$ where $\theta_0$ and $\delta_0$ are some specific real values of the parameters $\theta$ and $\delta$. Can I conclude that a PSNE exists for $\theta=\theta_0$ and any value of $\delta$?

In particular, here, I'm wondering whether the claim may follow from the fact that $v_i(a_i; \delta)$ enters additively, does not depend on $a_{-i}$, and it is monotone.

  • $\begingroup$ What does it mean that $v$ is monotone decreasing in $\sum_{j=1}^J a_{i,j}$? This is not a variable of $v$. $\endgroup$ – Giskard May 12 at 20:20

No. Consider the matching pennies game $$ \begin{array}{c |c} & H & T & \\ \hline H & (1,-1) & (-1,1)\\ T & (-1,1) & (1,-1)\\ \end{array} $$ which we know no PSNE exists.

Define the function $v_i(\cdot,\cdot)$ by $$ v_i(H,0) = 5 \\ v_i(x,y) = 0 \quad \forall (x,y) \not = (H,0) $$

At $\delta = 0$, we have the game $$ \begin{array}{c |c} & H & T & \\ \hline H & (5,4) & (-4,1)\\ T & (-1,5) & (1,-1)\\ \end{array} $$ which has a PSNE of $(H,H)$.

For any other value of $\delta$, we're back in the original game so no PSNE exists.

Redefine $H = 0, T = 1$ and you have your monotone decreasing requirement.


Seems like this class of games is very general. (Or I don't get the definition.) Note that $\theta$ is not even used, it is just some parameter that has value $\theta_0$ throughout.

As long as there exist two games $G_{\delta}$ and $G_{\delta_0}$, were both have the same number of players $J$, each player has two strategies in both $G_{\delta}$ and $G_{\delta_0}$, and $G_{\delta}$ has a Nash-equilibrium but $G_{\delta_0}$ does not, then one can get negation by defining $$ v(a; \delta_0) := u_{\delta_0}(a) - u_{\delta}(a) $$ where $u_{\delta}$ is the vector of payoffs in $G_{\delta}$, and $u_{\delta_0}$ is the vector of payoffs in $G_{\delta_0}$ given strategy profile $a$.

There is still the possibility that either all or no games have a Nash-equilibria, but this is not true on the class of these two strategies per player games, as shown by Walrasian Auctioneer's answer.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.