Independent scholarship · Open accessThe monthly digest

Indexed text attachment

games-and-information-in-higher-categories.tex

Source for Games and Information in Higher Categories

View paper · Download original file

Lines 3281–3380 of 6559. Source text is displayed literally.

3281safe play at one vertex forces safe play at the next. Truncations $\K^n$ of the common-knowledge hierarchy3282do not approximate $\CK$ as far as behaviour is concerned.3283 3284What does? Monderer and Samet's answer \cite{MS89} is to weaken the modality rather than truncate the3285iteration. Let $\mathsf{B}^p_iE$ be the event that $i$ assigns probability at least $p$ to $E$, and define3286\term{common $p$-belief} from the operators $\mathsf{B}^p_i$ as common knowledge was defined from $\K_i$. Then3287equilibrium behaviour under common $p$-belief of the game converges to behaviour under common knowledge as3288$p\to1$. The family $(\mathsf{B}^p)_{p\in[0,1]}$ is a \emph{graded} modality, and the theorem says that3289strategic behaviour is continuous in the grade although not in the depth.3290 3291The same discontinuity appears on the universal type space. $T^*$ was built in \cref{thm:universal} as an3292inverse limit and carries the product topology, in which two types are close when their beliefs agree to3293high finite order.3294 3295\begin{theorem}\label{thm:WY}\Scl[\cite{WY07,DFM06}]3296For every type $t$ in the universal type space and every action $a$ that is rationalizable for $t$, there3297are types arbitrarily close to $t$ in the product topology for which $a$ is the \emph{unique}3298rationalizable action (under a richness assumption on payoffs). Consequently no refinement of rationalizability3299is robust to perturbations of higher-order beliefs, and the coarsest topology on types in which3300rationalizable behaviour is continuous is strictly finer than the product topology.3301\end{theorem}3302 3303The universal type space exists because of a continuity property of the functor $\Giry$; strategic3304behaviour is discontinuous in exactly the topology that this property provides. The semantics of types and3305the semantics of play live on the same set and disagree about its topology. One may regard this as bad3306news for refinements, or as an instrument. \emph{Global games} \cite{CvD93,MS98} use it as an instrument:3307adding small private noise to a coordination game with multiple equilibria produces a game with a unique3308rationalizable outcome, obtained by an infection argument like that of the mail game, and the outcome3309selected is the risk-dominant one. Note that this is consistent with \cref{prop:which-type}: the argument3310uses only best-response structure, and risk dominance is the selection criterion that lives on3311$\NFG_{\mathrm{inc}}$.3312 3313\section*{Notes}3314 3315Sequential equilibrium is due to Kreps and Wilson \cite{KW82a}, perfect equilibrium to Selten \cite{Sel75},3316and perfect Bayesian equilibrium in its usual form to Fudenberg and Tirole \cite{FT91b}. Conditional3317probability systems go back to R\'enyi \cite{Ren55} and were brought into game theory by Myerson3318\cite{Mye86}; McLennan \cite{McL89a,McL89b} studied the topology of the spaces of conditional systems and3319of consistent assessments, and Kohlberg and Reny \cite{KR97} characterized consistency by an independence3320condition on relative probabilities. Lexicographic and non-Archimedean representations are in3321\cite{BBD91,Ham94,Hal10}. The reputation papers are \cite{KW82b,MR82a,KMRW82}, and the general bounds are3322due to Fudenberg and Levine \cite{FL89}; Diamond's model of reputation acquisition in debt markets3323\cite{Dia89} is an application in which a borrower's incentive to choose safe projects grows with her3324record. The electronic mail game is \cite{Rub89}; common $p$-belief is \cite{MS89}; the structure theorem3325for rationalizability is \cite{WY07} and the strategic topology \cite{DFM06}. Global games were introduced3326by Carlsson and van Damme \cite{CvD93} and applied to currency crises by Morris and Shin \cite{MS98}.3327 3328\section*{Exercises}3329 3330\begin{exercise}3331In \cref{ex:seq-entry}, find all Nash equilibria of the strategic form, including mixed ones, and3332determine which are sequential.3333\end{exercise}3334\begin{exercise}3335List the thirteen strata of the space of conditional probability systems on a three-element set, and3336match each with a face of the hexagon of \cref{fig:cps}.3337\end{exercise}3338\begin{exercise}3339Represent the conditional probability system on $\{1,2,3\}$ in which state~1 is infinitely more likely3340than the others and $p_2:p_3=2:1$ (a) as a lexicographic probability system and (b) as a distribution with3341values in $\R(\varepsilon)$.3342\end{exercise}3343\begin{exercise}3344Extend \cref{ex:reputation} to three towns and verify the threshold $p>\tfrac18$ for deterrence in the3345first.3346\end{exercise}3347\begin{exercise}3348Draw the knowledge graph of the electronic mail game, and compute $\K_1E$, $\K_2\K_1E$ and3349$\K_1\K_2\K_1E$ for the event $E$ that the state is $b$.3350\end{exercise}3351\begin{exercise}3352Show that $\mathsf{B}^p_i$ is monotone, satisfies $\mathsf{B}^p_iE\cap\mathsf{B}^p_iF\subseteq\mathsf{B}^{2p-1}_i(E\cap F)$,3353and reduces to an operator containing $\K_i$ when $p=1$. In what sense is the family a graded comonad?3354\end{exercise}3355%% ---- end ch/10-dynamic.tex ----3356\part{Asymmetric Information}3357%% ---- begin ch/11-moralhazard.tex ----3358\chapter{Moral Hazard}\label{ch:moralhazard}3359 3360Part~II is about transactions between parties one of whom knows, or will know, or will do, something the3361other cannot observe. The uninformed party---the \emph{principal}---designs the terms; the informed3362party---the \emph{agent}---responds to them. Everything turns on the fact that a contract can depend only on3363what is observed. In the language of Part~I, a contract is a function that must factor through an3364observation map, and contract design is the choice of a context for someone else's decision.3365 3366\section{Five models and one diagram}3367 3368The classical models of asymmetric information differ in the order of four events: Nature's choice of a3369state $\theta$ observed only by the agent; the signing of a contract; the agent's action or message; and3370the realization of an outcome. A convenient classification, which the source text \cite[ch.~7]{Ras06}3371makes systematic, is the following.3372 3373\begin{center}\small3374\renewcommand{\arraystretch}{1.25}3375\begin{tabular}{@{}p{0.17\textwidth}p{0.40\textwidth}p{0.33\textwidth}@{}}3376\toprule3377\textbf{Model} & \textbf{Order of events} & \textbf{What the principal cannot see}\\\midrule3378hidden action & contract $\to$ action $\to$ noise $\to$ outcome & the action\\3379hidden knowledge & contract $\to$ $\theta$ $\to$ report or action & a state learned after contracting\\3380adverse selection & $\theta$ $\to$ contract & the type, known before contracting\\