Indexed text attachment
games-and-information-in-higher-categories.tex
Source for Games and Information in Higher Categories
View paper · Download original file
Lines 5001–5100 of 6559. Source text is displayed literally.
5001$p_2$---prices are strategic complements---and the equilibrium is5002\[ p_1=p_2=c+t,\qquad\text{with profits }\tfrac t2\text{ each.} \]5003The margin is the transport cost: firms earn exactly what their distance from each other protects.5004\end{example}5005 5006Anticipating this, firms choosing locations before prices face two forces: moving towards the rival steals5007customers at given prices, and intensifies price competition. With quadratic transport costs the second5008dominates and the firms locate at the opposite ends, \emph{maximal differentiation}; with Hotelling's5009original linear costs the price game has no pure-strategy equilibrium when the firms are close, so his5010argument for minimum differentiation was incomplete \cite{dGT79}. The location stage is a commitment to5011soft pricing, and the sign of the strategic effect is the one predicted in \cref{ch:order} for5012complements.5013 5014\section{Comparative statics without calculus}5015 5016Because differentiated-product price games are supermodular, \cref{thm:supermodular} applies wholesale.5017If any parameter raises each firm's marginal profitability of a higher price---an increase in its own5018cost, in a rival's cost, in demand, in a tax---then the greatest and least equilibrium price vectors rise.5019No assumption of concavity, differentiability or uniqueness is used, and the conclusion holds for any5020number of firms and any pattern of substitution that keeps the complementarity. The classical approach5021differentiates first-order conditions and signs the result by a stability condition, which amounts to5022assuming the conclusion. The lattice-theoretic approach gets it from the order structure of the model,5023and extends it to games with discrete choices, such as adoption of a technology with network effects, in5024which the greatest and least equilibria may be far apart and the comparative statics of each is still5025monotone.5026 5027\section{Durable goods}5028 5029A monopolist sells a durable good to buyers who each want one unit and can wait. Having sold to those5030who value it most, she would like to cut the price and sell to the rest; and having done that, to cut it5031again. Buyers who foresee the cuts will not pay the high price today.5032 5033\begin{example}[Two periods]\label{ex:coase}5034Buyers' values are uniform on $[0,1]$; production is costless; both sides discount the second period by5035$\delta=\tfrac12$. If the monopolist could commit to prices she would charge $\tfrac12$ in both periods and5036earn $\tfrac14$. Without commitment, suppose buyers with values above $\hat v$ buy in the first period. In5037the second period she faces values uniform on $[0,\hat v]$ and charges $p_2=\hat v/2$. The marginal5038first-period buyer is indifferent, $\hat v-p_1=\delta(\hat v-p_2)$, so $p_1=\tfrac34\hat v$. Maximizing5039$p_1(1-\hat v)+\delta\,p_2\cdot\tfrac{\hat v}{2}$ gives5040\[ \hat v=0.6,\qquad p_1=0.45,\qquad p_2=0.3,\qquad\text{profit }0.225<0.25. \]5041Both prices are below the monopoly price and profit is lower: the monopolist is hurt by her own later5042self.5043\end{example}5044 5045\begin{theorem}[Coase conjecture]\label{thm:coase}\Scl[\cite{Coa72,GSW86}]5046In the infinite-horizon version, as the time between offers shrinks to zero, in every stationary5047equilibrium the opening price converges to the lowest buyer value (or marginal cost) and the5048monopolist's profit converges to the competitive level.5049\end{theorem}5050 5051The seller of a durable good competes with her future selves, and when they can move quickly they compete5052her profits away. The remedies are commitment devices: leasing instead of selling, so that she bears the5053loss when the price falls; most-favoured-customer clauses that oblige her to refund earlier buyers;5054limited capacity; a reputation for not cutting prices, in the sense of \cref{ch:dynamic}.5055 5056Three earlier themes converge here. First, this is the renegotiation problem of5057\cref{sec:renegotiation,ch:moralhazard2}: the commitment plan is not subgame perfect, because after the5058first sale both the seller and the remaining buyers gain from a new deal. Second, it is the screening5059problem of \cref{ch:mechanism} without commitment. Declining prices sort buyers by their values, exactly5060as a menu would; the revelation principle told us that with commitment the seller can do no better than a5061single posted price, and \cref{thm:coase} tells us that without commitment she does much worse. The5062revelation principle's assumption that the designer can bind herself is not a technicality. Third, it is5063the bargaining problem of \cref{ch:bargaining} with one-sided private information: a seller making5064repeated offers to a single buyer of unknown value is formally the same game, and the Coase conjecture5065says that the uninformed party's power to make all the offers is worthless when offers can be made5066quickly.5067 5068\section{Entry}5069 5070The threat of entry disciplines an incumbent only if she cannot deter it, and the theory of deterrence5071is a catalogue of the commitment and information effects already studied. An incumbent may install5072capacity that makes post-entry aggression credible \cite{Dix80}; this is \cref{ex:stackelberg} with the5073follower's option to stay out. Whether she should over-invest or under-invest depends on whether the5074post-entry game is one of strategic substitutes or complements and on whether investment makes her tough5075or soft: the fourfold taxonomy of \cite{FT84}. She may price low to signal low cost (\cref{ch:signalling}),5076or fight early entrants to maintain a reputation (\cref{ex:reputation}). In each case the conclusion of the5077equilibrium analysis is more guarded than the informal story. Capacity deters only if it is sunk; a low5078price deters only if it separates, and then entry occurs exactly when it would have anyway; a reputation5079deters only if there is some prior chance that it is deserved. The models do not say that predation and5080limit pricing are myths. They say what must be true of the world for them to work.5081 5082\section*{Notes}5083 5084The quantity and price models go back to Cournot and Bertrand; their reconciliation through capacity5085choice is due to Kreps and Scheinkman \cite{KS83}. Collusion in repeated price competition follows5086\cite{Fri71}. The search paradox is Diamond's \cite{Dia71}. Spatial competition is from Hotelling5087\cite{Hot29}, corrected by d'Aspremont, Gabszewicz and Thisse \cite{dGT79}. The durable-goods problem was5088posed by Coase \cite{Coa72} and resolved for stationary equilibria by Gul, Sonnenschein and Wilson5089\cite{GSW86}; the two-period example is in the manner of \cite{Bul82}. Capacity as commitment is5090\cite{Dix80}, and the taxonomy of strategic investment is \cite{FT84}. The standard text for this chapter5091is \cite{Tir88}, and the source text's treatment is \cite[ch.~14]{Ras06}.5092 5093\section*{Exercises}5094 5095\begin{exercise}5096For the linear market of \cref{ex:cournot} with $n$ firms, find the least discount factor at which grim5097trigger sustains the monopoly outcome under quantity competition, and compare with the Bertrand threshold5098$1-\tfrac1n$.5099\end{exercise}5100\begin{exercise}