Sumit Joshi

Networks of collaboration in oligopoly

In an oligopoly, prior to competing in the market, firms have an opportunity to form pair-wisecollaborative links with other firms. These pair-wise links involve a commitment of resources andlead to lower costs of production of the collaborating firms. The collection of pair-wise linksdefines a collaboration network. We study the architecture of strategically stable networks.Our analysis reveals that in a setting where firms are ex-ante identical, strategically stablenetworks are often asymmetric, with some firms having a large number of links while others havefew links or no links at all. We characterize such asymmetric networks; the dominant grouparchitecture, stars, and inter-linked stars are found to be stable. In asymmetric networks, thefirms with many links have lower costs of production as compared to firms with few links. Thuscollaboration links can have a major influence on the functioning of the market.

Bilateralism and Free Trade

We study a setting with many countries; in each country there are firms that can sell in the domestic as well as foreign markets. Countries can sign bilateral free-trade agreements that lower import tariffs and thereby facilitate trade. We allow a country to sign any number of bilateral free-trade agreements. A profile of free-trade agreements defines the trading regime. Our principal finding is that, in symmetric settings, bilateralism is consistent with global free trade. We also explore the effects of asymmetries across countries and political economy considerations on the incentives to form trade agreements.

Turnpike Theorems in Nonconvex Nonstationary Environments

This paper provides a comprehensive development of turnpike theory in a stochastic aggregative model with time-varying nonconvex technology. A new approach to turnpike theorems is developed that exploits the monotonicity of optimal programs and utilizes a supermartingale process generated by stochastic Euler equations. This extends the classical turnpike theory to general nonconvex, nonstationary environments. Copyright 1997 by Economics Department of the University of Pennsylvania and the Osaka University Institute of Social and Economic Research Association.

Convergence to symmetry in dynamic strategic models of R&D: The undiscounted case

Abstract Will R&D increase or decrease the asymmetry between firms over time? We examine this issue in the context of a dynamic, alternate-move duopoly model of non-cooperative R&D. The asymmetry we consider is with respect to the initial stocks of technological knowledge which provides one firm a greater potential for current and future profits in the product market. Utilizing a value loss process constructed from the Optimality Equation of dynamic programming, we show that for the undiscounted case the asymmetry between the firms disappears over time. We obtain this convergence-to-symmetry result by imparting a temporal character to R&D through the notion that R&D investment cannot be changed instantaneously, by allowing research externalities, complementarity or substitutability between own and appropriated R&D, and either increasing, decreasing or constant returns to scale in the production of technical knowledge from own and rival R&D.

Recursive utility and optimal growth under uncertainty

Abstract This paper extends the theory of optimal growth under uncertainty to a general class of recursive utility preferences. With this general description of preferences, and in a framework which incorporates time-varying productive technologies and non-stationarities in the evolution of the stochastic environment, the monotonicity properties of optimal programs with respect to the various parameters of the model is demonstrated and then utilized to develop the turnpike theorems.

The privacy cost of the second-chance offer

This paper examines a generalization of a two-stage game common on eBay: an ascending-price auction followed by price discrimination (the second chance offer). High bids in the auction lead to high price offers during price discrimination, and a financial disadvantage in the second stage. The disadvantage depends on (a) the amount of information revealed to the seller in the first stage, and hence the extent of privacy protection provided and (b) whether the bidder is non-strategic (ignores the possibility of price discrimination) or rational. A privacy cost of one mechanism over another is defined and studied.For the non-strategic bidder, the second chance offer provides a zero payoff. Addition of privacy protection (anonymity and bid secrecy) decreases revenue and increases expected payoff, with higher bidders benefiting more. Privacy protection can, however, decrease an individual bidders payoff by shielding potential buyers from the seller and thus causing an opportunity loss.If the bidder is rational, price discrimination results in a lower revenue than consecutive auctions, and is a bad strategy for the seller. Additionally, rational behavior provides more advantage to the bidder than does anonymity protection.

Existence in undiscounted non-stationary non-convex multisector environments

Abstract An omission in growth theory concerns the existence of optimal programs in models that simultaneously include non-stationary non-convex technologies, non-convex preferences, multisectors and undiscounted utility sums. Non-stationarities rule out the classical route employing golden rule programs; non-convexities vitiate classical duality methods based on separating hyperplane theorems; multisectors destroy the regularity in the dynamics of the one-sector growth model; and divergent utility sums render inapplicable the Weierstrass approach based on compactness-continuity arguments. This paper proves the existence of optimal programs in such generally specified models, by utilizing the non-convex duality theory of Balder to replace the bilinear functional that links a topological dual pair by a class of non-linear functions that are of needle type at the origin of the commodity space.

Sanctions in networks: “The Most Unkindest Cut of All”

The extensive literature on sanctions has mainly focused on a dyadic interaction between sender and target. In contrast, this paper examines sanctions when the sender and target are embedded in a network of linkages to other agents. The sender can assemble a sanctioning coalition of neighbors to sever their links (execute multi-link cuts) to the target and her allies. Efficacy of sanctions is now crucially dependent on the network architecture. We characterize the structural properties of networks in which a sender can effectively sanction a target in the short run (when links can only be deleted) and the long run (when links can be both deleted and added).

Dynamic Cooperation in R&D

It has long been recognized that dynamic processes underlie technical advance.1 Moreover, common economic parlance has it that firms in industries considered ‘strategic’ by developed countries engage in technology-based competition. What we have additionally come to realize, however, is that the structure of the ‘technical enterprise’ as a whole may be both more complex than earlier economic accounts led us believe and changing fast (e.g., Fusfeld, 1986, 1994). Gone is the conceptualization of the individual firm marshalling its own technological capabilities in order to gain over its rivals. The spread of technological capabilities around the world and the intensifying competition have obliged even the largest firms to look further and beyond their internal research and development (R&D) laboratories in order to keep up with the latest developments of rapidly evolving technologies as well as ease the excessive burden of exploding R&D expenditures on their budgets. Thus, in addition to direct competition, firms have increasingly utilized various forms of vertical and horizontal cooperative agreements (strategic alliances) to pursue their competitive objectives.

Randomization as a strategy for sellers during price discrimination, and impact on bidders' privacy

A previous paper demonstrates that if a seller always uses auction bids to later price discriminate against losing bidders, his revenue decreases dramatically. In this paper, we examine whether the seller obtains an advantage if he randomizes his strategy -- that is, if he does not use privacy-infringing information all the time, but only with probability ?;. Using both Bayesian techniques and genetic algorithm experiments, we determine optimal strategies for bidders and sellers in a two stage game: Stage I is a first price auction used to elicit information on a bidders valuation; Stage II is, with probability ?;, a price discrimination offer, and, a fixed price offer P; else. Our results show that the seller does not benefit from randomized price discrimination. Further, low valuation bidders benefit more from the sellers use of privacy-infringing information than do the high valuation ones, as they may wish to signal that they cannot afford a high second-stage offer. To our knowledge, our use of genetic algorithm simulations is unique in the privacy literature.