Spring 2011 Seminar Series

DATE: May 12th
LOCATION: E62-550
TIME: 4:15pm
Reception immediately following in the same room

SPEAKER:
George Shanthikumar

TITLE
When Does More Information-gathering Lead to More Production? The Case of the Newsvendor

ABSTRACT
Empirical research has found that advances in Information Technology (IT), accompanied by IT investments, largely explain the massive productivity gains of the 1990s. It is plausible, moreover that they partially explain the observed shifts in the composition of the US labor force in favor of IT-using labor. There is, however, very little in the way of clarifying theory. Do the shifts and productivity gains occur because a firm finds that its information-handling efforts and its IT capital are complements to its non-IT inputs, or are they substitutes for those inputs? We take initial steps by studying a model with just two persons, an Information-gatherer (IG) and a Producer. The Producer’s payoff depends on his product quantity and a randomly changing state. The IG provides signals which imply a posterior distribution on the state, and the signals become more reliable when the IG works harder. The Producer chooses a quantity that maximizes expected payoff under the posterior which the IG has supplied. We ask: when does more IG effort increase the average quantity chosen by the Producer (“Complements”) and when does it decrease that quantity “Substitutes”)? We focus on (1) an IG whose posteriors are scale/location transforms of a base distribution, where the average scale parameter drops when the IG works harder, and (2) a Producer who is a classic newsvendor. We find that the average ex post quantity delivered to users (when the newsvendor orders optimally) unambiguously rises when the IG works harder, but the average ex ante order rises if production cost exceeds a critical value and falls if it is below that value. A similar pattern emerges for another class of Producer, who suffers a penalty that depends, nonlinearly and asymmetrically, on the gap between the ideal quantity and the quantity chosen. That class includes a “generalized” newsvendor whose penalties for unsold inventory and unfulfilled demand are not linear and who may be viewed as “regret-averse”.