The past decade has seen the explosive emergence of online advertising as a major source of revenue for Internet publishers. Analyses of this phenomenon are mostly conducted in the sway of Google’s hugely successful search advertising program. In the early days of the Internet, before Google, virtually all advertising revenues were related to simple display ads. Yet by 2008 search advertising accounted for over $10.5 billion of the $23.4 billion in total online advertising, and pundits were forecasting continued growth at rates of 12 % per year over the next five years.1 Internet advertising markets have broken sharply from the advertising markets for traditional media. In the older media, every consumer that received a particular magazine, listened to a particular radio program, or watched a particular TV show would read, hear or see the same advertisement. An advertiser that wanted to reach an audience with particular characteristics could do so only within narrow limits. For example, a beer company might advertise on televised football games and a maker of fashion clothing might advertise in women's magazines. Although publications do some tailoring of their offerings, as when a newspaper has different local editions, audience mix is nevertheless

Forthcoming in the American Economic Review The LeChatelier principle, in the form introduced into economics by Samuelson, asserts that at a point of long-run equilibrium, the derivative of long-run compensated demand with respect to own price is larger in magnitude than the derivative of short-run compensated demand. We introduce an extended LeChatelier principle that applies also to large price changes and to uncompensated demand as well as to a wide range of concave and nonconcave maximization problems outside the scope of demand theory. This extension also clarifies the intuitive basis of the principle.

Manufacturing is undergoing a revolution. The mass production model is being replaced by a vision of a flexible multiproduct firm that emphasizes quality and speedy response to market conditions while utilizing technologically advanced equipment and new forms of organization. The authors' optimizing model of the firm generates many of the observed patterns that mark modern manufacturing. Central to the authors' results is a method of handling optimization and comparative statics problems that requires neither differentiability nor convexity.

In a repeated partnership game with imperfect monitoring, we distinguish among the effects of (1) reducing the interest rate, (2) shortening the period over which actions are held fixed, and (3) shortening the lag with which accumulated information is reported.All three changes are equivalent in games with perfect monitoring.With imperfect monitoring, reducing the interest rate always increases the possibilities for cooperation, but the other two changes always have the reverse effect when the interest rate is small.

In Section 2, we review some important results of the received auction theory, introduce a new general auction model, and summarize the results of our analysis. Section 3 contains a formal statement of our model, and develops the properties of affiliated random variables. The various theorems are presented in Sections 4-8. In Section 9, we offer our views on the current state of auction theory. Following Section 9 is a technical appendix dealing with affiliated random variables.

We study the class of multiround, multiproduct clock procurement auctions that reduce offered prices at each round. When prices stop declining, the remaining bidders become the winning sellers. For single-minded bidders, each such auction has five properties not shared by Vickrey auctions: each is obviously strategy-proof and group-strategy-proof, sets prices that are Nash equilibrium winning bids in the related first-price auction, preserves winner privacy about values, and can be extended to satisfy a budget constraint. In simulations of the US incentive auction, a heuristic clock auction from this class achieves quick computations, high efficiency, and low prices.

We present a signaling model, based on ideas of Phillip Nelson, in which both the introductory price and the level of directly "uninformative" advertising or other dissipative marketing expenditures are choice variables and may be used as signals for the initially unobservable quality of a newly introduced experience good. Repeat purchases play a crucial role in our model. A second focus of the paper is on illustrating an approach to refining the set of equilibria in signalling games with multiple potential signals.

In the 19th century, the railroad and telegraph were at the center of a set of technological advances, physical investments and managerial innovations that transformed American industry (Alfred Chandler). Later, the automobile and telephone played a similar role in another transformation. Today, the high-tech industries include computers, telecommunications and electronics. Working on our remarkably powerful computers (even as they rapidly become obsolete), coauthoring papers by electronic mail and fax, and conversing on our portable cellular telephones, we are struck by what appears to be a self-supporting and reinforcing dynamic to the technological improvements across the electronics industries. An advance almost anywhere in the sector seems to call forth more advances across the sector. These advances are occurring contemporaneously with a broad pattern of other changes, not only in the electronics industries, but in manufacturing more generally, and not just in hardware, but in methods and organization as well. A new paradigm has begun to emerge. In contrast to traditional manufacturing firms, modern firms frequently (1) make greater use of flexible, programmable

When a job-seeker and an employer meet, find a prospective joint surplus, and bargain over the wage, conditions in the outside labor market, including especially unemployment, may have limited influence. The job-seeker's only credible threat during bargaining is to hold out for a better deal. The employer's threat is to delay bargaining. Consequently, the outcome of the bargain depends on the relative costs of delays to the parties, rather than on the payoffs that result from exiting negotiations. Modeling bargaining in this way makes wages less responsive to unemployment. A stochastic model of the labor market with credible bargaining and reasonable parameter values yields larger employment fluctuations than does the standard Mortensen-Pissarides model.