Essays in organizational economics.

다국어 초록 (Multilingual Abstract)

This dissertation consists of three essays in organizational economics. The three essays in turn focus on how the organization of a firm is driven by the probabilities of the tasks the firm is facing, the nature of the uncertainty the firm confronts, and how cheaper communication technology facilitates contingency decision making. In particular, all three essays use discrete modeling approaches, where expertise, inputs, and assets are assumed to be discrete units.
In the first essay, this approach enables me to study the impact of "combinatorial power" of knowledge and inputs: If each production task requires two of four possible inputs, and every worker can provide two inputs, than one worker can only solve one out of four tasks. However, two workers together can solve all four tasks. This "combination of expertise" can result in convex returns to inputs and knowledge. I show how the probability distribution over such a collection of tasks and the complexity of the tasks drive the organizational structure of the firm. I find that hierarchies are optimal if a few simple tasks re-occur frequently. Teams are optimal when firms face high task uncertainty or high task complexity, but these teams differ along other organizational dimensions such as capital investments.
In the second essay I use a related approach to consider the uncertainty a firm may face about a task at hand and thus about the required inputs. There is no ordering among the discrete inputs in the sense that "more" always results in a higher expected outcome. Instead production is successful if and only if the right inputs are provided. Thus, production becomes a matching problem with an uncertain target. This "input uncertainty" is distinct from the typically considered "value uncertainty," i.e., a perturbation of the outcome given a "best effort" input. I explore the different implications that input and value uncertainty have on the choices firms make. I show that firms organize differently given their different ex-ante knowledge about the inputs needed for successful production, and that they write different sets of incentive contracts reflecting the distinct sources of outcome uncertainty. I also show that output reveals different information about a worker's type in a firm facing input uncertainty than in a firm facing value uncertainty.
Finally, in the third essay, I comment on the paper cite{HartMoore2006} which provides a model of authority-based hierarchies. In that paper, agents are assigned assets which they can use to implement ideas. Each asset is assigned a chain of command, and the asset is claimed by the highest-ranked agent that has an idea. Ideas can only be implemented if all necessary assets are available. Importantly, Hart and Moore assume that an agent exercises authority over the assets assigned to him whenever he has an idea---whether or not he is able to implement it. In particular, if one agent outranks another agent on one asset but is outranked by the same agent on another asset, then these two agents would block the implementation of both their ideas. I call such a "crossing of authority" an authority tie. The original assumptions imply that authority-ties are resolved by not letting either agent implement their idea. As a consequence, it is never optimal for an organization to allow potential authority ties. The advantage of this authority tie-break rule is that it is not contingent on who else in the organization may have had an idea. This case corresponds to an organization with a strictly local decision making mechanism. In my comment I suggest that other tie-break rules are possible, and potentially result in first-best---albeit at a higher communication cost.