Essays on network externalities and aggregate persistence.

다국어 초록 (Multilingual Abstract)

This dissertation studies the effects of network externalities on aggregate persistence in the areas of labor, money, and science. It is argued that network externalities provide a new explanation for, respectively, spatially persistent unemployment, dollarization hysteresis, and dominant scientific paradigms.
The first essay, entitled “Local Trade Networks and Spatially Persistent Unemployment”, studies the effect of network externalities on the spatial distribution of employment. Network externalities occur in the form of strategic complementarities in employment, i.e., the probability for a given agent of being employed increases with the average employment rate within this agent's trade network. It is shown that, if labor can be hired in continuous quantities, then the long run distribution of employment is uniform, and independent of any trade network topology. When labor has binary support, however, local trade networks are found to generate spatial unemployment clusters that can persist indefinitely.
The second essay, “Network Externalities and Dollarization Hysteresis: The Case of Russia”, develops a model in which network externalities in the demand for currency can lead to multiple steady state levels of dollarization. This model is estimated using a new source of data on dollar currency holdings in Russia. The results confirm the existence of multiple steady states, and suggest that a permanent decrease in dollarization in Russia could be obtained by a temporary appreciation of the ruble or a temporary increase in enforcement of the law that makes it illegal to carry out transactions in dollars.
In the final essay, “Market Failures in the Economics of Science”, science is modeled as a market for scientific articles that is governed by a citation mechanism, due to which scientists have an incentive to supply those articles that they expect other scientists to be willing to cite. This implies the existence of network externalities, due to which the probability that a given scientist adopts a given paradigm increases with the average support for this paradigm by other scientists. While under certain conditions such network externalities can speed up scientific progress, they can also lead a discipline to become locked into a suboptimal paradigm, thus implying a market failure.