This thesis evaluates the effects of regulation on the real economy in Industrial Organization, Banking and Finance.

Chapter 1 shows that a regulation designed to promote competition in one market can decrease competition in another market. I build a data set on the construction of fiber-optic networks in Japan from 2005 to 2009. Using linear regressions and probit regressions, I find that when access pricing regulation is more binding, construction of new optical-fiber networks by facilities-based entrants decreases.

In Chapter 2, I apply structural estimation techniques to this problem. Although I can learn the immediate effects of usage on investment decisions in reduced-form analysis, it is hard to learn the long-run effects on an oligopoly equilibrium without a dynamic structural model. By modeling competition in building fiber-optic networks as a dynamic game, we can analyze dynamic behavior of firms, and estimate profit and sunk cost. The estimation result shows that unbundling decreases profit from of building fiber lines for new entrants, and new entry decreases as a result.

The results of Chapters 1 and 2, suggest that unbundling regulation during periods of new technology diffusion may reduce the price of service but also decrease competition in the infrastructure market.

Chapter 3 investigates the mechanism by which bank failures affect the real economy from the lenders' perspective. Using a natural experimental setting of unique bank failures in Japan, this chapter identifies the credit crunch effect of bank failures. The main findings are the following. First, bank failures decrease the investment by client firms by approximately 30%. Second, the high investment growth/level firms deal with unhealthy banks. These choices generate a self-selection bias of 30–80%. Third, there is no evidence that the effect of a bank failure is related to the firms' accessibility to other financial sources. These findings suggest that large bank failures can trigger further recessions, if regulatory authorities do not prevent them.