This dissertation provides empirical evidence consistent with auditors transferring some information from one client to another and, as a consequence, with same industry rivals being reluctant to share the same auditor due to information spillover concerns.

In the first essay, using exogenous shocks to the auditing industry, including large auditor mergers and the collapse of Arthur Andersen, I document a reluctance of rivals to engage the same auditor due to information spillover concerns. This reluctance is more evident in concentrated industries where barriers to mobility, proxied by differentiation and capital expenditure levels, are low. More secretive manufacturing firms are also more reluctant to share their auditor with a rival. I also find weak evidence that the concern for information spillovers is lessened when rivals are dissimilar in terms of sales or when they are headquartered in the same state, where other conduits for information spillover, including employee turnover, are present. Last, I find some evidence that auditors extract rents from clients concerned about information spillovers.

In the second essay, I present evidence consistent with auditors being a conduit for information spillovers. My results indicate that firms sharing the same auditor are more alike in their investment, research and development, advertising and SG&A decisions compared to firms not sharing the same auditor. Causality is established using the collapse of Arthur Andersen as a natural experiment that broke channels for information spillovers for former clients of the auditing firm. Results are still valid after the enactment of Sarbanes-Oxley (SOX), indicating that the core auditing practice of auditors is a conduit for information spillovers. I also find some evidence that information spillovers are more prominent at the auditor office level. Last, additional evidence from patent citations indicates that information spillovers through sharing the same auditor may lead to dissemination of technological innovations among client firms.

We investigate the effects of information and diversification on cost of capital in a noisy rational expectations model. Assuming a factor structure for risky asset payoffs and two classes of investors, informed and uninformed, we show that in large economies the APT (Ross, 1976) holds and i) information from private signals about idiosyncratic shocks has no effect on cost of capital and ii) information from private signals about systematic factors affects cost of capital only through factor risk premiums; there is no effect on factor loadings. These results imply that there are no cross-sectional effects of information on cost of capital within large economies.