The unit of analysis used thus far is the county not the firm. We emphasize county-level analysis of misreporting in the
paper for two reasons. First, the county, as opposed to the firm, is a natural unit of analysis in our context because our
proxies for perceived costs such as distance from the SEC office and past SEC activity in the county do not display cross-
firm variation for firms located in any given county. Second, our design is consistent with the economics literature
that demonstrates clustering of crime or other economic activity for a spatial unit such a precinct, county or a state
(e.g., Audretsch and Stephan, 1996; Glaeser et al., 1996)