Sample and descriptive statistics
We use financial accounting and segment data from Compustat, stock market data from CRSP, earnings announcement
and management forecast data from IBES, Section 404 material weakness data from Audit Analytics, and restatement data
from Andy Leone. Following Dyreng and Lindsey (2009), we set the following variables to zero if they are missing in
Compustat: advertising expense, research and development expense, tax loss carryforwards, intangible assets, special items,
and long-term debt. We also employ their method to correct for errors in foreign tax expense, foreign pre-tax income, pretax
domestic income, total pre-tax income, federal current tax expense, and worldwide current tax expense.
Our sample period runs from 1994 to 2010. We begin our sample period in 1994 with the enactment of SFAS No. 109, to
have consistent accounting for income taxes throughout the sample. For a firm-year observation to enter our sample, it has
to have non-missing data for each control variable and at least one IIQ proxy. Two of the IIQ proxies (Earnings Announcement
Speed and Management Forecast Accuracy) are available throughout the entire time period (1994–2010), whereas No Error
Restatement and No Material Weaknesses are available only from 1994 to 2005 and 2004 to 2010, respectively. The number of
firm-years in our annual Cash ETR regressions range from 9,096 (using Management Forecast Accuracy) to 33,246 (using
Earnings Announcement Speed). The sample size is reduced when we investigate Cash ETR Volatility (which uses variables
measured over five-year periods). Finally, we exclude all financial firms (SIC codes 6000–6999) from the sample.