6. Although large firms may have incentives to reduce political costs by exploiting latitude in
accounting discretion (Watts and Zimmerman, 1978), they are likely to be under closer
scrutiny by outsiders than small firms. Such close scrutiny may potentially reduce
managers’ opportunities to exercise their accounting discretion.
7. For example, the 2004 survey found that 7 percent of companies do not have a majority of
independent directors on their boards, down from 21 percent in the 2002 survey; 59 percent
have boards with more than two-thirds of their directors entirely independent from
management, up from 43 percent in the 2002 survey; and 86 percent of companies have fully
independent audit committees in 2004, compared with 56 percent in 2002 (Globe and Mail, 12
October 2004).
8. The TSX undertook significant changes in 2003 and 2004. Thus, many firms not qualifying
for the index criteria were eliminated.
9. Unreported results indicate strong associations among the interactions of corporate
governance and the time indicator variable. Therefore, we do not include them together in
the model, but we include each individual interaction variable in combinations with
governance variables and other control variables.
10. We also rerun the regression by classifying the sample as “strong governance” if the score in
each governance category is above the median. The results are not materially affected by
this alternative classification.
11. Another possible target is meeting analysts’ earnings forecasts. Degeorge et al. (1999) find
that while managers appear to manipulate reported earnings upwards to meet analysts’
forecasts, earnings management to avoid losses and earnings declines proves predominant.
12. We also control for auditor tenure (Audit_T), measured as the number of consecutive years
that the client retained the same auditor since 1991, since prior research finds that earnings
management through discretionary accruals and special items decreases as audit tenure
increases. The coefficient on this variable is not statistically significant, and thus the result is
not reported.
13. Accordingly, using data for the eight-year period from 1995 to 2002, we estimate earnings
persistence for each available sample firm-year as the first-order autocorrelation in earnings
changes over the sample period prior to the sample year. Similarly, we estimate earnings
variability as the standard deviation in earnings changes during the same sample period.
14. The S&P/TSX was amended significantly in 2003 and 2004 to drop more than 40 companies
that did not meet new index criteria. Many of them were the smallest companies in the index
with the lowest governance scores in 2002.