Summary and conclusions
We examine the ability to detect discretionary accruals using several variants and extensions of theJones (1991)model
of discretionary accruals and estimation samples based on two alternative indicators of similarity: industry membership
(industry peers) and size (size peers). Our examination is motivated by the practical problem of sample attrition when
estimation samples are based on industry membership, particularly the SIC4 industry definition, and particularly for non-U.S. data.
Our main finding is that estimation samples based on similarity in size as measured by lagged total assets perform at
least as well as industry membership-based estimation samples, and often better, in detecting both seeded discretionary
accruals and observed discretionary accruals (as proxied by the existence of a restatement or an Accounting and Auditing
Enforcement Release). The superior discretionary accruals detection power of lagged asset-based estimation samples applies
to both U.S. data and non-U.S. data. For non-U.S. samples not constrained by the availability of industry peers, lagged asset-based estimation sample detection rates are similar to the detection rates observed for samples where we can perform a
controlled comparison of detection rates for industry-based peers and lagged asset-based peers.
We provide evidence of a tradeoff between increasing explanatory power ofnormalaccruals models and increasing detection
power forabnormalor discretionary accruals. While both size-based and industry-based estimation samples produce reasonable
explanatory power in estimating normal accruals, the industry-based samples achieve higher explanatory power. This result is
corroborated by additional analyses showing industry-based samples are characterized by normal accruals with greater congruity
than are size-based samples, although neither sample selection criterion yields normal accruals that are wholly or even
substantially congruent. In contrast, size-based estimation samples, in particular, samples based on similarity in lagged total
assets, oftenyield higher detection rates for abnormal or discretionary accruals. Viewed as awhole, these results suggest the greater
detection power of the size-based estimation samples is due to the greater stability of the accruals model regression estimated
using size-based samples, as compared to using industry-based samples, and this greater stability comes at the cost of lower
explanatory power.
Defining estimation samples (peer firms) based on similarity in lagged assets instead of industry membership has
substantial practical value in estimating discretionary accruals models because application of the size-based criterion
imposes no incremental sample loss, beyond the sample losses resulting from estimating the variables in the models. For
U.S. data, this means avoiding sample attrition of anywhere from 1–3% (SIC2 definitions) to 22–30% (SIC4 definitions).
The benefits, in terms of increased sample sizes, are much greater for non-U.S. data, where using lagged assets instead of
industry membership to identify estimation samples avoids sample attrition ranging from 32% to 93% (depending on
industry definitions and weighting schemes).
33
We do not imposeα1¼α2¼α3for each firm.
F. Ecker et al. / Journal of Accounting and Economics 56 (2013) 190–211 210
We believe our finding concerning the discretionary accruals detection power of lagged asset-based estimation samples
is important both in the U.S. context and in the non-U.S. context. For both settings we show that using lagged asset-based
samples rather than industry-based samples to estimate discretionary accruals models increases sample sizes, often
substantially, and generally results in equal or better detection of discretionary accruals. The overall effect is more dramatic
for non-U.S. settings and more valuable, because in those settings, entire countries dropped in an industry-based estimation
sample design can be retained in a size-based estimation sample design