Joint with Balakrishnan Blouin and Guay
The Accounting Review
ONLINE APPENDIX 1
Examples
In this appendix, we provide two detailed examples of firms’ tax planning and how their
planning maps into our measures of tax aggressiveness.
Google
In 2005, Google appears to have transferred the licensing rights for its intellectual
property to Ireland. By shifting its valuable intangible rights to Ireland, Google effectively shifts
income out of the U.S. (where revenues used to all be reported) and into a more lightly taxed
jurisdiction. In Google’s 2005 3rd Quarter 10-Q, Google states:
Our provision for income taxes increased to $408.7 million, or an
effective tax rate of 27% in the nine months ended September 30,
2005 … The decrease in our effective tax rate was primarily
because proportionately more of our earnings in 2005 compared
to 2004 are expected to be recognized by our Irish subsidiary and
such earnings are taxed at a lower statutory tax rate than in the
U.S.
We expect our effective tax rate to be approximately 30% for
2005. However, if future earnings recognized by our Irish
subsidiary are not at the levels we expect, our effective tax rate
will be higher than our expectations.
However, evidence suggests that Google was still in the process of negotiating its transfer
pricing related to the licensing rights with the IRS into 2006 (see Drucker, 2010b). During the
4th quarter of 2005 (see their 8-K filed on January 31, 2006), Google determined that it had either
over-allocated income or under-allocated expenses to its Irish operations thereby whipsawing its
quarterly GAAP effective tax rate form 27% to 41.8%:
Our effective tax rate for the fourth quarter increased to 41.8%
this quarter, and to 31.6% for the year, above our previously
announced expectation of approximately 30% for the year.
Primarily because the proportion of total expenses allocated to
our international operations was greater than we anticipated,
more of our profits were taxed at a higher domestic tax rate;
this resulted in a greater effective tax rate compared to our
expectations. We expect our effective tax rate for 2006 to be
approximately 30%.
Google did not meet its fourth quarter 2005 earnings target. Although Google executives
attempted to explain that the unexpected increase in the effective tax rate was due to “complex”
tax matters, analysts appear to have presumed that Google’s lower earnings was due to slower
than anticipated growth leading some to downgrade the stock:
"We're downgrading Google primarily because of concerns
about weaker-than-expected international revenue growth,"
said Ben Schachter, a UBS Securities analyst. "I think they
are investing heavily in that area, and that is the right thing
for the company in the long term. But in the near term, it will
put pressure on its margins for the next couple of quarters."
Another analyst reported the following on his blog after the conference call
(http://paul.kedrosky.com/archives/002512.html):
Why the tax rate was higher than expected -- Google doesn't
make that question easy to answer. Here is CFO George
Reyes doing obfuscatory tax-talk on the conference call:
[Tax rate] estimates are complex, and 2005 was the first year
we realized any reduction to our effective tax rate as a result
of profits earned overseas under our international structure.
At the end of the year, we must true up the tax provision for
the year, which could, and in the case of Q4, did have a
disproportionate impact on the fourth quarter.
In calculating our true-up for the year, a portion of expenses
allocated to international operations was greater than we
expected. Primarily as a result of this, a greater percentage
of our profits were taxed at a higher domestic tax rate, which
resulted in a greater effective tax rate compared to our
expectations.
Basically, Reyes is saying that Google spent more than it
planned to on its international business, and there wasn't
corresponding international revenue to match against it. As a
result, it had a higher percentage of its income in the U.S.,
and that meant a higher overall tax rate.
Okay, but why didn't international perform? I mean, Google
had to know it was "over-spending" internationally. Here's
Eric Schmidt's canned response to that question when asked
it twice on the call:
We think the opportunity ahead of us particularly in
international markets is just exceptional. And what you are
seeing is the investments that need to be made to sort of
harvest that opportunity.
Fair enough. Eric's basically saying "trust us", we're spending
madly on international markets, but we know what we're
doing. It's not a terrible answer, but it's also one that requires
more faith in a young public company than some people
might feel inclined to offer.
Overall, we conclude that there was significant confusion on the behalf of analysts regarding the
separation of the tax related issues from growth estimates. Note that the fact that Google’s GAAP
effective tax rate was affected so severely in the 4th quarter is consistent with some incremental
cost allocation rather than a sudden drop in revenue. Furthermore, no evidence of any drop in
revenue growth can be gleaned from the 2005 financial statements.
Google’s GAAP effective tax rate did drop from 38.6% in 2004 to 31.6% in 2005. In 2006,
Google must have ultimately negotiated a favorable transfer pricing agreement as its effective tax
rate dropped to 23.3%. To provide some context for our tax aggressiveness measures, we estimate
Google’s TA_GAAP and TA_CASH using the methodology described in Section II. We then scale
all TA measures by the applicable size industry mean GAAP ETR and CASH ETR. Scaling allows
us to compare the measure across firms and over time. We then sort the scale measures and ranked
them by year into deciles (low deciles suggesting relatively more tax aggressiveness).
Using this methodology, in 2004 Google’s level of tax aggressiveness fell into decile 8
suggesting that it was relatively unaggressive. However, in 2005, it moved up to decile 5 and
then, in 2006, to decile 2 (for both TA_CASH and TA_GAAP). These rather large changes in
ranking imply a sizeable shift in the level of Google’s tax planning. Google remained in decile 2
through 2013 (overlapping a period where Google undertook some significant foreign tax
planning as described in Drucker, 2010b).
Forest Laboratories
Forest Labs has been identified as a firm that undertook the Double Irish and Dutch
Sandwich transaction during 2005/2006 as described in the introduction. Yet, Forest Labs
appears to have begun its extensive international tax planning agenda in fiscal year 2000. In its
2000 10-K, Forest reports net foreign revenue of $36.61 million but foreign pre-tax income of
$67.827 million. Compare these amounts to what was reported in its 1999 10-K: $37.044
million of revenues yielding $20.576 million in foreign pre-tax income. This large shift in pretax
income without any corresponding change in revenues suggests the presence of significant
intercompany payments – likely royalty payments attributable to the transfer of intellectual
property into Ireland.
Unfortunately, Forest Labs does not report an Exhibit 21 until 2004. In its 3/31/2004
Exhibit 21, it reports three foreign subsidiaries. After Forest Labs undertakes the tax planning
described in Drucker 2010a, it reports six foreign subsidiaries (2006 Exhibit 21). This list
includes the two new Dutch subsidiaries created for the Dutch sandwich. Interestingly, there is
no mention of any Bermuda subsidiary. However, as Exhibit 21 only requires firms to report the
foreign subsidiaries incorporation jurisdiction, the Bermuda activity must be included in one of
its Irish subsidiaries.
In terms of our measures of tax aggressiveness, following the methodology described in
the Google example, Forest Labs was in the 9th decile in 1999. Once it began its multinational
tax planning, it dropped all the way to decile 2 in 2000 and has remained in either decile 2 or 3
(for both TA_GAAP and TA_CASH) ever since.
ONLINE APPENDIX 2
Discussion of Limits of Alternative Measures of Tax Aggressiveness
In this appendix, we discussion some of the drawbacks of other measures currently used to
capture tax aggressiveness.
In addition to GAAP and Cash ETRs discussed in Section II, the literature also uses
measures of tax shelter probability, haven activity, FIN 48 and discretionary permanent differences
to capture tax aggressive behavior. We discuss the relative strengths and weaknesses of each
measure in turn in this appendix.
Wilson (2009) and Lisowsky (2010) create measures of the estimated probability that a
firm has entered into tax shelter. Both measures are derived by estimating a probability from the
coefficients of a logit model of attributes of firms discovered engaging in shelter activity. These
measures, however, rely on a very small sample of firms whose shelter behavior was a) detected
by the tax authorities and b) potentially litigated, and are therefore unlikely to be representative of
the broad set of firms that engage in extensive tax planning.1 Frank, Lynch, and Rego (2009)
develop a measure of discretionary permanent book-tax differences, DTAX, which relies on the
premise that certain permanent differences are more aggressive than timing differences. Although
anecdotal evidence suggests that the optimal tax planning opportunity is one that creates permanent
differences because of their financial statement benefits, there is little evidence to support this
conjecture (see Hanlon and Heitzman, 2010).
Dyreng and Lindsey (2009) document that having material subsidiaries located in tax
havens is associated with lower firm tax burdens. Tax havens are jurisdictions that structure a tax
1 Wilson (2009) relies on court records discussed in the popular press to identify his shelter firms. Lisowsky (2010),
which is an extension of Wilson (2009), identifies shelters using proprietary IRS data. His sample of shelter
transactions comes from the Office of Tax Shelter Analysis, which identifies shelters through the audit process.
regime to take advantage of firms’ desire to reduce their tax burdens. Generally, tax havens have
low or no tax rates and have very little information sharing of tax information with other
jurisdictions thereby making it more difficult for one jurisdiction to determine whether a firm is
artificially stripping its earnings into the haven. However, a haven could also include countries
that have modified their tax laws to attract foreign capital (e.g., Ireland). Either type of haven could
be utilized for tax avoidance purposes.
De Waegenaere, Sansing, and Wielhouwer (2015) provide a theoretical analysis suggesting
that the FIN 48 unrecognized tax benefit is the best financial-statement-generated measure of the
level of firm tax aggressiveness. However, their assertion is contingent on the quality of firms’
compliance with the FIN 48 reporting regime (e.g., its usefulness as a measure of tax
aggressiveness would be eliminated if firms use the FIN 48 accrual to manage earnings, as
suggested by Hanlon and Heitzman, 2010). Another potential drawback to FIN 48 is that this
information is only available beginning in 2007. Finally, Lisowsky, Robinson, Schmidt (2013)
provide a continuum of the ability of specific measures of firms’ tax attributes to capture tax
aggressiveness. However, they do not empirically test whether FIN 48 accruals capture aggressive
tax planning more generally, only that they are associated with disclosed listed transactions.
Although the shelter probabilities, DTAX, tax haven usage and FIN 48 are likely
correlated with aggressive tax planning, none of the measures satisfactorily captures the full
array of tax planning activities. The shelter probabilities rely on detecting specific law–breaking
tax-related transactions, and although such transactions are clearly aggressive, the shelter
probability measure does not capture many commonly employed legal tax planning efforts (e.g.,
transfer pricing). Our objective is to develop a measure of tax aggressiveness that includes both
legal tax planning (e.g., aggressive transfer pricing) strategies as well as inappropriate or “grey”
tax shelter activity. Further, our intention is to study variation in aggressive tax planning that
stems from both timing and permanent differences, as opposed to the DTAX measure which
includes only aggressive permanent differences. Irrespective of our concerns with these existing
measures of tax aggressiveness, we provide evidence later in the paper that some of our results
are robust to using DTAX and tax shelter probabilities as alternative proxies for tax
aggressiveness. Although the use of tax havens seems ubiquitous, there are many firms that do
not have significant off-shore activity. Simply relying on the use of a tax haven to detect tax
avoidance would preclude researchers from detecting tax aggressiveness in a sample of
domestic-only firms. Finally, FIN 48 captures the management’s assessment of the extent of the
firm’s tax aggressiveness. Measurement of the FIN 48 reserve requires management to assess not
only the probability that the firm will prevail in a court of law but also the median outcome of
any settlement activity. De Waegenaere et al. (2015) point out that the more managerial
discretion influences the FIN 48 balance, the less useful the measure is to capture tax
aggressiveness.