 |
| Matthew Frank |
I will start with a few conclusions.
The advance pricing agreement, or APA, program provides a valuable service to the IRS and to taxpayers alike, and is staffed by smart, hard-working, public-spirited professionals. The program should receive robust support from the IRS to expand its operations. The IRS charges taxpayers a hefty fee to use the program so the program mostly pays for itself; expanding it would not come at the expense of the public fisc.
The uptick in APA completions in 2020 versus the prior year — 127 completions versus 120 — was a terrific accomplishment given the COVID-19 backdrop. The increase in staffing in mid-2020 — 20 professionals were added to the team — should help the program going forward, but the program has a lot of catching up to do and still has seven fewer staff than the 104 persons it had in 2012.
It takes too long to complete an APA. Yes, this is a perennial complaint but the situation in recent years hasn't improved notwithstanding the program's reorganization in 2012, which was supposed to streamline and speed the process. No doubt the delay in completing APAs is due to understaffing, but it's also a matter of case management and taxpayer unpreparedness.
With more resources and improved timeliness, the program should look to expand its appeal. Much of the program is now devoted to the same taxpayers and indeed the same APAs that it has serviced in the past. APA renewals in 2020 constituted more than half of APAs completions — 59% in 2020, up from an already high 57% in 2019.
More broadly, excluding the relatively new U.S.-India docket, APA renewals account for more than 96% of the growth in the APA program inventory over the last 20 years. This shouldn't be. An opportunity is lost if the program is not attracting new taxpayers, particularly when the need for transfer pricing certainty is as great or greater than at any time in the past.
Some modest suggestions for the program, directed both to taxpayers and to the IRS, are outlined in the final section of this discussion.
The Year 2020 in Review
The program has been marked throughout its 30-year history by its constancy, discussed below, and while some metrics move year-to-year, the year 2020 fell into line in many ways notwithstanding the COVID-19 circumstances.
The program completed 127 APAs in 2020, compared to 120 completions in 2019 and an average completion total of 108 APAs in each of the prior five years. Eighty-five percent of the completed APAs were bilateral APAs, compared to an average of 75% in the prior five years. Fifty-nine percent of the completed APAs were renewals, compared to a prior five-year average of 58%.
The average age of a new bilateral APA completed in 2020 was 50.8 months, compared to an average of 46.7 months in the prior five years.
For bilateral APA renewals, the average age in 2020 was 34.1 months, significantly better than the 39.6 month average for bilateral APA renewals completed in the prior five years.
The program received 121 APA requests in 2020, a drop from the prior five-year average of 141 APA requests but equal to the 121 requests received in 2019. As usual, bilateral requests accounted for the lion's share, about 88% of the total. The prior five-year average was 83%.
With APA completions exceeding new APA requests, the number of pending APAs in inventory at the end of the year decreased by six, to 448 cases from 454 cases at the end of 2019. This is the second year in a row the number of pending cases in inventory has declined slightly — which coupled with the recent staffing increase may bode well for processing times going forward.
In terms of the geographic mix of the bilateral inventory and APA completions, 2020 was in line with historical trends. As always, the U.S.-Japan bilateral flow dominated, accounting for 41% of the APA bilateral requests filed during the year and 52% of the bilateral APAs completed. More on the geographic makeup of the inventory and APA completions below.
So, in short, the program in 2020 versus the prior five-year period saw:
- An 18% increase — 127 versus 108 — in the number of APAs completed;
- A 14% decrease — 121 versus 141 — in the number of APA requests filed;
- A 9% increase in the age of new bilateral APAs completed but a 14% improvement in the age of bilateral APA renewals completed;
- A small — six-case — decline in the number of cases in year-end inventory;
- A higher percentage of bilateral APAs — 85% versus 75% — and a higher percentage of renewals — 59% versus 58% — among APAs completed; and
- A geographic mix of cases in inventory that is fairly constant.
The headline from the 2020 report is likely to be that APA completions increased to an impressive 127 cases notwithstanding the COVID-19 pandemic. This was the third highest number of annual completions in the 30-year history of the program, and the highest number since 2013. This is good news.
The APA Program: A 30-Year View
The long view confirms what is known to anyone who is familiar with the program. It shows that the program has grown, reflected by the number of APA requests, the number of APAs completed and the size of its annual year-end inventory.
In its first 10 years of operation, the APA program received on average 49 APA requests a year, in its second decade, an average of 103 APA requests a year and, in its most recent decade, an average of 127 APA requests a year.
The table below shows the growth in APA requests filed each year, including the split between unilateral APA requests and bilateral APA requests beginning in 2001. (All references to bilateral APAs in this article include multilateral APAs as well, which represent a very small fraction of the total.)
The APA program's completion rate averaged 29 per year during its first decade, 68 per year during its second and 109 per year during the decade just concluded. The table below shows the number of APAs completed each year, and the split between new APAs and APA renewals.
As the number of APA requests has generally outpaced the number of APA completions — 2020 being an exception — the size of the APA year-end inventory has trended higher. There were 448 cases in inventory at the end of 2020, 10 fewer than the 2018 all-time high.
Things That Remained the Same
I wrote an article in 2006 on the program's 15th anniversary, noting that:
This statement continues to be accurate. A few highlights:One of the striking things to an observer of the US program over the last 15-plus years is the constancy. The taxpayers, the transactions, and the transfer pricing methods (TPMs) have remained remarkably the same.
Foreign companies dominate.
The program continues to be dominated by foreign-parented companies. Of the first 50 APA applications filed in the early 1990s, 60% were filed by foreign-parented companies. The same percentage held true for the next 50 applications, and have remained largely the same throughout the program's history, continuing to the present.
In the 10-year period 2001-2010, an average of 62% of all APAs completed each year involved a foreign parent transacting with a U.S. subsidiary versus 31% that involved a U.S. parent transacting with a foreign subsidiary. The rest involved transactions between brother-sister corporations or other structures.
In the more recent 10-year period, 2011-2020, the percentage is 61% foreign parent versus 26% U.S. parent. APAs involving a foreign parent outnumber APAs involving a U.S. parent by more than 2 to 1.
The comparable profits method dominates.
The use of the comparable profits method, or CPM, continues to dominate. While some have argued recently that the CPM is a method of last resort, this has never been true in the APA program.
In the nine-year period 2012-2020 for which data is most readily available, the CPM approach has been used on average for more than 80% of the transactions involving tangible or intangible property, each year ranging from 75% to 89% of the cases.
The geographic composition of the APA inventory remains familiar.
Another constant — with one big exception — has been the geographic composition of the APA bilateral inventory which has always been, and continues to be, dominated by Japan, with significant participation also by Canada and Korea, and to a lesser extent the United Kingdom.
India is the exception, of course, having come onto the U.S. bilateral APA scene only recently and generating a large number of APA requests — approximately 100 — over the last five years.
The table below shows the geographic composition of the U.S. bilateral APA inventory for each of the most recent eight years, 2013-2020. The absence of a line for some countries in some years does not mean that there were no cases in inventory involving that country, only that the IRS did not disclose the total. The countries in the "All Other" basket include most prominently Switzerland, France, Australia, Italy, the Netherlands, Belgium, Denmark and other Nordic countries.
APA terms have grown slightly.
The length of APA terms has grown slightly over the years. In the 10-year period 2001-2010, the simple average APA term was 5.4 years. In the more recent 10-year period, the simple average term has grown to 6.4 years. The extra 12 months are welcome.
Things That Changed
While constancy prevails generally in the operation of the APA program, there have been two striking changes in program operations over the last 20 or 30 years. One is the profile of APA taxpayers.
When the U.S. program began, taxpayers commonly were motivated to seek an APA as an alternative dispute resolution mechanism to address conflicts with the IRS field. (One of my early APAs, years before I became APA director, arose in precisely this context following an extraordinarily contentious exam.)
Over the last many years, however, it is increasingly common for taxpayers to enter the U.S. APA program even though they have a good relationship with the IRS exam team — or even because they have a good relationship — to allow the IRS to serve as a counterweight to the overly aggressive demands of foreign tax authorities. Taxpayers are coming to the program to seek IRS help against foreign tax authorities rather than to seek protection from the IRS field.
The second major shift in the APA program is the relative demotion of unilateral APAs to a sideline feature.
The absolute number of unilateral APAs has remained fairly constant over the last 20 years, averaging 31 APA requests a year and an average inventory of about 60 cases — which makes sense because a unilateral APA takes on average between two and three years to complete. But with the growth of the program overall, unilateral APAs are accounting for a diminishing percentage of the program's work.
Unilateral APA requests represented 40% to 45% of all requests during the first decade —1991-2000 — and 36% of all requests in the period 2001-2010. Those percentages have been cut nearly in half, to 19% in the more recent 10-year period — 2011-2020 — and to just 15% in the last five years.
Equally striking is that new (nonrenewal) unilateral APA requests and completions are becoming rare. The table below shows that while total unilateral APA completions have remained fairly stable in absolute terms, the number of new unilateral APA completions continues to fall.
The program has averaged just 7.6 new unilateral APA completions each year over the last five years, down nearly 40% from the average of 12.4 completions during the five-year period a decade before.
The decline of unilateral APAs is a function, first, of the expansion of APA programs in other countries, e.g., India and China, which make bilateral APAs realistic in cases where they were not available before.
A second reason for the decline in unilateral APAs as a percentage of the total is the very significant uptick in transfer pricing enforcement by tax authorities outside the U.S. As mentioned above, this foreign tax authority aggressiveness has propelled many taxpayers to come into the U.S. program with a bilateral APA as essential to their objective.
The other reasons for the decline in unilateral APAs are the length of time it takes to secure one (more on the timeliness issue below) and the general sense that unilateral APAs are discouraged.
Taxpayers should not be discouraged, however. Unilateral APAs remain available where they make sense, as they often do.
When the APA program examined its unilateral APA inventory of completed and pending cases many years ago — totaling more than 200 cases — it determined that about one-quarter of unilateral APA requests were submitted in cases where bilateral APA mechanisms were not available, e.g., because the counterparty was a nontreaty country.
Another quarter were submitted in cases where bilateral APAs were simply impracticable, e.g., because the transactions involved a U.S. hub dealing with many foreign counterparties.
The remaining 50% fell into neither category but the vast majority of these were determined to be unilateral APAs for good reason.
Taxpayers should not hesitate to seek unilateral APA coverage where it makes sense. Although these cases represent a smaller percentage of total inventory, the program remains open to them.
Things That Require Attention
There are two program features that require attention: the relative stagnation in the number of new APA requests and the length of time it takes to complete agreements.
New APAs Versus Renewals
The program needs to be, and I expect surely is, mindful of the slowing uptake by companies seeking new (nonrenewal) APAs.
The chart below shows the division in APA completions between new APAs and APA renewals over the last 20 years. The table hardly requires elaboration — it shows that APA renewals are becoming the dominant focus of the APA program.
The story is similar looking at the total inventory of APAs, rather than just completions.
The figures show that recent growth in the APA program inventory is dominated by renewals. In the 20-year period from 2001 to 2020, excluding the recent influx of U.S.-India bilateral APA requests, the number of cases in inventory has grown by 141 cases, 96% of which represents renewals.
Not an Argument Against Renewal APAs
Renewal APAs provide a valuable service. For the IRS, renewal APAs provide an opportunity to address multiple years of transfer pricing issues in an efficient manner, often by persons who are familiar with the file and can build on a base of experience with the taxpayer and the competent authority counterpart.
APA renewals are also an opportunity for the APA team to understand how APAs are implemented in practice — to test and confirm expectations and, where expectations are disappointed, to learn from that experience and to improve on it.
For taxpayers, APA renewals provide the obvious benefit of certainty, usually faster and with less cost than the first time around, allowing taxpayers to get a bit of a return on a large upfront investment associated with the original APA.
Reliance on an expired APA as a kind of soft precedent is not a substitute for an APA renewal, so one cannot say, either to the IRS or to a taxpayer, that the renewal exercise is not worth the time. The IRS revenue procedure makes plain that an expired APA is not precedent of any kind and cannot be used as such.
The recently decided Coca-Cola Company U.S. Tax Court case[1] underscores that taxpayers cannot rely on IRS prior acceptance of a transfer pricing method as implied consent to the continuation of that method.
I repeat, therefore, I am not arguing against APA renewals or suggesting that the APA program is misapplying its resources by devoting so much time to them.
My point is that the APA program must be equipped to handle a much larger volume of APA work if it is going to contribute meaningfully to the betterment of tax administration beyond serving its existing customers.
A U.S. General Accounting Office report in 2000 signaled disappointment that the APA program was being used by only 10% to 15% of the large corporate taxpayers with material transfer pricing issues. The expectation was that the APA program would reach a higher percentage. I do not know what the current percentage is, but I am confident it should be higher than it is.
Timeliness
The most serious challenge facing the APA program is the time it takes to complete an APA. It is discouraging many taxpayers from using the program.
This has been a long-standing complaint. Twenty years ago a GAO report detailed the timeliness problems at the APA program. The problems were reported also in the Wall Street Journal. One article, in 2000, observed that: "Even IRS admirers often are dismayed by lengthy delays."[2] A second, published a year later, remarked on the "long-running concerns about how long it takes to reach agreements."[3]
These comments about APA program tardiness were made against a backdrop when the program had a stated objective to complete unilateral APAs within 12 months, and to be ready to start formal negotiations on bilateral APAs within nine to 12 months. The program was not meeting the targets and was faulted for this, but the situation has not improved.
The table below shows the average age of completed unilateral APAs by year over the last 30 years, and reveals a steady upward trend in the amount of time it takes to get a unilateral APA completed.
The table below shows the average age of completed bilateral APAs by year over the last 30 years.
The table shows that the 44 new bilateral APAs completed in 2020 took on average 50.8 months to complete. Over the prior four years, the corresponding average completion time for these cases was between 46.9 months and 50.5 months. Adding 2020 produces a five-year weighted average of 48.6 months, not what the program hopes to achieve.
The age of the bilateral renewal APAs completed in 2020 and over the five-year period 2016-2020 was much better. Bilateral renewal APAs completed in 2020 took on average 34.1 months, bringing the five-year weighted average for 2016-2020 to 37.6 months and trending in the right direction.
It is difficult to gauge with confidence how long an APA is going to take based on the times reported for completed APAs. Completion times are lagging indicators — lagging by a number of years.
Projecting how long it will take to complete an APA filed today requires consideration of a number of additional factors. Some of these are case specific and beyond the scope of this discussion, e.g., the complexity of the case, the quality of the submission and the persons working the case.
At a more general level, there are two main factors that drive the average age of APA completions: staffing levels and the rate of completions each year as a percentage of inventory. Regarding the latter metric, the APA program has been falling behind due to its understaffing in recent years.
In 2013, the program completed 56 new bilateral APAs, representing 32% of its 2012 year-end new bilateral APA inventory of 154 cases. That percentage fell to 27% in 2014, then to 21% in 2015. It fell further and varied between 18% and 19% for the four years 2016-2019. In 2020, it was again 18%.
If the APA program is completing on average 18% of its prior year-ending new bilateral APA inventory in an environment in which the inventory is not growing rapidly, it will take a while to work through the backlog.
The recent increase in staffing is a reason to be optimistic however. The strong tie between staffing levels and program production has been demonstrated throughout the program's history, in both directions. There is good reason to believe the connection will hold true now and will result in a large improvement in 2021 and beyond, so long as the program can maintain its recently supplemented staffing levels.
Three Modest Suggestions
The time it takes to complete APAs is an impediment to attracting taxpayers into the program, which is a disservice to the IRS and to tax administration. The recent staffing increase should help address the problem. I favor more staffing, at least to return to 2012 levels. I'll offer three other suggestions as well.
1. Keener Appreciation of Opportunity Cost
It is important to recognize that the APA program functions to improve tax administration within a much larger tax administration bureaucracy. As important a role as the APA program plays, it is only a supporting role. Over 40% of all U.S. imports and exports each year involve transactions between related parties. I suspect only a modest percentage of those transactions are covered by APAs. The rest are left to the regular IRS examination process.
If the APA program is going to maximize its contribution to tax administration overall, it should look — with full taxpayer assistance — to cover more ground by moving more quickly. It is tempting, and thoroughly understandable given the sums involved, to ask in each case whether a particular transfer pricing analysis could be improved by asking more questions and doing more work. The answer will often be yes.
But that is not the crucial question. The crucial question is whether the analysis is responsible and an improvement on the examination alternative.
In most cases, even at the early stage of the process, the APA program is — or with proper taxpayer assistance should be — equipped to reach a solid transfer pricing conclusion, better than one would reasonably expect to develop in examination. At that point, the APA program should look to conclude the matter, and devote its time and attention to other matters where it can achieve similarly superior results.
This view is informed by the fact that most taxpayers come into the APA program to achieve certainty, not to gain a tax advantage. One can have confidence in this proposition for four main reasons.
First, aggressive taxpayers shy away from a process that requires them to shine a spotlight on their own transfer pricing practices.
Second, a large percentage of the U.S. APA inventory consists of bilateral requests for taxpayers operating in other relatively high-tax jurisdictions — in Japan, Korea, Canada, Germany. These APAs do not typically involve significant tax arbitrage opportunities. They are driven by the taxpayer's desire to stay out of trouble.
The third reason for confidence that most taxpayers are not trying to achieve an undue tax advantage is that APA program data confirm that taxpayers in the program do not act this way. This was confirmed by an analysis performed in 2004. At that time, the APA program was responding to an inquiry by the Senate Finance Committee looking into the operations of the program and seeking to find out, among other things, whether taxpayers were obtaining undue benefits by coming into the program.
The analysis performed by APA program staff — who did an outstanding job marshalling a huge amount of information from boxes retrieved from storage — examined more than 100 APAs that required a taxpayer's income to fall within a defined range and looked to see where in fact taxpayers tended to fall within that range.
The analysis calibrated each range to have 100 basis points, 0 points representing the minimum amount of U.S.-required income and 100 points representing the maximum, and measured actual reported results against this normalized range to generate a percentile score. The results showed, in the aggregate from 1993 through 2002, that taxpayer results fell evenly throughout the range, with the simple average of all results falling near the middle at 49%.
The same was true on a single year basis for 2000, 2001 and 2002 — the three most recent years for which data was available at the time. The average percentile scores in those years were 46%, 54% and 49%, respectively. This showed that even after getting an APA, taxpayers as a whole showed no intention of minimizing the amount of income reported in the U.S.
A fourth reason for confidence that taxpayers are not seeking APAs to minimize U.S. income is that many taxpayers, especially now, are coming into the program to ward off challenges from foreign tax authorities that argue the taxpayers are leaving too much income in the U.S.
This is perhaps the biggest development in the U.S. APA program I have observed over its 30-year history. These are not taxpayers seeking to minimize U.S. tax; they are taxpayers seeking to defend U.S. tax and enlisting the U.S. APA program to help them do so.
These are four reasons, I suggest, that the IRS should be comfortable moving APA cases to resolution more quickly.
Lastly, the APA program should take comfort from two things. First, its APAs do not set policy or make precedent. When an APA expires, it disappears. If there is anything in an APA that gives the IRS reason for regret, it will be short-lived.
Second, within a year or two of completing an APA, the taxpayer is likely to be back at the door, seeking a renewal. Both parties — the taxpayer and the IRS — know that the taxpayer's implementation of the prior APA will be examined in the renewal process.
This discourages gamesmanship. If the taxpayer does not seek a renewal, the IRS examination team will have the benefit of the APA file to pursue the next examination cycle.
The only APAs that fall outside the paradigm addressed above are APAs that involve big one-time transfers of value to tax-advantaged jurisdictions, such as the transfer of crown jewel intangibles to a low tax country. Those are a small percentage of the APA program inventory, and are not the reason for the 48-month average age of new bilateral APA completion statistics.
I'm mindful, of course, that APAs involve large dollar amounts with real tax impact.
I recall running the numbers on a profit split case in the program years ago and calculating that each one-tenth of a percentage point in the arm's length profit split formula equaled tens of millions of dollars in tax over the proposed term of the APA. This was a big case but not an extraordinary one for the program.
I accept that the program deals with large sums and that almost any APA could be improved with time and effort, but with limited resources — because, being realistic, the program is always going to be resource constrained — it's critical to perform intelligent triage through risk assessment and resource allocation.
Taxpayers complained 20 years ago that the APA program takes too long to make decisions. It's understandable why this was, and why it continues to be, given the large stakes involved. But it is a complaint that should be addressed, consistent with the need for prudent tax administration.
2. Enforcing Mutual Obligations
It is important for all parties to the APA process to commit themselves to meeting more ambitious goals. There was a time, as I mentioned, that the APA program was explicit that its goal was to complete unilateral APAs within 12 months and to be ready to commence competent authority discussions on bilateral APAs within nine to 12 months.
The program should rededicate itself to those timelines and demand from taxpayers the quality of data and quickness of delivery that can make achieving those timelines realistic.
The program has always had to deal with uneven APA submissions. The standardization of APA requests prescribed in the current revenue procedure was intended to raise the floor but while many submissions are excellent, others fall short because of bad writing or deficient content or both. (Yes, there are APA requests that are simultaneously impenetrable and vacuous.)
The program might consider taking firmer steps to police the quality of submissions, and the quality and quickness of taxpayer answers to APA team questions in exchange for APA program assurances of accelerated review and resolution.
3. Use of Term Tests
A small suggestion to end. The year 2020 highlighted the value and appropriateness of term tests by which compliance with the approved APA method is tested by reference to taxpayer-aggregate results over the full term of the APA rather than on an individual or three-year average basis. These tests reduce complexity of administration and reduce the need to accommodate short-term market disruptions.
Term tests have been common in the past. They were used on average for 25 covered transactions a year in the mid-2000's. The APA program has not released figures in more recent years but I know they continue to be used. The APA program should encourage treaty partners to use them even more. A lot of APA discussions that were complicated by the COVID-19 pandemic in 2020 could have been simpler if the parties understood that APA compliance would be measured on a full-term basis.
The benefits are self-evident and the risks are slim. Taxpayers in my experience have shown little interest in abusing the discretion afforded to them by term tests and, frankly, their opportunity to do so is limited because often only a couple years are left on the term when the APA is signed. If there are concerns nevertheless, anti-abuse language can be added as a critical assumption or the matter can be dealt with by appropriate action on renewal.
Conclusion
The APA program is a terrific innovation that has justly earned widespread support over its 30 years of operation. Its success has inspired many other countries to adopt and promote similar programs.
It is run by a first-rate set of professionals who tackle some enormous transfer pricing challenges and who have labored under difficult conditions — a large inventory of cases managed by a workforce that declined 26% from 104 persons in 2012 to 77 persons before the spate of hiring in mid-2020.
The imperative now is to move APA cases quicker to expand the program's ability to attract and complete new cases, so it can broaden its contribution to tax administration. In 2000, the General Accounting Office issued a report concluding that "Delays that discourage participation in the APA program could also lead to more disputes that are costly to both the taxpayer and the government."[4] It is in everyone's interest to support the APA program to allow it to achieve its goals, to avoid the costly disputes about which the GAO warned twenty years ago.
Matthew Frank is a partner at Steptoe & Johnson LLP and was director of the IRS Advance Pricing Agreement Program from 2003 to 2008.
The opinions expressed are those of the author(s) and do not necessarily reflect the views of the firm, its clients or Portfolio Media Inc., or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.
[1] The Coca-Cola Company v. Comm'r
, 155 T.C. No. 10, 2020 U.S. Tax Ct. LEXIS 27 (2020).[2] The Wall Street Journal, Oct. 11, 2000, A1.
[3] The Wall Street Journal, Nov. 14, 2001, A1.
[4] U.S. General Accounting Office, Report to The Honorable Byron L. Dorgan, U.S. Senate, Tax Administration: IRS' Advance Pricing Agreement Program, GAO/GGD-00-168, page 11 (August 2000).
For a reprint of this article, please contact reprints@law360.com.