Statement of Reuven S. Avi-Yonah, Irwin I. Cohn Professor of Law, University of Michigan Law School

Testimony Before the Subcommittee on Select Revenue Measures
of the House Committee on Ways and Means

March 05, 2008

My name is Reuven S. Avi-Yonah. I am the Irwin I. Cohn Professor of Law and Director of the International Tax Master of Law Program at the University of Michigan Law School. I hold a JD (magna cum laude) from Harvard Law School and a PhD in History from Harvard University. I have 19 years of full and part time experience in the tax area, and have been associated with or consulted to leading law firms like Cravath, Swaine & Moore, Wachtell, Lipton, Rosen & Katz and Cadwalader, Wickersham & Taft. I have also served as consultant to the US Treasury Office of Tax Policy and as member of the executive committee of the NY State Bar Tax Section. I am currently Chair of the ABA Tax Section Committee on VAT, a member of the Steering Group of the OECD International Network for Tax Research, and a Nonresident Fellow of the Oxford University Center on Business Taxation. I have published eleven books and over 80 articles on various aspects of US domestic and international taxation, and have fourteen years of teaching experience in the tax area (including basic tax, corporate tax, international tax and tax treaties) at Harvard, Michigan, NYU and Penn Law Schools.

I would like to thank Chairman Neal and the Committee staff for inviting me to testify today on the international aspects of the tax treatment of derivatives.

1. The Use of Derivatives to Avoid Withholding Taxes.

Since the beginning of the income tax, the US has imposed a withholding tax (currently 30 percent) on payments of "fixed or determinable, annual or periodical" (FDAP) US source income to nonresidents. However, recent developments such as the rise of derivative financial instruments have seriously undermined our ability to tax FDAP at source, which frequently means that it is not subject to tax at all. This is true even if the income ultimately inures to the benefit of US residents.

The major categories of FDAP are dividends, interest, royalties and rents. Of these, royalties and rents are typically not subject to source-based taxation because of our tax treaties. Interest is likewise rarely subject to tax at source because of the portfolio interest exemption (IRC 871(h)), which exempts most payments of interest to nonresidents from withholding tax. Interest payments that do not qualify for the portfolio interest exemption are frequently exempt by our tax treaties. Thus, the main category of FDAP that is still subject to withholding tax is dividends. Generally, even dividends paid to residents of countries with whom we have a tax treaty are subject to tax at 15 percent, while dividends paid to residents of countries with which we do not have a tax treaty are subject to the full tax of 30 percent.[1]

How can derivatives be used to avoid the withholding tax on dividends? A simple example involves a derivative called a total return equity swap (TRES). In a TRES, a foreign investor enters into a contract with a US investment bank under which the investor pays the bank an initial amount, say 100. In return, the bank agrees to pay the investor an amount equal to the dividends paid by a US corporation in a given period of time (e.g. 5 years). In addition, the bank agrees at the end of the five years to pay the investor any appreciation in the value of the US corporation's stock over 100, and the investor agrees to pay the bank if the stock declines under 100. The bank then turns around and invests the 100 in the stock of the US corporation.

What are the tax consequences of this arrangement? The capital gain or loss at the end of the five years is not subject to US tax even in the absence of a TRES because we generally do not tax capital gains of nonresidents. However, the TRES does make a difference to the taxation of any dividends paid during the five years. If the investor had invested directly in the stock of the US corporation, any dividends would have been subject to tax at a rate of at least 15 percent. However, dividends paid by the US corporation to the US investment bank are not subject to withholding tax because they are not paid to a nonresident. The investment bank does include them in income, but it gets an offsetting deduction for paying the dividend equivalent to the foreign investor. The dividend equivalents in turn are not subject to withholding tax because under a regulation adopted in the early 1990s, payments on "notional principal contracts" (such as the TRES) are sourced at the residence of the recipient.[2] Thus, because the investor is foreign, the dividend equivalents are not US source and therefore are not subject to withholding tax.

Recent press stories have suggested that the use of TRES to avoid withholding tax on portfolio dividends has increased exponentially since the arrival of hedge funds. Disturbingly, the stories suggested that some of the TRES held by hedge funds ultimately inure to the benefit of US residents, but are not reported to the IRS under the cover of tax haven bank secrecy. Recent reports that US residents have been discovered to hold secret accounts in Liechtenstein and the Isle of Man highlight this concern.

Using TRES to avoid the withholding tax on portfolio dividends is an old technique, although the amounts involved seem to be growing dramatically in recent years. A newer version involves using derivatives to convert dividends (subject to withholding tax) to interest (not subject to tax because of the portfolio interest exemption). A good example is the recent investment by the sovereign wealth funds of Abu Dhabi in Citigroup ($7.5 billion) and of China in Morgan, Stanley ($5 billion).

It has been reported that “under the terms of the Citigroup transaction, [Abu Dhabi Investment Authority] agreed to purchase $7.5 billion of equity units. The equity units are structured using Citigroup's patent-pending Upper DECS strategy. Citigroup has agreed to pay an 11 percent yield on the units, with slightly over 6.5 percent classified as interest payments. The remaining payment is a contract payment on the purchase contract. The result to Citigroup is a tax savings in the neighborhood of $175 million per year for the three years before the conversion to common shares begins. Morgan Stanley sold $5 billion of equity units through its PEPS structure to China Investment Corp., and will pay a total annual rate of 9 percent on them. Morgan Stanley classifies 6 percent of the payments as interest on a debt and will benefit from a tax savings of approximately $100 million per year during the preconversion period of the deal.”[3]

Both of these transactions rely on the IRS’ feline PRIDES ruling of 2003.[4] Feline PRIDES is Merril Lynch’s version of the Citigroup Upper DECS and the Morgan Stanley PEPS. In all three transactions, the US entity issues “equity units consisting of a forward contract for the purchase of equity shares and notes equivalent to the price of the stock. The amount of stock deliverable under the forward contract is determined on the settlement date and has a market value equal to the settlement price. The forward contract and the notes are treated as separate transactions, and the purchase price is allocated between them according to their fair market values on the settlement date. The holder of the investment units may, but is not required to, separate the obligations of the units. The holder is also obliged to remarket the notes. The issuer will make regular payments under the terms of the transaction and treat the lion's share of the payments as interest payments on the notes.”[5]

Under the 2003 Revenue Ruling, the IRS treats the forward contract and the notes as two separate instruments. As a result, the payments on the notes are treated as interest, and when the notes are held by a foreign investor, they are entitled to the portfolio interest exemption and not subject to withholding tax even though they are mandatorily convertible into stock, and even though the investment is treated as equity for bank regulatory purposes. Presumably, the foreign investors intend to sell the units to a US party before conversion to equity.

2. What can be done?

The immediate response to both the TRES and feline PRIDES structures is to suggest that the IRS should revise its rules to close the relevant loopholes. In the case of TRES, the IRS could revise the regulations and treat dividend equivalent amounts under the TRES as dividends for withholding tax purposes, just like it does in the case of securities loans from a foreign lender to a US borrower, where payments of dividend equivalents from the US to the foreign person are treated as dividends and subject to withholding tax.[6] In the case of feline PRIDES, the IRS could revoke its 2003 ruling and treat the structure as equity, as suggested by Prof. David Weisbach.[7]

However, the problem with adopting such narrow loophole-closing measures is that they invite taxpayers to find new ways to achieve the same goals. For example, if dividend equivalent amounts under a TRES linked to a specific stock are treated as dividends, the investment banks will issue TRES linked to a basket of stocks that will behave similarly to the targeted stock. The IRS may then adopt “substantially similar or related property” rules, and the game will go on. Likewise, if the feline PRIDES ruling were revoked, I have no doubt that other, more complex instruments could be invented that achieve the same goal.[8]

A more ambitious reform proposal would be to revise the regulations and treat the source of all payments on notional principal contracts as the residence of the payor, rather than the residence of the recipient. This would align the source rule for derivatives with the source rule for dividends and interest. However, as long as the portfolio interest exemption is in the Code, derivatives could still be used to convert dividends into interest, and then the source would not matter because they payment would be exempt.

In general, it seems strange to insist on levying withholding tax on dividends, which are not deductible, while not levying it on interest and royalties, which are. The payment of a dividend does not reduce our ability to tax the underlying corporate income, but the payment of deductible interest and royalties does.

Thus, one possible response to the use of derivatives to avoid withholding tax is to say “who cares?” As noted above, most forms of FDAP are already not subject to withholding, and if derivatives are used to avoid the withholding tax on portfolio dividends and our treaties are revised to eliminate withholding on direct dividends, perhaps the time has come to give up on the withholding tax altogether. It collects negligible revenue while imposing high transaction costs on withholding agents, as indicated by the voluminous regulations governing withholding under IRC 1441-1446.

3. A possible solution.[9]

Nevertheless, it seems unlikely that Congress will give up on withholding taxes on foreigners. And there is one compelling argument in favor of imposing such taxes: They may be our best chance to prevent US residents from avoiding tax on US source income earned through foreign intermediaries.

The basic problem stems from the portfolio interest exemption. In 1984, the United States unilaterally abolished its withholding tax (of 30 percent) on foreign residents earning "portfolio interest" income from sources within the United States. "Portfolio interest" was defined to include interest on U.S. government bonds, bonds issued by U.S. corporations (unless the bondholder held a 10 percent or more stake in the shares of the corporation), and interest on U.S. bank accounts and certificates of deposit. This "portfolio interest exemption" is available to any nonresident alien (that is, any person who is not a U.S. resident for tax purposes), without requiring any certification of identity or proof that the interest income was subject to tax in the investor's country of residence.

The result of enacting the portfolio interest exemption has been a classic race to the bottom: One after the other, all the major economies have abolished their withholding tax on interest for fear of losing mobile capital flows to the United States. The table below shows current withholding rates in EU member countries and in the United States on interest paid on banks accounts, securities (government and corporate bonds), and dividends paid to foreign residents in the absence of a treaty. 

---------------------------------------------------------------------
Country              Bank Accounts      Securities     Dividends
---------------------------------------------------------------------
Belgium                  0                  10            15
Denmark                  0                   0            15
France                   0                   0            15
Germany                  0                   0            15
Greece                  10                  10             0
Ireland                  0                   0             0
Italy                   10                  10            15
Luxembourg               0                   0            15
Netherlands              0                   0            15
Portugal                15                  15            15
Spain                    0                   0            15
United Kingdom           0                   0            15
United States            0                   0            30
---------------------------------------------------------------------

As the table indicates, most developed countries impose no withholding tax on interest paid to nonresidents on bank deposits and government and corporate bonds. Withholding taxes are imposed on dividends, despite the fact that dividends (unlike interest) are not deductible and, therefore, the underlying income has already been taxed once.

The standard economic advice to small, open economies is to avoid taxing capital income at source, because the tax will be shifted forward to the borrowers and will result in higher domestic interest rates. However, the countries in the table include large economies (the United States, Germany, and the United Kingdom) in which the tax is not necessarily shifted forward. Rather, the principal reason for the lack of withholding taxes in most of the countries included in the table above is the fear that if such taxes were imposed, capital would swiftly move to other locations that do not impose a withholding tax. Thus, the Ruding Committee, writing about the European Community, concluded in 1992 that "recent experience suggests that any attempt by the [European Union] to impose withholding taxes on cross-border interest flows could result in a flight of financial capital to non-EC countries."[10] 

The experience of Germany is a case in point: In 1988, Germany introduced a (relatively low) 10 percent withholding tax on interest on bank deposits, but had to abolish it within a few months because of the magnitude of capital flight to Luxembourg. In 1991, the German Federal Constitutional Court held that withholding taxes on wages, but not on interest, violated the constitutional right to equality, and the government therefore was obligated to reintroduce the withholding tax on interest, but made it inapplicable to nonresidents. Nevertheless, nonresidents may be German residents investing through Luxembourg bank accounts and benefiting from the German tradition of bank secrecy vis-a-vis the government.[11]

The current situation is a multiple-player prisoner's dilemma: All developed countries would benefit from reintroducing the withholding tax on interest, because they would then gain revenue without fear that the capital would be shifted to another developed country. However, no country is willing to be the first one to cooperate by imposing a withholding tax unilaterally; thus they all defect (that is, refrain from imposing the tax) to the detriment of all.

In global terms, this outcome would make no difference if residence jurisdictions were able to tax their residents on foreign- source interest (and dividend) income, as required by a global personal income tax on all income "from whatever source derived."  However, as Joel Slemrod has written, "although it is not desirable to tax capital on a source basis, it is not administratively feasible to tax capital on a residence basis."[12]  The problem is that residence country fiscal authorities in general have no means of knowing about the income that is earned by their residents abroad. Even in the case of sophisticated tax administrations like the IRS, tax compliance depends decisively on the presence of either withholding at source or information reporting. When neither is available, as in the case of foreign-source income, compliance rates drop dramatically.

In the case of foreign-source income, withholding taxes are not imposed for the reasons described above. As for information reporting, even though tax treaties contain an exchange of information procedure, it is vitally flawed in two respects: First, the lack of any uniform world-wide system of tax identification numbers means that most tax administrations are unable to match the information received from their treaty partners with domestic taxpayers. Second, there are no tax treaties with traditional tax havens, and it is sufficient to route the income through a tax haven to block the exchange of information. For example, if a Mexican national invests in a U.S. bank through a Cayman Islands corporation, the exchange of information article in the U.S.-Mexico tax treaty would not avail the Mexican authorities. The IRS has no way of knowing (given bank secrecy) that the portfolio interest that is paid to the Caymans is beneficially owned by a Mexican resident covered by the treaty. 

The resulting state of affairs is that much income from portfolio investments overseas escapes income taxation by either source or residence countries. Latin American countries provide a prime example: It is estimated that following the enactment of the portfolio interest exemption, about US $300 billion fled from Latin American countries to bank accounts and other forms of portfolio investment in the United States. Most of these funds were channeled through tax haven corporations and therefore were not subject to taxation in the country of residence. For all developing countries, various estimates of the magnitude of capital flight in the 1980s average between US $15 billion and US $60 billion per year. Nor is the problem limited to developing countries: Much of the German portfolio interest exemption benefits German residents who maintain bank accounts in Luxembourg, and much of the U.S. portfolio interest exemption benefits Japanese investors who hold U.S. treasuries and do not report the income in Japan. Even in the case of the United States, it is questionable how much tax is actually collected on portfolio income earned by U.S. residents abroad other than through mutual funds. One estimate has put capital flight from the United States in 1980-82 as high as US $250 billion. More recently, Joseph Guttentag and I have estimated the “international tax gap” (the tax owed by US residents on income earned through foreign tax havens) at $50 billion.[13] 

Thus, in the absence of withholding taxes or effective information exchange, income from foreign portfolio investments frequently escapes being taxed by any jurisdiction. This is particularly significant because the flows of portfolio capital across international borders have been growing recently much faster than either world gross domestic product or foreign direct investment. It is currently estimated that international capital flows amount to US $1 trillion a day; although this figure is much larger than income from capital, it gives a sense of the magnitude at stake.

This situation has led knowledgeable observers like Richard Bird to write that "the weakness of international taxation calls into question the viability of the income tax itself...If something is not done to rectify these problems soon, the future of the income tax is bleak"[14] Other authors have written papers like "Can Capital Income Taxes Survive in Open Economies?" and "Is There a Future for Capital Income Taxation?" Unless something is done about this situation, the answer to those questions is likely to be "no." 

However, the present may present a unique opportunity to remedy this state of affairs because of the EU’s adoption in 2003 of a “Savings Directive”. Under the Directive, the EU adopted a "coexistence" model based on two options, only one of which can be chosen by a member state: Either to cooperate in an exchange of information program, or to levy a 20 percent withholding tax on interest payments made by paying agents within its territory to individual residents of another member state. Under the exchange of information system, the member state agrees to provide automatically, at least once a year, information on all interest payments made by paying agents in its territory in the preceding year to individual beneficial owners residing in every other member state. Under the withholding tax system, the member state agrees to impose a 20 percent withholding tax on all interest payments made by paying agents within its territory to individual beneficial owners residing within the European Union. However, the withholding tax is not imposed if the beneficial owner provides a certificate drawn up by his country's tax authorities attesting that they have been informed of the interest to be received. The withholding tax must be credited against the tax liability in the beneficial owner's country of residence. 

 The EU Savings Directive only applies to payments within the EU. However, its adoption presents a golden opportunity. As explained above, the problem of nontaxation of cross- border interest flows stems to a large extent from the unilateral enactment of the portfolio interest exemption by the United States in 1984. As observed above, the nontaxation of cross-border interest flows is a repeated prisoner's dilemma: each player (the European Union, the United States, and Japan) refrains from taxing for fear of driving investment to the others, even though they would all benefit from imposing the tax. However, it is well-established that such repeat prisoner's dilemmas can be resolved if parties can signal to each other in a credible fashion their willingness to cooperate. 

The EU Directive represents just such a signal. The European Union is telling the United States that it is willing to go forward with taxing cross-border interest flows. Thus, if the United States were to commit itself to taxing cross-border interest by repealing the portfolio interest exemption, the prisoner's dilemma could be resolved and a new, stable equilibrium of taxing -- rather than refraining from tax -- would be established.

The prospects for agreement in this area are particularly good because only a limited number of players need to be involved. The world's savings may be parked in tax havens, but the cooperation of such tax havens is not needed. To earn decent returns without incurring excessive risk, funds have to be invested in an OECD member country (and more particularly, in the European Union, the United States, Japan, or Switzerland). Thus, if the OECD member countries could agree to the principles adopted by the European Union in its Savings Directive, most of the problem of taxing cross-border portfolio interest flows could be solved.

My proposal is therefore as follows: the United States should move within the OECD for a coordinated implementation of the principles contained in the EU Savings Directive. However, while in the European Union context, exchange of information could play a large role, because there are few traditional tax havens in the European Union, in a global context withholding taxes have to be the primary means of enforcement. As noted above, tax havens with strong bank secrecy laws render it very difficult to have effective exchange of information among OECD member countries. If the investment is made through a tax haven intermediary, exchange of information is likely to be useless unless the tax authorities in the payor's country can know the identity of the beneficial owner of the funds that are paid to the tax haven intermediary.

I would therefore propose that instead of the "co- existence" model of the European Union, the United States, and following it the OECD, should adopt a uniform withholding tax on cross-border interest flows, which should also be extended to royalties and other deductible payments to portfolio investments (for example, all payments on derivatives). To approximate the tax rate that would be levied if the payment were taxed on a residence basis, the uniform withholding tax rate should be high (at least 40 percent). However, unlike the withholding taxes that were imposed before the current "race to the bottom" started in 1984, the uniform withholding tax should be completely refundable. To obtain the refund, as in the EU Directive, a beneficial owner need only show the tax authorities in the host countries a certificate attesting that the interest payment was reported to the tax authorities in the home country. No actual proof that tax was paid on the interest income is required: from an efficiency, equity, and revenue perspective, it is sufficient that the home country authority has the opportunity to tax the income from overseas investments in the same way as it taxes domestic-source income.  Thus, even if the home country has a generally applicable low tax rate on its residents (or even a zero tax rate, as long as it applies to all bona fide residents), the resident could obtain a refund by reporting the income to the tax authorities in his home country.

Both the proposed withholding tax and the refund mechanism would not require a tax treaty. However, it would be possible for countries to reduce or eliminate the withholding tax in the treaty context when payments are made to bona fide residents of the treaty partner. In those cases, the exchange of information in the treaty should suffice to ensure residence-based taxation. Because most OECD members have tax treaties with most other OECD members, the proposed uniform withholding tax would in general apply only to payments made to non- OECD member countries (including the tax havens).

Were this type of uniform withholding tax enacted by OECD members, it would go a very long way toward solving the problem of under-taxation of cross-border portfolio investments by individuals. Such under-taxation is unacceptable from an efficiency, equity, or administrability perspective. Moreover, unlike the under-taxation of direct investment, this type of under-taxation is illegal (which is why it is so hard to assess its magnitude). By adopting a uniform withholding tax, the OECD could thus strike a major blow at tax evasion, which (as described above) is a major problem for the US as well.

4. Conclusion

Under current conditions, the US collects remarkably little revenue from the withholding tax on non-residents. That is because most forms of FDAP other than dividends are exempt, while the withholding tax on dividends can be avoided by the use of derivatives, in the ways described above.

One possible reaction to this state of affairs is to repeal the withholding tax. However, even if it were acceptable for the US not to tax foreigners on passive income from US sources, the risk of doing so is that it will enhance the ability of US residents to derive income from US sources through foreign intermediaries located in tax havens.

Narrower fixes to the problem posed by derivatives are possible, but as long as the portfolio interest exemption remains in the Code, they are unlikely to address the fundamental problem. Thus, in my opinion the best course to pursue is to repeal the portfolio interest exemption in coordination with the OECD, and instead impose a refundable withholding tax, along the lines set out above.

[1] Dividends paid to controlling shareholders are generally subject to tax under treaties at 5%, but some of our recent treaties reduce this tax to zero.

[2] Treas. Reg. 1.863-7.

[3] David D. Stewart, Sovereign Wealth Fund Deals Take Advantage of IRS Ruling, 2008 TNT 41-5 (Feb. 29, 2008).

[4] Rev. Rul. 2003-97, 2003-34 IRB 1.

[5] Stewart, supra.

[6] For a discussion of the different treatment of TRES and securities loans see Reuven Avi-Yonah and Linda Z. Swartz, U.S. International Treatment of Financial Derivatives, 74 Tax Notes 1703 (1997).

[7] Stewart, supra.

[8] For some ideas see Gregory May, Flying on Instruments: Synthetic Investments and the Avoidance of Withholding Tax, 96 TNT 239-32 (1996).

[9] This section is based in part on Avi-Yonah, Memo to Congress: It’s Time to Repeal the U.S. Portfolio Interest Exemption, 17 Tax Notes Int’l 1817 (Dec. 7, 1998).

[10] Report of the Committee of Independent Experts on Company Taxation (Commission of the European Communities, 1992), p. 201.

[11] As indicated by recent press reports, since Luxembourg is now subject to exchange of information under the Savings Directive (discussed below), German capital has shifted to Liechtenstein.

[12] Joel Slemrod, Comment, in Vito Tanzi, Taxation in an Integrating World (1995), 144.

[13] Guttentag and Avi-Yonah, Closing the International Tax Gap, in Max B. Sawicky (ed.), Bridging the Tax Gap: Addressing the Crisis in Federal Tax Administration, 99 (2005).

[14] Richard Bird, Shaping a New International Tax Order, 42 Bulletin for International Fiscal Documentation 292 (1988), 303.