no job name · daniel n. shaviro is the wayne perry professor of taxation at new york university...

The Financial Transactions Tax vs.The Financial Activities TaxBy Daniel N. Shaviro

Table of Contents

I. Introduction . . . . . . . . . . . . . . . . . . . . . . . 453II. The FTT and the FAT: A Brief Overview . . 454

A. Prior Intellectual History of the FTT . . . 454B. Efficiency Problems With an FTT . . . . . . 457C. The European Commission’s FTT . . . . . . 458D. The FAT Variants Discussed by the IMF

Staff . . . . . . . . . . . . . . . . . . . . . . . . . . 461E. Combining the IMF’s FAT Proposals . . . 467

III. Rationales for Financial Sector Taxation . . 467A. Raising Revenue . . . . . . . . . . . . . . . . . 467B. Ensuring a Fair and Substantial

Contribution . . . . . . . . . . . . . . . . . . . . 468C. Reducing Undesirable Market Behavior . 470D. Achieving Coordination . . . . . . . . . . . . 471E. Other Relevant Considerations . . . . . . . . 471

IV. An Alternative FTT Rationale . . . . . . . . . 472V. Conclusion . . . . . . . . . . . . . . . . . . . . . . . 473

I. Introduction

On her deathbed, Gertrude Stein reportedlyasked, ‘‘What is the answer?’’ but on hearing noreply, she added, ‘‘In that case, what is the ques-tion?’’1 In evaluating what new tax instruments, ifany, should be levied on the financial sector in theaftermath of the 2008 financial crisis, we would dowell to emulate Stein’s focus on what question isbeing asked. We need to know what purposes are tobe served by a tax on the financial sector before wecan evaluate how best to advance those purposes.

The European Commission, in its recent proposalthat the European Union adopt a financial transac-tions tax (FTT) that is directed mainly at secondarysecurities trading,2 is commendably clear about theobjectives that a financial sector levy might serve. Itmentions (1) raising revenue; (2) ensuring an ‘‘ad-equate (fair and substantial)’’ contribution from thefinancial sector; (3) reducing undesirable market

1See Janet Malcolm, ‘‘Someone Says Yes to It,’’ The NewYorker, June 13, 2005, at 148, 164.

2An FTT proposal has recently been introduced in Congressby Rep. Peter A. DeFazio, D-Ore., and Sen. Tom Harkin, D-Iowa.See Wall Street Trading and Speculators Tax Act, H.R. 3313 andS. 1787.

Daniel N. Shaviro is the Wayne Perry Professorof Taxation at New York University School of Law.This report is based on a presentation at the Con-ference on Taxing the Financial Sector, held at theAmsterdam Centre for Tax Law on December 9,2011. Shaviro is grateful to other conference partici-pants for their insights and to the following peoplefor their comments on an earlier draft: Alan Auer-bach, Joseph Bankman, Joseph Grundfest, MichaelKeen, Daniel Kessler, Victoria Perry, John Vella, andother participants in tax policy colloquia that wereheld at NYU Law and at Stanford Law School.

In Europe and the United States, there has beenmuch debate regarding whether, in response to the2008 financial crisis, governments should enact afinancial transactions tax (FTT) or a financial activi-ties tax (FAT), which are commonly viewed asmutually exclusive alternatives. This report evalu-ates those alternative instruments, focusing on re-cent proposals by the European Commission andthe IMF. It concludes that the case for enacting aFAT is considerably stronger than that for an FTT,mainly because the FAT focuses on a broad netmeasure, rather than a narrow gross measure, offinancial sector activity.

The report further concludes that a rationale forthe FTT not emphasized by the European Commis-sion — to address wasteful overinvestment in seek-ing trading gains at the expense of other traders —could conceivably support its enactment, althoughit is uncertain whether the social benefits wouldexceed the costs. The questions raised by thatrationale are independent of whether a FAT hasbeen enacted.

Finally, the desirability of enacting an FTT maybe affected by broader political economy con-straints on revenue raising and on the pursuit ofgreater tax progressivity by alternative (includingclearly superior) means.

Copyright 2012 Daniel N. Shaviro.All rights reserved.

tax notes®

SPECIAL REPORT

TAX NOTES, April 23, 2012 453

(C) T

ax Analysts 2012. A

ll rights reserved. Tax A

nalysts does not claim copyright in any public dom

ain or third party content.

behavior and thereby stabilizing markets; and (4)achieving coordination between different memberstates’ internal taxes.3

In my view, however, the commission is lesspersuasive in arguing that those considerationssupport enacting an FTT — particularly in relationto the alternative it identifies, which would be toenact instead some variant of a financial activitiestax (FAT),4 as recently proposed by the staff of theIMF.5 I will argue that the considerations identifiedby the commission — some of which are morecompelling than others — along with broader taxpolicy objectives, strongly support enacting a FAT(at least, assuming no other alternatives) while alsoraising serious questions about an FTT’s desirabil-ity. Indeed, the case that a properly designed FAT issuperior to the FTT is sufficiently compelling — notto mention unrebutted by the commission’s analy-sis — as to leave one wondering why the commis-sion came out as it did. As for the FAT, which todate has been somewhat underexplained, I willexpand on, and in at least one respect, modify theIMF staff’s analysis, while also explaining how onemight combine the most appealing features of thealternative versions that it describes.

As it happens, however, there is potentially adecent rationale for enacting an FTT — albeit onethat does not relate to extracting a ‘‘fair contribu-tion’’ from the financial sector or to easing the riskof another 2008-style economic crisis. Instead, itrelates to investors’ incentive to seek trading gainsat the expense of rival investors, whether by actingfaster than their rivals on new information or byspecial talent (or luck) in ‘‘anticipating what aver-age opinion expects the average opinion to be.’’6The competitive pursuit of trading gains can vergeon being a zero-sum game. Moreover, even whensome social benefit results from speeding the proc-ess whereby markets incorporate new information

into asset prices, the private gain from being onemicrosecond faster than one’s rivals may so exceedthat benefit as to make a tax on the activity poten-tially appealing — at least if the substantial designobstacles that an FTT would face can be sufficientlywell addressed to create optimism about its goodeffects’ outweighing its social costs.

Given how little that possible rationale for anFTT has to do with the main objectives identified bythe commission,7 I believe that the FTT-or-FATquestion is misguided. I will argue that a FATshould be enacted in any event — leaving asidepossible further alternatives — for reasons pertain-ing to undertaxation of the financial sector andincentives for bad risk taking. But the case for asuitably redesigned FTT should rise or fall onwholly separate grounds, and largely without re-gard to whether a FAT is in place.

In this report, I first discuss the FTT and FATmodels that have featured in historical and morerecent discussion, including by the commission andthe IMF. Then I evaluate the objectives cited by thecommission, along with further tax policy objec-tives, and assess their relevance to the FTT-vs.-FATchoice that is being debated in Europe. Next Idiscuss the alternative rationale that potentiallysupports adopting an FTT. Finally, I offer a briefconclusion.

II. The FTT and the FAT: A Brief Overview

A. Prior Intellectual History of the FTTFTTs have a long and varied history. They are

commonly traced back to a proposal by JamesTobin8 that countries impose special taxes purely onone type of financial transaction: ‘‘spot conversionsof one currency into another, proportional to thesize of the transaction.’’9 The commission’s FTTproposal would exempt currency conversions,

3See Committee on Economic and Monetary Affairs, Euro-pean Parliament, ‘‘Draft Report on the Proposal for a CouncilDirective on a Common System of Financial Transaction Tax andAmending Directive 2008/7/EC,’’ (Feb. 10, 2012), at 7; EuropeanCommission, ‘‘Commission Staff Working Paper: ExecutiveSummary of the Impact Assessment,’’ accompanying the docu-ment ‘‘Proposal for a Council Directive on a Common System ofFinancial Transaction Tax and Amending Directive 2008/7/EC’’(Sept. 2011), at 3-4. A further objective is to create internationalmomentum toward the general adoption of FTTs.

4Several countries have adopted other bank taxes, oftenaimed at some measure of bailout risk and expected bailoutcost. I will generally ignore those provisions, because theirconsideration has effectively been proceeding on a separatetrack from that of the FTT and FAT.

5IMF staff, A Fair and Substantial Contribution by the FinancialSector: Final Report for the G-20 (June 2010).

6John Maynard Keynes, The General Theory of Employment,Interest, and Money 156 (1964).

7While the European Commission did not mention any suchrationale for enacting an FTT, the Committee on Economic andMonetary Affairs of the European Parliament, in its 2012 draftreport amending the commission’s work, notes (at 15) investors’‘‘turn from long-term investments to short termism’’ as anotherrationale. Also, the FTT that was recently proposed in theUnited States has been rationalized by its sponsors on groundsthat are considerably closer to the line of argument that Isuggest. For example, it is called the ‘‘Wall Street Trading andSpeculators Tax Act,’’ and its lead House sponsor, DeFazio,emphasizes that Wall Street is a ‘‘gambling casino’’ and that anFTT would ‘‘rein in speculation on Wall Street.’’ See Josh Boak,‘‘Wall St. Trading Could Generate Taxes,’’ Politico, Oct. 4, 2011,available at http://www.politico.com/news/stories/1011/65140.html.

8James Tobin, The New Economies: One Decade Older 88-92(1972). Tobin, ‘‘A Proposal for International Monetary Reform,’’4 Eastern Econ. J. 153-159 (1978).

9Id. at 155.

COMMENTARY / SPECIAL REPORT

454 TAX NOTES, April 23, 2012

(C) T

ax Analysts 2012. A




which are the sole target of the so-called Tobin tax,while applying to transactions that the Tobin taxwould not have reached — specifically, sellingsecurities, such as corporate equities and bonds, onsecondary markets (that is, only after their initialissuance).

Thus, the commission’s proposal is actuallycloser to being a securities transactions tax (STT)than either a Tobin tax or an all-purpose FTT on allfinancial transactions.10 That idea has considerablyolder antecedents. As early as 1808, English stampduties on legal documents transferring title to prop-erty, including land or stock, started to be based onthe value of the property being transferred,11 mak-ing them a recognizable FTT precursor. John May-nard Keynes,12 after noting that England imposedtransfer taxes on securities trades, argued that theUnited States also should adopt ‘‘a substantialGovernment transfer tax on all [such]transactions . . . with a view to mitigating the pre-dominance of speculation over enterprise in theUnited States.’’13

At first glance, Keynes’s and Tobin’s analyses arevery similar, despite addressing markets for distinctfinancial assets.14 Nonetheless, one can discern atleast a slight difference in their emphasis. Both relyon Keynes’s15 famous comparison of financial in-vestment to a newspaper’s beauty contest in which

the competitors have to pick out the six pret-tiest faces from a hundred photographs, theprize being awarded to the competitor whosechoice most nearly corresponds to the averagepreferences of the competitors as a whole; sothat each competitor has to pick, not thosefaces which he himself finds prettiest, butthose which he thinks likeliest to catch thefancy of the other competitors, all of whom arelooking at the problem from the same point ofview.

The analogy reflects that in short-term asset trading,fundamental value (based on the risk-adjustedpresent value of expected long-term cash flows)may matter less than ‘‘what average opinion ex-pects the average opinion [regarding resale value]to be.’’16

Given that distinction between long-term funda-mental value and short-term trading value, Keynesproposed to ‘‘appropriate the term speculation forthe activity of forecasting the psychology of themarket, and the term enterprise for the activity offorecasting the prospective yield of assets over theirwhole life.’’17 He argued that speculation will oftenpredominate in financial markets, especially if trad-ing is easy and cheap, and that effectively turnsfinancial markets into casinos, in which luck andmood shifts drive the action and asset prices fail tofunction as good signals of fundamental value:‘‘When the capital development of a country be-comes a by-product of the activities of a casino, thejob is likely to be ill-done.’’18 By contrast, a world inwhich one could not trade so cheaply and readily‘‘would force the investor to direct his mind to thelong-term prospects and those only,’’19 thusstrengthening the relationship between asset pricesand fundamental value.

With speculation playing so central a role infinancial markets, Keynes argued that Wall Street’ssuccess in directing

new investment into the most profitable chan-nels in terms of future yield, cannot be claimedas one of the outstanding triumphs of laissez-faire capitalism — which is not surprising if Iam right in thinking that the best brains ofWall Street have been in fact directed towardsa different object. . . . It is usually agreed thatcasinos should, in the public interest, be inac-cessible and expensive. And perhaps the sameis true of Stock Exchanges.20

Tobin similarly saw currency exchange marketsas working well in one sense but not another. Theywere highly efficient ‘‘in a mechanical sense: trans-actions costs are low, communications are speedy,prices are instantaneously kept in line all over theworld, [and] credit enables participants to takelarge long or short positions at will or whim.’’21

However, their efficiency in the ‘‘deeper economic-informational sense’’ was ‘‘very dubious.’’ With

10As discussed below, what makes the European Commis-sion’s proposal not just an STT is that it also applies toderivative transactions.

11HM Revenue & Customs: Code of Practices 9 (2011). StampTaxes Manual, available at http://www.hmrc.gov.uk/so/manual.pdf.

12See Keynes, supra note 6, at 160.13The United States levies very modest STT-like securities

transaction fees to fund operations of the SEC, but they aretypically ignored by analysts because they are so low.

14This similarity reflects, of course, Keynes’s enormous in-tellectual influence on Tobin, whom The New Palgrave Dictionaryof Economics calls ‘‘the leading proponent of Keynesian econom-ics in the second half of the twentieth century.’’ Donald D.Hester, Tobin, James (1918-2002), in Steven N. Durlauf andLawrence E. Blume (eds.), The New Palgrave Dictionary of Eco-nomics (2008).

15See Keynes, supra note 6, at 156.

16Id.17Id. at 158.18Id. at 159.19Id. at 160.20Id. at 159.21See Tobin, ‘‘A Proposal for International Monetary Re-

form,’’ supra note 8, at 157-158.



(C) T

ax Analysts 2012. A




little available factual basis for confidently (muchless reliably) projecting proper long-term currencyvalue relationships, ‘‘the markets are dominated —like those for gold, rare paintings, and — yes, oftenequities — by traders in the game of guessing whatother traders are going to think.’’ This createdepisodes of severe short-term currency price vola-tility, leading to the transmission to domestic econo-mies of ‘‘disturbances originating in internationalfinancial markets. National economies and nationalgovernments are not capable of adjusting to mas-sive movements of funds across the foreign ex-changes, without real hardship and withoutsignificant sacrifice of the objectives of nationaleconomic policy with respect to employment, out-put, and inflation.’’22 An ‘‘internationally agreeduniform tax’’ on currency trades, based on the valuebeing traded, might reduce volatility by ‘‘throw-[ing] some sand in the well-greased wheels’’23

through the expected reduction in transaction vol-ume via the tax-induced increase in trading costs.

Tobin based his case for the Tobin tax on theclaim that it would reduce market volatility. Thus,one could in principle refute his argument bydemonstrating empirically that it would not in-crease asset price stability. Moreover, while Tobinshared Keynes’s concern about speculation’s impacton efficient resource allocation, as manifested in hisskepticism about financial markets’ efficiency in the‘‘deeper economic-informational sense,’’ he usedthat more as a response to the concern that theTobin tax might undermine efficient price revela-tion than as an affirmative motivation for the tax.

Keynes, while mentioning asset price and result-ing macroeconomic ‘‘instability due to speculation,’’did not so closely link the diagnosis and the pro-posed cure, given the issue of resource misalloca-tion.24 Accordingly, his argument for an STT mightmore readily survive empirical refutation of theclaim that it would increase asset price stability.And whether it would succeed in improving theextent to which financial markets ‘‘direct new in-vestment into the most profitable channels in termsof future yield’’25 might be hard either to confirm orrebut empirically. There is not even general agree-ment about the importance of secondary stockmarket prices, with some arguing that they play no

basic informational role in the economy26 and donot significantly affect real resource allocation.27

There also is a distinction between Keynes’s andTobin’s explanations of how short-term, beauty-contest-driven trading undermines the aspects offinancial market performance that they emphasize.In general, lower trading volumes and longer hold-ing periods are logically linked, all else being equal,since if you are trading less, you presumably areholding the financial assets in your portfolio, onaverage, for longer periods. But they are conceptu-ally distinct. Tobin cares about trading volume,which he believes (as we will see, with at bestmixed empirical support) leads to sharper pricefluctuations, including through price bubbles andpanics. Keynes appears to be focusing more onaverage holding periods.

A possible reason for that focus, if one sharesKeynes’s concern about the distinction betweenspeculation and enterprise, is as follows. Exceptinsofar as one anticipates profiting from a stock’sactual cash distributions, whether from dividendsor the ultimate liquidation proceeds, one cannothelp being subject to the beauty contest problem,even if one waits 50 years to sell. After all, the valuethat potential buyers (as well as lenders) ascribe tothe stock will always reflect the ongoing influenceof ‘‘what average opinion expects the average opin-ion to be.’’28 Keynes’s claim accordingly must bethat the longer you first hold the stock beforeseeking to sell it, the greater the likely effect of newinformation about fundamental value, even withongoing noisy random variation from the beautycontest’s vicissitudes. Thus, people who expect tohold stocks longer will pay more attention to enter-prise and fundamental value, even if they can neverignore the concerns of speculation. Another argu-ment, frequently made but not yet attracting con-sensus either for or against it, is that the prevalenceof speculation may worsen corporate governance,by encouraging both the shareholders who mightotherwise more carefully monitor managers and themanagers themselves to embrace ‘‘short-termism’’at the expense of focusing on long-term valueenhancement.29

Those considerations suggest caution about tooswiftly embracing Keynes’s or Tobin’s grounds for

22Id. at 154.23Id. at 158.24See Keynes, supra note 6, at 161.25Id. at 158.

26Joseph E. Stiglitz, ‘‘Using Tax Policy to Curb SpeculativeShort-Term Trading,’’ 3 J. of Fin, Services Res. 107 (1989).

27Lynn A. Stout, ‘‘The Unimportance of Being Efficient: AnEconomic Analysis of Stock Market Pricing and SecuritiesRegulation,’’ 87 Mich. L. Rev. 645-651 (1988).

28See Keynes, supra note 6, at 156.29See, e.g., Stout, ‘‘Are Stock Markets Costly Casinos? Dis-

agreement, Market Failure, and Securities Regulation,’’ 81 Va. L.Rev. 687 (1995).



(C) T

ax Analysts 2012. A




advocating an FTT, even if one believes (as I do)that the beauty contest metaphor offers an impor-tant and convincing insight into how financialmarkets actually function. The journey from under-mining one’s confidence in financial markets’ effi-ciency in the ‘‘deeper economic-informationalsense’’30 to embracing a particular policy responsemay not be as clear as Keynes and Tobin suggest.

B. Efficiency Problems With an FTTEven if an FTT has desirable qualities for the

reasons identified by Keynes or Tobin, it also hassignificant defects from the standpoint of efficiency.Indeed, without a significant positive rationale, theefficiency arguments against adopting it would bevery compelling. Consider the following:

• An FTT applies to transactions’ gross, ratherthan net, proceeds. Suppose I buy 100 shares ofSiemens stock for $10,000 and then sell thesame shares for the same amount. Under anSTT, I will be taxed twice, despite not havinggained any profit. An STT thereby discourageseconomic activity without (at least directly)advancing the distributional aim of makingthose who have fared better pay more. Bycontrast, taxes on net proceeds — for example,income taxes and VATs — while also discour-aging economic activity, at least have the ad-vantage of directly serving that distributionalgoal.

• An FTT imposes cascading taxes on interbusi-ness transactions. The more that the produc-tion process involves taxable sales from onebusiness to another before the ultimate sale toa customer, the greater the tax burden on aproduct. For good reason, EU member stateshave mainly rejected these taxes since the riseof the VAT in the 1950s. Economic theoryconfirms that under plausible conditions, pro-duction efficiency is maximized, without anyloss of the ability to achieve desired distribu-tional goals, by not imposing those taxes.31

• Taxing sale transactions, while not taxing thedecision just to hold particular financial assets,creates needless inefficiency unless — asKeynes and Tobin argue — the sales imposeexternal costs on others or are correlated withotherwise unobserved flaws that merit taxdiscouragement. In complete markets whereeveryone is a price taker and there always areavailable counterparties, a would-be buyer or

seller is seeking to improve his own expectedwelfare and is not adversely affecting anyoneelse’s. In thin markets, where counterparties ata ‘‘fair’’ price are hard to find, one’s willing-ness to buy or sell may create positiveexternalities for others by enabling them totransact more easily at such a price. Thus,unless there is more to the story, tax-penalizing sales is hard to rationalize.

• Realization-based income taxes already dis-courage sales of appreciated assets, which mayyield taxable gain to the seller that couldotherwise be deferred or even permanentlyavoided. Accordingly, if taxing sales is unde-sirable, an STT does not merely start from zeroin undesirably discouraging them, but mayactually worsen preexisting distortions for ap-preciated assets. (On the other hand, the STTmay offset inefficient income tax encourage-ment of sales of loss assets.)

• Depending on their design, STTs risk beinghighly avoidable in at least two ways. The firstis location. If there is an STT on sales inCountry X but not on those in Country Y,taxpayers may find it a lot easier to change thesale location than to change more meaningfulchoices such as where individuals live orwhere tangible business activity occurs. Swe-den recently learned that the hard way whenits FTT, within a four-year period, inducedmore than half of all domestic securities trad-ing to move to London.32

• The second way in which an STT is potentiallyhighly avoidable concerns the rules for defin-ing both specific financial instruments andtaxable transactions such as sales. Now thattransactions using derivatives have becomecommon — for example, through the use ofswaps that depend on the performance ofSiemens stock in lieu of buying or selling thatstock — an STT is unlikely to be very effectiveunless the taxation of derivative transactions isadequately aligned with that of the ‘‘primary’’transactions they may replace.

To illustrate, suppose that in the absence of anSTT, I would borrow $10 million and use the fundsto buy Siemens stock. If the STT only applied toliteral sales, I could avoid it, while replicating theeconomics of this transaction as follows. Presum-ably with a financial firm like a bank as my counter-party, I could simply arrange a swap based on anotional principal amount (NPA) of $10 million. On

30See Tobin, ‘‘A Proposal for International Monetary Re-form,’’ supra note 8, at 158.

31See generally Peter A. Diamond and James A. Mirrlees,‘‘Optimal Taxation and Public Production: I — ProductionEfficiency,’’ 61(1) Am. Econ. Rev. 8 (1971).

32Marion G. Wrobel, ‘‘Financial Transaction Taxes: The Inter-national Experience and the Lessons for Canada,’’ Parliament ofCanada (1996).



(C) T

ax Analysts 2012. A




the transaction date, no cash would change hands(leaving aside the likelihood that the counterpartywould insist on my posting collateral to secure mypotential liability under the swap). On the swapsettlement date, I would owe the bank an amountequal to the interest that I would have owed on anactual $10 million loan during the swap term. Thebank would owe me the dividends and apprecia-tion that $10 million of Siemens stock would haveyielded during the swap term. Accordingly, I wouldend up in exactly the same position (leaving asidetransaction cost differences) as if I had actuallymade a debt-financed purchase of $10 million ofSiemens stock, yet there would not have been anactual (or at least literal) sale:

• STTs, like income and wealth taxes but unlikeconsumption taxes such as VATs, discourageinvestment and saving. They are thus subjectto the same critique as income taxes for creat-ing needless distortion without necessarilymaking possible greater progressivity.33 How-ever, even if one favors taxing investment andsaving, it is unclear (barring rationales such asthose advanced by Keynes and Tobin) why onewould favor this particular mechanism, ratherthan one like a wealth tax or an income tax,which depends on the amount saved or thereturn to saving rather than on the grossamounts involved in sale transactions.

• STTs can be expected to increase the cost ofcapital when firms seek to raise funds byissuing tradable securities. Even if the initialissuance is not taxed, investors may well an-ticipate that the tax will affect resale prices.

C. The European Commission’s FTTThe commission’s FTT clearly was ‘‘designed

with an eye to [addressing] known weaknesses inFTTs by ensuring that its scope is broad along anumber of dimensions.’’34 However, breadth in-tended to address avoidability is only one of thetwo main design principles that one can infer byexamining the commission proposal’s main fea-tures. The other is attempting to create both theappearance and the reality of a tax that falls on the‘‘financial sector’’ — a term that, as I will discuss inPart III, requires a bit of unpacking — while osten-

sibly ensuring that households and small tomedium-size business enterprises will ‘‘hardly beaffected.’’35

1. Addressing avoidability. The commission’s STTbroadly defines covered financial transactions toinclude not only the trading of equity and commer-cial debt, but also the ‘‘conclusion or modificationof derivatives agreements’’ and the ‘‘purchase/saleor transfer of structured products’’ along with se-curitizations.36 The proposal uses a broad definitionof financial institutions, which must be involved ina transaction in order for the tax to apply. Thecovered institutions include not just conventionalbanks (other than the European Central Bank andnational central banks, which are exempted), butalso ‘‘investment firms, organized markets, creditinstitutions, insurance and reinsurance undertak-ings, collective investment undertakings and theirmanagers, pension funds and their managers, hold-ing companies, financial leasing companies, specialpurpose entities . . . and other persons carrying outcertain financial activities on a significant basis.’’37

Among the proposal’s most notable features is itsuse of residence-based jurisdiction. Past FTTs, likethe one that worked out so poorly for Sweden in the1980s, have typically applied to transactions thatwere executed domestically — meaning that all onehad to do to avoid them was move the place of saleabroad. The commission’s STT, by contrast, wouldapply on a residence basis. Thus, if any party to afinancial transaction is ‘‘established in the territoryof a Member State,’’ the tax applies. Such establish-ment is itself defined broadly. It includes not onlyfinancial institutions that are registered in an EUcountry or are authorized to act (say, as banks) thereor that have headquarters in an EU country, but alsofirms that otherwise would be treated as foreign butthat have a branch in an EU country. However, EUresidence for purposes of the FTT apparently doesnot extend to foreign firms that have separatelyincorporated EU resident subsidiaries but choose totransact through other affiliates.

EU resident financial institutions, as defined un-der the proposal, generally would face the tax nomatter where a transaction was executed.38 More-over, if an EU resident (such as an individual)

33See, e.g., Daniel N. Shaviro, ‘‘Replacing the Income TaxWith a Progressive Consumption Tax,’’ Tax Notes, Apr. 5, 2004, p.91, Doc 2004-6003, or 2004 TNT 66-48; Joseph Bankman andDavid A. Weisbach, ‘‘The Superiority of an Ideal ConsumptionTax Over an Ideal Income Tax,’’ 58 Stan. L. Rev. 1413-1456 (2006);Shaviro, ‘‘Beyond the Pro-Consumption Tax Consensus,’’ 60Stan. L. Rev. 745-788 (2007).

34John Vella et al., ‘‘The EU Commission’s Proposal for aFinancial Transaction Tax,’’ Brit. Tax Rev. No. 6, 607 (2011).

35European Commission, ‘‘Explanatory Memorandum: Pro-posal for a Council Directive on a Common System of FinancialTransaction Tax and Amending Directive 2008/7/EC,’’ article 5(Sept. 2011).

36Id. at article 3.3.1.37Id.38The council directive states, however, that ‘‘in case the

person liable to pay the tax was able to prove that there is no


(Footnote continued on next page.)


(C) T

ax Analysts 2012. A




participates in a financial transaction that exclu-sively uses non-EU financial institutions, those in-stitutions will be treated as residents for purposesof the transaction and thus will have to pay thetax.39

A follow-up draft report by the Committee onEconomic and Monetary Affairs of the EuropeanParliament extends the FTT’s proposed reach sothat it also applies to any transaction that ‘‘involvesa financial instrument issue by legal entities regis-tered in the Union.’’40 Thus, if an American stock-broker sells Siemens stock to an American investoron the New York Stock Exchange, at least ostensiblythe transaction is subject to the EU’s FTT. Likewise,if non-Europeans use derivatives, such as a swap, inwhich one of the party’s payments is computedwith reference to stock in a company that is incor-porated in Europe, the FTT likewise ostensiblyapplies. In either case, however, collection of the taxappears unlikely.

Ignoring the enforcement problems that may beassociated with requiring tax remittance by finan-cial institutions that operate outside the EU, thestated breadth of application could indeed reducethe tax’s avoidability. However, one would expectnon-EU persons to react by shifting away from theuse of EU financial institutions to execute theirdeals. Thus, for example, U.K. banks that competewith rivals in the United States and Hong Kong toattract business from non-EU individuals will pre-sumably find themselves at a disadvantage. Also,the ability of multinational corporate entities toavoid the tax by using their non-EU rather thantheir EU-based affiliates to conduct transactionsmay prove to be a significant gap.41

One of the trickier questions posed by the pro-posal is how to determine the taxable amount in atransaction using derivatives. For a sale of $10million worth of equity, the taxable amount is ofcourse $10 million, to which a 0.1 percent tax (thatis, $10,000) is supposed to apply to each taxpayer.42

However, since (as discussed below) transactionsbetween entities that are members of the samecorporate group are subject to the STT, in those

cases transfer pricing issues will arise. With deriva-tives, however, determining the taxable amount isnot so easy.

Suppose that two parties effectively did the $10million swap involving Siemens equity that I de-scribed above. Unsurprisingly, the taxable amountwould be the $10 million NPA.43 The commissionrecognizes, however, that things will not always bethis simple: ‘‘For example . . . the notional amountof a swap could . . . be divided by an arbitrarilylarge factor and all payments multiplied by thesame factor.’’44 In the above case, that might involvean NPA of $1 million and requiring each party topay at 10 times the rates used in the precedingexample. That type of problem would be dealt withusing undetermined ‘‘special provisions.’’45

Without presuming to anticipate all the ways inwhich sophisticated financial engineers could pack-age economic returns that effectively depend on theperformance of $10 million of Siemens stock (orsomething that is correlated with it), one can beconfident that many possibilities will present them-selves. And in a world where the same financial betcan be expressed in so many ways,46 often it may beimpossible to identify or even define the ‘‘true’’underlying financial transaction.

The commission’s proposed response to thatproblem is twofold. First, when ‘‘more than onenotional amount is identified, the highest amountshall be used for the purpose of determining thetaxable amount.’’47 The standards for determiningthe list of possible NPAs remain to be considered.Second, presumably to respond to the concern thatthat might result in unduly tax-disfavoring deriva-tive transactions that could be decomposed in multi-ple ways, the tax rate for those transactions is only0.01 percent,48 or one-tenth that which otherwiseapplies. Presumably, that means the tax on the saleof $10 million in Siemens stock can be reduced from$10,000 to $1,000 by replacing it with the swap.49

link between the economic substance of the transaction and theterritory of any Member State,’’ the tax may not apply. Id. atarticle 3.3.1.

39Id.40See Committee on Economic and Monetary Affairs, supra

note 3, at 10.41See Vella et al., supra note 34, at 4.42See European Commission, ‘‘Proposal for a Council Direc-

tive on a Common System of Financial Transaction Tax andAmending Directive 2008/7/EC,’’ article 8 (Sept. 2011).

43See European Commission, supra note 35, at section 3.3.2.44Id.45Id.46Consider, for example, the put-call parity theorem, which

‘‘states that given any three of the four following financialinstruments — a zero-coupon bond, a share of stock, a calloption (‘call’) on the stock, and a put option (‘put’) on the stock— the fourth instrument can be replicated.’’ Michael S. Knoll,‘‘The Ancient Roots of Modern Financial Innovation: The EarlyHistory of Regulatory Arbitrage,’’ 87 Ore. L. Rev. 93, 95 (2008).

47European Commission, supra note 35, at article 6.48Id. at article 8.49However, that assumes that the swap is not characterized

for STT purposes as instead of or even also a sale. See EuropeanCommission, supra note 35, at article 3.3.1 (‘‘The scope of thetax . . . is also not limited to the transfer of ownership but ratherrepresents the obligation entered into, mirroring whether or not




(C) T

ax Analysts 2012. A




The Committee on Economic and Monetary Af-fairs of the European Parliament, in a draft reportamending the original commission proposal, ex-plains the lower tax rate for derivative transactionsas follows: ‘‘In the case of derivatives, the estima-tion of their value being much more difficult, thedecision to opt for the notional value — which canbe significantly higher than the real market value ofa derivative — justifies the choice of a lower rate.’’50

However, that does not entirely make sense. Con-sider the swap described that involved $10 millionin Siemens stock versus the interest on $10 million.In principle, each side of that swap should initiallybe worth zero. However, that is a net, rather thangross, value measure, and net value generally isirrelevant under the FTT. After all, a wholly debt-financed explicit purchase of $10 million in Siemensstock also has a net value of zero, and yet would betaxed as a $10 million transaction.

The committee may have in mind that oftenwhen parties enter into derivative transactions, theymay not be seeking to substitute for deals thatwould have involved the full NPA. Instead, theymay be seeking to hedge particular economic riskscheaply, such as by paying a modest premium for aposition that has a positive expected (althoughcontingent) payout. In that setting, charging the full‘‘normal’’ tax rate based on a given NPA might beviewed as imposing very high effective tax rates onthe ‘‘true’’ underlying transactions. Unfortunately,however, the fact that similar-looking derivativetransactions may actually be substitutes for verydifferent ‘‘primary’’ transactions makes it extremelydifficult for an FTT to replicate the tax burdens thatmight have been deemed appropriate if derivativesdid not exist.51

2. Seeking to direct tax burdens to the financialsector. The commission’s goal of directing bothactual and perceived tax burdens to the financialsector could not have been accomplished simply bymaking financial firms the only parties that remitSTT payments. After all, only businesses must remit

VAT payments to the tax authorities, yet VATs arewidely understood as taxing households based ontheir consumption. However, two further sets ofdesign features appear to reflect the commission’sstated distributional aim. The first concerns exclu-sions from the reach of the proposed FTT, while thesecond concerns what might be regarded as asurprising category of inclusions.

As noted above, the proposal exempts spot cur-rency transactions, thus preserving the free move-ment of capital.52 Also, consistent with the aim ofkeeping citizens and business outside the scope ofthe FTT, it excludes the primary issuance of debtand equity securities. It is unclear how broadly thatexclusion applies, such as when the underwriters inan initial public offering take on a lot of new stockbefore deciding how much of it they plan to sellpromptly. However, even with a broadly definedexclusion for securities sales that are related toprimary issuance, the tax would be expected toimpose a burden on that issuance, since the amountthat primary purchasers are willing to pay presum-ably will reflect the prospect of a future tax onresale.

Also exempted, presumably to limit the tax bur-dens directly imposed on households, are ‘‘insur-ance contracts, mortgage lending, consumer credits,payment services, etc.’’53 One wonders if creativetax planners could use those exclusions to avoid thereach of the FTT. Consider, for example, that creditdefault swaps are economically akin to insurance,because they offer the holder a payoff if the debtorfails to pay. Could insurance-like financial productsthat financial institutions and their customers trade,and that the proposal presumably is intended toreach, therefore be designed to escape the STT? Thatmight require the cooperation of an EU host coun-try if, in applying the exclusion, insurance is de-fined in terms of being subject to conventionalnational insurance regulation. But a country thatdecided to cooperate with aggressive tax plannersmight view itself as benefiting from the opportunityto attract transactions and business. In effect, itwould be engaging in tax competition with the restof the EU by not actually levying the same tax asother countries, even if it appears to be doing so.

This brings us to the perhaps surprising inclu-sions. If two or more EU resident financial institu-tions participate in a transaction, each is fullytaxable. Thus, if one European bank sold €10 mil-lion in Siemens stock to another, apparently eachwould pay €10,000 of tax. More complicated trans-actions with, say, 10 financial industry participants

the financial institution involved also assumes the risk impliedby a given financial instrument (‘purchase and sale’)’’).

50Committee on Economic and Monetary Affairs, supra note3, at 18.

51The FTT variant recently proposed in the United Statesapproaches derivatives differently than the FTT proposal. Whiletreating the issuance of a derivative as taxable, the U.S. proposalmakes no effort to determine NPAs, and instead simply appliesits general 0.03 percent rate to the fair market value of (andpayments made under) derivative instruments. See S. 1787 andH.R. 3313, supra note 2, at proposed section 4475(b), (e)(1)(D)through (F), and (h). In cases like the above-described Siemensswap, the proposed U.S. version therefore would permit tax-payers to greatly lower their FTT liabilities by substitutingderivatives transactions for ‘‘actual’’ ones.

52European Commission, supra note 35, article 3.3.1.53Id.



(C) T

ax Analysts 2012. A




would lead to the imposition of 10 taxes. Even in arelatively simple deal, that could matter a lot.Suppose I sell Siemens stock to you (we are both EUresidents), and each of us uses a broker. Will threetaxes be imposed, rather than one, if I first transferthe stock to my broker, who then transfers it to yourbroker, as intermediate stages in the transaction?54

Moreover, the rule taxing all financial industryparticipants applies not just to arm’s-length dealsbetween unrelated firms, but also to transactionstaking place between entities of a group55 — per-haps including distinct branches, like those in dif-ferent countries, rather than just separatelyincorporated affiliates. The proposed FTT wouldtherefore create cascading taxes within the financialsector that could not be avoided through commonownership. That is a design virtue in one sense,because it avoids creating inefficient tax incentivesfor consolidation. But it is a vice in another sensebecause it strengthens the presumably inefficientcascading tax.

D. The FAT Variants Discussed by the IMF StaffI now turn to the alternative of enacting a FAT

instead of an FTT. Purely on the level of rhetoric, itis difficult to imagine a question that initiallysounds as tedious as whether we should tax finan-cial transactions or activities, and thus endorse theF-blank-T acronym with the T or the A in themiddle. In fact, the key element of the choicebetween the two taxes can be presented a lot moreconcisely than that. An FTT targets the gross pro-ceeds, while a FAT targets some variant of the netproceeds (that is, the gross proceeds minus speci-fied cash outlays) that financial firms generatethrough their business activities.

Suppose we were evaluating that tax designchoice for the food industry, rather than financialinstitutions. Then the question would be whetherretail stores, wholesalers, farmers, and the likeshould be taxed on their gross sales proceeds (in-cluding those arising from transactions betweenseparate entities within the food industry)56 or only

on some net measure of industrywide profits orvalue added. We should keep in mind, however,that this is a question of tax design, not of whethertaxes should be higher or lower. While the grossproceeds tax would nominally have a much largertax base, it presumably would use a much lowerstatutory rate if the two alternatives were meant toimpose the same overall burden on the industry orto raise the same amount of revenue.

While a gross proceeds tax on the food industrymay appear bizarre,57 it should help to make moreintuitive the key difference between an FTT and aFAT. An FTT targets a transactional measure ofoverall gross activity in the financial sector; a FAT,the profits or — in an accounting if not a socialsense — its value added.

What might a FAT look like? Here we againencounter the Gertrude Stein issue that I noted atthe start. That is, the answer depends on the ques-tion, which initially is why one would considerimposing a profits tax that is particular to thefinancial sector. After all, income taxes generallyapply to all industries (although as we will see, theydefine the profit concept differently than do the FATvariants that have recently been discussed).

In that regard, it is best to go to the source.Contemporary discussion of the FAT, including itsacronym and name, dates from an important pub-lication by the IMF staff, published in June 2010 inresponse to a request from the G-20 leaders that theIMF describe a ‘‘range of options . . . as to how thefinancial sector could make a fair and substantialcontribution toward paying for any burden associ-ated with government interventions to repair thebanking system.’’58 The report argues that an FTT‘‘does not appear well suited to the specific pur-poses set out in the mandate from G-20 leaders.’’59

It therefore advances three alternative FAT variants,each with a distinct design reflecting particularpurposes.

54In a letter directed to clients, the British law firm CliffordChance describes what it views as a common scenario in whicha simple stock sale might lead to the imposition of six, ratherthan just three, taxes. See Clifford Chance, ‘‘Financial Transac-tion Tax: Update,’’ (2011), available at http://www.cliffordchance.com/publicationviews/publications/2011/10/financial_transactiontaxupdate.html.

55See European Commission, supra note 35, at article 3.3.1.56If transactions between food industry entities were not

taxed, then the tax on its gross sales proceeds from transactionsoutside the sector would cause it to resemble a retail sales taxjust on the food industry. The main difference would be itsreaching food sales that were production inputs to other indus-tries, rather than just sales made directly to consumers.

57However, one could imagine there being plausible motiva-tions for a gross proceeds tax on the food industry. Suppose theindustry imposed negative externalities — relating to its envi-ronmental effects or to the publicly borne healthcare costsresulting from obesity or poor diets — and that no more directmeasure of the harm being caused than sectorwide transactionalactivity was available. In that scenario, one would want toimpose taxes to compensate for the externalities even if the foodsector was merely breaking even. In the absence of a betterproxy, the gross proceeds tax might be worth consideringdespite its undesirable imposition of a cascading tax on trans-actions within the food industry.

58IMF staff, supra note 5, at 4.59Id. at 19. The report offers three reasons for considering the

FTT ill suited to the G-20 leaders’ mandate: (1) the volume offinancial transactions is not a good proxy for the firm-levelbenefits and societal costs resulting from the prospect of finan-cial firm bailouts; (2) the FTT ‘‘would not target any of the key




(C) T

ax Analysts 2012. A




The IMF’s three models, which it calls FAT-1,FAT-2, and FAT-3, are not wholly distinct choices,like doors 1 through 3 in the famous TV game showLet’s Make a Deal. Rather, they relate to each otherlike Matryoshka dolls, one nested inside another.FAT-1 is the broadest, while FAT-2 and FAT-3 usenarrower bases so that they can target particularelements of the FAT-1 base that it would includenondistinctively.

1. FAT-1. This is essentially a special or modifiedVAT on the financial sector. To explain it, we willstart from a standard VAT, which is a tax on ‘‘salesof real goods and services less purchases of non-labor inputs.’’60 Thus, a grocery store would includeall of its proceeds from sales to consumers, whileeffectively deducting (in the form of a credit thatwould be computed at the VAT rate) its outlays toother VAT-paying businesses. However, the wagesthat it paid would not be deducted — reflecting thatits workers, unlike the businesses to which it madedeductible payments, would not face VAT liabilityon the amounts received. Likewise, financial flows,such as interest payments that the grocery storemade to banks that had helped fund its operations,would neither be deducted by the store nor in-cluded by the banks.

The VAT, given that it taxes ‘‘value added’’ in thesense of sales minus purchases without regard towages or financial flows, is ‘‘implicitly a tax on thesum of wages and ‘profits’ defined in cash flowterms (that is, with full expensing of investmentand no deduction for financial costs).’’61 Profits inthat sense refers to returns in excess of the normalrate of return on investment, which effectively isexempted by allowing the business’s capital outlaysto be expensed.

VATs normally do not apply to financial sectorfirms, reflecting that financial flows, such as interestpayments, generally are excluded from it. TheFAT-1, however, is designed to extend to the finan-cial sector the basic VAT concept of taxing the sumof its wages and profits in the above sense.62

Despite the conceptual overlap, however, the FAT-1not only requires a different method than a plain-vanilla VAT, but (as we will see) is rationalizeddifferently.

Why do VATs — along with retail sales taxes(RSTs), which many U.S. states and localities im-pose — generally ignore financial cash flows, likeinterest payments on a loan? In the business sector,the combination of exclusion on the lender side andnondeductibility on the borrower side has zero netimpact if the tax rates for borrowers and lenders arethe same. For example, if Siemens pays DeutscheBank $10,000 of interest and both pay tax at a 25percent rate, the net tax revenue produced frombringing that cash flow within the reach of theGerman VAT would be zero: Deutsche Bank’s$2,500 tax liability would be offset by Siemens’s$2,500 tax recoupment.63 Exempting that interbusi-ness transaction is a wash.

But what about the fact that for VAT purposes,Deutsche Bank does not merely get to ignore finan-cial flows in computing its liability, but is entirelytax exempt? If all that Deutsche Bank did wasengage in interbusiness transactions, the VAT ex-emption would raise revenue. After all, if DeutscheBank were a full-fledged VAT taxpayer, then inaddition to paying tax and getting credits for finan-cial cash flows that (net of the effects on businesscounterparties) are a wash, it would also get re-funds or credits for purchasing nonfinancial inputslike buildings and desks. In sum, VAT exemption isa detriment, rather than a benefit, to financial firmsinsofar as they are purely engaged in business-to-business transactions.

But why doesn’t the VAT apply to transactionsbetween financial (and other business) firms andhouseholds? Suppose I pay interest on a vacationloan to a bank, which would lead to VAT liability ifI had been paying for the vacation itself. Here theproblem is that imposing the VAT on interest flowsbetween businesses and consumers would raise thequestion of what a consumer is. At least from thenormative standpoint that underlies support forconsumption taxes like the VAT, one could arguethat no one is really a consumer by reason of saving(or dissaving) and thereby earning a positive (ornegative) financial return. After all, earning orpaying interest is not itself an act of consumption,

attributes — institution size, interconnectedness, and substitut-ability — that give rise to systemic risk’’ potentially necessitat-ing bailout; and (3) the real incidence of the tax might fall onconsumers, rather than on earnings in the financial sector. Id. at19-20.

60Id. at 66.61Id.62See id. (stating that ‘‘it would be appropriate’’ to design the

FAT-1 similarly to the basic VAT concept).

63Because after-tax interest rates presumably reflect supplyand demand, one might further expect nominal or pretaxinterest rates to adjust to the choice of tax rule, such that atequilibrium, Siemens and Deutsche Bank, and not just the taxauthorities, would end up in the same after-tax position eitherway. The main reason an income tax, unlike a consumption taxsuch as a VAT, cannot so readily ignore interbusiness interestflows even if all parties tax rates’ are the same is that theborrower’s interest expense may need to be capitalized ratherthan deducted — for example, if it contributes to creating adurable asset. Thus, interbusiness interest flows may yield nettax revenue in an income tax system that would be overlookedif they were ignored.



(C) T

ax Analysts 2012. A




and consumption taxes typically are neutral regard-ing the timing of consumption, a goal that generallyrequires ignoring time-value-based returns to sav-ing.64

If the financial sector served simply as an uncom-pensated middleman between households thatwere borrowers and those that were savers, hand-ing along interest payments from the former to thelatter without getting paid for it, then the VATexclusion for financial transactions would still beimmaterial. After all, as an uncompensated middle-man, the sector would have neither wages norprofits. But of course that is not the case. Even thesimplest community bank generally charges ahigher interest rate on loans than it offers onsavings deposits, not just because of default risk,but also because of the services it renders in itsmiddleman role. In effect, it is bundling its servicefee with the interest charge that one might naivelyhave thought merely reflected the time value ofmoney.

Even that bundling might not matter if we be-lieved that households’ financial transactions havezero consumption content. But that characterizationwould be highly questionable. Suppose I have achoice between saving with two banks. The first oneoffers free checking and free ATM use but pays mea low interest rate. The second charges me for bothservices, but offers me a higher interest rate. Thereis a powerful argument that the convenience ofbeing able to write checks and get money from anATM contributes to my consumption.65

If banks charged separately stated fees for allservices provided that offered consumption valueto the purchasers, it would be conceptually simpleto subject those fees to the VAT, without permittingthe payer to deduct or credit them. But with bun-dling plus the difficulty of identifying the consump-tion component of using financial services, real-world VATs (as well as RSTs) have generally settledfor simply exempting the financial sector. Thus,they get a VAT subsidy insofar as consumer trans-actions are concerned, and an overall or net subsidyif that tax benefit outweighs the detriment fromexempting business-to-business transactions.66

Less widely recognized is the fact that financialfirms may get a similar tax subsidy for consumertransactions under the income tax. That subsidyresults from their ability to give their customersimplicit deductibility for consumer fees, such as forthe personal convenience of using checks andATMs, by using bundling in the form of offeringwhat would otherwise be below-market interest onthe money deposited in checking and savings ac-counts. Tax experts often ignore that point, perhapsbecause they are unduly fixated on whether finan-cial firms are expressly tax exempt (even thoughVAT exemption is a detriment for business-to-business transactions). In practice, it is plausiblethat the net income tax subsidy to the financialsector exceeds that under the VAT, assuming similarrates under the two taxes, given that financial firmscan deduct (or at least capitalize) their outlays fornonfinancial inputs. Thus, the common focus on theVAT as the main source of subsidies may be mis-guided.

With all that in mind, let’s return to the FAT-1,which seeks to impose a VAT-like tax on financialsector wages and profits. How does one accomplishthat technically, given the bundling problem? Oneway would be to require that financial sector firmsinclude and deduct all cash flows, such as interestpayments but also loan principal, rather than just(as with VATs and RSTs) including cash flows fromtransactions that involve ‘‘real’’ rather than finan-cial goods and services.67 That would prevent bun-dling from affecting financial firms’ FAT-1 liability,since the same tax would apply regardless of thechoice between financial and nonfinancial labels oncash flows. For convenience, I will call this the cashflow tax version of FAT-1.

However, one could also implement the FAT-1 byusing a tax base for financial firms that was some-what like a broad-based corporate income tax, suchas in how it ignores flows of loan principal, but withthe following adjustments: (a) wage nondeductibil-ity; (b) expensing for all outlays to other businesses,including those that would be capitalized under theincome tax; and (c) the allowance of an interestlikededuction for the taxpayer’s equity. Providing aninterestlike deduction for equity that reflected thenormal rate of return would ensure that only profitsabove that rate of return, plus the amount of thefinancial sector’s wages, would remain in the taxbase. I will call this the allowance for corporateequity (ACE) version of the FAT-1.

One could further modify the ACE version of theFAT-1 by having a notional time value of moneydeduction apply to corporate debt as well as equity,

64See Shaviro, ‘‘Replacing the Income Tax With a ProgressiveConsumption Tax,’’ supra note 33, at 104.

65The point is better described as relating to the value ofuntaxed leisure. The convenience of being able to write checksand get money from ATMs increases my opportunities to deriveuntaxed imputed income from enjoying leisure.

66A recent study by a major accounting firm argues thatbanks’ tax exemption under EU VATs actually disadvantagesthem on balance, because of the business-to-business issue. SeePricewaterhouseCoopers LLP, ‘‘How the EU VAT ExemptionsImpact the Banking Sector’’ (2011). 67IMF staff, supra note 5, at 66.



(C) T

ax Analysts 2012. A




in lieu of providing deductions for actual interestpaid or accrued. (However, for banks that paidbelow-market interest, such as on retail checkingand savings accounts, one would need to limit thededuction to the amount actually paid, as a devicefor indirectly including the offsetting implicit serv-ice fees.) For financial instruments that paid morethan the notional time value of money, that wouldmake the distinction between debt and equity, andthus the question whether a given payment tofinancial instrument holders constituted interest ora dividend, irrelevant for purposes of applying theFAT-1. I will call this the allowance for corporatecapital (ACC) version of the FAT-1.

No matter which of those versions is used, thereis an important respect in which the FAT-1 wouldnot merely result in effectively extending the VAT tofinancial firms. Assuming that nonfinancial firmsdid not get a refund (under either the VAT or theFAT-1) for their payments to financial firms, theFAT-1 would impose a net tax on interbusinesstransactions. As the IMF staff report makes clear,the rationale for that — leaving aside the issue ofuntaxed personal consumption from financialtransactions — would be to extract from the finan-cial sector a ‘‘fair and substantial contributiontoward paying for any burden associated withgovernment interventions to repair the bankingsystem.’’68 In effect, the idea is that the detriment ofovertaxing — or indeed increasing existing VATovertaxation of — business-to-business financialservices is outweighed by the benefit of addressingVAT (and income tax) undertaxation of business-to-consumer financial services. The report proposesthat concern about cascading tax liabilities on inter-business transactions be addressed by charging theFAT at lower than the generally prevailing VAT ratein order to limit the damage.69

One problem with the FAT-1, however, is thatbesides not directly targeting the behavior thatgives rise to the possibility of bailout, it also onlyimperfectly offsets business-to-consumer under-taxation (given the business-to-business overtaxa-tion problem). Accordingly, even if its rate was setso as to raise exactly the right ‘‘excess’’ amount ofrevenue from the financial sector as a whole (that is,relative to simply offsetting the sector’s VAT andincome tax subsidies), liability under the FAT-1would be misallocated.70

Thus, the FAT-1 is certainly not a first-best financ-ing mechanism for requiring a ‘‘fair and substantialcontribution’’ from the financial sector. The case forit would have to rely on one’s inability to do betterin targeting VAT and income tax gaps and thecreation of bailout risk. The case for using the FAT-2or the FAT-3, in lieu of the FAT-1, rests on thepossibility that by narrowing the base in a targetedway, one could lower the general efficiency costs orbetter address the bailout risk problem.2. FAT-2. The FAT-2 takes on that challenge asfollows. Suppose we want to charge the financialsector for expected bailout costs plus the value of itsVAT and income tax subsidies (via the effectiveexclusion of consumer services), but that we cannotdo so in a first-best fashion by directly taxing thethings that we have in mind. We might be reassuredif we could design the tax to create as little ineffi-ciency as possible.

From that standpoint, consider the idea of taxingwhat economists call rents. As I have explainedelsewhere:

In lay terminology, rents are what you payyour landlord each month. Economists, how-ever, use the term to denote ‘‘payments toresource deliverers that exceed those neces-sary to employ the resource.’’ . . . An examplewould be Michael Jordan, back in the day,when he could earn $30 million per yearplaying basketball and no more than, say,$100,000 doing anything else with his time.The existence of this $29.9 million excess ofwhat Jordan could earn by playing basketballover the next best use of his time potentiallyhas an important tax policy implication. If heplanned to work [with the same intensity] inany event, at whatever occupation paid himthe most, one could tax away all of the extrareturn (leaving only, say, an extra cent) and hestill would play basketball rather than doinganything else. A very high tax would thereforeresult in no economic distortion of behavior,contrary to what one normally expects.71

The FAT-2 reflects the premise that it wouldtherefore be highly efficient to tax financial sectorrents. To be sure, that is equally true for rentsderived outside the financial sector. However, non-financial sector rents already are subject to VAT inmost countries, and in any case the IMF staff wasspecifically tasked with exploring how taxes mightbe increased on the financial sector. What is more,the staggering growth of the sector’s profits andhigh-end compensation over the last two decades68Id. at 4.

69Id. at 67.70Several countries, like Denmark, France, and Italy, have

taxes that at least resemble the FAT-1 by applying to the payrollof financial sector firms or of VAT-exempt firms generally. 71Shaviro, Decoding the U.S. Corporate Tax 22 (2009).



(C) T

ax Analysts 2012. A




has prompted the widespread belief that rents arerife here.72 Indeed, a recent paper estimated that inrecent years, rents have accounted for between 30and 50 percent of the wage differential between thefinancial sector and other sectors in the economy.73

How does one design a tax on financial sectorrents with the FAT-1 as the starting point? Recallthat as a VAT-like instrument, it is a tax on the ‘‘sumof [financial sector] wages and ‘profits’ defined incash flow terms.’’74 Suppose we change the FAT-1tax base (whether in the cash flow tax, ACE, or ACCversion) by making wages deductible. Then, transi-tion issues aside,75 you have a tax just on profits,defined in cash flow terms — that is, on extra-normal rates of return. While those can arise inpractice even without an expected above-normalrate of return, when the taxpayer ends up winninga risky bet, suppose that risky outcomes even out inthe long run (and that taxpayers can effectivelyaverage out their high and low rate of return years,for purposes of the tax). Then what remains is a taxon truly above-normal rates of return — that is, onrents. So the FAT-2 could seemingly be narrowedinto a tax on financial sector rents (rather than rentsplus wages) simply by modifying it to treat allwages as deductible.

However, modifying the FAT-1 by making allwages deductible would create two problems. First,owners of financial sector firms that were generat-ing extra-normal profits could remove those profitsfrom the FAT tax base simply by paying the moneyto themselves as wages. Second, high-end wages tofinancial sector employees (whether they areowners or not) may themselves represent the veryrents that we want to tax. Accordingly, the proposedFAT-2 would not permit all wages to be deducted.Instead, it would allow deductions only for somemeasure of ‘‘ordinary’’ wages, such as those paid torank-and-file workers, that are thought not to re-flect the payout of financial sector rents.

The IMF staff report concedes that identifyinglabor rents ‘‘is extremely difficult in practice, [sug-gesting] that a pragmatic approach would need tobe adopted. That might be done, for instance, bycomparing the earnings of top earners in the finan-

cial sector with those of top earners in other sec-tors.’’76 In effect, deductions for wages paid tohighly compensated employees would be limited tosome measure of what it was thought that peoplewith similar qualifications (such as education lev-els) working with similar intensity in other profes-sions typically were earning. Thus, roughlyspeaking, FAT-2 wage deductions for paying com-pensation to investment bankers might be limitedto the levels that were attributed, say, to similarlyexperienced lawyers.77

3. FAT-3. A focus on financial sector rents promptsone to ask why they have apparently been so highin recent decades. Several explanations are possible.For example, financial sector rents may reflect bar-riers to entry in the rarefied world of high-endfinance, or they may result from financial firms’ useof opacity to confuse and dupe customers about thevalue of particular financial products and the pricesfor and availability of economically comparableproducts.

A further possibility, however, is that financialsector profits often reflect fake rents rather thanactual ones, in the same sense that I could generate‘‘rents’’ from playing roulette at the casino if I couldbet someone else’s chips, pocketing all the winningswhen the ball landed on red but not having to payfor the chips when it landed on black. That re-sembles what financial firms increasingly did in theyears leading up to the 2008 financial crisis.Through means such as the placing of highly lever-aged bets on appreciating assets like real estate andstocks, the firms widely followed ‘‘‘nickels in frontof a steamroller’ strategies, under which one earnsextra-normal returns most of the time but occasion-ally experiences dramatic losses.’’78 The bets wereeffectively ‘‘heads I win, tails you lose’’ in character,given that if the downside tail risk eventuated (suchas via declining stock or real estate prices), peopleother than the financial sector bettors themselves —either the taxpayers who paid for bailouts, or every-one who was hurt by a resulting macroeconomicdownturn — would bear enormous losses.79

Countering financial firms’ socially dangerousincentive to place ‘‘heads I win, tails you lose’’ bets

72See, e.g., John Cassidy, ‘‘What Good Is Wall Street?’’ TheNew Yorker, Nov. 29, 2010, available at http://www.newyorker.com/reporting/2010/11/29/101129fa_fact_cassidy.

73Thomas Philippon and Ariell Reshef, ‘‘Wages and HumanCapital in the U.S. Financial Industry: 1909-2006,’’ NationalBureau of Economic Research working paper 14644 (2009), at 2.

74IMF staff, supra note 5, at 66.75In addition to taxing rents, the FAT-2 (as well as the FAT-1)

could result in taxing old capital that was on hand when the taxfirst began to apply. Old capital may return positive cash flowsthat either FAT reaches, without cost recovery if that is limitedto new (i.e., post-effective date) capital.

76IMF staff, supra note 5, at 68.77Iceland recently enacted a FAT-2, with a tax rate of 5.45

percent.78Douglas A. Shackelford, Daniel Shaviro, and Joel Slemrod,

‘‘Taxation and the Financial Sector,’’ 63 Nat’l Tax J. 781, 787(2010).

79Shaviro, ‘‘The 2008 Financial Crisis: Implications for In-come Tax Reform,’’ NYU Law School, NYU Center for Law,Economics, and Organization, working paper 09-35 (2009), at 13.Also, financial sector employees with incentive compensationmay have made ‘‘heads I win, tails you lose’’ bets in whichshareholders bore a significant downside. See id.



(C) T

ax Analysts 2012. A




is primarily the job of financial regulatory policy, aswell as of bank taxes (such as those explicitlyfinancing bailout or resolution funds) that dependon some measure of the systemic risk that an actoror set of actions appears to pose. However, theFAT-3 operates from the premise that those ruleswill be imperfect and that the mechanism for ex-tracting a ‘‘fair and substantial contribution’’ fromthe financial sector can contribute at this level aswell.

To that end, the IMF staff proposes modifying theFAT-2 tax base so that rather than reaching allprofits, as defined relative to the normal rate ofreturn, it would tax only very high profit rates.80

The premise would be that those extra-high profitsare indirect evidence of tail risk that the bettor is notbearing. Accordingly, under the FAT-3, the cost ofcapital deduction, as computed under an ACE orACC approach, might be 15 percent, rather thansomething that (depending on actual market inter-est rates) might well be only one-fifth to one-thirdas great.

That alone would do nothing to focus the tax’simpact on risky bets, which depends instead on thedistinct issue of refundability when a financialsector firm falls short of achieving a taxable return.To make this point more clearly, we will first returnto the FAT-1 in its cash flow version. If positive netcash flows (as defined under its rules) are taxable,then presumably — reflecting common practiceunder existing VATs — negative net cash flowswould be refundable. Thus, under a 10 percentFAT-1, if a firm with $1 million in net cash flows (ascomputed under the system’s rules) would pay taxof $100,000, then one with a net cash flow ofnegative $1 million would receive $100,000. Thatwould cause the FAT-1 to apply symmetrically totaxpayer bets that could either win or lose, reflect-ing that risk discouragement is not among itsgoals.81

Now suppose that we switch to the ACE (orACC) version of the FAT-1, under which flows ofloan principal are ignored, but a cost of capitaldeduction is allowed for corporate equity (or allfirm capital). Presumably, refundability would ap-ply with regard to cost of capital deductions no lessthan actual negative cash flows, given the lack ofany change in the tax’s scope. That presumablywould continue to hold if we switch to the FAT-2.After all, merely because one is allowing ‘‘ordinary’’

wage deductions in order to focus the tax on rentsdoes not imply departing from the FAT-1’s neutral-ity as between risky investments and those thathave a relatively fixed expected positive return.

What makes the FAT-3 different in that regard isnot increasing the tax-free rate of return to a muchhigher level, like 15 percent, but eliminating thetax’s gain-loss symmetry, or more generally itslinear rate structure. Note that the FAT-3 surelywould not provide a refund to financial firmsearning less than the tax-free rate of return. Thus,suppose that under a FAT-3 with a tax rate of 20percent, a financial firm has corporate equity of$100,000 and net cash flow (as computed for FAT-3purposes, with only ordinary wages being de-ducted) of either (a) $40,000 or (b) zero. If thetax-free rate of return was 15 percent, leading to anexempt return deduction in the amount of $15,000,then under (a) the firm would pay tax of $5,000, butunder (b) it presumably would not get any refund.82

Otherwise, the FAT-3 would turn into a subsidy forfinancial firms that earned merely normal returns.

To achieve the FAT-3’s goal of discouraging risktaking, on the view that high observed profit ratesof financial firms are associated with hidden tailrisk that implies loss externalization, all one needsis some version of that gain-loss asymmetry (ormore generally, having a nonlinear tax rate thatrises in some fashion with the ex post rate ofreturn). Making financial firms’ profits tax free untilthey reach the ‘‘extraordinary’’ (such as 15 percent)level seems unnecessary, and it means that somerents would escape the tax even though the logicunderlying the FAT-2 is in no way refuted bypositing that there is also a ‘‘fake rents’’ problem ofthe sort targeted by the FAT-3. I would thereforeadvocate doing less to narrow the FAT-2 base thanthe IMF staff suggests in its discussion of the FAT-3.All one needs to provide to discourage risky bettingthat may reflect hidden tail risk is some sort ofmechanism for applying a higher tax rate to largerates of return than to smaller or negative ones. Forexample, although without any implication that thisis necessarily the best way of doing it, one couldsimply provide that the ACE or ACC deduction forthe normal rate of return was nonrefundable.

80IMF staff, supra note 5, at 68.81Under an income tax, loss refundability can lead to tax

planning games to generate payments for the governmentbased, for example, on fake tax losses that reflect takingadvantage of the realization requirement. Under a cash-flow-based tax, however, this is not as great a problem.

82The FAT-3 might, however, allow carryovers of unusedtax-free rate of return deductions as between tax years, thusproviding the equivalent of income averaging. This would focusit on firms that had average returns above the target level, ratherthan on firms that had volatile annual cash flows or a limitedability to engage in self-help by shifting cash flows betweenadjoining years.



(C) T

ax Analysts 2012. A




E. Combining the IMF’s FAT ProposalsAs the previous section showed, there are dis-

tinct rationales for the FAT-1, the FAT-2, and theFAT-3. However, those rationales are not logicallyinconsistent. Indeed, one may believe that each hassome validity, and thus that any FAT being sup-ported for adoption should have elements of each.

Suppose we conclude that the financial sector isboth deriving true rents that can efficiently be taxedaway and placing socially undesirable ‘‘heads wewin, tails you lose’’ bets that yield high observedprofitability when the hidden tail risk fails to even-tuate. One might want both to tax all observedabove-normal returns and to have a nonlinear taxrate that either rises with profitability (perhaps witha degree of multiyear averaging) or that treatslosses (or ACE/ACC deductions that create orincrease losses) unfavorably. We could call thissystem the FAT-2-3.

Even having done that, one might still want toaddress, through the FAT-1, any remaining netsubsidies to the financial sector. That, in turn, mightsupport taxing not just financial sector profits, butalso any wages that were deductible under theFAT-2. Arguably, however, the FAT-1 rate should belower than that for the FAT-2-3, given the cascadingproblem when it applies to business inputs, as wellas the FAT-1’s imperfect fit with the subsidies that itoffsets.

The idea of imposing a multi-rate FAT-1-2-3 maysound more complicated than it is. All it wouldrequire, once the FAT-2-3 had been designed, isimposing a tax — at a rate below those generallyapplicable under the FAT-2-3 — on the amount ofordinary wages that was deductible under theFAT-2-3 component.

To illustrate, suppose that the tax rate under theFAT-2-3 was 20 percent (leaving aside the nonlinearrate features that discouraged risk taking), and thatone wanted a tax rate of only 10 percent on theFAT-1 component of financial sector ordinarywages. Suppose further that a financial firm had netcash flows, for FAT-2-3 purposes, of $20 million, butalso had paid $15 million of wages that weredeductible in making that computation. The taxunder the FAT-1-2-3 would equal $4 million on theFAT-2-3 part of the base, plus $1.5 million on thewage component, for a grand total of $5.5 million.Indeed, although this is perhaps just a semanticpoint, one could define the FAT-2-3 as merely aspecial case of the FAT-1-2-3, in which the tax ratefor the ordinary wage base happens to be set atzero.

In evaluating whether the tax rate for the ordi-nary wage base component should be positive, oneimportant question is what other means could beused to address the financial sector’s net subsidy.

One alternative approach might be to impose ahigher corporate tax rate on financial sector firms oractivities than on the rest of the corporate sector.However, that would build on the inefficienciesassociated with the existing corporate income tax. Asecond approach might be to tackle the VAT andincome tax exclusions more directly, within one orboth of those systems. There is extensive literatureon that topic.83 Even short of identifying a funda-mental fix, one might consider lesser measures,such as imputing for income tax purposes a mini-mum interest rate on checking and savings ac-counts.84

III. Rationales for Financial Sector TaxationWith that FTT and FAT background in place, let

us now consider the objectives that increased finan-cial sector taxation might serve. I will start withthose identified by the commission and move on toothers that may be relevant.

A. Raising RevenueThe first objective identified by the commission is

to raise revenue. In that respect, there is no doubtthat an FTT can be a powerful tool, at least ifavoidance is sufficiently addressed. For example,the commission suggests that its FTT proposalmight raise at least €16.4 billion if its tax rate was0.01 percent, and at least €73.3 billion if its tax ratewas 0.1 percent.85

An FAT can potentially raise comparable rev-enue, although that would require significantlyhigher statutory tax rates.86 For example, the FAT-1

83See, e.g., David F. Bradford, ‘‘Treatment of Financial Serv-ices Under Income and Consumption Taxes,’’ in Economic Effectsof Fundamental Tax Reform (1996); Harry Grubert and JamesMackie, ‘‘Must Financial Services Be Taxed Under a Consump-tion Tax?’’ 53 Nat’l Tax J. 23 (2000); Alan Schenk, ‘‘Taxation ofFinancial Services (Including Insurance) Under a U.S. Value-Added Tax,’’ 63 Tax L. Rev. 409 (2010).

84To aid the political feasibility of imputing a minimum rateto low-interest checking accounts, the tax might be collectedfrom the bank, rather than the depositor. That presumablywould have the same economic incidence as requiring inclusionby the depositor, although it might change the applicablemarginal tax rate.

85European Commission, supra note 3, at 9-10. The JointCommittee on Taxation reportedly estimated that the DeFazio-Harkin FTT proposal would raise more than $350 billion from2013 through 2021 (an average of almost $40 billion per year),with a tax rate of only 0.03 percent. See Zeke Miller, ‘‘DemocratsPush Financial Transactions Tax, Say It Would Raise $350Billion,’’ Business Insider, Nov. 7, 2011.

86The European Commission concludes that the FTT hasgreater revenue potential, but that is based on comparing it to aFAT ‘‘at an illustrative tax rate of 5 percent,’’ which it stateswould raise between €9.3 billion and €30.3 billion, ‘‘dependingon assumptions on relocation and design.’’ European Commis-sion, supra note 3, at 5. That likely refers to something like FAT-1.



(C) T

ax Analysts 2012. A




tax base might cover about 3 to 4 percent of GDP inthe European Union, and the FAT-2 tax base about 2percent.87 Since GDP in the EU exceeded $12 trillionin 2010, and presumably should continue growingeven with recessionary concerns, it is not unreason-able to posit overall tax bases in the neighborhoodof $500 billion for FAT-1 and $250 billion for FAT-2.Thus, even with significant behavioral and taxplanning responses, either instrument might be ableto raise revenues approximating those availablefrom an FTT without setting the statutory rate atanything approaching confiscatory levels.

It would be a mistake to consider the FTT’spotentially high revenue yield, relative to the statu-tory rate applied, as evidence in its favor. As hasbeen noted:

Supporters [of FTTs] often tie this feature tothe oft-quoted tax policy mantra favoring‘‘broad-based, low-rate’’ taxes over narrow-base, high-rate taxes. But the logic behind thismantra does not apply to any and all broad-based taxes, regardless of their underlyingefficiency properties. Thus, for example, eco-nomically well-informed proponents of retailsales taxes generally agree that ‘‘broadening’’the base by including business-to businesssales, rather than just those to consumers, andthus creating a gross receipts (or turnover) tax,would reduce, rather than increase, economicefficiency, by generating a cascading tax oneconomic production by multiple non-integrated firms.88

Likewise, ‘‘broadening’’ the income tax base bymaking it a tax on gross, rather than net, incomewould likely not be an improvement, even though itmight permit raising the same revenue with anominally much lower rate.

More generally, revenue-raising capacity is notwhere discussion of the FTT’s and FAT’s relativemerits should focus. After all, if one simply wantedto raise as much revenue as possible, there aremyriad ways of doing so. Consider, for example, aper-person or ‘‘head’’ tax on each resident indi-vidual, or a tax based on the number of letters ineach taxpayer’s name. Instruments like those couldsurely raise vast amounts of revenue, but theymight also rightly generate little support. The aim isnot just to raise revenue, but to do so in a manner

that is reasonably appealing from the standpoints ofefficiency and distribution. That, however, requiresexamining objectives other than just revenue rais-ing.89

The commission may have in mind a point aboutrevenue-raising potential that takes into accountpolitical considerations. Suppose that greater taxrevenues are needed, but that grave political ob-stacles impede obtaining them by any rationalmeans (or perhaps, as has recently been the case inU.S. politics, by any means whatever). Then the factthat an FTT could raise billions in revenue at anextremely low statutory rate, while also garneringplaudits as a supposed ‘‘Robin Hood tax,’’90 mightcount seriously in its favor. Even if one believedthat various alternatives (like the FAT) are better,those might reasonably be viewed as politicallyirrelevant if their chances of enactment are too low.The commission does not state any such argument,however.

B. Ensuring a Fair and Substantial ContributionThis goal, mentioned by both the IMF staff and

the commission, arguably requires more explana-tion than either has given it. I therefore offer somegeneral observations before turning to the FTT andFAT in particular.1. What is the financial sector? Just as only people,not intangible legal entities like corporations, canactually bear the corporate tax, so the financialsector cannot itself contribute substantially, muchless fairly, to the public fisc. Calls for fiscal contri-butions from the sector surely pertain to some set ofindividuals, but which ones?

The answer may initially seem clear. Calls for thefinancial sector to pay are surely directed, at least inthe main, neither at its customers (whom the com-mission expressly wants to hold harmless) nor evenat diversified shareholders who may happen tohold a smattering of the firms’ shares in theirbroader stock portfolios. The primary target pre-sumably is the people who control financial firms orwork for them as high-level, highly compensated

87I base that estimate on eyeballing the range of GDPpercentages for various EU countries. In the United Kingdom,for example, the FAT-1 would cover an estimated 6.1 percent ofGDP, and the FAT-2, 2.7 percent. For Germany, with its lessprominent financial sector, those numbers would be 3.6 percentand 1.5 percent, respectively. See IMF staff, supra note 5, at 70.

88See Shackelford et al., supra note 78, at 797.

89In support of viewing the FTT’s estimated revenue effectspositively, the Committee on Economic and Monetary Affairs ofthe European Parliament cites a recent report — StephanyGriffith-Jones and Avinash Persaud, ‘‘Financial TransactionsTaxes’’ (2012) — that says the commission’s proposed FTT couldactually have positive effects on GDP growth (on the order of0.25 percent annually), rather than the negative effects concededin the commission’s study. That estimate is based on the viewthat the FTT will have such favorable effects as reducingsystemic risk, providing funding for needed government out-lays, and permitting other distortive taxes to be reduced. Seesupra note 3, at 15-16.

90See http://robinhoodtax.org/, a pro-FTT website that la-bels the FTT the ‘‘Robin Hood Tax.’’



(C) T

ax Analysts 2012. A




employees. Members of that group have been aninescapable cultural, political, and economic pres-ence in Europe and the United States for decadesnow, alternately (or perhaps simultaneously) at-tracting public fascination, emulation, envy, andanger. Consider the film Wall Street’s notoriousGordon Gekko, as well as the ‘‘Masters of theUniverse’’ in Tom Wolfe’s novel Bonfire of the Vani-ties — both dating from 1987 — or, from 2009, thewhite-hot controversy on both sides of the AtlanticOcean regarding bonus payments to executives atrecently bailed-out financial firms.

There is potential ambiguity, however, regardingwhether our concern about fair and substantial taxcontributions from members of that group, takinginto account the bailout side of the equation, ismeant to apply at the individual or the group level.New financial sector ‘‘Masters of the Universe’’ arecontinually cycling in to replace the old, as thebusiness schools discharge more graduates and thesated minions of prior economic cycles retire totheir newly purchased country estates. Thus, if oneis concerned about beneficiaries of the 2008 bail-outs, a tax that takes effect in 2014 might come toolate to extract a fair contribution from them.Backward-looking taxes typically have at best ashort window in which they can hit their targets,unless they are explicitly retroactive. Only if we areconcerned about the possible next wave of bailouts,and about making particular actors bear the ex anteexpected cost rather than the ex post realized cost,does that problem become less critical.2. What are the grounds for favoring a fair andsubstantial contribution from the financial sectoractors? Unobjectionable though it may be to requirethat the people whom we identify with the financialsector make a fair and substantial contribution, itstill is important to specify why. At a minimum, thatinquiry may be needed to assess whether a taxinstrument can satisfy the underlying objective.

The reference to a fair contribution unmistakablysounds in equity or distributional concerns, ratherthan those of efficiency. Obviously, the fact that thepeople who are meant to be evoked by the financialsector are generally at the top of the income scalecontributes to the appeal of requiring a fair contri-bution from them. Handing them bailout funds thatare then recouped from average taxpayers may notbe an attractive distributional outcome. However,the question is not just one of progressivity versusregressivity, or else we would focus on ensuringthat rich people in general bear enough of the costof bailouts (or of government spending generally).It appears clear that something like the benefittheory of taxation, which holds that people shouldpay for the benefits that they derive from thegovernment’s activities — or perhaps the ‘‘polluter

pays’’ approach to assigning liability for harm — isplaying a primary role here.

I myself am not persuaded that benefit tax prin-ciples, as opposed to an ‘‘ability to pay’’ approach(or, better still from my standpoint, utilitarianismand related welfare economics), should govern dis-tributional analysis of tax policy. I also believe thatthe polluter-pays approach to assigning liability forharm is best rationalized in terms of efficiency,rather than on distributional grounds. The goal is togive potential polluters the right incentives regard-ing potentially harm-causing activity. But even forpeople who differ from me in that they subscribe toprinciples of distributive desert that are not entirelywelfare-based, implementing those principlesthrough a tax on the financial sector is inherentlychallenging. While I have already noted the diffi-culty of looking backward and imposing burdenson malefactors (or even mere freeloaders) who mayalready have left the scene, there is also a potentialdifficulty associated with looking forward.

The problem is transition. Suppose the financialsector is receiving huge subsidies because of theundertaxation of financial services to consumersunder the VAT and the income tax, along with theprospect of unfunded bailouts. That certainlywould be expected to increase the size of thefinancial sector. However, it would not necessarilyincrease the returns earned by specific financial-sector owners or high-ranking employees. If capitaland highly compensated labor are free to flow intothe financial sector as long as the returns there arehigher than elsewhere, presumably they will do sountil equilibrium is reached through elimination ofthe disparity in available returns.

To be sure, such an analysis may need to bemodified if we posit that people in the financialsector are earning rents. But insofar as the analysisholds, increasing the tax burden on the sector mightonly reduce the returns being earned by people inthe sector when it was first announced. The smallerfinancial sector that remained once the reduction ofsectoral subsidies caused others to exit would notnecessarily feature smaller returns for the peoplewho were still there.

Such a shrinking of the financial sector wouldclearly be desirable on efficiency grounds, if subsi-dies have made it too large. And that might be allthe more true if one views the financial sector’sextraordinary growth over the last two decades asreflecting some broader malfunctioning or evenpathology of our economic system that is not en-tirely explicable in terms of the subsidies. But oneshould clearly distinguish that line of argumentfrom demanding fair contributions on equitygrounds.



(C) T

ax Analysts 2012. A




How should one think, from a purely distribu-tional standpoint, about making the financial sectorsmaller or (if this was feasible through the FTT orFAT) less hyperprofitable at the top? A conventionaleconomic analysis might suggest that the best ap-proach is simply to focus on overall wealth distri-bution as the factor to balance (when necessary)against considerations of efficiency, rather than fo-cusing for distributional purposes on the size of anyparticular sector or the structure of economic re-turns within it. But if one is skeptical that thedecades-long rise of the financial sector entirelyreflects the standard story of markets driven by thepreferences of rational actors directing resources totheir highest and best uses, then one may come toview the efficiency failures that arguably havehelped to produce a bloated and overrich financialsector as having broader negative distributionalexternalities that should be addressed if possible.

I derive several conclusions from this discussion.Despite the equity-based rhetoric of the phrase ‘‘fairand substantial contribution,’’ the underlying con-cerns are mainly about efficiency — in particular,giving market actors appropriate incentives (suchas when they make risky bets) and addressingresource misallocations through subsidies to thefinancial sector. However, there also are plausibledistributional reasons for favoring a smaller finan-cial sector, and one in which the returns are not soheavily concentrated at the top. Those distribu-tional concerns suggest that it should not particu-larly matter whether a new tax on the financialsector targets the right set of individuals (defined interms of who received ‘‘unfair’’ benefits) as one agecohort succeeds the next in the sector’s front lines.3. Assessing the FTT and the FAT with respect tofair contribution. Enactment of either the FTT orthe FAT could result in a smaller financial sector. Inother respects, however, the taxes may differ sig-nificantly in their likely distributional effects. Start-ing with the FTT, if selling financial assets andissuing derivative financial instruments is newlytax-discouraged, the size of the financial sector maydecline insofar as it is engaged in providing relatedservices. However, as the IMF staff report pointsout, ‘‘a large part of the burden may well be passedon to the users of financial services (both businessesand individuals) in the form of reduced returns tosaving, higher costs of borrowing, and/or increasesin final commodity prices. . . . It is not obvious thatthe incidence would fall mainly on either the better-off or financial sector rents.’’91 The report furtherstates that shifting the incidence of the FTT frompeople in the financial sector to consumers is espe-

cially likely if the tax applies broadly.92 The com-mission, despite arguing that households and smallto medium-size business enterprises will ‘‘hardly beaffected’’ by the FTT,93 elsewhere concedes that a‘‘large part of the burden would fall on direct andindirect owners of traded financial instruments.’’94

The FAT has considerably more promise from thedistributional standpoint of targeting big players inthe financial sector.95 In particular, to the extent thatit focuses on rents (which is at least the intendedeffect of FAT-2), those deriving the rents appearlikely to bear the tax burden. If they are alreadyearning above-normal returns, then presumablythey are extracting what the market will bear andcannot react to the tax by demanding even higherpretax returns. However, the FAT’s application tocompensation that does not reflect the earning ofrents, including high-end compensation identifiedby FAT-2 that merely reflects ‘‘returns due to highproductivity, . . . would likely be passed on to pur-chasers of financial services.’’96

C. Reducing Undesirable Market BehaviorThe commission argues that ‘‘the FTT might be

an appropriate tool to reduce excessive risk-takingto the extent that short-term trading and highlyleveraged derivative trading creates systemicrisks.’’97 By contrast, it views the FAT as‘‘only . . . an indirect measure to tackle [excessive]risk-taking.’’98

Regarding the FAT, the question of interest is notdirection versus indirection, but the magnitude ofthe expected effect from having nonlinear rates. Aslong as a risky investment is treated asymmetri-cally, such that it bears a higher expected taxliability than it would if it were certain to achieve itsexpected return, it is being tax discouraged. How-ever, the magnitude, and indeed the basic signifi-cance, of the response is admittedly open toquestion. Further, insofar as the FAT rates are

91IMF staff, supra note 5, at 20-21.

92See id. at 20.93European Commission, supra note 35, at article 35.94European Commission, supra note 3, at 11.95It may be only fair to point out that a problem noted by the

IMF staff regarding the FTT — that a tax instrument’s ‘‘cumu-lative, cascading effects’’ when it applies to interbusiness trans-actions ‘‘can be significant and non-transparent’’ — potentiallyapplies to the FAT as well, given that it might apply in cascadingfashion to interbusiness transactions between financial andnonfinancial firms. Again, that problem would not be entirelyeliminated by charging the FAT ‘‘at lower than the generallyprevailing VAT rate in order to limit the damage’’ and thuscreating overall balance between cases of aggregate undertaxa-tion from a VAT-equivalence standpoint (in the absence ofcascading) and those of overtaxation. See supra note 5.

96Id. at 23.97European Commission, supra note 3, at 5.98Id.



(C) T

ax Analysts 2012. A




symmetric, they could conceivably induce in-creased pretax risk taking by financial firms.

Regarding the FTT, it is far from clear whethersystemic risk would decline. The commission em-phasizes the tax discouragement of trading strate-gies that may increase volatility. To similar effect,others note that an FTT ‘‘may reduce activity by‘noise traders,’ who trade on spurious informationsuch as past price movements and are thought todestabilize markets. . . . However, it may also sup-press activity by informed traders and arbitrageurs,whose trading tends to push prices towards theirfundamental values.’’99

More generally, the problem is that with thin,incomplete, or otherwise imperfect markets, bothpositive and negative externalities from tradingactivity may be rife. Thus, theoretical models sug-gest that ‘‘volatility may either rise or fall uponintroduction of an STT, depending on the marketmicrostructure,’’ and that the tax is unable ‘‘todiscriminate between discouraging stabilizing anddestabilizing trading activity.’’100 If one knew thatthe STT would disproportionately drive from themarketplace those engaged in ‘‘bad,’’ rather than‘‘good,’’ trading, there might be grounds for opti-mism. But if we could tell apart the two types oftrading, then a better approach still might be just totax (or regulate) the bad part of the marketplace.

Absent the ability to discriminate, it is hard to besure whether the tax has a greater discouragingeffect on the good trading or the bad. Presumablyfor those reasons, empirical studies generally fail tosupport the view that an STT can be expected toreduce either short-term price volatility or the oc-currence of price bubbles and crashes.101 And justanecdotally, it is clear that bubbles and crashes canplague asset markets in which trading is costly andthus relatively sporadic. A case in point is the U.S.real estate market in the first decade of the 21stcentury. In sum, there is little clear support for thecommission’s hope that the FTT would help stabi-lize markets.

D. Achieving CoordinationA final reason advanced in support of the com-

mission’s FTT proposal is that it might help inachieving coordination between different memberstates’ internal taxes. Even if the enactment of anFTT is otherwise a bad idea, that consideration mayhave some merit insofar as, in the alternative,member states might enact their own FTT variants.Uniformity would at least limit tax competition and

national variation between applicable FTTs withinthe European Union. FTTs and similar instrumentsare not widespread within the EU, although theFrench government recently indicated that it plansto enact one unilaterally, and Germany has said itmay follow suit. The United Kingdom, which hasEurope’s largest financial sector, appears unlikely todo so.

Even if particular countries at least briefly bitethe bullet of acting unilaterally, the adoption andretention of significant national-level FTTs maytend to be discouraged by concern about tax com-petition. Again, the recent Swedish experience sug-gests that initial exuberance may rapidly fade ifrestructuring to avoid the tax is rife. On balance,then, it is not clear how strongly this factor weighsin favor of the commission’s proposal, at least if oneis otherwise agnostic about the desirability of hav-ing any FTT at all.

E. Other Relevant ConsiderationsSeveral other considerations may be relevant in

evaluating whether to enact an FTT, a FAT, orneither. They include:

• VAT and income tax undertaxation of the financialsector — Assuming that this problem cannot beaddressed more directly, such as through theVAT and income tax themselves, the FAT hasan important advantage over the FTT in re-sponding to it. Whereas the FTT only reachesspecified transactions like securities tradingand (at a much lower tax rate) derivativestransactions under the commission’s proposal,the FAT applies to financial sector activitiesgenerally.

• Income tax bias in favor of debt relative to equity —Existing corporate income taxes generally fa-vor debt over equity, in particular by providingthat corporate taxpayers can deduct interestexpense but not dividends. The FTT mightexacerbate that bias. Although securities thatare traded in secondary markets would gener-ally be subject to the FTT whether classified forincome tax purposes as debt or as equity, thelatter instruments tend to be traded more.

• Tax competition from outside the taxing jurisdic-tion — As noted above, even an EU-wide FTTwould discourage the use of European finan-cial sector companies by non-EU persons toconduct taxable securities trades and deriva-tives transactions. By contrast, the FAT-2, atleast to the extent that the tax base succeededin identifying rents, should not have that effect,given that higher-ups in the resident financialsector firm presumably would bear it. On theother hand, the FAT-2 could discourage theorganizers of potentially taxable firms fromchoosing European residence.

99Thornton Matheson, ‘‘Taxing Financial Transactions: Issuesand Evidence,’’ IMF working paper (2011), at 20.

100Id.101Id. at 20-21.



(C) T

ax Analysts 2012. A




• Defining taxpayers — While the FTT may createserious problems in identifying taxable trans-actions, it is likely to be less problematic thanthe FAT in identifying taxpayers. As notedearlier, the commission’s proposal identifiespotential taxpayers very broadly, reflectingthat the stakes are lowered by the fact that onlythe taxable transactions engaged in by thoseentities would lead to FTT liability. Under theFAT, a financial sector firm (or branch within afirm) presumably is taxable on all its relevantactivities. Accordingly, the classification stakesmay be higher under the FAT than the FTT, andthe cost of error more significant. In that re-gard, particular difficulty may arise because‘‘financial [and other] institutions with differ-ent designations often perform overlappingfunctions and sell overlapping products.’’102

Moreover, one needs to address the existenceof ‘‘predominantly non-financial firms withfinancial units.’’103 On the other hand, coun-tries that exempt financial firms from VATshave apparently not found that prohibitivelydifficult.

• Progressivity in the presence of political constraints— I argued earlier that while the FTT shouldnot be preferred to the FAT on the grounds thatit can offer a higher revenue yield relative tothe statutory rate being applied, political con-straints on rational revenue raising mightmake that an advantage after all. The samepoint could apply to the goals of increasing theprogressivity of the overall tax system. TheFTT is far from being an optimal mechanismfor increasing progressivity, if that is the onlyreason for it, given the unmitigated distortionsthat it causes if one is not persuaded by theefficiency arguments for it. Moreover, it wouldlikely be less progressive in incidence than theFAT-2, since it does not target high-end finan-cial sector rents. However, if all other progres-sive tax changes are assumed to be politicallyunavailable, the FTT might be considered bet-ter than nothing, at least if one believes that itis borne by investors, who tend to be relativelyaffluent.

IV. An Alternative FTT RationaleEfficiency analysis in tax policy often assumes

that taxpayers will have suitable incentives whenguided by pretax profitability unless there are clearand tangible externalities, like those resulting fromfinancial firms’ ‘‘heads I win, tails you lose’’ betting

opportunities, or from industrial pollution. At leastone potentially important type of externality isoften ignored. Suppose I find two economic oppor-tunities, but can pursue only one of them. The firstwould offer an economic reward worth $10 million,but if I don’t secure it, someone else will. Thesecond would offer an economic reward worth $9million, but if I don’t secure it, it will go unclaimed.If the costs I would incur to get either (as well as theprobability of success) are the same, I will presum-ably seek the first reward. But that would causeaggregate social returns to be $9 million lower thanif I had chosen to pursue the second reward. Thatthe first reward but not the second would, in effect,come out of someone else’s pocket is an externalitythat I have no reason to care about.

The reason for commonly ignoring considera-tions of that kind is that outside the tidy boundariesof a hypothetical, the considerations may be pro-hibitively difficult to identify and measure. Exter-nalities are rife in the world around us, but wecannot take account of them all. Moreover, thegeneral economic success over many decades offree market economies relative to those that aremore centrally managed and thus limit competitionbetween businesses that are seeking to fill the sameniche, may be viewed as supporting the intuitionthat ignoring externalities of that kind leads toreasonably good results overall.

Nonetheless, there are times when it may bedesirable to take account of the externality thatresults when people compete for the same prize.One recent example is the literature arguing thathigh-return labor markets increasingly are winner-take-all ‘‘tournaments’’ in which many compete forrewards that only a few can win, potentially withimportant tax policy implications, such as strength-ening the case for progressive redistribution orhighly graduated marginal income tax rates.104

Likewise, in the literature on the economics ofinformation, an influential paper by Jack Hirshleifernotes that the socially optimal level of intellectualproperty protection is strongly affected, and poten-tially greatly reduced, by the occurrence of patent

102See Shackelford et al., supra note 78, at 794.103Id.

104See, e.g., Robert H. Frank and Philip J. Cook, The Winner-Take-All Society: How More and More Americans Compete for EverFewer and Bigger Prizes, Encouraging Economic Waste, IncomeInequality, and an Impoverished Cultural Life (1995); Martin J.McMahon and Alice G. Abreu, ‘‘Winner-Take-All Markets: Eas-ing the Case for Progressive Taxation,’’ 4 Fla. Tax Rev. 1 (1998);and Shaviro, ‘‘1986-Style Tax Reform: A Good Idea Whose TimeHas Passed,’’ Tax Notes, May 23, 2011, p. 817, Doc 2011-8373, or2011 TNT 100-9.



(C) T

ax Analysts 2012. A




races.105 In a patent race, multiple inventors arecompeting to be the first person to perfect andpublish a type of invention that many people haverealized might be feasible. If I get there first, just oneday ahead of the runner-up, I will receive the entirepatent reward, even though the social benefit frommy efforts is limited to that from people getting theinformation a day sooner.106

Now consider financial market profits from se-curities trading. Each trade, insofar as it reflectsdiffering predictions regarding future value, endsup having a winner and a loser. In the Keynesbeauty contest scenario, the game has no aggregatesocial value (other than its being enjoyed as aconsumption activity) and merely leads to zero-sumtransfers from the losers to the winners. But even ifpublic securities trading also helps to ensure theproper allocation of resources through the ongoingincorporation of new information into securitiesprices, there is a Hirshleifer element to the profitfrom being the first to discover and trade on newinformation that affects value. That is, the privategain exceeds the social gain under the same analysisas that applying to patent races. It is clear, more-over, that huge resources are devoted to seekingtrading profits.107

Suppose we therefore conclude that it would besocially desirable to tax-discourage and therebyreduce the effort that people invest in the pursuit oftrading gains. If the thing we want to discouragecannot be observed directly, we might insteadchoose to tax proxies that are correlated with it inpractice. If trading securities more rather than less(in terms of the value traded) is indeed an empiri-cally robust proxy for the socially excessive pursuitof trading gains, then the FTT could yield socialbenefit at that margin even if the act of trading itselfhas no negative externalities in the aggregate.

In my view, that line of argument may establish aplausible motivation for enacting an FTT. So mayKeynes’s argument, discussed earlier, if one be-lieves that more frequent trading (and thus shorteraverage holding periods) worsens resource alloca-tion or promotes damaging short-termism. Asnoted above, however, those arguments are hard toassess definitively. And even if one does accept

them, the next step is weighing the expected effi-ciency gain against such efficiency costs of the FTTas its discouraging trades that the parties wouldvalue, imposing cascading taxes within the businesssector, and inducing wasteful tax avoidance behav-ior.108

One also could perhaps devise alternative meansof pursuing the Hirshleifer and Keynes objectives.One possibility might be to have the income tax rateon gains from selling publicly traded financialassets start relatively high and decline graduallywith the length of the holding period. U.S. federalincome tax law does that in one respect: by impos-ing a long-term capital gains rate (currently 20percent) on capital assets that have been held for atleast a year, but imposing a 35 percent top rate onassets held for less than a year. The capital gains taxthen entirely disappears if one holds the stock (orany other appreciated asset) until death. That ap-proach, however, uses only two specific time win-dows, each involving a sharp discontinuity. Onecould imagine replacing it with an approach — justfor publicly traded stock — under which the appli-cable tax rate declines more smoothly and gradu-ally.109

No matter how the FTT analysis comes out,however, it has little evident relationship to the riskof future financial crises, to any of the other maingrounds advanced by the commission in support ofan FTT, and to the question whether a FAT has beenenacted. Thus, the best case for enacting an FTT liesoutside the ‘‘FAT or FTT’’ framework that hasdominated much recent discussion, particularly inEurope.

V. ConclusionThere are several good grounds for raising taxes

on firms in the financial sector:• Various financial services to consumers are

treated preferentially by income taxes andVATs. The income tax preference may begreater overall, although it is less commonlyemphasized.

• The financial sector as a whole is implicitlysubsidized by the prospect of bailout if finan-cial firms’ failure endangers national or globalmacroeconomic performance. Bank regulation,along with bank taxes other than the FTT and

105Jack Hirshleifer, ‘‘The Private and Social Value of Infor-mation and the Reward to Inventive Activity,’’ 61 Am. Econ. Rev.561 (1971).

106See generally Joseph E. Stiglitz, ‘‘Using Tax Policy to CurbSpeculative Short-Term Trading,’’ 3 J. Fin. Serv. Res. 101 (1989).

107See, e.g., Lawrence H. Summers and Victoria P. Summers,‘‘When Financial Markets Work Too Well: A Cautious Case for aSecurities Transactions Tax,’’ 3 J. Fin. Serv. Res. 261 (1989); Stout,‘‘Are Stock Markets Costly Casinos? Disagreement, MarketFailure, and Securities Regulation,’’ 81 Va. L. Rev. 611 (1995).

108For a view that is similarly uncertain regarding the bottomline, see Lee A. Sheppard and Martin A. Sullivan, ‘‘Should theU.S. and U.K. Enact a Securities Transfer Tax?’’ Tax Notes, Sept.14, 2009, p. 1055, Doc 2009-20010, or 2009 TNT 175-3. For a morefavorable view, see Dean Baker, The Benefits of a FinancialTransactions Tax (2008).

109Among the difficulties that approach would involve ishow to net capital losses against different-period capital gains.



(C) T

ax Analysts 2012. A




FAT, may fail to entirely solve this problem,even if it ameliorates it.

• The key actors in financial firms may oftenhave opportunities to derive rents that can beefficiently and progressively taxed.

• Those same actors may often have the oppor-tunity (despite financial sector regulation) tobenefit from choosing investments that offerextra-normal returns most of the time whileoccasionally experiencing dramatic losses thatwill end up being someone else’s problem.

Insofar as those concerns could be addressed byenacting either an FTT or a FAT, I have argued thatthe case for a FAT is much stronger. The variant Ipropose, which for want of a better name I call theFAT-1-2-3, would tax financial firms’ ordinary wagebase at a low rate and their profits at a higher ratethat is in some sense nonlinear so as to discouragerisk taking.

A final point of interest concerns the almostentirely separate case for an FTT that is rationalizedas a mechanism for discouraging the socially exces-sive pursuit of trading profits. Here the case forimposing a tax on securities values traded wouldrest on the view that that is a decent proxy for theactual underlying concern, rather than on the claimthat the trading itself imposes net negative exter-nalities. However, whether an FTT that was thusrationalized would be desirable on balance dependson how one assesses the trade-off between itspotential benefit and its undoubted efficiency costs,as well as by the availability and merits of alterna-tive tax instruments.

Finally, the desirability of enacting an FTT maybe affected by broader political economy constraintson revenue raising and on the pursuit of greater taxprogressivity by alternative means. Even if the FTTis clearly inferior to other tax changes that couldsimilarly raise revenue and reduce wealth inequal-ity, it might be not only better than doing nothing,but also the best tax instrument realistically avail-able at the time.



(C) T

ax Analysts 2012. A




no job name · daniel n. shaviro is the wayne perry professor of taxation at new york university...

Documents