On the Feasibility of a on Foreign Exchange Transactions

REPORT

To the Federal Ministry for Economic Cooperation and Development, Bonn

by

Goethe-Universität, [email protected]

Frankfurt am Main February 2002 On the Feasibility of a Tax on Foreign Exchange Transactions

 SUMMARY II

 CHAPTER 1: MOTIVATION 1

 CHAPTER 2: INSTRUMENTATION 14

 CHAPTER 3: ORGANIZATION 29

 CHAPTER 4: IMPLEMENTATION 47

 CHAPTER 5: REACTION 63

 APPENDICES AND REFERENCES 70

 INTERVIEW PARTNERS 71

 FORMULAE 74

 AND CRISES 76

 SPREADS AND 80

 ABBREVIATIONS 83

 REFERENCES 84 On the feasibility of a tax on foreign exchange transactions Summary

A tax on foreign exchange transactions is expected to realize different objectives: 1. The stabilization of exchange rates, 2. the exploitation of new revenue sources 3. the redistribution of resources, in particular be- tween financial and producing sectors, and between nations (in particular between the North and the South) , and 4. aspects directed toward transforming the world economic order, in particular with the aim of controlling the process of globalization. Systemic change is not pursued in this report. On the contrary: The proposals of a tax on foreign exchange transactions are contingent on avoiding a negative sys- temic impact. Aspects relating to distributional issues are discussed in an ancillary manner only. They fall ulti- mately in the realm of politics. The present study focuses mainly on the feasibility of a tax on foreign exchange transactions with emphasis on the objectives of exchange-rate stabilization and fiscal revenue. The scope for political decisions on a tax on foreign exchange transactions is constrained by the fact that the tax has to be introduced and accounted for by existing political decision bodies, in particular national and su- pranational parliaments. The tax will therefore have to be unilateral and partial, not multilateral and universal. Moreover, will fall to the jurisdiction that is accountable for legislating the tax, not to international organizations. Multilateral international bodies may how- ever be given some or all of the revenue in a second step—via budgetary grants. The analysis has brought to the fore the following re- sults: Political restrictions Switzerland. 1. As to „politically feasible“ 7.The exchange-rate instruments to curb foreign normalization should be exchange speculation, the applied unilaterally, but only by discussion focuses on non- transition, developing and remunerated reserve requirements emerging economies, and by those of foreign transactions industrialized countries that aim at and/or deposits. There is also a pegging their currency to one of the proposal by Zee to use an larger currency areas (or a basket asymmetrical cross-border capital of ). tax on of foreign capital 8. The combination of two (albeit not on ). Both in the form of a Tobin-cum-Circuit- instruments are interesting as parts Breaker Tax possesses significant of an arsenal devoted to coping allocative and distributive with currency speculation. advantages over the prevailing 2. In particular non-remunerated orthodox policies to stabilize reserve requirements are likely to exchange rates. form part of the future global Market structure and economic financial architecture. restrictions 3.The proposal of an 9. The analysis of the structure of asymmetrical tax on cross-border foreign exchange markets reveals capital flows appears to be laden further elements that restrict the with severe and complex scope for a . Such administrative problems however. restrictions are mainly economical This renders its realization highly and relate to issues such as the unlikely. level of taxation (, tax base), 4.Both instruments are and to the distribution of tax essentially distinct from a tax on revenue. foreign exchange transactions as Tax rate examined in this report. 10. If the purpose of the tax is to Concepts be borne by traders/banks, the bid- 5. To reconcile the objectives of ask spread will limit the margin for exchange-rate stabilization and re- the rate. With the bid/ask being venue generation, a unilateral and roughly one basis point for more „politically feasible“ Tobin tax liquid markets, the tax rate should (PFTT) on foreign exchange not exceed half or one basis point. transactions with a small rate, 11. If one assumes the tax to be combined with a high-rate shifted, it entails that it be borne surcharge on externalities resulting substantially by non-banks. from speculation, the „Exchange However non-banks represent only Rate Normalization Duty“ appears 13.3 percent of total turnover. It to be most promising. Both taxes implies a leverage effect that could would be technically intertwined. multiply the tax burden on non- 6. The Tobin tax proper can be financial agents by a factor of up to implemented unilaterally by 7.5. member states of the OECD, 12. If the bid-ask is accepted to individually or (preferably) as a limit the scope of taxation, the uni- group. It could also be form tax rate of a PFTT should then implemented by the European be derived from the conditions Union in cooperation with prevailing in the more liquid markets. This confers a relative Operations advantage to less liquid markets 17. There is a number of positive where the spread is higher. But this factors that prevent foreign should be conceded because it exchange transactions from favors the lesser industrialized migrating off shore into tax havens. countries. The tax rate should then These are the concentration of fall in the range of one half to one business onto one main financial basis point. The financial sector center as a „natural monopoly“ then likely bears part of the tax. within the time zone, complexities 13. At this level of taxation, there relating to foreign exchange is no need to relieve exporters, transactions and their derivatives, importers or direct investors from as well as significant network tax. externalities. However this is Tax base contingent on a that respects the specific features of 14. The tax base could consist of liquidity trading and recognizes the all spot transactions, and outright effective margins of trading as a forwards and swaps up to one constraint on tax rates. month. Options and other financial derivatives will not attract the tax Implementation directly, but they are taxed 18. The problems relating to the indirectly through the spot and implementation of a PFTT are forward transactions they trigger. rather complex. First of all, there Partitioning tax revenue must be clear principles to guide tax policy. They concern the 15. The PFTT is inappropriate as a definition of the tax base and the national revenue raising taxpayer. A reasonable approach instrument. This results from the will be to base the tax on a “market fact that foreign exchange principle”, whereby all traders transactions are carried out by accredited in European financial time zone. Moreover there is a centers, centrally operating clear-cut trend toward centralizing automated broker systems and these transactions at one center clearing/settlement systems will be within the zone. This implies the taxable. The same is true for tax to be implemented for the EU as foreign exchange trading by non- a whole, including of course its banks (such as Volkswagen or main trading place, London, but Daimler-Chrysler). also Switzerland as a potential rival within the zone, but outside the EU. 19. Generally speaking there are two options to define tax liability: at 16. Tax revenue is collected by the trading desks, or at the time of central banks, but it falls to the re- settlement. Both procedures gion as a whole, not to national appear to be feasible, although authorities. It could be they both have advantages and redistributed to national disadvantages. governments via formula-based transfers, but it should preferably 20. If the tax is levied at the go into a European Fund for trading desks, there may be a Economic Development to be cumbersome reporting necessity, managed at the level of the EU. It is which is inappropriate in view of of course conceivable the revenue the electronic platforms that to be used for other „global public characterize the market. This could goods“ as well. be avoided by an automated, centralized tax collection at the two options to implement such a stage of clearing or settlement. The tax, one starting form the trading latter still poses problems as the desk, the other operating at the available information is not passed level of settlement. Both appear to on to the settlement stage now. be promising. Moreover, only spot transactions 25. The knotty problems are not at would be recognizable when all technical. They are related to settling claims. political will, to international 21.However the further cooperation, and to legal concentration and automatization enforcement. of foreign exchange trading, in Reactions particular the introduction of a 26. The introduction of a PFTT will continuous link gross settlement entail a number of very different system will significantly improve reactions among actors on foreign the conditions for levying the tax at exchange markets. It is to be settlement stage. This is why I expected generally that trading prefer this latter approach. It volumes will decline, and that the requires a different principle of bid-ask spread will widen. This taxation though, which I have raises the question of who will bear dubbed the “access principle”. It the burden of the tax. This question defined tax liability from the access is largely open and controversial. to official national gross settlement systems, with contractual 27. Most affected by the tax are „backward chaining“ by which those engaging in covered interest operations prior to settlement are rate arbitraging. However, given the included. extremely thin margins of this business, it is likely that the higher 22.It seems obvious that this risk will be largely shifted forward procedure will also require some though higher premia. It implies reporting at the desk for those the tax to be borne also by the institutions that do not participate production sector and by house- in official and centralized clearing holds (both private and public). and settlement. It could be waived however if those institutions would 28. The proper speculators in the join such systems and/or convey market, for instance funds, relevant information into the will be hurt comparably less centralized clearing and settlement because they operate with machinery. significantly higher margins than liquidity traders. The tax adds only 23. The ominous literature on tax a relatively smaller charge onto avoidance strategies related to the their business. However they will introduction of a PFTT is by far have to fear the anti-speculative overstating the risks. The high surcharge, which will not play a degree of concentration of foreign significant role in the trading exchange operations tends to work decisions of all other groups. against it. Developments in foreign exchange markets will further 29.Among the institutional enhance compliance with the tax. investors, insurance companies will bear a comparably higher tax 24.I consider a PFTT to be burden because of their longer- feasible in technical term investment strategies and the terms—provided that the nature of consequent lower rotation of liquidity trading is respected and turnover relative to assets. This is taken into account. There are even different for the group of countries. investment funds, for instance, Revenue where trading is comparably more 30. Revenue estimates of a PFTT of intensive. Within this latter group of one basis point result in an amount investors, it is most likely that of 17 to 20 bill. for the EU those institutions can more easily plus Switzerland. shun the tax that specialize in financial assets of industrialized Chapter 1: Motivation What is the purpose of a tax on foreign exchange transactions?

The proposal of a tax on foreign exchange transactions goes back to Tobin (1972).1 It has repeatedly being discussed, but realization is still pending.2 The discussion of the Tobin tax is however motivated by extremely different objectives. The more important ones are the following: » The reduction of exchange-rate volatility through „throwing sand into the gears“ (Tobin) of world financial markets. This includes two sub-elements 3: ü A greater reorientation onto economic fundamentals, and ü the freedom of central banks from being compelled to intervene in foreign exchange markets in order to stabilize their currency. » Fiscal motives. These may be divided into two: ü The mere exploitation of new revenue sources, and ü an indirect approach to taxing globalized capital income in view of difficulties to tax them under a national . » The redistribution of resources, in particular among the North and the South, as well as the symbolic association of the tax with principles of social justice. » Expectations of controlling or altering the process of globalization through constraints imposed onto the international financial system.

1 Tobin presented his proposal initially in 1972 in his Janeway Lecture at the University of Princeton; it was published in 1974 as The New Economics One Decade Older, pp. 88-92. Tobin has repeated his proposal several times such as in Tobin (1978, 1984, 1991, 1996) and Eichengreen, Tobin, and Wyplosz (1995). In recent discussions with the author, Tobin has retained his proposal although he distances himself from groups that usurp his concept as a mean to combat globalization. 2 However the French Parliament has enacted a Tobin tax in 2001 (Loi de finances pour 2002 - n° 3262, Art. 986. I), though it hinges on all other member states of the adopting such a tax. 3 Tobin (1996, pp. xii-xiii)

One cannot expect that only one pol- other. This ni creases the tax burden icy instrument—such as the Tobin on frequent short-term currency trad- tax—could realize all these objectives ing compared to longer-term invest- at the same time. First it is to be dis- ments in foreign currencies. According cussed, which objectives are at all re- to Tobin this would reduce an erratic alistic, and in which form a tax on for- volatility of the because eign exchange transactions could traders would again be forced to focus achieve their realization. on fundamental data instead of being seduced by transient market senti-

ments (Tobin (1991), p. 16). The idea » Stabilization of exchange rates. is to limit short-term capital move- The reduction of exchange rate volatil- ments without hindering international ity was Tobin’s original intention of the in goods and services, and di- tax. He argued before the background rect investments. of a collapsing fixed-exchange-rate regime (Bretton Woods) that had re- peatedly led to against Table 1: the US dollar. At the time the dollar Foreign interest (in percent) required was the almost exclusive world cur- to match a domestic investment with rency. There were only three motives a 5-percent return (for different tax rates) to exchange dollars as the interna- Required tional mean of payment against na- Holding foreign tional currencies, two of them „honor- period able“ (for financing exports/imports of in percent goods and services, and of direct in- Tax rate 0.5 pc 0.1 pc vestment) and one questionable: for speculation. One day 541.3 50.7 The idea to reduce speculation in fi- Three days 92.6 18.5 nancial markets through taxation goes One week back to Keynes (1936). Keynes com- 37.0 10.7 pared speculative activities to casino One month 12.1 6.4 operations and argued that "...casinos should, in the public interest, be inac- Three months 7.7 5.5 cessible and expensive" (p. 159). One year 6.1 5.2 Tobin transposes this idea onto for- eign exchange markets where he Five years 5.6 5.1 wants to throw "sand in the wheels" in the form of financial transactions taxes. More specifically Tobin sug- The main advantage of the tax is in- gests an international and multilateral deed that it can target short-term cur- tax on all spot transactions from one rency transactions very effectively. currency into another, which is propor- The differential impact of the tax on tional to the size of the transaction short- and long-term transactions can (Tobin 1978, p. 490). Initially he be expressed arithmetically and is thought that a uniform tax rate of one shown in Appendix 2. The formula al- percent would be appropriate, but lows calculating the interest rate on more recently he changed his pro- foreign investments required to match posal by reducing the rate to about a domestic investment in spite of the 0.25 to 0.1 percent (Tobin 1996, p. tax. For instance, if the interest rate is xvii). 5 percent for a domestic investment and the tax rate on currency transac- The tax would be due every time a tions is 0.5 percent (0.1 percent), the currency is exchanged against an- foreign interest rates required for ef-

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 2 fective between two cur- tem. Today the predominant part rency areas for different holding peri- of currency transactions consists ods is exhibited in Table 1. of liquidity trading among financial institutions. This serves primarily It is clearly shown that the required for hedging against exchange-rate foreign interest rate must be the risks, which is essentially stabiliz- higher, the shorter the holding period ing. It is extremely rare that liquid- of the foreign investment becomes. ity is also used for speculative This is the essence of Tobin’s argu- 6 purposes. A tax on currency ment: the tax would encumber short- transactions would primarily hurt term speculative transactions more liquidity trading and thus jeopard- heavily than longer-term foreign in- ize the functioning of the world fi- vestment that would essentially be de- nancial system. termined by fundamental data. » A comparison of net interest rates Ideally financial transactions associ- between currency areas—as ated with direct foreign investments shown in Table 1—does not fully and the trade of goods and services 4 describe the impact of taxation on would be exonerated from the tax. speculation. Speculators act with a However this requires substantial ad- very short-term perspective. For ministrative “red tape”, which would instance, if an investor expects a certainly favor evasive practices and depreciation of only 5 percent in a avoidance strategies, especially in de- discernible short period (say, a veloping countries, whereby specula- week or a month), he or she will tive financial transactions would be not refrain from speculating if a tax declared to represent the financial of 0.1 or 0.5 percent is levied. As corollary of real economic transac- 5 the foreign-exchange crises of the tions. This is why Tobin accepts the 1990s in , in South- tax to apply to all financial transac- East Asia, in transition countries tions without any discretion as a sec- and within the European Monetary ond best solution. System have demonstrated, the Tobin’s argumentation has a number repercussions of speculation can of weaknesses, which I have dis- cause exchange-rate changes that cussed more extensively elsewhere go well beyond the 5 percent (Spahn 1995, in particular Chapter 5 mark.7 „The four dilemmas of the Tobin tax“). It is for these and other considera- The main problems are the following: tions8 that I have come to the conclu- » International financial markets have remarkably changed since 6 Even the term „speculation“ is vague and of- the end of the Bretton-Woods sys- ten abused ideologically. I use the term in a technical, intentionally value-free, connotation. Some reflections on “speculation” are found in 4 In fact the French parliament took provisions Appendix 3. 7 to exonerate such transactions. However it ig- Some examples of currency crises with a nores or underestimates (in spite of earlier dramatic impact on the exchange rate can be negative experiences with capital controls) the found in Appendix 3. administrative intricacies of such exemptions 8 and the potential for evasion strategies. These „other considerations“ include doubts 5 that a reduction of liquidity would contribute to Especially in developing and emerging stabilizing exchange rates. Theory and prac- economies the distinction between financial tice commend that less liquid markets are ex- transactions of dissimilar kinds is extremely posed to higher and more abrupt price volatility difficult to make. Insistence on such a distinc- than more liquid markets. I shall come back to tion would only encourage corruption, as offi- this point later in this Chapter when discussing cial documentation is often cheap to obtain. systemic aspects.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 3 sion that a tax on foreign exchange Table 2: transactions, as originally conceived Development of transaction volumes by Tobin, is an inappropriate instru- on international foreign exchange markets ment for mitigating exchange-rate volatility. The higher the tax rate, the Daily average in In- more aggravating the repercussions Year bill. US Dollars crease on the world financial system will be- (April) at constant $-rates in (of April 2001) come; the smaller the rate, the less percent suitable it will be to deter speculation. 1989 570 –

However this does not imply the con- 1992 750 31.6 clusion that the objective of stabilizing exchange rates cannot be achieved 1995 990 32.0 through a tax on foreign exchange 1998 1,400 41.4 transactions. Such a tax must only be designed in a different way as pro- 2001 1,210 –13.6 posed by Tobin. Theoretical consid- erations by Tornell (1988, 1990) have encouraged me to ponder on ways to Purely arithmetically, the multiplication use taxation as a mean to achieve ex- of such transaction volumes with even change rate stabilization. It led me to smallest tax rates produces significant a “two-tier” tax, whereby a proper amounts of revenue. For instance, Tobin tax (albeit with a very small tax with roughly 250 business days per rate) is combined with an exchange- year, a tax base of $300 trillion could rate normalization duty (ERND). The be calculated for 2001, and a tax rate former is apposite primarily as an in- of only 0.1 percent would then pro- come-generating device while the lat- duce a yearly revenue of about $300 ter takes up the function of stabilizing billion.11. exchange rates. The proposal is fur- I consider such calculations dubious ther explained in Chapter 2. for several reasons. » First they do not account for the » The exploitation of new revenue structure of the market and its sources. Fiscal aspects are not in the possible alterations that could re- forefront of Tobin’s proposal.9 How- sult from introducing the tax. ever he also realizes that a tax on for- » Furthermore, total transaction vol- eign exchange transactions could umes contain distinctive financial raise substantial revenue. This results instruments: spot transactions and from the enormous amounts of foreign outright forwards, foreign ex- exchange transactions that have change swaps, options and fu- evolved over the last decades.

According to a tri-annual survey on 11 the developments of foreign exchange The French parliament, in its legislation of markets published by the Bank for In- last year, proceeded in a similar fashion. It ternational Settlement (BIS) in Basel, based its estimate on the daily transaction vol- ume of its financial center (56.5 bill. eu- the following transaction volumes are ros), accounting for a yearly deduction of 10 obtained as depicted in Table 2 : 1,017.75 bill. (or 4.1 bill. per day), which corresponds to transactions of the trade bal- ance and direct foreign investment that are ex- 9 onerated by law, and thus calculated a tax „Raising revenue has never been my main base of about 50 bill. euros per day. At the motivation.“ (Tobin 1996, p. xvi). maximum rate provided by the law of 0.1 per- 10 See BIS, Quarterly Review, December cent, this was expected to generate revenue of 2001, p. 39. 50 million euros per day (or 12.5 bill. per year).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 4 tures. It is unclear in which way markets and the world economy.13 such different instruments would 2. Legal avoidance by using non- bear on such revenue estimates. taxable substitutes and fi- » Moreover the number and struc- nancial innovations. This includes ture of participants in foreign ex- a possible retreat into a single change markets has changed world currency for handling liquid- dramatically in recent years. This ity trading (the US dollar), with cor- has had a lasting impact on the responding differential compensa- tax base.12 These developments tions via financial derivatives. will be discussed further in Chap- 3. An increase in the effective- ter 3. ness of currency trading. This is » In addition, the development of achieved through net clearing sys- foreign exchange markets is char- tems, the institutional consolida- acterized by dramatic technologi- tion of the financial sector, the cal changes, for instance the in- outsourcing of financial transac- troduction and further develop- tions into non-financial institutions ment of automated brokerage sys- etc. tems, net clearing and settlement 4. The legal and illegal relocation (e.g., „front-loaded“ and „continu- of financial operations into tax ha- ous settlement“). Such develop- vens. ments had a significant impact on the structure of markets and they Further considerations as to possible are likely to prevail in the future. evasive strategies under a tax on for- Their repercussion on the base of eign exchange transactions can be a tax on foreign exchange transac- found in Chapter 4. tions deserves further scrutiny Apart from the immediate goal of (see also Chapter 3). achieving fiscal revenue with a tax on » Finally—and not the least—a tax foreign exchange transactions, the tax on foreign exchange transactions is occasionally also considered to (as any other tax) will entail eva- work as a presumptive tax on capital sive strategies, which would nec- income. The argument is that under essarily curtail the tax base. These globalization it will become more and reactions are essentially of four more difficult to capture capital income types: from international investments with a national income tax. Such income fre- 1. The simple refraining from ex- quently escapes the income tax by ecuting a taxable transaction. This dislocating into tax havens whose entails an often-overlooked dead- governments are known to be unco- weight loss, which can also find its operative on source taxation. To the expression in higher economic extent that the Tobin tax would reduce risks. Such effects may possibly the net return on foreign capital in- correspond to the objectives of the vestments, a sort of presumptive in- critics of globalization, but it could come tax would become feasible, have fatal consequences for the whereby the transactions tax would stability of international financial generate some form of compensation, albeit crude.

12 For instance, the reduction in the volume of 13 transactions in 2001, compared to 1998, is This reaction also limits the scope for the chiefly explained by changes in the market rate of the Tobin tax. I shall come back to this structure (Galati 2001). in Chapter 3.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 5 » Aspects of social justice and re- tied, not without reason, to the ac- distribution. cess to international exchange markets. The tax on currency transactions pos- sesses a politically strong symbolic Whenever distributional aspects are in force, which was hardly recognized by the game, the potential costs of redis- Tobin. This is why it is potentially ac- tribution have to be expressed in claimed both by the public and by poli- terms of efficiency losses. If the intro- ticians. Behind this symbolism there duction of a Tobin tax would result, for are rudimentary prejudices against instance, in heavy costs of foreign ex- “easily earned money” through finan- change trading and international capi- cial trading as opposed to “hard- tal markets, this would undoubtedly earned money” through labor. Adding have a negative impact on real eco- the notion of „speculation“ as a short- nomic variables such as investment, cut for „fraud“ plays this antagonism jobs, and incomes. The explanation of further up. such complex relationships is not easy however, and the inter-linkages are The political muscle of the Tobin tax is more difficult to convey that simplistic strengthened even more by the fact clichés of social justice, in particular that foreign exchange markets are as the financing of such redistribution (rightly so) considered to represent a programs is expected to be carried by colossal exclusive undertaking of n-i others (the “speculators”). Moreover dustrialized countries. The currencies the liberalization of international capi- of the OECD countries are involved in tal markets is often seen to deprive 89 percent of all transactions. The re- the Third World anyway. mainder falls almost exclusively to emerging economies (including the But even though there would be no currencies of Hong Kong and of Sin- negative impact on the world econ- gapore). The currencies of developing omy, the achievement of the goals of countries play practically no role. This international justice through a Tobin is why a tax on foreign exchange tax with corresponding transfers is not transactions can be seen to burden necessarily assured: mainly the richer industrialized na- » On the one hand it is possible— tions. Its revenue could be reserved and even probable—that the fi- for foreign aid to developing countries nancial industry would shift the tax implying a more equitable distribution on currency transactions onto pro- of opportunities within a globalizing ducers and consumers, and not world economy. 14 carry it itself (see Chapter 5). Thus the Tobin tax is hailed as a » On the other hand it is question- “righteous tax” in a double sense: able whether development policy » On the one hand as a compensa- is only a question of money and tion between the financial industry, can be accommodated merely by often depicted to be non-pro- financial transfers to the Third ductive (Karl Marx), and the pro- World. ducing sectors of the economy In particular this latter point would de- within industrialized countries; and serve a more comprehensive treat- » On the other hand as compensa- ment, which is of course not within the tion between developed and less- terms of reference for this study. developed economies (in short: Nevertheless it has to be mentioned between North and South), whereby the degree of economic 14 This would even be intended if the Tobin tax development is regarded to be were to serve as a presumptive income tax.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 6 that development aid is often de- for a tax on currency transactions flected toward military spending, os- however: tentatious public consumption, non- Exchange rate volatility is not only a sustainable or unsuitable investment problem per se, it also affects the dis- projects, and even into private ac- tribution of wealth between industrial- counts of local oligarchies, which fails ized and developing economies, and it to realize the proper goals of devel- can thus thwart an aid policy that re- opment policies. lies essentially on budget policies. I consider structural reforms—within This point is often ignored because recipient and donor countries—to be development policy all too often looks basically more important for economic toward financial flows, and hardly ever development and an equitable and fair toward and their changes in distribution of opportunities than the valuation.15 mere extension of international finan- Tobin himself has repeatedly argued cial transfers. In Third World countries that the reduction of exchange rate this includes, for instance, the recogni- volatility aims not only at the reorienta- tion of human rights, reforms of edu- tion toward fundamental flows (such cation and health policies, of public as foreign trade, direct investments), administration (“good governance”), but also at the independence of the the struggle against corruption and the central banks from the vagaries of for- dismantlement of state monopoly eign exchange markets. In the ab- power and price controls that convey sence of such independence, a de- benefits to a small minority at the ex- valuation of its currency will compel pense of the broader population. In central banks to either offer high inter- donor countries the main problems est rates on overnight deposits (e.g., are, for instance, the collusion with lo- the strategy of the Swedish National cal potentates, the transfer of inap- Bank at the beginning of the 1990s), propriate technologies, or an ineffi- or to sacrifice foreign exchange re- cient policy of price subsidies, which serves in order to take out their own renders it difficult for developing coun- currencies from the market (or both). tries to overcome their dependency from primary production. To this com- This indicates one of the most impor- pounds the protection of mar- tant asymmetries between industrial- kets—for instance for agricultural and ized and developing countries in the textile products—, which is inefficient era of globalization: While the leading and intricate from an equity point of industrialized economies were able to view. ascertain the independence of their central banks from exchange rate Structural reforms in these areas are volatility (obvious in the statutes of the often independent from financial as- ECB, but also apparent during phases pects. Conversely, specific financing of „benign neglect“ of American mone- models may contribute to preserve in- tary policy), the international consen- efficient and unfair arrangements. sus of economic actors and politicians Where this is the case, development expects from developing and emerg- aid would even jeopardize the realiza- tion of the very objectives of develop- 15 ment policy. This statement has to be taken cum grano salis however. Responsible politicians will rec- ognize both aspects and have acted accord- » Redistribution of national wealth. ingly—for instance through debt relief for the highly indebted poor countries (HIPC). Never- An important aspect of regional justice theless I have reasons to stress that point in a globalizing world has to be em- again—in view of more recent discussions with phasized in the context of a proposal politicians in the realm of development policy.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 7 ing economies to defend, in the case ests, global warming, the loss of bio- of speculation, their currencies by sur- diversity), addressing global health rendering their arduously conquered problems (HIV/AIDS), malaria), re- foreign exchange reserves. Once search (to address basic questions of these reserve have been exhausted, global significance), and other goals of the countries are all too often pushed overarching interest (“global public into debt in foreign currencies.16 goods”). Whenever the of a coun- The use of the attainable proceeds try has to surrender foreign exchange from a Tobin tax is ultimately a politi- that does not correspond to real capi- cal question that cannot be answered tal formation, this is tantamount to an in this study. However it is important international transfer of net wealth, to discuss who (i.e. which institutions, whose magnitude can easily dwarf the national or supranational) would col- flow of development aid. Arduously lect the Tobin tax eventually, because earned wealth can fade away within a this will also limit the possibilities of its short period of time due to currency use (see Chapter 3). speculation. A tax on foreign ex- change transactions could not only protect the freedom to act of these » Systemic transformation. The tax countries’ central banks; it could also on currency transactions is finally generate some additional income in called for by political groupings whose the case of speculative attacks. aims are not always lucid, but are un- doubtedly motivated by the desire to I think it is important to stress in par- transform “the system”. This is com- ticular the connectivity between the prehensible to the point of controlling potential loss of net wealth of develop- the process of globalization more ing countries, and budgeted develop- generally (its “humanization”). Why ment aid. Stable exchange rates thus should we forsake to impose onto in- become a precondition for effective ternational financial markets some development policies, which ought not rules as they have been adopted for be overlooked. long in the realm of national politics for As to aspects of distributional justice it banks, exchanges and other fi- must at last be emphasized that fi- nancial institutions? In this context, nancing development may not be the the Tobin tax can yet be only one only objective of a tax on foreign ex- element of a more comprehensive change transactions. In this context regulatory concept for global financial other global policy goals have been markets. It would be wrong to expect mentioned that could also be ad- the tax to solve all pending problems dressed by using the proceeds from a of the global financial architecture at Tobin tax. The tax yield could equally once. be spent for achieving general policy Systemic aspects of globalization are objectives (such as the war against indeed most prominent in international terrorism, the trafficking of humans, financial markets. It is there that na- arms, and drugs), ecological goals tional boundaries have come to play (such as the protection of tropical for- almost no role. Financial information is available to all market participants si- 16 multaneously, in real time, and ubiqui- The recent experiences of with its tously. Prices of financial products fixed-exchange rate regime—particularly vul- nerable to speculation (see Appendix 3)— have melted down under extreme render it obvious that the orthodoxy is pre- competition that leaves minute resid- pared to push a whole economy into insol- ual markups in the order of basis vency. points (hundredths of a percent). If

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 8 this model of global financial markets would lead to a more efficient alloca- is transposed onto the markets for tion of resources.17 goods and services, in particular the Others start from the intuitive—but labor market, it becomes understand- misleading—equation “commodity = able that an unfettered globalization is money = commodity” of Marx’s the- easily experienced as a threat, which ory—namely a one-to-one relationship would provoke political resistance. between transactions of goods and How these concerns of citizens are to money. They point to the fact that the be addressed will again be a task for volume of trading in international cur- politicians. Scientific analysis can offer rency markets has reached a multiple only limited assistance here. In par- of the value of foreign trade and for- ticular distributional issues are prone eign direct investment, and has lost to escape scientific analysis. It is any direct relationship with transac- however hardly contentious among tions of the real economy.18 economists that market forces are in a Such argumentation is misleading be- better position to unfold economic and cause it pays no heed to the nature of social welfare than administrative in- liquidity. According to my view a re- terventions by national and suprana- quest to limit the volume of financial tional governments. In this vein the transactions to the financial equivalent process of globalization is seen to be of activities in the real economy is tan- inescapable and irreversible—unless tamount to asking for a reduction of one is prepared to bear the immense oxygen in the air to the bare minimum cost burden that socialist countries necessary for life. had been willing to assume for dec- ades. This can certainly not be ex- Liquidity is based on the fact that so- pected for all nations, which implies called “market makers” are willing to that countries opting out from global- set prices everywhere at any time for ization must fall behind economically typically large sums of foreign cur- in relative terms. rency, and to carry out conforming transactions. In a similar way as we According to my view, the yearning to would normally not think about the turn back the wheels of history by availability of oxygen in the air, liquid- means of a Tobin tax either comes ity thus creates freedom of action for from an ideologically-based position of exporters, importers, and direct inves- structural conservatism. It deems the financial industry, once again, to be 17 the “spearhead” of capitalism or glob- Also Tobin mentions that "vast resources of alization, which one hopes to domes- intelligence and enterprise are wasted in finan- ticate by a Tobin tax. Or it is (at best) cial speculation, essentially in playing zero- based on a misunderstanding of the sum games" (Tobin 1991, p. 18), which re- functioning of such markets, and of flects unawareness as to the character of li- quidity trading. international liquidity trading in particu- 18 lar. As an example the Intergroup „Capital Tax, Fiscal Systems and Globalisation“ of the Euro- It is interesting to note that the volume pean Parliament notes that „as a means of of international financial transactions comparison, the total yearly exchange of has been regarded to be excessive by goods and services is evaluated at 4,500 bil- lion dollars, equivalent to less than a week on authors of very different political the currency market. Today, most of the trans- trademark. For instance Summers and actions on the currency market have no link Summers (1990), p. 881) expect a tax with exchange of goods and services or in- on foreign exchange transactions to vestment and are purely speculative“ (Declara- eliminate „wasteful trading” and „ex- tion in preparation of the International Confer- ence „Financing for Development“ in Monter- cessive financial engineering“, which rey; emphasis added).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 9 tors. logical tensions will be discharged in an earthquake. The introduction of a Liquidity does not only trim down price Tobin tax could even provoke a se- and settlement risks, and it reduces vere liquidity shock at a global scale if the costs of hedging; it also prevents it is effected thoughtlessly and disre- destabilizing insolvencies. gards the structure of markets.20 This implies however that the market Some prominent authors have argued maker is able to close his/her open however that a financial transactions position in a foreign currency immedi- tax with a very small rate would affect ately in order to eliminate risks and to liquidity trading only negligibly but remain solvent. Open foreign ex- eliminate a behavior that is character- change positions are thus handed on ized as destabilizing “noise trading” like “hot potatoes” until they finally (Summers and Summers 1990). This reach a market participant who is will- behavior is described in a new brand ing to hold a corresponding counter- of the financial literature “that ques- position. In this way one transaction tions more generally the efficiency of initiated in the real economy can trig- 21 financial markets”. Noise traders act, ger a whole chain of subsequent fi- unlike informed “rational” traders, on nancial transactions. The demand to the basis of misinformation such as limit financial transactions to the “nec- technical investment analyses (chart- essary scale” overlooks that this scale ing techniques), or rumors. Their be- cannot be based on the value of eco- havior can drive prices off their fun- nomic transactions. damental equilibrium value, which It is unquestionable that a Tobin tax renders markets more risky and vola- would reduce the volume of foreign tile (Shliefer and Summers 1990). In- exchange transactions. The disadvan- formed traders cannot counter these tage is however that the tax cannot destabilizing tendencies (DeLong, distinguish between liquidity trading Shliefer, Summers, and Waldmann and speculation. It would particularly 1988, Summers and Summers 1989). hit the stabilizing liquidity trade be- This is because arbitrageurs usually tween wholesalers. This would lead to operate without reference to funda- “thinner” markets with less liquidity. mental data by optimizing their deci- Stabilizing arbitrageurs would with- sions exclusively in view of a given draw from currency trading and only price, and by realizing only a local op- resume their activities if the actual ex- timum. This actual price can deviate change rate deviates from its “intrin- significantly from its fundamental sic” value (respectively, its value as- value and thus create “speculative sumed to be realistic) by more than bubbles”. To the extent that the Tobin the tax rate.19 This would increase the tax could reduce such noise trading it volatility of exchange rates because would contribute to stabilizing foreign the price-discovery process will be in- terrupted and prices can no longer re- 20 flect all information available on the This argues in favor of a low tax rate when introducing the tax. I even imagine starting with market. Prices will then adjust abruptly an algorithm for calculating the tax that con- whenever the deviation from a pre- tains a zero tax rate initially. This would not sumed equilibrium rate has become produce any tax revenue but allow to identify too great—as gradually accruing geo- the potential tax base on a recurrent basis and to monitor its reactions to a modest increase of the rate in a heuristic fashion. 19 21 "With the tax in place, arbitrage investors This literature "has developed the perspec- would wait for larger price discrepancies be- tive that the financial markets may not be as fore entering the market" (Kiefer (1990), p. efficient as previously thought" (Kiefer (1990), 891). p. 889).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 10 exchange market even at very low avoid a negative impact of a Tobin tax rates. on the functioning of world financial markets. This thesis is important because it would resolve the apparent contradic- » Summary. Taxes on currency tion between exchange rate stabiliza- transactions pursue very different ob- tion and revenue raising at low tax jectives. The more important ones re- rates. I am however not convinced late to stabilizing exchange rates, pro- that the thesis is necessarily correct. ducing fiscal revenue, and redistribut- Liquidity traders would usually not use ing resources between the financial chart techniques in making their deci- and producing industries, and be- sions, although they are of course not tween countries (in particular between immune against exploiting rumors. North and South). The tax is also ex- Chartists are mainly found among the pected to aim at systemic changes customers of foreign exchange traders that are directed against the process outside liquidity trading (for instance of globalization. institutional investors such as invest- Systemic changes are not considered ment funds or hedge funds). Their po- in this report. On the contrary: the tentially speculative actions are prone proposals for a tax on foreign ex- to affect the exchange rate more change transactions are subject to heavily than the more neutral behavior avoiding systemic effects as much as of arbitrageurs. But the presence of possible. the latter group is much more pro- nounced than the former. It means Distributional objectives are also not that the tax would still affect liquidity pursued in this study because they operations more heavily than noise remain the privilege of politicians. trading. The centerpiece of the study on the As to the exploitation of rumors and its feasibility of a tax on foreign exchange effect on the exchange rate, it is ques- transactions is formed by the two ob- tionable whether a Tobin tax could jectives stabilization of exchange rates counter the destabilizing effects that and fiscal revenues. this may trigger. In this respect it is true what has been said before: the tax is ineffective whenever the ex- pected variation of the exchange rate is higher than the tax rate even by a small amount. The worst would be if the tax would reduce noise trading only partially or not at all, but lessen or even eliminate stabilizing liquidity op- erations. This must lead to greater volatility in world financial markets.22 Systemic objectives will not be con- sidered in this study. On the contrary I believe that developments in world fi- nancial markets have to be appraised positively. It is all the more important to control systemic effects in order to

22 Moreover the entailing contraction of the tax base is hardly in the interest of those who ex- pect revenue from the tax.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 1 ® Page 11 Chapter 2: Instrumentation What concept to choose for a tax on foreign exchange transactions?

Before posing the question of the feasibility of a tax on foreign exchange transactions it is necessary to clarify how the concept of the tax to be realized should look like. The Tobin tax as originally conceived is not the only option of a tax on foreign exchange transactions. Alternative proposals have been developed starting from points where the Tobin tax exhibits clear conceptual weaknesses. It is therefore necessary to clarify the question of the concept and its instrumentalization before posing the question of its operationalization and implementation. The purpose of this Chapter is to develop a basically functioning and politically feasible concept of a tax on foreign exchange transactions. This requires to first reduce the complexity that goes with the original proposal by Tobin.

» Reduction of complexity. The alledged non-feasibility of the Tobin tax is often explicated by a number of economic, legal, administrative and political complexities.

1 For an analysis of these weaknesses see in particular Spahn (1995), Shome and Stotsky (1996), Nadal-De Simone (1997), or (in German) Buch, Heinrich and Pierdzioch (2001).

» Under economic aspects, it is criti- equally to expect that there is cor- cized that the Tobin tax would responding interjurisdictional co- operation in this matter. Most of 1. produce similar inefficiencies the governments of OECD coun- as the multi-phase tax on gross tries (in particular the United transactions of commodities that States) reject the idea of a Tobin applied in the Federal Republic of tax at present. before 1968. It was cen- sured to entail a cascading effect, » Under political aspects it is a to- which led to a varying tax burden tally open question who would be on final products in accordance entitled to collect the tax revenue. with the number of transactions. Tobin himself thought of interna- This must provoke organizational tional organization (IMF, World- reactions (such as the vertical in- bank, ). But there tegration of processes), which un- could also be supranational or- doubtedly distorts economic deci- ganizations to be newly created, sion making; international public foundations and non-government organiza- 2. produce particularly heavy dis- tions (NGO). In principle it would tortions of this kind as long as it is also be possible to distribute the impossible to introduce the tax tax revenue onto national tax au- universally, since this would dislo- thorities, however the distribution cate parts of foreign exchange formula is likely to be highly con- markets to non-cooperating tax troversial. havens. Moreover the tax could be avoided by moving into trading Even if those questions would with other instruments such as have been resolved it is still open short-term foreign securities for which purpose such revenue (treasury bills etc.) or through for- should be used, who would be the ward and trading. recipient, who would administer it, and how a legitimate democratic 3. put a one-sided burden onto and administrative mechanism for the competitiveness of the respec- controlling such institutions should tive financial centers and squeeze look like.2 If one thinks in particular worldwide liquidity trading, which of a transfer of resources toward would trigger unsolicited systemic developing countries, it is open in changes. which way these countries could » Under legal aspects it is often as- participate in the decision proc- serted (without further examina- ess.3 tion) that the tax would disagree As to legal questions, these do not fall with the idea of capital liberaliza- into the terms of reference of this tion in accordance with the Liber- study. It has to be emphasized how- alization of Capital Movements ever that the freedom of capital Code of the OECD or, in the case movements is not abolished or put in of the EU, the Maastricht Treaty.

» Under administrative aspects it is 2 reckoned that the international Patomäki (2001) argues that a global tax on community of important financial foreign exchange transactions would also centers is unlikely to cede sover- strengthen national sovereignties, and he con- ceptualizes an institutional framework for its eignty in the area of taxation and administration that aims at the democratization consent to the creation of a world of the process of globalization. tax organization, which would be 3 This question is unsatisfactory even now as reasonable for operating a univer- to the development of international financial sal and multilateral tax on foreign markets. See for instance Griffith-Jones exchange transactions. It is (2001).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 13 jeopardy just because a tax is levied gitimate parliamentary institutions. on such transactions. Taxes are lev- The tax would therefore work uni- ied on various transactions of goods laterally and partially, not multilat- and services and are typically re- erally and universally. garded by economists as conforming » The same legislature that is ac- with the market. In most respects they countable for the tax will also de- are superior and to be preferred to cide on the apportionment and use other types of government interven- of the tax proceeds. It implies the tion such as capital controls, or mone- revenue to fall to a national gov- tary policy involvement of the ortho- ernment (respectively jointly to all dox type. The key question under le- governments that engage in a co- gal aspects is whether taxes are neu- ordinated effort), but under no cir- tral with regard to the competitiveness cumstances would it go to interna- of different actors in the market. 5 tional organizations. A tax on for- In the following I shall assume that eign exchange transactions to be legal problems would not emerge introduced unilaterally by the within the framework of the European European Union would be subject Union as long as certain preconditions to decisions of the European are fulfilled. This point of view is sup- Council and the European Parlia- ported by a study for the Parliamen- ment as well as national parlia- tarian Working Group “Tobin tax” of ments that would have to ratify the the .4 law. This study asks in particular for a All proposals that go beyond such small rate of the tax on foreign ex- limitations must be considered unreal- change transactions, which „à la lu- istic, and be excluded as politically mière de ses objectifs, n’entrave pas “unfeasible” for the time being. les mouvements de paiements de In the following I shall call this limited manière déraisonnable ou dispropor- decision space “politically feasible”. In tionnée.“ This must have repercus- order to account for eventual legal sions on the fiscal objectives of the scruples, I shall also posit a much tax. They will be addressed later on. lower tax rate than was considered by Tobin and his supporters.6 » Limitations. In view of a multiplicity While I have been using the terms of unresolved political questions that „tax on foreign exchange transactions” relate to the universal and multilateral and “Tobin tax“ synonymously so far, I character of the Tobin tax, this report shall speak of a „politically feasible will confine itself to limited scenarios. Tobin tax (PFTT)“ from now on, In order to be realistic these scenarios which shuns the critical objections will have to acknowledge the follow- against a multilateral and universal ing: tax. » The tax on foreign exchange Before entering into examining eco- transactions cannot be introduced nomic and technical aspects that are universally. It is required that the important for the realization of a tax be planned by an existing au- PFTT, I shall first discuss a few vari- thority, a government or a coordi- ants of taxes on foreign exchange nating body such as the European Council, and be introduced by le- 5 This could eventually be achieved in a sec- 4 ond round via the explicit earmarking of tax The study has been prepared by Lieven A. proceeds or through budgetary decisions of Denys, professor for European at the (the) parliament(s). Faculté de Droit of the Vrije Universiteit Brus- 6 sels. It dates of April 18, 2001. See also Clunies-Ross (2000).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 14 transactions that are politically feasi- PFTT. ble and are relevant in the context of a Chile has used NNR to fend off short- PFTT. term capital inflows and to avoid artifi- cial short-term appreciations of its » Relevant variants of taxes on for- currency. At the same time the device eign exchange transactions. The was expected to secure solvability in the case of a reversal of foreign capi- following concepts are related to taxes 7 tal flows, and to deal with the pressure on foreign exchange transactions: of depreciation. 1. “Non-remunerated reserve re- It is debatable whether this strategy quirements (NRR)” on short-term for- was successful, because nobody is in eign exchange deposits. a position to say whether speculative 2. The taxation of cross-border capi- capital inflows have been successfully tal transactions with a „cross-border suppressed by the measure. The capital tax (CBCT)“. mere indication that capital inflows 3. A two-tiers „Tobin-cum-Circuit- have endured and the Chilean peso has appreciated in spite of the meas- Breaker Tax (TCCBT)“. It consists of a PFTT as a base and an „exchange- ure is not sufficient for prove. The rate normalization duty (ERND)“ that fundamental data of the Chilean responds to exchange rate volatility. economy were extraordinarily positive during that period, and the country was therefore highly attractive for non- » „Non-remunerated reserve re- speculative foreign capital regardless quirements“ (NNR)—NNR have of the NNR. Under fiscal aspects the been used in various forms as unilat- Chilean NNR were quite successful eral policy instruments, even in the however, and this in spite of the fact Federal Republic of Germany (intro- that market participants were develop- duction of a „Bardepot” for foreign ing successful strategies to evade the deposits in 1971). Spain has em- loss of interest payments. This had ployed the instrument in 1992 during repeatedly forced the central bank to the peseta crisis. The best-studied adopt new measures in order to close case appears to be the one of Chile loopholes of the system. during the 1990s8, but also Colombia9 This is not the place to discuss the and Slovenia have used the instru- experiences with NNR further. It is ment. Since a part of the inflow of however conceptually helpful to stress currencies has to be kept in the form the following points with regard to a of mandatory non-interest bearing PFTT: deposits, lost opportunity costs act like a tax. This is why the measure is » NNR concern primarily the holding of stocks, not transactions (flows), often discussed in the context of a whereby stock will of course vary in response to transactions. The 7 For further variants, that are interesting basi- comparison with a tax on foreign cally under theoretical aspects, see Spahn exchange transactions is therefore (1995, Appendix 2). strictly not fitting. 8 See for instance Nadal-De Simone and » Compared to the volume of foreign Sorsa (1999). The reserve rate in Chile was at exchange transactions, the parts first 20 percent, then 20 percent, but it has been reduced intermittently to 0 percent, of the capital balance that are sub- whereby the system has however been main- ject to the tax are significantly tained in principle. smaller. This finds its reflection in 9 The experiences with capital controls in Latin a much higher “tax rate” compared America have been discussed, inter alia, in to feasible rates of a PFTT. As- Agosin and Ffrench-Davis (1996). suming that the rate of interest for

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 15 foreign capital is 10 percent p.a. an interest rate of 10 percent p.a., and the required NRR 30 percent, a tax rate of one percent would be this would give a “tax” burden of 3 equal to 10 percent of the income percent, a significantly higher rate from an investment of one year, than the rates discussed in the but put a confiscatory 125 percent context of a PFTT. charge on income from an invest- ment of one month.10 This renders » More important is the fact that the the approach similar to the Tobin „tax burden“ (contrary to the Tobin tax. tax) is directly (and not inversely) proportional to the holding period 2. Moreover Zee wants the CBCT of the asset. For instance a short- withholding tax to be credited term investment for one week against transactions that are not would carry a charge of 3/52 per- related to capital imports. - cent, while the charge would be 3 ers receive the through percent for an investment over a the value-added tax11; this is a year. This proportionality of the simple administrative procedure. charge runs counter Tobin’s con- Recipients of interest payments, ceptual idea. This is why NNR are dividends, repatriated profits etc. accompanied (also in Chile) with a would receive a reimbursement host of exceptions aimed at exon- when filing for the income tax; this erating exporters and importers as is of course more complicated in well as long-term investors. These administrative terms than in the exceptions engender substantial case of VAT credits for exporters. administrative costs, they gener- I shall refrain from a more compre- ate “loopholes”, and they are still hensive analysis of the proposed in- incapable of eliminating the sys- strument. In comparison to a PFTT temic bias of tax-burden propor- however, the following points seem be tionality for short-term investors. worth emphasizing: The speculator with an extremely short-term commitment is affected » Transnational capital movements the least, while other investors are not necessarily identical with would bear the brunt of the foreign exchange transactions. charge. They could also be settled exclu- sively in foreign currencies (e.g. 12 the US dollar). This renders the » The „cross-border capital tax“ CBCT similar to NRR. It is not the (CBCT)—The proposal of Zee (2000) transaction as such that is subject to tax cross-border capital flows with a to the tax, but ultimately only the CBCT has not yet been tried— (positive) variations of the net contrary to the NNR. However it fulfills capital balance, adjusted for cer- the requisites of “political feasibility” tain positions of the current bal- and is therefore an interesting alterna- ance. tive. The objective of the tax is— similarly to the NNR—to fend off capi- 10 tal imports that could become an ex- See Appendix 2 for the calculation (formula change rate risks because of their (4)). short-term nature. The CBCT avoids 11 The tax rate for exports would thus be –1 the systemic bias of the NNR. It is percent rather than 0 percent under VAT that therefore a proper transactions tax. applies the destination principle. 12 The proposal consists of two parts: Within the EMU there are even transna- tional capital movements that are carried out in 1. At first all private capital inflows “domestic currency” (the euro). In developing into the country a fraught with a and emerging economies it is not unusual to withholding tax. Assuming again employ the US dollar as a mean of transaction.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 16 » The CBCT covers only capital have to pay a tax when repatriating imports, not capital exports. It their profits, but the initial import of works thus asymmetrically. The capital may have been subject to the objective is to limit capital inflows, tax. But the foreigner had of course no not to establish a “mouse trap” for possibility to credit the tax against his foreign investors. By emphasizing or her income tax.15 the destabilizing effects of capital Summarizing it can be said that the imports (and by refraining from proposal of a CBCT has a number of checking capital flight altogether), advantages compared to NRR, in par- Zee takes up an important aspect ticular because it respects the inverse that also motivated my earlier pro- relationship between the holding pe- posal of 1995 (see below).13 riod and the tax burden. In administra- » The settlement of payments (and tive terms the CBCT is however likely payments backed by borrowing) to exhibit similar complexities as the can be effected entirely outside a NRR, perhaps even more as regards currency area. They may appear the integration of the tax with the per- as changes of assets and liabilities sonal income tax. in the books of accounts of eco-

nomic agents, but never trigger any cross-border transactions.14 » The Tobin-cum-Circuit-Breaker Tax (TCCBT). My own proposal » Movements of the capital balance (Spahn 1995, 1996) can be under- are significantly smaller in volume stood to solve the basic conflict be- than the transactions (payments) tween the two objectives of a PFTT: that are targeted by the PFTT. This is explained partially by the ü Either the tax rate is too high: This fact that there may be many (li- may fend off speculators, but it exhib- quidity) transactions behind net its significant negative consequences capital movements. A CBCT will for the allocation of resources by re- only target the result in economic ducing (and even possibly eliminating) terms. It is also explained by the international liquidity trading. It is asymmetrical construction of the doubtful whether the tax would tax (there are no charges on capi- achieve the goal of exchange rate tal exports as a result of imports of stability under these circumstances, goods and services, or of income because hedging operations would transfers abroad). And finally it become more costly and pricing n-i hinges on the various exemptions formation is taken out of the market. entitling economic agents to a tax ü Or the tax rate is very low: In this credit under the CBCT. case international liquidity trading In particular the tax credits render the could continue to operate fairly unim- scheme extremely complex. For in- peded, but the tax is unlikely to deter stance foreign investors would not speculation.

13 My own proposal also aims at checking very 15 short-term capital inflows through the norma- This statement is correct only in principle, lization duty, while longer-term investments go but is more complicated in practice. Undoubt- untaxed with almost 100 percent probability. edly a tax on capital imports will not be credit- Contrary to Zee’s proposal however, my own able against foreign income tax according to approach is symmetrical in that it covers both double-taxation treaties. But one may expect capital imports and exports. the financial industry to develop products whereby the capital import tax is initially 14 Such books of account can of course be swapped onto a domestic agent (who credits kept „unofficially“ if this allows avoiding the tax the tax against income tax), and re-swapped (in particular in emerging and developing thereafter. Foreigners are thus exonerated via countries). the pricing of such swap instruments.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 17 There are also fiscal objectives asso- foreign exchange markets), and ciated with my proposal. The idea is to this externality is charged with a avoid a contraction of the tax base tax of the Pigou-type. The aim is following the introduction of a TCCBT, to reduce the externality—similar and to bolster the revenue potential of to approaches of environmental the tax in periods of speculation. policy. The tax rate of the addi- tional charge on the negative ex- The TCCBT consists of two integrated ternality may be very high. The parts that relate to the same tax basis, additional charge has only a regu- namely „relevant foreign exchange latory function and no fiscal objec- operations (RFXO)“, which I shall tive. On the contrary: If it achieves have to specify more precisely later its goal and holds back specula- on. tion, the charge cannot yield any 1. The first tier is a classical Tobin revenue. tax (PFTT), however with a tiny The additional charge, which I have rate. This tax rate is uniform for dubbed „exchange-rate normalization each , but it could duty (ERND)“ on another occasion eventually vary for different mar- (Spahn 1996), deserves some further kets in accordance with the size of 16 reflections on its specific features: the bid-ask spread. It is decisive that the tax rate be comparably neutral on liquidity trading. This » Which is the tax base of the tier of the TCCBT has mainly fiscal ERND? “Negative externality” has to functions, but it can also contribute be rendered operational for taxation, to stabilizing exchange rates to the and to be standardized for legal pur- extent that it eliminated destabiliz- poses. To this effect I was inspired by ing noise trading.17 the EMS (European Monetary Sys- 2. The other tier consists of an addi- tem) that was operational in the EU tional charge that is specifically from 1979 until the introduction of the tailored to discourage speculation. euro. In order to guide monetary poli- Under normal circumstances this cies, it worked with a tax is dormant, but it is triggered 1. target rate, the ECU reference automatically whenever there is 19 rate, and speculation on foreign exchange markets.18 Thereby speculation is 2. a target zone (or „corridor“).20 valued to represent a negative ex- Once the exchange rate would devi- ternality (such as a “pollution” of ate from the target zone, the central banks of the respective countries were compelled to bring the rate back 16 I shall come back to the question of the tax into the corridor by means of mone- rate, but I shall finally accept a uniform tax rate tary policies.21 in spite of varying spreads for different cur- rency pairs.

17 See Chapter 1 and in particular the contri- 19 bution by Summers and Summers (1989). This rate corresponded to a weighted arith- metical average of 12 European currencies. 18 This process is controlled through an algo- 20 rithm within the settlement system for foreign This corridor was defined relative to the exchange, i.e. the formula pre-exists in the target rate, and was initially 2.25 percent (for form of a computer program and is recognized some currencies 6 percent), and later 15 per- by all market participants a priori. So the pric- cent. ing conditions are always known in advance. 21 An advantage of the tax is the fact that there is The system is sketched in a simplistic way no need for discretionary decisions—neither here. In fact there was an obligation for infra- on the side of legislators, nor on the side of the marginal interventions whenever the exchange actors in foreign exchange markets. rate would tend towards an intervention point, although it was still in the corridor. Moreover

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 18 I chose a similar approach to defining Different from the EMS, the target rate negative externalities that result from of the ERND is an adjustable moving speculation on foreign exchange mar- average of daily official exchange kets. It requires standardizing “specu- rates relative to a reference or anchor lation” in a technical sense. currency (such as the US dollar or the euro). The workings of the ERND are Speculation usually finds its expres- sketched in Chart 1. sion in an abrupt change of the ex- change rate (see the examples in Ap- pendix 3). If there is a well-defined » How to define the corridor, the official target for the exchange rate anchor currency and the tax rate? and an equally well-defined corridor, it The EMS had defined the upper and is possible to distinguish “normal lower boundaries of the corridor in an transactions” as those that are carried ad hoc fashion, i.e. pragmatically. The out within the price corridor. These same should apply to the ERND. transactions are considered non- However the width of the corridor speculative and are exempt from the could be determined with regard to surcharge. However if there are devia- empirical data on the daily fluctuations tions from the corridor, this is consid- of the exchange rate relative to the ered to represent a negative external- target rate under normal circum- ity (i.e. “pollution”) and will be charged stances, perhaps with some safety with a Pigou tax.22 margin in order to trigger the duty not too often. These daily fluctuations the instruments of and their could vary for different pairs of cur- coordination were regulated in a more detailed fashion. 22 exchange reserves to support the exchange This procedure could be considered ot rate, the country under represent the „dual solution“ of the EMS that would achieve revenues with the dual scheme. operated with monetary instruments. Instead We know from environmental policies that of reacting to speculation by means of a sub- taxes and subsidies can be equivalent instru- sidy (for instance high interest rates for over- ments to cope with correcting externalities and night deposits), the dual scheme would use a a misallocation of resources. tax policy instrument. Instead of using foreign

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 19 rencies, which could lead to the con- » Must the ERND not fend off po- clusion to use different bandwidths for tential capital investors? different currency pairs. Since the The contention that the ERND would scheme operates unilaterally, this can deter potential investors—in particular be determined individually by each because of its high statutory rates— country. The same is true for the re- seems to be totally unwarranted. spective anchor currency (respectively currency basket) and the tax rate to ü On the one hand the effective tax be chosen. burden is significantly lower than the statutory rates (see previous The statutory tax rates of the ERND paragraph), although it could in- should be high, within a range of 50 to deed increase with the degree of 100 percent. deviation from the target zone.

ü On the other hand will the ERND » Isn’t a tax rate of 100 percent con- be activated only in those in- fiscatory? stances when the exchange rate wreaks havoc. This is an eventual- It is important to realize that the tax ity whose probability is extremely rate is not applied to the value of the low and uncertain as to its timing. transaction as such (as for the PFTT), In relation to a future trading hour, but only on the externality, i.e. that the probability of an activated part of the price that lies outside the ERND is practically (and mathe- corridor. It means that the effective tax matically) zero. burden is variable—with a burden of 0 percent within the corridor and at its If there are investors to fear the margin, and an effective tax burden ERND, they are exclusively those with that increases with the degree of de- a short time horizon who are betting viation from the target zone. The ef- on speculative rents. It is exactly this fective tax burden for statutory ERND group of investors that schemes such rates of 50 and 100 percent are de- as NRR and the CBCT aim at deter- picted in Chart 2. ring from investing in a country. The ERND appears to be much more effective in this respect as well as more targeted, because it will be activated imme- diately and auto- matically (via a previously instal- led software). It thus works in the form of an auto- matic “circuit breaker” against foreign exchange speculation. The variable effective burden on a Long-term inves- transaction that is carried out at prices tors are likely to disregard the tax al- outside the corridor also conforms to together however, because its activa- the conditions for a neutral (respec- tion is improbable over a longer period tively allocation-enhancing) Tobin tax of time. (Tornell 1988, 1990).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 20 On the contrary: One may even ex- » Which countries could benefit pect long-term investors to be en- from an ERND and what is its rela- couraged to invest in a country that tionship with the orthodoxy? employs an ERND. This is because The concept of an ERND is primarily the surcharge acts as a sort of insur- suitable for transition, emerging and ance. It guarantees practically that the developing economies that aim at exchange rate will never vary more accessing free international capital than the set bandwidth in one day. In markets. For industrialized countries comparison to the potential alternative the concept is only apposite for those without the tax (a possible “free fall” of that do not belong to one of the larger the rate) this must be considered a currency areas (US dollar, euro), but considerable advantage. The same is seek to peg their money to an anchor true for importers and exporters. They currency. The ERND does not appear should be prepared to pay even a to be reasonable for the USA or Euro- premium for this exchange-rate guar- land. Exchange rate volatility between antee either in the form of (an occa- the dollar and the euro could perfectly sional) ERND, or in the form of a cur- be controlled through the coordination rency option.23 between the ECB and the Fed.

The advantage of the ERND as a uni- » Wouldn’t the ERND lead to devia- lateral measure consist in particular in tions from the equilibrium ex- that national Legislatures can deter- change rate? mine the conditions for their own scheme at their discretion, without Given the design of the ERND, varia- having to coordinate with other au- tions of the target rate are perfectly thorities. The period for the moving possible. For instance, if the period for average, the calculation of the target the moving average consists of 20 rate, the width of the corridor, and the business days, it is possible to incor- tax rate could all vary from currency to porate in the exchange rate all fun- currency. damental data that become available over a month. Therefore the ERND Whatever the advantages or disad- does not permit a “leaning against the vantages of an ERND, it is always wind” policy because it allows the ad- appropriate to evaluate it against the justment of the exchange rate to fun- benchmark of current alternatives—in damentals in a continuous fashion. particular against the practices of the Appreciations and depreciations of the orthodoxy. Orthodox politicians prefer currency are possible and accepted. It monetary interventions by central is only in the case of abrupt exchange banks rather than tax policies.24 This rate changes—typical for periods of calls for the following annotations speculation—that the ERND works as (Spahn 2001): a circuit-breaker. 1. Tax policy is an instrument con- This defuses equally the argument forming to the market. This is true that governments could hide behind for monetary policy only if it is the scheme to carry out irresponsible long-term oriented, but not for the policies at the expense of investors. short run when it may be com- pelled to hectic reactions under 23 speculative pressure. An advan- It can be expected that the financial industry tage of the tax policy approach is will assume the ERND risk for exporters and importers, who are often more dependent on notably that the rules have to be short-term payment requirements than long- term investors, in the form of currency options 24 net of tax. In this way the implicit insurance There are of course also „harder“ interven- premium would become an explicit cost of the tions such as capital controls, which I shall not insurance scheme. use as a benchmark for that matter.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 21 established a priori. They are sure.26 known to all participants and are ü Finally, the greater independ- therefore predictable. This does ence of the central bank from ex- not apply to an extemporizing change rate turbulences will also monetary policy. render the country less dependent 2. The unpredictability of monetary from foreign capital investors and policy in times of speculation has the support by international or- a number of severe drawbacks ganizations such as the IMF. If that even encourage speculation: there is need to intervene in cur- rency markets beyond the loss of ü At first will hectic interventions currency reserves, it tends to in- signal policy distress to market crease the foreign indebtedness of participants. Tax policy is a proce- the country, which renders its cur- durally neutral measure that acts rency all the more vulnerable to automatically as a built-in stabi- speculation because the confi- lizer. dence in the economy is dwin- ü Then will monetary interven- dling. Moreover the country is bur- tion encourage speculative behav- dened with debt service, and—if ior if market participants can rea- devaluation is ultimately unavoid- sonably expect the central bank to able—it looses additional net be politically compelled to bail out. wealth because its foreign indebt- The ERND will procure the neces- edness, measured in domestic sary room for monetary absti- currency, will increase in propor- nence by the central bank and al- tion to the rate of devaluation. low it to focus mainly on domestic 25 Given these considerations and policy objectives. against the background of reoccurring ü Moreover it is not negligible currency crises, which are tackled in that the ERND will produce reve- vain with orthodox instruments over nue during periods of exchange and over again, it is incomprehensible rate turbulences, whereas mone- why tax policy interventions in foreign tary policy intervention is tanta- exchange markets should have such mount to subsidizing foreign ex- a bad reputation. change trade—as mentioned be- The TCCBT possesses significant fore. This is true for its ni terest allocative and distributive advantages rate policy, which aims at render- compared to orthodox instruments as ing domestic assets more attrac- long as negative repercussions on tive, and for its direct interventions international liquidity trading can be in foreign exchange markets, avoided. How this could be achieved whereby valuable reserves are will be discussed in the following sacrificed just to stabilize the ex- Chapter. change rate. Monetary policy in- terventions—preferred by the or- thodoxy—also entail a negative impact onto the real economy of countries under speculative pres- 26 This is true in particular for interest-rate policies "... as with simply raising interest rates to defend a weak currency, it is virtually im- 25 This was one of the most important goals of possible to burn the speculators without simul- Tobin with his tax on foreign exchange trans- taneously affecting other sectors of the econ- actions. Monetary policy abstinence during omy" (Garber and Taylor (1995), p. 178). As periods of an activated ERND is even manda- has been argued previously, monetary policy tory for its effectiveness, because traders interventions may also lead to a transfer of net would otherwise attempt to shift the burden of wealth at the expense of the countries con- the tax onto the central bank. cerned.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 22 » Summary. This Chapter has dis- plex administrative problems however, cussed the basic conditions for a po- which will render its realization less litically feasible tax on foreign ex- likely. Both instruments are however change transactions, and developed significantly different from a tax on an effective set of instruments. foreign exchange transactions and will no longer be examined in the remain- The scope for decisions is politically der of the report. constrained in that the tax has to be introduced and be accounted for by “Politically feasible” appears to be a existing decision-making bodies such combination of a unilaterally acting as national and supranational parlia- Tobin tax with a low rate as a “true” ments. The tax works therefore unilat- tax on foreign exchange transactions, erally and partially, not multilaterally in combination with a surcharge that and universally. Moreover the tax reacts on speculative variations of the yield will fall to whoever will be re- exchange rate. Both are technically sponsible for legislating the tax, not to intertwined. international institutions. They could The proper Tobin tax could be used get hold of the tax revenue only in a by groups of OECD countries, for in- second step—via budgetary transfers. stance by the EU. The surcharge Among the measures that are “politi- should be used unilaterally only by cally feasible” in this sense range the transition, emerging, and developing going practice of mandatory non- countries as well as those industrial- interest bearing deposits on foreign ized countries that remain outside the exchange transactions (respectively major currency areas, but aim at peg- foreign exchange positions) and a ging their currencies to an anchor proposal by Zee, who would levy an currency (or a basket of currencies). asymmetrical tax on capital imports. The combination of two taxes in the Both instruments are interesting as form of a Tobin-cum-circuit-breaker parts of an arsenal aimed at combat- tax has significant allocative and dis- ing currency speculation. In particular tributive advantages over an ex- mandatory deposits are likely to con- change rate policy that is based on stitute an important element within a orthodox monetary policy. future global financial architecture. The proposal of a tax on capital im- ports appears to be laden with com-

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 2 ® Page 23 Chapter 3: Organization What can be said about the structure of foreign exchange markets and its development?

In order to discuss the problems of implementing a PFTT (and implicitly of the ERND) it is necessary to understand the structure and function of international foreign exchange markets and its developments.

» Characteristics of the . After the breakdown of the fixed-exchange-rate regime of Bretton Woods, a global and highly differentiated international foreign exchange market has emerged. The more important currencies can now be exchanged every time, from any place, in substantial amounts, and at little costs. In April 2001 the daily turnover of foreign exchange was in the order of 1.2 bill. US dollars, three years before it had still been roughly 1.5 bill. US dollars.1 The reduction in volume during these three years was 19 percent in current, and 14 percent in constant dollar exchange rates (base: April 2001). The amounts traded on foreign exchange markets are impressive: single transactions of 200 to 500 million US dollars are not abnormal, i.e., it is essentially a market for wholesalers. Price setting often occurs 20 times per minute; and the exchange rate can alter several thousand times a day, albeit usually in very small steps, which expresses the smoothing and risk-reducing effects of liquidity.2

1 See BIS (2001) and Galati (2001). 2 See also the exemplary „snapshots“ of an electronic trading desk in Appendix 4. The seven most important currencies hind Singapore. The partition of the and their market shares are depicted market according to trading locations in the following Table 3 (standardized is found in Chart 3 for the year 2001. at 100 percent).3 The of these The comparison of the distribution by currencies of total transactions was traded currencies and trading loca- roughly 90 percent in 2001. tions renders it obvious that activities on foreign exchange markets have to Table 3: be seen as totally detached from the Volumes of daily foreign exchange transac- currencies of respective market tions according to currencies in bill. US places. The globally leading trading dollars (for the months of April) place London controls roughly one third of the market, although the 1998 2001 was involved only in 6.6 percent of all transactions. The US dollar 43,7 45,2 city-state Singapore, with a currency Euro *) 18,8 that is involved in only 1.1 percent of all , achieves a trading volume Yen 10,1 11,4 higher than the Federal Republic of Germany, although the Deutsche Pound sterling 5,5 6,6 Mark was involved in 15 percent of all Swiss franc 3,6 3,0 transactions in 1998. This character- izes the foreign exchange market as a Canadian 1,8 2,3 truly global market, in which national dollar boundaries and the emission of cur- rencies play practically no role any Australian 1,6 2,1 longer. Against this background, all dollar measures equivalent to capital con- *) Before the introduction of the euro, the na- trols (limitation of currency imports tional currencies of Euroland were recorded and exports, split exchange rates, individually. The more important ones were the Deutsche Mark (15.1 percent) and the French mandatory deposits on capital im- Franc (2.6 percent). The other EMS currencies ports, or Zee’s capital import tax) must and the ECU represented 8.7 percent in 1998. have to be considered as highly un- Adding up these figures to compare them with promising, because the respective the share of the euro would be incorrect. currency area will tend to detach itself from international capital markets, which undoubtedly entails more dis- The geographical distribution of the advantages than advantages in the global foreign exchange market has long run. hardly changed during the last three years: After all, the trading place The strong concentration of foreign Great Britain (London) dominates the exchange trading is explained by the market, with one third of the volume. It following factors: is followed by the USA (New York) 1. A currency (traded on the spot with 16 percent, and Japan (Tokyo) market, for instance) is a homoge- with 9 percent. The Federal Republic neous good, also in standardized of Germany, with roughly 5 percent of variants such as outright forwards, the market, holds the fifth place be- and in swaps and options that combine spot and forward trade. Therefore the financial center of a 3 Since foreign exchange trades always trigger currency area does not possess two transactions —one in the base currency, any particular advantage in the the other in the target currency—the BIS adds up transactions to 200 percent. In this report, form of specific information that the usual standardization is adopted for rea- would be relevant for the trading sons of comparability. location alone. Data on foreign ex-

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 25 change markets are always avail- distribute trading onto three trad- able globally. ing zones. Each trading zone will develop its financial center for for- 2. The concentration onto a small eign exchange transactions. In number of financial centers is Europe the center is London, in mainly explained by technological the United States New York has factors. The trading technology is acquired this role. In the Asia- characterized by falling average Pacific rim, Tokyo dominates, but costs for increasing volumes of Singapore has achieved a compa- transactions. This leads to a “natu- rably strong position as a secon- ral monopoly” with a consequential dary financial center.6 Secondary concentration of the market world- trading locations in Europe are wide. London, New York and Frankfurt, Zurich, and Paris. They Tokyo have particular locational are likely to cede their business to advantages for historic reasons. the time-specific financial center London has the additional plus London in the longer run, because that its time zone overlaps not only of its cost advantages as a “natu- with one, but also with two other ral monopoly” in wholesale cur- time zones. Singapore and Hong rency trading. Currency transac- Kong have benefited from the tra- tions with non-banks will continue ditional linkage with the British fi- 4 to be effected locally, but they are, nancial center. and will be more and more, settled 3. Given the global 24-hours real- through correspondence banks time trade5, it is almost natural to that are located at the central trad- ing place.

4 This is not meant to diminish the political success of these financial centers’ govern- ments, since the relationship with Great Britain alone may not have been sufficient for devel- opment (compare for instance India as a coun- 6 terexample). The consolidation in Asia is far from being as advanced like in the US and in Europe. Never- 5 Trading is however interrupted by week- theless one may expect similar tendencies to ends/holidays. prevail there in the longer run.

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 26 Chart 4: The development of the number of financial institution reporting to the BIS for selected countries (1992-2001)

Japan

United Kingdom

United States

Switzerland

Germany

4. The concentration tendencies are makers, who will continuously of- not only perceptible geographi- fer bids and asks for the main cur- cally; they can also be identified rency pairs is said to be only 20 by the number of dedicated finan- worldwide (Galati 2001, p. 42). cial institutions. In this regard there equally exists a concentra- Table 4: tion that can be illustrated by the The most active banks in foreign exchange number of institutions reporting to markets, and their market shares in percent the statistics of the BIS. This is (April 2001) depicted in Chart 4. Only Tokyo exhibits a diverging tendency until Citygroup 9.74 1998, which seems to have been broken since, however. Deutsche Bank 9.08 5. Finally there is also an increasing Goldman Sachs 7.09 concentration within the banking JP Morgan 5.22 system. In the year 1998, for in- stance, 75 percent of the trading Chase Manhattan Bank 4.69 volume in the US fell on 20 banks; in the year 2001 there were only Credit Suisse First Boston 4.10 13 actors. The corresponding fig- UBS Warburg 3.55 ures for the United Kingdom are 24 and 17. In Frankfurt only do- State Street Bank & Trust 2.99 mestic banks carry out foreign ex- change transactions; foreign Bank of America 2.99 banks have all transferred their Morgan Stanley 2.87 foreign exchange operations to Dean Witter London. The number of market

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 27 6. The ten most important banks and bank will accept the amount, but their respective market shares in immediately close the ensuing foreign exchange trading are open positions by plowing the cur- shown in Table 4 (according to rency back into the market. This Euromoney, May, 2001). “hot-potato trading” continues until a partner is found who is willing to The consolidation within the bank- hold the position definitely. ing industry has led to a significant reduction in the number of trading 8. Currency trading is not uniformly desks. It can be expected that this distributed over the day. Activities trend will continue in the future. vary substantially in relation to Trading will then further concen- trading of the respective time trate on the main financial centers zones. This can be illustrated by of each time zone. the number of electronic contacts that result in a 24-hours rhythm. 7. It is characteristic of foreign ex- The daily cycle for foreign ex- change markets that they are pre- change activities expressed by the dominantly used by banks/traders. number of electronic contacts is In 2001 their transactions were shown in Chart 5 (according to 86.8 percent of the total. Trading Reuters; see next page). with non-banks (such as import- ers/exporters, direct investors, in- 9. Concentration tendencies are rein- vestment funds, life insurers etc.) forced by synergies that exist be- was comparably small. The struc- tween various segments of the ture of the market in accordance market. They will be discussed in with market partners is repre- the following section. sented in Table 5.

» Instruments of foreign exchange Table 5: markets. The structure of the Transactions in foreign exchange foreign exchange market in accordance markets are carried out by using vari- with groups of trading partners (April 2001) ous instruments. While spot transac- Party Bill. Share tions were predominant only 12 years US $ in ago (they represented 54 percent of percent the total in 1989), they play only a Reporting 689 58.7 much smaller (and diminishing) role traders today (32 percent in the year 2001). The more mi portant instruments are Other reporting 329 28.0 foreign exchange swaps and outright financial institutions forwards. In addition there is a smaller Non-financial 156 13.3 part of foreign exchange and interest institutions rate derivatives that are traded OTC („over the counter“), i.e. on the basis of bilateral contracts. Of this total, 42.5 percent were lo- cal, and 57.5 percent cross- The relationship of the different seg- boarder transactions. ments of foreign exchange markets is represented in the following Table 6.7 One has to realize that an initial

transaction from outside liquidity trading will trigger a large number

of subsequent transactions. For 7 instance if an exporter transfers The data are daily averages for the months his dollar proceeds to a German of April according to the statistics of the BIS bank in exchange for euros, this (BIS 2001).

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 28 Chart 5: The daily cycle of activities in foreign exchange markets according to the average number of electronic contacts (1992-93)

Table 6: one for the purchase, another for The structure of foreign exchange markets the sale (bid/ask). For instance the in accordance with the main instruments bid/ask for euros against US dol- (for the months of April) lars in interbank trading was Bill. US Percentage 1.12108 / 1.12120 on the 14th of dollars shares January 2002 (opening of the market in Tokyo).8 If, on this day, 1998 2001 1998 2001 a bank A in Tokyo has sold an Spot 568 387 38.1 32.0 amount of $100 million to a bank B in Frankfurt in exchange for euros, Forward 128 131 8.5 10.8 the bank B had to pay an amount Swaps 734 656 49.3 54.2 of 112.108 million euros to bank A on the 16th of January. Had bank Total 1,490 1,210 100 100 A bought an amount of $100 mil- *) lion from bank B on this day, it Total 1,400 1,210 would have had to pay an amount of 112.120 million euros. *) At fixed exchange rates (April 2001). Per- Spot transactions are largely stan- centages do not add up to 100 percent due to 9 unreported other financial instruments. dardized and therefore extremely

8 As to specific market segments, the One may note that the spread was only 12 following has to be noted: „pips“ (or about one basis point) in this case. 9 » On spot markets, two currencies There are however variants such as the „pre- spot“ or „ante-spot“, and even „cash“, whereby are exchanged directly. There are settlement is anticipated, and the trade is ef- always two prices for the currency, fected retroactively. Such transactions are however insignificant as to their trading vol-

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 29 liquid. The US dollar/euro ex- York. This can result in significant change market is by far the most risks for the buyer of US dollars. liquid of the world. 11 » For „outright forwards“ the object » The liquidity of specific, more of the trade is the promise to de- rarely traded currency pairs is sig- liver/buy foreign exchange at a nificantly lower than for standard particular date that is agreed upon currency pairs. This finds its ex- in advance. This date can be three pression in a wider spread of the or more business days after the bid/ask (see Appendix 4). There conclusion of the trade. For each are also cases in which the trans- maturity date there exists another action is carried out through a “ve- exchange rate, which will normally hicle currency” in order to benefit deviate from the rate of spot from the higher liquidity of these transactions. Outright forwards are markets and to reduce transac- settled only at maturity, i.e. no tions cost despite of this double payment is made when concluding currency conversion. This „cross- the contract.12 rate trading“ will cause a duplica- » Forwards are used, by non- tion of the number of trades, which financial customers, for the hedg- will compound the spreads ac- ing of exchange rate risks that re- cordingly. Ac cordingly the PFTT late to future financial operations would apply several times in this associated with real economic case. transactions. They can also be » As is clear from this example, spot used for speculation however. trades will normally have to be set- These trades have to be custom- tled two days after the trade.10 ized to the particular needs of a There are always two national set- client as to amount, currency, and tlement systems required. In the maturity of the payment (“customi- previous example, settlement is zation”). The less typical and stan- likely to have been effected dardized the maturity date and/or through Fedwire in the USA, and the currency pair are, the more RTGSPlus in Germany. expensive the transaction will be- come, because the product has to » The settlement of the foreign ex- be generated on less liquid mar- change trade has to be made si- kets. multaneously in both currencies if a settlement risk (“Herstatt risk”) is Financial institutions adjust their to be avoided. Since payment is activities in the forward market to made in different time zones, set- standardized markets as regards tlement will first occur in Asia, then maturity and currency pairs. For in Europe, and finally in the USA. the more important currencies This entails the risk that, for n-i there are standardized markets for stance, euros are paid out in one-, two-, three-, six- and twelve- Frankfurt before the corresponding month monies. Such standardized amount of US dollars has been markets are again more liquid than credited to an account in New for customized products, which entails a smaller spread (lower costs of intermediation). umes. For the pricing of such products, market participants will take the spot rate as a refer- ence point (with conforming surcharges or 11 rebates). Traders use the term „outright“ in order to 10 clarify that it is a simple transactions and not This time is (still) required to confirm the one that is part of a foreign exchange swap. contract and to settle payments. The time is 12 likely to be reduced further through technologi- It is possible however that there is a transfer cal developments. of securities as collateral before payment.

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 30 » There is a close relationship be- there is no need to use a specific tween the spot rate of a currency market for trading directly.13 and its forward rate, which is ef- » In foreign exchange swaps, a cur- fected through the so-called “cov- rency is swapped against another ered interest parity”. If there is a for a certain time, after which the possibility for arbitrage, any larger trade will be reversed. A swap spread between the spot and the consists therefore of two parts that forward rate must, for a given in- result from one operation, but, terest differential, immediately each on its own, are executed at trigger transactions that would re- two different dates although being duce the spread to normality. recorded as one single transac- Such arbitrage operations cannot tion. Often swaps consist of a spot be held to represent speculation. and a forward trade (but also of a If, for instance, the interest rate on combination of two forwards with three-months deposits in euros is different maturities) that go in op- 5 percent p.a., but for three- posite directions. For roughly two months deposits in yen only 2 per- thirds of all swaps based on the cent p.a., and if there were no spot rate, the second part of the premium nor discount for three- trade is executed within a week. months monies in the euro/yen The foreign exchange swap is a forward market, this would trigger standard instrument in OTC trad- a risk-free arbitrage transaction. In ing that consists of an exchange this case, actors would borrow yen and its reversal, whereby it does at 2 percent, convert it into euros, not constitute two converse pay- and realize 5 percent interest for 3 ments. For reasons of taxation this months, after which time the trade operation must however be con- can be reversed risk-free. In this sidered one single transaction. case the arbitrage benefit would » The swap is equivalent to a short- consist in a quarter of the differ- term borrowing arrangement com- ence between the yearly ni terest bined with a simultaneous loan on rates of the two currencies. This is a collateralized basis. This renders the reason why interest-rate dif- swaps an appropriate instrument ferentials are skimmed off, in for- for liquidity management and for ward markets, by a premium or hedging against exchange rate discount relative to the spot rate. risks, but also for speculation. The There may only be minute differ- attractiveness of the swap lies in ences caused by transactions the fact that trader find it often costs. In order to limit these costs, necessary to go into another cur- forward transactions are typically rency temporarily without having effected “off shore” on Eurocur- to bear the risk of an open position rency markets. in the currency that is held tempo- » For later reflections it has to be rarily. This is different in spot and emphasized that—for a given n-i forward markets. In these markets, terest differential—any specula- trading one currency position tion, for instance on forward mar- against another changes the risk. kets, will trigger arbitrage transac- » The pricing of swaps is based on tions that affect also the spot mar- the rates of the two combined in- ket. It is equally important for these considerations that forward contracts for important currency 13 pairs are often arranged “off This aspect also poses a problem for the shore”, i.e. outside the respective capital import tax of Zee (200). It cannot be maintained, as Zee argues, that the CBCT is currency areas. It implies that largely immune against .

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 31 struments. Interest rate differen- tilities, and it is typically employed tials of both instruments are simply for hedging purposes. However exchanged. This is effected options can also be used for through so-called “swap points” on speculation, but only if there are the basis of interests on assets conforming speculative operations and liabilities on offshore markets, in both the spot and forward mar- which allows a transformation into kets at the same time. The latter a spread based on the exchange aim at driving the option “in the rate. money”. It is only then that the holder of the option can benefit » As in the case of forwards, swaps from the operation. If an option is will not trigger immediate transac- not executed, it has to be as- tions. There is only a so-called sumed that it was bought for hedg- compensation of the differential ing purposes only (or that a specu- that has to be secured by capital lation has not worked out!) or collateral. As in the case of for- wards it has to be reckoned that » The pricing of an option is rather swaps (via borrowing in forward complex, but this need not be dis- markets, and the spot sale of cur- cussed here. It is based on the rencies) will also trigger compen- empirical volatility of a currency, sating arbitrage transactions in the i.e. options for currencies with high . fluctuations of the exchange rate are more expensive than those for » In the case of currency options, currencies whose exchange rate is the buyer obtains the right, but not comparably stable. Prices of op- the obligation, to purchase or sell tions reflect therefore the costs of a particular currency at a particular 14 exchange rate volatility in an ex- date at a price that has been plicit and observable form. previously agreed upon. An option is executed only if it is in the inter- » As regards the significance of op- est of the holder of such a right. In tions for a PFTT, it has to be noted this respect the option differs from that there are no direct transac- a forward transactions. The cur- tions associated with the contract, rency option will always involve unless the option is executed later two currencies also. A “put” option, on. However, as in the case of for instance of euros against yen, forwards and swaps, there are in- is equivalent to “call” option, in this direct transactions that are trig- case: of yen against euros. gered by the trade on the trader’s side who will aim at closing open » The bulk of currency options is commitments. For instance, if a traded OTC, i.e. through bilateral trader has sold a call option in eu- contracting between two part- 15 ros against US dollars (= put op- ners. The overwhelming part tion in US dollars against euros) of consists of generic „plain vanilla a notional amount of $20 million at contracts“, which are standardized a strike price that lies at the cur- for the major currencies, amounts rent forward rate („at-the-money and maturities. The instrument forward“), the probability that the works as a sort of truncated insur- value of the option will increase or ance against exchange rate vola- fall is 50 percent. In order to elimi- nate the risk, the trader will imme- 14 diately sell 50 percent of the no- In a few cases the option can also be exe- tional value in dollars on the spot cuted before maturity. market (i.e. $10 million) and buy a 15 A limited number of standardized options is corresponding amount of euros in also traded on the , which is not considered here. the market to be solvent in the

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 32 case of the option being executed. pricing, whereby they indirectly act However the risk that the option is as arbitrageurs by responding to struck will change continuously small changes of the exchange over time. If the euro appreciates rate. against the dollar, the probability » Pricing of liquidity is always based of the option being executed will on information that is based on the increase. The trader will have to actual rate. Traders/banks typi- purchase more euros on the spot cally ignore fundamental data or market. Other factors affecting his charts when trading, because this risk are changes in interest rates would drive them into open posi- and exchange rate volatility. tions. On the contrary: They are These must also be considered usually “blind” vis-à-vis noise trad- when hedging again the risk of the ing, and they react exclusively to option being executed.16 changes in the last digits after the comma, the so-called “pips”. This » The structure of foreign exchange entails the focusing on extremely markets and the development of short trading periods. Trad- exchange rates. ers/banks may however not be immune against rumors in the As regards the influence of the market market, which could eventually structure on the exchange rate, the trigger speculative activities, albeit following has to be noted: not necessarily. » First it has to be reemphasized » The daily liquidity cycles are highly that it would not be reasonable to important for the development of peg the volume of currency trading the exchange rate. If there were to the value of real economic ac- strong fluctuations during phases tivities. The bulk of foreign ex- of diminished liquidity (for instance change transactions is effected to because of speculation), actors procure liquidity. This is not would generally react barely, wait- equivalent to speculation. On the ing until the markets in London contrary: Liquidity renders a spe- and New York will open. The cific and positive contribution to greater liquidity then improves the stabilizing exchange rates, be- conditions for the price discovery cause it creates opportunities for process, and markets usually calm ubiquitous real-time access to for- down rapidly. This behavior of ac- eign exchange markets at a lim- tors illustrates the importance of ited price risk. Liquidity is primarily liquidity for the stabilization of ex- used for hedging purposes, change rates. If a tax on foreign whereby various instruments can exchange transactions would re- be used. duce liquidity more generally, this » Traders/banks do not engage in would lead to greater uncertainty speculation on foreign exchange and ultimately greater exchange markets, because they typically rate volatility. close their “open positions” imme- » Contrary to liquidity traders, actors diately. For liquidity trading they outside the financial industry will are guided by market makers for typically base their decisions on other criteria. This may well lead to speculative behavior and noise 16 This is effected through a parameter „delta“ trading. in the Black-Scholes formula, which is used for pricing an option. The delta measures those » As to pure speculation, i.e. the price changes of an option that result from deliberate acceptance risky of small changes of the underlying exchange open positions, which are often re- rate. This is also called a “delta hedge”.

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 33 inforced through leveraging via 2. On the other hand, there may implicit lending (for instance a for- be an abrupt restructuring of port- eign exchange swap), the so- folios whenever investors loose called hedge funds have a particu- confidence in a particular currency larly bad press. These funds often and reduce the share x radically take significant risks through open (even to 0). One might also call positions, by contrast to their mis- this speculation. I personally re- leading label. The result of such gard it as an attempt to protect, in speculations can of course go in the interest of investors, the value both directions. These funds may of the portfolio against expected however reckon to be on the safe losses. The repercussions of such side if they can expect govern- behavior on the exchange rate ments to intervene in the market may of course hardly be distin- (see also the reflections on specu- guishable from speculative trad- lation in Appendix 3). Fortunately ing. the presence of these funds on in- » As to the assessment of commit- ternational foreign exchange mar- ments by longer-term investors in kets has recently been declining particular currencies, we have to according to the BIS.17 realize that these employ very dif- » The behavior of these and other ferent methods in trading. Gener- actors in foreign exchange mar- ally they will adjust their behavior kets is more decisive on the de- to economic fundamentals and po- velopment of exchange rates than litical risk factors, but when trading the activities of traders/banks. In- shorter-term they would also em- terestingly, two different kinds of ploy mechanical extrapolation reactions can be observed: methods such as chart tec h- niques. This can lead to noise 1. On the one hand, institutions trading and provoke a systematic that follow a longer-term strategy drifting off from equilibrium of the (such as investment funds or life exchange rate. It increases the insurers) would orient their behav- volatility in foreign exchange mar- ior on a fixed grid for their portfolio kets because sooner or later lar- structure, the so-called “gatekeep- ger corrections of the rate become ers”. For instance a fund decides unavoidable.19 that its assets (say, in Brazilian reais) should represent x percent of its portfolio. If there is a de- » Summary and consequences. The valuation of the real, the corre- presentation of the market structure sponding share will fall below the and of instruments used in currency mark, to the effect that assets in exchange markets was determined by real will be increased. This con- two objectives: tributed to a strengthening of the real and countervails the tendency 1. On the one hand, the analysis of devaluation, although—theoreti- leads to further, economically mo- cally—one could classify such be- 18 havior as “speculative”. buyer, while he/she would sell reais whenever their share will go beyond the fixed mark fol- lowing an evaluation. 17 19 Galati (2001, p. 45) argues that this is likely Insofar as the PFTT charges exactly this to be a consequence of the LTCM debacle of type of transactions, it can contribute, at small fall 1998. tax rates, to stabilize exchange rates even 18 without the ERND. The latter would assume an The „speculative character“ of this stabiliz- auxiliary role as a circuit breaker against very ing operation could be seen in the fact that the short-term speculative attacks that the PFTT portfolio manager expects a reevaluation of alone cannot cope with. the real and therefore enters the market as a

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 34 tivated limitations of a PFTT. non-financial, customers of the These will be discussed in the re- real sector of the economy— mainder of this Chapter. exporters/importers and direct n-i vestors, but also portfolio inves- 2. On the other hand, there are con- tors, life insurers, and so on. It is sequences for the technical im- then incompatible to exempt ex- plementation of a PFTT, which will porters/importers and direct inves- be addressed in the following tors from the tax, as is the case for Chapter. a number of proposals of a Tobin Which are these further, economically tax (for instance for the legislation motivated limitations of a PFTT? of the French parliament).20 If the significance of liquidity trading 3. If one further considers that non- for a globalizing economy is recog- financial institutions are responsi- nized, in particular its stabilizing— ble for only 13.3 percent of the because risk-diminishing and cost- trading in foreign exchange mar- reducing—functions, the potentially kets, this would imply a relatively damaging effects of a PFTT on high burden on the real sector due international financial markets cannot to a leverage effect. For a tax rate be ignored. This has significance for of 10 basis points on all transac- the level of taxation. Furthermore it tions, for instance, a burden of 75 points toward some consequences for basis points is calculated on the the distribution of tax revenue. non-financial customers of trad- As to the tax rate, the following seems ers/banks (= 10 / 0.133). The net capital return of a direct investor to be compelling: must fall accordingly.21 1. If one enters a market that oper- ates with a spread of one to three 4. If one accepts the profit margin as basis points (see Appendix 3) with a limitation of the tax rate trying to a tax whose rate is 10 basis points find a compromise by which the fi- (as in the French legislation), one nancial industry would be willing to risks to smash up (or even elimi- bear at least part of the tax, i.e. at tax rates that do not exceed the nate) this market. It is unimagin- able that trades with a (gross) margin, there would still be a fur- profit margin of one basis point ther dilemma: The most liquid would be carried out if this margin markets display the largest tax were taxed with 900 percent of its base, but they operate with the gross benefit. If one wants the tax smallest profit margins. Less liquid to be borne by traders, their profit market operate with greater mar- margin will limit the tax rate. If the gins, but they realize substantially smaller transactions. If this is actual spread is one basis point, and the tax rate is half a basis taken into account and if one point, the presumptive tax on their wants to tax the less liquid mar- gross income would still be as kets more heavily, it follows that high as 50 percent.

2. If one assumes, however, (as the 20 The exemption of transactions that are supporters of higher tax rates do) triggered directly by the exporter/importer and that the tax can be shifted, one the investor is only partial anyway, because it overlooks that the counterparts of cannot indemnify for charges that are com- most of the currency transactions pounded, in a cascading fashion, in the price are again currency traders who of the trade. 21 work with similar profit margins. Of course, this may be intended by those This implies that the tax can ulti- who advocate for an indirect presumptive mately be shifted only onto final, income tax on capital income from interna- tional investments in a globalizing world.

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 35 the PFTT would have to have dif- dance with the counterpart of the ferent rates for different currency trade. If the tax were levied only markets . asymmetrically on the “euro leg” of the trade, as in the case of a uni- Such differentiation is of course lateral introduction of the tax, this not operational. I therefore plead would automatically lead to half for a PFTT with a uniform tax rate the charge as long as the other leg that focuses on the most liquid of the trade remains untaxed.24 market. This may be advanta- geous for smaller markets, but this 6. Moreover there is a discussion on should be accepted by design.22 whether, and which, transactions Whereas, previously, I had pro- should be exempt from tax (or be posed a tax rate of two basis taxed at a zero rate). In this con- points (Spahn 1996), I now plead text one often declares currency for a tax rate of only one half to interventions of the central bank, one basis point. At that order of public transfers (for instance to the magnitude there is no need to IMF, but also official public aid to formally exonerate exporters/im- developing countries, etc.) sacro- porters, for instance through the sanct without further elucidation. I value-added tax, or a fortiori think this would complicate the tax through the income tax as asked unnecessarily, because it would for by Zee (2000).23 lead to a host of exemptions that are difficult to administer and to 5. Kenen (1996, p. 114/15) has ar- control. A transactions tax is by gued that wholesale traders in the nature inappropriate to pursue market would have to be charged public, social, ecological, and half the standard tax rate since it other sensible policy objectives. would be levied on both ends of a Even if a currency transfer to de- trade. Otherwise the burden on veloping countries would be exon- wholesales would be twice as high erated from the tax (as requested than on the clients of the financial by Kenen 1996, p. 115/16), this sector. This is correct only for- would still not remove the accumu- mally. It is to be expected that a lated compound effects of the tax large part of the burden on whole- incorporated in the price through salers would be shifted onto the tax shifting. And finally: A minute non-financial sector anyway, via tax of half or one basis points larger spreads. The statutory rate charged at the final stage is hardly does not say anything about the perceptible by the customer, and it effective incidence of the tax. I should also be carried by govern- therefore believe that such a dis- ments that engage in development tinction is unnecessary, even more aid. so as this would entail the need to differentiate transactions in accor- I have argued earlier (1995) that foreign exchange transfers of cur- 22 rency boards should be exempt It would implicitly relieve the tax law from from the Tobin tax. This argument exempting certain transactions, for instance those involving the currencies of developing is still on firm grounds and should countries (Kenen 1996). Such transactions play a minor role anyway. Currency transac- 24 tions of central banks and international organi- This translated the proposed tax rate of 0.5 zations would also be axedt for operational – 1.0 basis point on one side of the transaction reasons, but they could obtain a tax reim- into a rate of 1.0 – 2.0 basis points in Kenen’s bursement ex post. definition (which he would divide by two for wholesalers). This leaves sufficient room for 23 Huffschmidt (2000) makes a proposal sim i- the American legislature in case it would want lar to Zee’s as to an exoneration of export- to introduce a similar unilateral PFTT on the ers/importers. dollar leg of the trade in the future.

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 36 be the only essential exception. of course its main center London Such interventions can also easily (and the non-EU financial center be identified and therefore the tax Zurich). This implies that the reve- is easy to administer. 25 nue cannot be assigned to na- tional tax authorities, but it falls to 7. As to the tax base, I think that it all cooperating countries in the should include only spot transac- time zone collectively. The tax tions as well as forwards (and the- revenue for Europe must therefore refore, indirectly, swaps) up to a be considered to represent a maturity of one month (or its stan- 26 “pool” whose means are either dardized equivalent ). These transferred to the member states transactions constitute the “rele- through formula-based grants, or vant foreign exchange operations go directly into a “European Fund (RFXO)” mentioned in the previ- 27 for Economic Cooperation and ous Chapter. Development”. Of course there are 8. Because of the concentration of other possibilities to make use of trade in a time-zone specific finan- the funds for “global public goods”. cial center, the PFTT is inappro- The decisive point is that such priate as a national policy instru- funds are multilateral European, ment. It means that the calcula- and not bilateral national. Since I tions of the French parliament, do not trust the success of nego- that tie the revenue to the transac- tiations on formula-based trans- tion volume of its financial center fers28, the only feasible solution is Paris, do not correspond to the likely to be a European fund. reality of markets. The PFTT can 9. Positive factors for the implemen- only be realized within the time tation of the PFTT are the concen- zone for the EU in toto, including tration on one financial center as a „natural monopoly“ as well as the 25 complexity of currency trade and Simplicity of administration is not the main the substantial positive network objective for this restriction (although it should be aimed at also for a PFTT). I think that the externalities that this entails for the feasibility of the tax requires a high degree of various market segments and for- transparency, and any exemption, how ever eign exchange instruments. It ex- well motivated, is likely to lead to political cludes, for all practical purposes, pressures that aim at obtaining special treat- ment. This would be highly counter-productive, that trading desks will be moved to inefficient and probably unfair because it cre- other financial centers following ates loopholes and opens up the door for the unilateral introduction of the corruption. PFTT. London (and its European 26 It could be a market reaction to define stan- financial satellites) exhibits such dard contract (instead of one month, as now) unique advantages as a foreign as one month + one day, for instance, in order exchange center of three overlap- to formally escape the tax. This is why I speak ping time zones that a dislocation of a “standard equivalent to one month”. of trade into other time zones is 27 It is understood that the transition from spot not to be worried about. To estab- to forward transactions is continuous, and it lish a rivaling financial center would not be sensible to tax spot transactions within the European time zone, but alone, while outright forwards of three days would be exempt (see also Kenen 1996). outside the EU and Switzerland Since 75 percent of all transactions are settled within a week however (BIS, 2001), this defini- 28 tion of RFXO would suffice to cover the over- I simply remind the difficulties of coordinat- whelming part of all transactions. Financial ing tax policies in the EU, for instance in the innovations, such as currency options, would context of a clearing system for the value- be taxed indirectly through the corresponding added tax in line with the origin principle, or hedging activities on spot and forward mar- the problems relating to the introduction of a kets. withholding tax on interests.

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 37 would entail prohibitive high costs, complex network of foreign ex- and is totally absurd. It would change trade and move with their mean to transplant a whole net- computers to exotic places such work at once and in one piece. as the Bahamas is utterly ridicu- The dislocation of single terminals lous and entirely grotesque. But is not sufficient (see also Kenen much depends, of course, on how 1996). The very idea that individ- the tax policy is implemented and ual actors of foreign exchange whether it respects the peculiari- markets could break out of the ties of liquidity trading

Paul Bernd Spahn ® [email protected] Zur Durchführbarkeit einer Devisentransaktionssteuer ® Kapitel 3 ® Seite 38 Chapter 4: Implementation How could the PFTT be realized?

In this Chapter, I shall examine the question how a PFTT could eventually be realized. The focus will be on technical aspects that have to be observed when implementing the tax. This centers around the question, which provisions must be made in order to limit potential evasive reactions to the tax.

» Principles of taxation. A consistent approach to taxing foreign exchange transactions requires proper principles of action. For instance one has to decide at which point the tax should be levied—when concluding a contract at the trading desk, when entering the trade into the books of account, or when the trade is finally settled. Moreover it has to be clarified who should be responsible for withholding the tax and to whom the tax is to be paid. These questions require both theoretical and practical considerations. It is also important that the tax be simple to administer, i.e. tax collection should be tailored to the conditions of the market in order to keep the costs of administration as low as possible. The answer to these questions hinges not only on the transaction technology, but also on the legal possibilities to assess potential taxpayers and to enforce probable sanctions. One must keep in mind that the PFTT is a unilaterally levied tax (for instance by the EU). This could offer options for taxpayers to avoid the tax by a single legal action that transfers his/her activity to a place outside the jurisdiction of the PFTT. It is also illusory to count on the willingness of non-taxing governments to cooperate in this matter as long as they can expect to draw benefits from these evasive actions. Peter Kenen (1996) has presented a 1. First, foreign exchange transac- remarkably detailed proposal for the tions are always netted before be- realization of a PFTT. In his paper he ing entered into an official settle- discusses a number of important prin- ment system, i.e. all claims and ciples of taxation that are relevant in obligations in the various curren- this context. cies that materialize over the day will continuously be cleared “in As to the tax object, Kenen raises the house”. Only the net position at question, at which point the tax should the end of the trading day will be be levied: (i) when the dealing is put into an official clearing or set- struck, (ii) when the contract is issued tlement system in order to close and entered into the books of account the open position. at the back office, or (iii) when the trade is settled, i.e. payment is ef- 2. And second, the national settle- fected. He dismisses the book be- ment systems are unable to dis- cause he (rightly) deems that this can tinguish transactions in accor- be kept at any place on earth and the dance with their underlying busi- tax could therefore be avoided.3 How- ness. A payment order to the ever the back office is the place where German settlement system the contract is usually verified and RTGSPlus for instance could be one finally confirmed. It is often also the leg of a foreign exchange transac- place where a proper judicial paper tion (say, the settlement of a liabil- (or electronic) trail will start.4 ity in euros that corresponds to a purchase of yen); yet it could Moreover Kenen discounts the possi- equally correspond to a payment bility to levy the tax at the point of set- in euro for a commercial transac- tlement. He mentions two reasons for tion within the European Monetary this: Union.

3 Bookkeeping of foreign exchange transac- For these reasons, Kenen decides tions is of course made electronically and the that the point of taxation should be the various trading desks are connected with the trading desk. back office through networking. There is a trend toward concentrating all bookkeeping By contrast, Rodney Schmidt (1999, onto one platform (for instance Citibank admin- 2001) has recently argued that the tax isters all its worldwide foreign exchange trades could be levied at the point of settle- in London), but this type of concentration does ment (more precisely: payment), not offer an advantage for levying the tax be- cause bookkeeping could of course equally be which would better correspond to the effected in New York. nature of foreign exchange markets 4 For the OTC business, and in particular for and possess the better perspectives the „open outcry“ by which traders or brokers in the future. He points to a number of handle multiple businesses often simultane- particularities of foreign exchange ously at the telephone, there is a risk of mis- markets and their developments that apprehension that will have to be clarified by aim at discounting Kenen’s argu- the back office through bilateral exchanges of data (and eventually on the basis of telephone ments. tapings). This is the exact point where the tax An examination of the advantages would have to be levied because auditing is possible only from this point on. Kenen, who and disadvantages of both ap- opts for the trading desks as the point of taxa- proaches, different as they are, re- tion, realizes this difficulty and requires that the quires a deeper analysis of the trans- final contract confirmed by the back office and actions technology used in foreign all pertaining documentation be transferred back to the trading desk. He also mentions the exchange markets, which I shall focus possibility that a trading desks could be o-l on in this Chapter. Before doing so I cated in a country whose legislation forbids have to discuss some further taxing this transmission of information, but he over- principles. looks that this could also be true for the back office. As to the definition of the taxable sub-

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 40 ject, i.e. of the persons or companies banks. This is why Kenen opts for the liable to pay the tax (the taxpayer), market principle of the tax. Kenen develops two principles: There is however a further, quite in- » The national principle.5 In this teresting variant of the so-called na- case the head offices of the firms tional principle. One must not neces- are required to collect all data on sarily opt for the legal headquarter of foreign exchange transactions that a firm for taxation purpose; one could are made by their desks world- also opt for the accreditation or licens- wide, and the tax would be levied ing of foreign exchange trading at a on the global transactions of the particular financial center. In this case, firm by the residence country of American banks would become sub- the head office. ject to European tax legislation when 6 taking up a license to carry out foreign » The market principle. In this case exchange transactions, say, in Lon- the tax is levied on foreign ex- don. It would apply to the totality of change transactions where they their foreign exchange transactions occur, and they are paid to the 7 worldwide. I have discussed this op- country in which the trading desk tion in my paper (Spahn 1995) and is located. still think it could be an interesting This would mean concretely that, un- model once a decision is made to levy der the national principle, all British the tax at the trading desk. traders/banks residing in London Of course the modified form of the would have to pay tax for their world- national principle entails significant wide currency operations if the EU problems of law enforcement (as is would introduce such a tax unilater- the case for the income tax with the ally. However American firms would citizens’ principle) because legal obli- not pay the tax even though their for- gations8, effective auditing, and the eign exchange transactions would be persecution of illegal activities are all effected predominantly in London. difficult outside the realm of the tax. Under the market principle, all trading This is an important objection to both in London would be subject to the tax 9 variants of the national principle. For irrespective of the nationality of the this reason, the significance of the trader/bank—whether British, Ameri- market principle in Kenen’s terminol- can, or other. However offshore trad- ogy is certainly more appropriate, ing by British traders/banks (i.e., trad- whereby all traders/banks would be ing outside the EU) would remain tax- liable to pay the tax to the place of free. their accreditation, but only on their It is understood that the national prin- ciple would lead to significant distor- 7 This interpretation of the national principle tions in the market, by which Euro- has also a corollary of income taxation: the pean traders/banks would suffer a citizens’ principle that used by the USA. It serious competitive disadvantage. requires all (“accredited”) US citizens to pay This cannot be accepted by the city of income tax on their worldwide incomes to the American , independent from London (not by the EU for that mat- the fact whether they reside in the US, or not. ter). It would mean that foreign ex- 8 On the question of legal obligations see be- change trading in the EU would exclu- low. sively be carried out by non-European 9 Kenen mentions further objections against the national principle, for instance that financial 5 In public finance one also speaks of the institutions would have to carry an undue ad- „residence principle” (for instance for the in- ministrative burden when having to report their come tax). worldwide activities. This argument is not con- vincing. Large corporations (such as Citicorp) 6 This principle is called „source principle“ in record their worldwide foreign exchange op- the case of the income tax. erations centrally even now.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 41 local transactions. „gross trading in these claims“ (i.e. foreign exchange trading in the Both principles focus on the trading spot market) „will be effectively desk however. If one follows the eliminated in favor of T-bill swaps proposition of Rodney Schmidt and in currencies with liquid (same-day) levies the tax at the point of settle- T-bill markets. The swapped T-bills ment, one would probably have to will be immediately sold for depos- define a further taxing principle, which its” (p. 179). I would call “rule of access”. Under this principle all institutions would be- In concrete terms this would mean come liable that make use of an offi- that traders would no longer trade in cial settlement system. The obligation central bank monies, but would start to pay the tax follows from access to transacting predominantly or exclu- official systems (such as RTGSPlus or sively in short-term public securities. TARGET) and will be linked to the Alternatively they could use money netting operations that precede set- market funds and mutual funds. tlement (net clearing, automated bro- Trading with short-term treasury bills kerage systems) through contractual or money market certificates is indeed “chaining”. How this could operate will common, but such instruments must be taken up further below. still be regarded to represent an aliud If the tax were levied at the desk ac- compared to spot transactions of for- cording to the market principle, finan- eign exchange. Securities are not only cial institutions would have an incen- subject to the exchange-rate volatility tive to migrate to tax havens, accord- of their respective currency, but they ing to Kenen, but the governments of also bear a market risk, even a sover- these countries would have no incen- eign risk. Central bank money serves tive to create such havens because exclusively to hedge against immedi- refraining from taxing would not rep- ate exchange-rate risks. As long as resent an option to provide competi- this is the case, and the maturities of tive advantages to their own compa- potential substitutes in the form of nies.10 If there were still tendencies of securities cannot be fully synchro- dislocation, Kenen hopes to contain nized with spot trading (which is diffi- these through certain hindrances cult to imagine), trading in securities (such as a penalty tax on transactions must always be more expensive than with off-shore financial centers). This spot transactions. point will be taken up later. This is partly explained by the fact that Dislocation of desks is not the only there could be problems for price set- strategy to circumvent the PFTT. It ting of such securities, which renders may also be achieved through un- additional hedging transactions nec- taxed financial instruments. For in- essary.11 Moreover such swaps al- stance Garber and Taylor (1995) have ways require several consecutive pointed out that transactions: a money transaction when purchasing the security/money market fund, and another when selling it. Only the intermittent swap of securi- ties would go tax-free. This is why securities may be used as collateral in foreign exchange trading, but not as a primary mean of transactions. If can be expected that such trading 10 It should however be mentioned that there would not survive in view of further are also indirect advantages (e.g., creation of jobs) that could motivate countries to attract such firms through . 11 Se also Kenen (1996, p. 119).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 42 developments in the trading technol- Kenen illustrates the structural rigidi- ogy, if they would ever come to bear. ties of networks by referring to a They would simply be too expensive. trader who decides—first on his/her own—to dislocate his/her trading Nevertheless the argument of Garber desk. He/she would then have to do and Taylor has to be taken seriously, all dealings with actors of the former in particular with regard to the proba- trading place, which alone entails sig- bly more important money market nificant additional costs. Moreover funds (not mentioned by Gar- Kenen proposes to charge all transac- ber/Taylor). Should these strategies tions with a penalty rate (also pro- penetrate the market—contrary to my posed by Spahn 1995), for instance expectations—, it is always possible 500 basis points instead of his regular to bring them into tax net by a legisla- 2.5 basis points. However he argues tion that would qualify such operations against the background of a tax that is to be abusive. This is also possible as levied universally at the ten most im- to further financial innovations that portant foreign exchange centers in could surface as surrogates of foreign 12 the world. These centers could en- exchange transactions. force the tax collectively through a penalty rate. » Measures against potential dislo- The example is not very helpful for a cations of trading desks. unilateral imposition of a PFTT, say, I share Kenen’s point of view that the by the EU. It is simply unacceptable risk of a dislocation of trading desks is that transactions of European trading centers with centers such as New small, although I build this on different York and Tokyo would be charged arguments. As has become clear from 14 Chapter 3 I have come to believe— with “penalty rates”. This is however notably after discussions with repre- totally unnecessary because any dis- sentatives from the financial indus- location of a trading desk entails sig- try—that the dislocation of trading nificant costs. The costs depend on desks entails prohibitively high costs. the target country for an eventual dis- As discussed before the comparative advantages of London as a financial office rents, high salaries for financial profes- center and a regional “natural monop- sionals, and—in another segment of the mar- oly” for foreign exchange trading are ket—in spite of a tax on the transactions of so enormous compared to all other stocks. This advantage is even more pro- financial centers, including those out- nounced for foreign exchange transactions. side its time zone, that a PFTT with a 14 By contrast Huffschmidt (2001) is in favor of tiny rate could constitute only a negli- such penalties even in these instance. His gible cost factor. 13 argument: The euro does not constitute a mean of payment outside the EMU, so any transaction to places outside the EU must be considered to be speculative. („Da entweder 12 It may be useful to remind at this point that Steuerhinterziehung oder Finanzspekulation even the income concept is all but simple. If die wesentlichen, wenn nicht die einzigen this tax was conceived in rather simplistic Gründe für derartige Transaktionen von Euro terms initially, it has become one of the most hier zu Euro im Ausland sind—denn im Aus- complicated of all taxes today. This results land sind Euro weder als Recheneinheit noch from the fact that the legislature had to imple- als Zahlungsmittel noch als Wertaufbewah- ment rules to contain strategies rungsmittel nützlich—sollte die Steuer bei der of all sorts over time. It is almost certain that Überweisung von Euro in die USA ansetzen, the PFTT will also further tax avoidance was erfassungstechnisch keine Probleme strategies, which—according to my percep- bereitet.“). He therefore stipulates the Tobin tion—would be much easier to cope with than tax to be levied in Europe as an “exit tax” in the case of the comprehensive global in- whenever the euro leaves its constituency. come tax. This position is totally at odds with the nature 13 of liquidity trading and deserves no further One may also consider that London domi- comment. nates as a financial center in spite of high

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 43 location. It is of decisive whether the zone specific monopoly rents only target place lies inside or outside the conjointly. If they would try to increase time zone. their respective market shares through rivaling with each other via » If one assumes the dislocation to “tax competition”, this would mean the take place within the time zone, failure of a PFTT for Europe. namely out of the EU (plus Swit- zerland)15 to another European lo- cation (such as Andorra or War- » The PFTT as a „payments tax“. saw), it should be apparent how limited realistic options of disloca- The considerations of this Chapter tion are under this restriction. were so far contingent on the PFTT Moreover many non-EU members being levied at the trading desk. This in the time zone (such as Poland) requires a sophisticated reporting sys- are interested in joining the EU in tem, which is further complicated by the future and drop therefore out the fact that the desks have nothing to as potential competitors. This form do with the settlement of the trade and of dislocation can totally be disre- that the proper auditable paper or garded. electronic trail often begins in the back office. » If one assumes however that the target place is outside the time In contrast to this, Rodney Schmidt zone (for instance the Bahamas (1999, 2001) has pointed to the inter- or, more likely, one of the existing esting possibility, rejected by Kenen foreign exchange centers), this (1996), that the tax could be levied at would of course be feasible in the “end of the chain”: at the point of technical terms (and certainly be payment or settlement. Since pay- less onerous than starting up a ments and settlement systems are new trading place from scratch), highly concentrated —unlike the but it would also mean to forego decentralized trading desks—, this all time-zone specific and other method of assessing foreign ex- advantages that distinguish Lon- change transactions would conform don and its European financial much better to the conditions of elec- satellites. Therefore this option is tronic markets and they would be eas- also likely to play a negligible role ier to administer. Tax collection at the in foreign exchange trading, pro- point of settlement could be largely vided however that the PFTT op- made automatic and an extensive and erates with a very moderate rate. sophisticated reporting system would not be required. These considerations demonstrate that a trade-off has to be struck main- Every country has normally its own taining the net advantages of trading payment and settlement systems for foreign exchange in London and its financial transactions of all kinds. European branches in spite of the These systems possess different insti- PFTT. They also emphasize the ne- tutional, legal, and technical compo- cessity of a coordinated fiscal ap- nents. proach of all European financial cen- In the following I shall focus on cas h- ters. These can preserve their time- less payment systems (that dominate 16 foreign exchange markets) and I 15 It is important to include Switzerland as an important financial center outside the EU, but within the European time zone, because of 16 I do not think it to be necessary to tax retail potentially unfair competition. If Switzerland transactions of foreign exchange, for instance would be uncooperative in this matter, the EU currency exchange by tourists. Such transac- could of course consider to use “penalty rates” tions are insignificant compared to the transac- when trading with Zurich. tions in wholesale markets.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 44 define a “payment” as the instruction transfers play a particular and increas- to transfer a money amount on the ing role. Whereas electronic trading basis of a legal obligation, and “set- accounts only for 0.1 percent of the tlement” the definitive and irrevocable number of all financial transactions in transfer of the money that is the pur- the USA (including domestic settle- pose of this payment. For instance if ments)18, their share of the total vol- someone pays with a check for a pur- ume is more than 80 percent (Cross chase, payment is effected when the 1998). Electronic settlement systems check is handed out. However settle- play a decisive role for trades be- ment is effected when the check is tween banks, foreign exchange trad- cleared and cashed in, or credited to ers, and institutional investors. the account of the recipient.17 In the USA there are two competing For the settlement of financial transac- payments and settlement systems: tions of all kinds, there are different CHIPS (Clearing House Interbank

BOX 1: CHIPS AS AN EXAMPLE OF BOX 2: FEDWIRE AS AN EXAMPLE OF A CLEARING SYSTEM AN RTGS-SYSTEMS In contrast to Fedwire, settlement on ac- If a payment is effected through Fedwird, a counts of the central bank system is ef- regional Federal Reserve Bank will debit fected only once a day under CHIPS. Dur- the central bank account of the sending ing the day, all claims and obligations of bank, and will credit the amount to the the participating financial institutions that recipient bank. In this way it is assured that result from business activities are adminis- there is an immediate and simultaneous tered by CHIPS and continuously cleared transfer of central bank money (assets in through netting. At the end of the day the the central bank system). A Fedwire- net balance will be settled collectively for payment is settled through the crediting of all participants through Fedwire (in the the amount to the central bank account of form of central bank money). The central the recipient bank. In order to limit risks, bank accounts of CHIPS participants with the Federal Reserve will charge for over- net claims will be credited, and the ac- drafts during the business day if these go counts of participants with obligations be beyond a certain predetermined limit. debited. If a participating bank with net obligations against CHIPS transfers central bank money to its customer before the end Payments System), a private settle- of the day, it assumes a settlement risk, ment system of the New York Clear- because CHIPS could eventually refuse to ing House; and Fedwire, a service settle for some reason. In order to cover provided by the American central this risk, the current deficits of each partic i- banking system. In the United King- pant are limited (net debit caps) and the dom, the pound-sterling leg of a for- participants are also obliged to participate eign exchange transaction is usually collectively in covering any possible losses settled through CHAPS (Clearing that have to be collateralized with capital or House Association Payments Sys- securities. tem). The functioning of a clearing system is explained in Box 1.19 options: cash, automated clearing, In the Federal Republic of Germany, electronic transfers (in particular for the settlement of payments is effected interbank trading). Each of these sys- tems has its own rules. Electronic 18 The larger part consists of cash, checks, and credit card payments. 17 If cash is used, payment and settlement are 19 The exposition of the two Boxes is based on of course simultaneous actions. Cross 1998.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 45 mainly through RTGSPlus. The short- success or failure of a particular hand RTGS stands for „real-time transaction. Rapid access to central gross settlements”. The real-time set- bank money through national RTGS tlement of each single gross transac- systems is crucial in this context, but tion eliminates the settlement risk. also the existing clearing systems The functioning of an RTGS system is help to limit this settlement risk. described in Box 2 for Fedwire as an In order to facilitate the automated example. access to central settlement systems, The introduction of the euro in 12 the interface is largely standardized. countries of the European Monetary This does not only apply to TARGET Union has not changed the structure (the link between the national systems of national payments and settlement in the European Monetary Union), but systems significantly. Every country also to the access through automated settles its payments through an indi- private brokerage systems (such as vidual RTGS system. However the FXNET). interface between the different RTGS Until the mid-1990s, the private bro- systems has been standardized in kerage systems functioned mainly order to facilitate cross-border pay- OTC via direct broking per telephone ments between member countries of („open outcry“). They have since been the monetary union. This interface, largely automated and consolidated. the “link” (but occasionally also the One speaks of „automated order- totality of the national systems, includ- matching systems“. The most impor- ing the link) is called TARGET (Trans- tant ones are the Electronic Broke- European Automated Real-time Gross 20 21 rage System EBS and Reuters . settlement Express Transfer system ). Today roughly 50 to 70 percent of all In contrast to RTGS, the clearing sys- foreign exchange transactions are tems (such as CHIPS or CHAPS) cleared and settled through these function as netting systems between systems. In 1998 the share had been the participants of the clearing system 40 percent, and in 1995 only 10 per- before settlement through central cent (Galati 2001, p. 43). banks. All individual positions are These developments have contributed cleared (i.e. “netted”), which are then to reducing the volume of foreign ex- settled through central bank money change trading significantly. This is once a day. This does not exclude the partly explained by improvements in settlement risk, in contrast to the the price-discovery process. The RTGS systems of central banks. trader can observe the bid/ask on the In the case of a foreign exchange screen and react quickly electronically transaction, there are two simultane- ous transfers in opposite directions. It always implies the use of two settle- 20 EBS was founded in 1993 by 12 of the lar- ment systems of two countries. This ges trading companies and it is now the lead- entails an additional settlement risk if ing provider of electronic brokerage services. distinct, i.e. non-integrated, payment At the same time it is licensed to provide bilat- eral netting through FXNET (see below). and settlement systems are used— 21 especially if they operate in different Reuters offers a number of information pro- ducts and also manages the clearing and time zones. In this case one also settlement of transactions with financial in- speaks of a “Herstatt risk”—according struments such as foreign exchange, money to a spectacular case of this kind in market funds, stocks, etc. The electronic con- the 1970s. versations on foreign exchange markets run through the platform „Dealing 3000-Direct“. Foreign exchange markets have to Reuters is also a participant of consortia such rely on payment and clearing sys- as ATRIAX (of Citibank) or FXALL, which have tems, which ultimately decide on the developed separate platforms for foreign ex- change dealings.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 46 on a mouse click. Misapprehension is may note however that such systems significantly reduced compared to the are primarily service providers of the open outcry, and back offices are software industry. They are not trad- automatically informed. The auto- ers/banks themselves, and are there- mated order-matching systems are fore not subject to public banking su- therefore extremely reliable and fast. pervision.25 In addition the expansion of such sys- Apart from bilateral netting, there were tems has reduced the possibility of also systems specializing in multilat- traders/banks to engage in so-called eral clearing among traders/banks „leveraged trading“ (Galati 2001, p. (ECHO or Multinet). These have been 44). This entails a reduction of the acquired (and temporarily deacti- volume of interbank trading that is vated) by a real-time-PVP system that triggered by a primary transaction of a is being developed at present, and is final customer of the non-financial likely to start operating in the fall of sector.22 this year: CLS (Continuous Link Set- tlement). I shall come back to this Traders/banks employ a number of path-breaking development on foreign typical netting (or clearing) systems exchange markets later on. before settlement. Foreign exchange accounts At present the routing from the indi- vidual trading desk to the central » are first settled “in house” on a RTGSPlus of the Bundesbank can be continuous basis, i.e. Deutsche schematically described as follows: Bank, for instance, would clear all claims that develop on foreign ex- Bank A in Germany and bank B in the change accounts of its customers USA exchange information on a trade with corresponding customer obli- through SWIFTNet Services and set- gations („in-house clearing“); and tle the payment through a standard- ized SWIFT platform that has direct » are also cleared bilaterally among access to RTGSPlus. Both banks main- trader banks. For instance tain accounts with the Deutsche Deutsche Bank maintains a for- Bundesbank in this case. If this is not eign exchange account with Citi- true for the foreign bank B, the trade group (and reciprocally), where will be settled through a third bank mutual claims and obligations are (correspondence bank) that maintains continuously cleared during the an account with the Bundesbank and day. These operations avoid settlement risks via „payment versus payment Today it is part of the EBS Dealing Resources (PVP)“ through continuous clearing. of Citicorp. Bilateral net clearing is provided for 24 23 24 SWIFT (Society for Worldwide Interbank instance by FXNET or SWIFT . On Financial Telecommunication) is a platform for the electronic exchange of financial data. It 22 started its operations in 1977, and it counts on Galati (2001, p. 44) develops the following more than 7,000 customers in 192 countries example: Assume that a non-financial cus- with some 1.3 bill. messages per year. SWIFT tomer asks his trader bank to sell $100 million has substantially contributed to the standardi- against yen. If the trader expects the dollar to zation of the clearing and settlement process. fall, he would probably sell more than $100 million hoping to buy back the excess balance 25 There is a clear-cut trend to engage soft- as a . This triggers transactions ware firms for complex information processing that exceed the original amount of $100 mil- services. One may mention in this context lion. Electronic order-matching systems handle COGNOTEC (another brokerage technology) such transactions in a neutral and non- or Currenex (an internet-based trading sys- speculative way, which does not entail an tem). There is of course competition among extension of the trading volume. these firms, but recently a clear trend towards 23 cooperation and even consolidation of the FXNET is a consortium of 13 of the largest sector is noticeable. trader banks; it started its operations in 1987.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 47 can clear a claim she has against It is much more difficult to cover the bank B (for example through SWIFT internal and bilateral clearing opera- via bilateral netting). tions within and among financial insti- tutions. Clearing and net settlement This illustrates that the official national among financial institutions and sys- RTGS systems represent something tems has become a conventional pro- like the “kernel” of a payment system cedure worldwide. In particular in- that is based on central bank money. house clearing is something difficult to Service providers such as electronic monitor. order-matching systems will gather around this core. However RTGS However these problems will be sub- sees only the “tip of the iceberg” in stantially mitigated by more recent foreign exchange trading because developments in foreign exchange traders will have netted, before enter- markets, which entail a further con- ing the settlement system, all possible centration of bilateral and multilateral in-house and bilateral transactions clearing before official settlement. through clearing in order to reduce This is because, in spite of a high de- costs and risks. gree of integration of payment and Moreover the settlement sys- RTGS even does tems, there is a re- not „know“ which sidual settlement kind of underlying risk for national, trade is settled but notably for when it accepts international pay- and executes a ments. This has payment instruc- motivated several tion. Payments in banks to develop a European curren- worldwide real- cy that result from time gross settle- a foreign ex- ment system as a change trade are private initiative: settled without the CLS bank that distinction toge- had been men- ther with all other 26 tioned before. It is payments. likely to determine future develop- The latter could of course be easily ments on foreign exchange markets, remedied, for instance through a sim- which render it necessary to offer ple identifier (0 or 1) that indicated some further explanations. whether the purpose is to settle a do- The CLS Bank is a foundation of a mestic or a foreign exchange trade. consortium of traders/banks with its There could also be an obligation to headquarters in London.28 Its objec- hold two accounts at the central bank tive is to eliminate the settlement risks and settle in accordance with the na- 27 for bilateral and multilateral clearing ture of the trade. operations in foreign exchange mar-

26 From the standard (MT 202) used for central would be settled. This would mean to define bank clearing, the purpose of a payment can- the tax base in a narrower sense. I have ar- not be identified. Not even reference numbers gued before that this could be accepted insofar are mandatory, and if they are used they vary as other foreign exchange transactions, such among banks. as swaps and options, also trigger spot trans- 27 The consequence would of course be that actions indirectly through arbitraging. different kinds of foreign exchange transac- 28 The bank is subject to American law how- tions cannot be distinguished at the time of ever. settlement. Only resulting spot transactions

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 48 kets altogether. This will be effected cant shortening of settlement delays, through a continuous and simultane- which will allow interest periods below ous crediting of “both legs” of a for- the 24-hours limit, which would elimi- eign exchange trade on the foreign nate such interest-free swings. For- exchange accounts of the trading ward transactions could automatically banks. The currency pair is settled be activated at maturity, etc. These gross, i.e. there is no netting within will all be significant advantages of the system. Only the balance will be CLS to attract business on a global settled officially for each currency scale. through RTGS systems of central Once the consolidation process of banks. international foreign exchange mar- Moreover each trade is identifiable kets will have come to an end, one according to the purpose of the trans- can again expect a widening of the action, in particular as to the curren- volume of transactions. This hinges cies used, the maturity of the pay- on the cutting of maturities to hourly ment, and the kind of trade. This al- fractions and on the extended use of lows a judicial tailoring of the PFTT gross payments. These trends are and the tax collection process to its promoted by centralized platforms precise tax base. such as CLS. The operation of CLS was initially The centralization of payment and intended to start last year already, but settlement systems as well as elec- technical problems and problems of tronic order-matching systems will project management have delayed the also facilitate the administration of an start until the fall of 2002. Representa- eventual PFTT: The tax could be de- tives of the bank expect the advan- termined through appropriate tax tages of the system to attract about 80 modules embedded in the transac- percent of all foreign exchange trans- tions software, and the proceeds actions in five years. Member banks could be levied automatically and would be able to make use of the sys- transferred to central banks through tem directly; others would access the RGTS. The central banks would thus system indirectly through correspon- become the collecting agencies for dence banks. the tax revenue. I believe that the concentration ten- Although these trends appear to be dency in clearing and settlement as compelling, a comprehensive cover- well as in automated order-matching age of the tax is still fraught with prob- systems will further escalate. This is lems: not only explained by the advantages » Despite these tendencies there of real-time gross settlement systems, will remain a substantial number of but also by some other problems of in-house clearing operations that international liquidity management call for a separate reporting sys- that could eventually be solved tem. This is facilitated by the con- through CLS and other centralized centration within the banking in- systems. For ni stance CLS provides dustry. Only the largest institutions automatic cost-reducing „self-collate- would have to report for taxing ralizing overdrafts“. This requires a purposes. Smaller institutions play sophisticated tracking system. I think almost no role as to their internal further of strategies by which interest- clearing potentials. The reporting free swing arrangements within mu- requirement should hinge on a tual daily liquidity trading are now be- minimum transactions volume, ing abused, which has led to a unilat- whereby a large proportion of the eral encumbrance of some (mainly small institutions would not have continental-European) financial institu- to report, or pay tax, at all. They tions. CLS is likely to lead to a signifi-

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 49 would be charged indirectly how- If transactions are not carried out ever when using correspondence through official or centralized payment banks or centralized settlement systems and are therefore impossible systems. to be taxed at well-defined points at payment/settlement, one has to oper- » Bilateral clearing between institu- ate with a reporting system for such tions that do not participate in cen- transactions. As said before this is tralized systems would also have more costly and creates an incentive to be subject to reporting. This is to participate in the less costly (and more costly than an automated risk-diminishing) CLS system either assessment through computerized directly or through correspondence systems, which creates some n-i banks, whereby the “rule of access” centives to make hook onto cen- would apply for taxation. tralized systems in spite of the tax. One should not underestimate foreign As has been discussed before in con- exchange transactions within the non- nection with the advantages of a fi- financial sector however.30 For in- nancial center such as London, eva- stance Volkswagen or Daimler- sive reactions of centralized settle- Chrysler maintain important foreign ment or of electronic order-matching exchange departments for internal and brokerage systems are highly clearing. These transactions would unlikely. The concentration onto a also have to fall under the tax law in small number of large corporations order to avoid the dislocation of for- renders this extremely difficult for eign exchange trading into the pro- them. Deutsche Bank, for instance, is 31 ductive sector. unlikely to fraud on the tax by carrying out illicit foreign exchange transac- For production companies—as for tions intentionally. But there are of financial institutions that do not hook course incentives to consider ways to onto centralized RTGS systems— avoid the tax legitimately. This is most taxation must be based on reporting.32 prominent for in-house clearing opera- However there is a particular problem tions, because this could lead to legal in that the market principle is more constructions by which clearing is difficult to apply for those companies sourced out to software firms residing than for financial institutions (that are outside the EU. These may not even licensed at a certain financial center). be banks subject to banking supervi- On could however ask for a similar sion. If a different method for taxing such non-American counterparts worldwide through transactions is chosen, for instance at contracting. A recently failed attempt by the German federal government to extend, in the the trading desk as proposed by interest of consumers , the information re- Kenen, one has to ascertain that there quirements onto suppliers outside its jurisdic- is no due to a mix of tions does not augur well for this approach assessment methods. I therefore pro- however. pose to use the same tax object for in- 30 Internal foreign exchange transactions of house clearing operations as for all production firms are not included in the statis- other foreign exchange transactions at tics of the BIS. The tax base is therefore likely to be higher in practice than assumed by most settlement, and define those clearing authors. operations legally to be equivalent to 31 settlement.29 The clarification of this issue is also required in view of the fact that automated brokerage and settlement systems are usually operations of non-banks, i.e. software companies. 29 I think in particular of the attempt of the 32 This could of course be waived if the firm American legislature (although highly contro- decides to participate in a generally accessible versial politically and legally) to oblige Ameri- centralized payment system and be subject to can firms through the Helms-Burton Act to taxation according to the „rule of access“. extend the trade embargo of to their

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 50 accreditation for carrying out foreign settlement system will facilitate tax exchange transactions by produces assessment and collection at pay- as for traders/banks. However produc- ment/settlement considerably. This is tion firms could dislocate their foreign why I prefer the PFTT in the form of a exchange transactions into non-taxing “payments tax”. In this case taxation jurisdictions. One must realize that it could be tied to the access to official is much more difficult to follow these settlement systems, with a contractual firms than financial institutions. “chaining” of the taxing obligation onto in-house clearing systems. The cen-

tral banks would collect the tax, which » Summary. is however to be “pooled” Europe- Questions relating to the implementa- wide for a common purpose. tion of a PFTT are non-trivial. First One may expect however that a more one has to fix general taxing principles comprehensive reporting requirement that define the taxable object and the is still needed for those institutions taxpayer. Thereby one can generally that do not participate in a centralized rely on the market principle, which payments system , nor have voluntarily would cover all traders/banks accred- accepted tax liability through “chain- ited at European financial centers ing”. If both assessment and collection (including Switzerland) as well as cen- techniques are employed simultane- tralized automated order-matching ously there could be some double- and settlement systems. The same taxation, which should be accepted should apply to producers such as however in view of the fact that it pro- Volkswagen or Daimler-Chrysler. vides an incentive to join one of the For tax collection there are in principle official payment systems. two possibilities: At the trading desk, The often-emphasized evasion reac- and at the point of settlement. Both tions to a PFTT are sternly exagger- procedures are technically feasible, ated. The high concentration of for- but each has its own advantages and eign exchange trading clearly runs disadvantages. counter the possibility to avoid the tax, and this trend will be reinforced even Assessing the tax at the trading desk entails a reporting requirement that further in the near future. I therefore does not conform with the nature of think the PFTT to be technically feasi- the market. Automatic assessment at ble—albeit under restrictive precondi- centralized clearing and settlement tions as to the tax rate in order limit points would be more appropriate, but economic distortions. a differentiated registering of individ- The real problems are not to be found ual taxing purposes (suc h as swaps, in the area of technology. The true options etc.) would be impossible be- nature of these problems could best cause the relevant information is not be portrayed by a quotation found in a handed down to the settlement stage. paper by Griffith-Jones (1996, p. 148), It implies that only spot transactions even though it is from an earlier paper could be taxed at present. This is of mine: likely to change in the near future „Generally speaking, there do not seem however—through new technologies to be major administrative problems as- that are being developed and will lead sociated with the operation of a Tobin to continuous gross settlement on a tax, although specific difficulties may PVP basis. arise in detail, in particular for the deriva- The further concentration of foreign tive markets. The main riddle relates to international cooperation and legal en- exchange markets and in particular forcement.” (Spahn 1995). the introduction of a continuous gross

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 4 ® Page 51 Chapter 5: Reaction How could the PFTT affect market behavior?

In this concluding Chapter some considerations are given to the possible reactions of actors in foreign ex- change markets in response to a PFTT. I shall also ven- ture a rough estimate on the possible revenue of a PFTT.

» Behavioral reactions. It ca be assumed that a PFTT would affect the various segments of the market in a very different way. On the one hand there are the wholesalers whose costs haven been falling dramatically as a consequence of technological developments, although they remain in fierce competition among each other at a global scale, and they rely on high transactions volumes to remain profitable in view of minute profit margins, despite of cost decreases.1 A tax on foreign exchange transac- tions must increase the bid-ask spread of traders and reduce the volume of trading. At the same time the av- erage maturity of foreign exchange transactions would increase in length. This is explained by a relatively strong decline of spot transactions relative to outright forwards. Both effects are intended by the PFTT.

1 „A more controversial feature of the new shape of the financial system is that the bulk of its participants now have a vested in- terest in instability because the advent of high-technology deal- ing rooms has raised the level of fixed costs. High fixed costs imply that high turnover is needed for profitability. But high turn- over tends to occur only when markets are volatile” (Walmsley, Julian (1988) The New Financial Instruments: An Investor’s Guide. New York: John Wiley & Sons; quoted according to Felix and Sau (1996, p. 231). The effects of a Tobin tax on net prof- tail transactions through commercial its and trading volumes of wholesalers banks or credit card companies for can be illustrated by a simple model tourism or for cross-border transac- calculation. This is represented in tions of private households. Chart 7.2 In the ambit of final customers and in But there is also trade with various particular in the retail segment of the final customers: exporters and import- business traders operate with signifi- ers, direct investors or portfolio inves- cant margins. These margins are the tors (such as hedge funds, investment higher, the lower the volume of trans- funds, insurance companies) and actions, the lower the liquidity in the other institutional investors. Also the particular segment of the market, the government and central banks will less price-elastic local demand, and take part as actors on foreign ex- the higher the degree of information change markets. Finally, a small pro- asymmetry that warrants some mo- portion of the trade is executed in re- nopoly rents. If the effect of the tax will be a widen- 2 ing of the spreads and a reduction of These calculations were made using the the volume in wholesale trading, as I Model described in Appendix 2. For the present Chart, the following (not implausi- would expect, this would affect differ- ble, but freely chosen) parameters were ent market participants in the following used that do not have any empirical rele- way (see also Felix and Sau 1996, p. vance: a 0 0.5; b = 0.95; g = 0.0005; t = 230ff.): 0.0001. The Chart has been normalized in » At the „shortest end“ of the mar- such a way that, for the maximum net profit margin before tax, the transactions ket, i.e. in particular for „covered- volume for the year 2001 is reproduced interest-rate arbitraging, transac- (in bill. $ per day). One could use the tions costs have always created model to calculate the tax elasticity even- something like a “neutral zone” tually, i.e. the reduction of the trading within which there are no profits volume in response to the tax rate. For from arbitrage and transactions this set of parameters it would be 77 per- will therefore not take place at all. cent for one basis point. I think the tax This zone would undoubtedly be elasticity to be significantly lower however widened by the transactions tax. (see below).

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 5 ® Page 53 On the dollar-euro market, the » Also risky trading of currencies of neutral zone represents now some developing countries will be af- 10 to 15 “pips”, or roughly one ba- fected by the tax. However in this sis point. A PFTT of one basis market there are often no mature point would extend the neutral forward markets that would allow zone by two thirds to 100 percent. covered-interest parity trading. For that reason this type of trade Therefore the risks are already would be hit particularly hard. For- higher in the primary trade than in eign currency traders would find it the previous case, which is of more difficult to pass on their “hot course expressed through higher potatoes” onto other traders with a spreads. A uniform PFTT whose small risk once they have taken tax rate is tailored to the most liq- them up. If a customer insists on uid market (for instance one to two selling/buying foreign currency in basis points) would then be a spite of these increased costs, the comparably small additional trader will ask for a higher pre- charge in relation to the already mium in compensation for the high margins of the primary bus i- higher transaction risk. He/she will ness. It is unlikely to lead to a sig- (have to) shift the tax burden to a nificant contraction of trade. The large degree if he/she wants to tax burden on developing coun- continue operating profitably. tries is therefore relatively small. Of course the volume of trading in » This statement has to be put in this market segment is almost in- perspective however. It must be significant in the global context of reminded that transactions costs currency trading, albeit not neces- in foreign exchange trading have sarily for the respective countries been falling significantly over the themselves. last years. Today the spreads in the dollar-euro market are in the » Interestingly, a similar argument order of one basis point; at the also applies to currency trading by time of my first study in 1995 the the so-called hedge funds that en- usual margins were about 4-5 ba- gage in highly speculative and sis points. Even though the spread risky businesses. Because the would be widened by 100 percent risks of such trading at uncovered through a PFTT, it would still be at interest parity (i.e., the deliberate least 50 percent below the mar- exposure to risks through “open gins of six years ago. Then the positions”), typically even lever- daily transactions volume was aged by borrowing, are extremely roughly one billion US dollars— high, this market segment must only 17 percent below the volume operate with large margins. This is of the year 2001.3 This is an ar- why a transactions tax with a rela- gument against the contention that tively small rate adding to an al- even a small PFTT would wreak ready large neutral zone for trad- havoc in world financial markets, ing is unlikely to exhibit deterring and the negative impact of the tax effects. on the volume of liquidity trade However Felix and Sau point to two appears to be utterly exaggerated. side effects in this context that could eventually play a role: First, a Tobin tax could contribute to 3 The comparison of trading volumes ac- lower volatility and therefore limit the ross time is of course problematical in scope of action for speculative trading view of continuous structural changes in by hedge funds. I personally doubt the the market (consolidation, introduction of validity of this argument because less the euro, etc.). liquid markets (as emphasized several

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 5 ® Page 54 times before) are typically character- final customers (in order to secure ized by higher volatility. profitability). As mentioned already before this would leverage the tax Second, the authors argue that the burden for this group of participants. central bank could use the greater Since final customers have only 13.3 freedom to act under the umbrella of a percent of the market, a tax of one Tobin tax to speculate against the basis point would quickly be trans- hedge funds and therefore reduce (or formed into 7.5 basis points onto final even eliminate) their profits. This ar- customers (see page 40). gument is yet less convincing, even adventurous. Greater freedom to act It is an open question to which extent of the central bank means primarily to such tax shifting will be successful be capable of concentrating on do- however. mestic policy objectives and to ignore » Tax shifting is the easiest in retail the exchange rate. The more absti- trading because demand is rela- nent a central bank, the better it is for tively price-inelastic and locally the stability of its currency. There are limited, which allows certain mo- also cases (such as the speculation nopoly rents. against the Bank of England in 1992) that demonstrate a central bank to be » Tax shifting is easier for investors powerless against speculative hedge of smaller and medium-sized funds because an important instru- companies than for multinational ment regularly used by hedge funds is firms. The latter possess a much problematical in her hands: the lever- stronger position toward foreign age effect through borrowing. exchange traders given their higher trading volumes. Eventually I fear that central banks that engaging multinational firms can even run in counter-speculation would drive the own foreign exchange depart- exchange rate from its intrinsic value ments, which would intensify com- over time. They would fall prey to de- petition. pending on foreign debt more and more deeply, and would ultimately » Also institutional portfolio investors have to give in under market pressure make a distinction as to their anyway. Unfortunately the empirical readiness to take on the tax bur- evidence lends support to this thesis den. all too often (see also Appendix 3). For instance insurance companies I interpret “freedom of central banks” take a very long-term perspective primarily as having the option to re- and they guide their behavior by frain from intervention in foreign ex- institutional rules as to the compo- change markets, i.e. not to act in re- sition of their portfolios (through sponse to alien interference. Under no “gatekeepers” mentioned above). circumstances should it be interpreted Moreover the volume of transac- as “freedom to counter-speculate”. tions is comparably small in rela- The neutrality of central banks in for- tion to their stock of assets, in con- eign exchange trading is even a cru- trast to other portfolio investors cial if a shifting back of the Tobin tax such as investment funds. This onto the central is to be avoided, renders insurers more ready to which would run counter the objective take on the tax burden, especially of the tax. as they can expect to shift the tax burden onto their clients over a

long period of time. » Who bears the tax? It should be In contrast, investment funds pur- clear form the previous discussions sue a strategy that is much more that wholesalers are likely to shift a short-term oriented because their substantial part of the tax onto their

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 5 ® Page 55 relative success is continuously link settlement). Moreover it is not being monitored against certain clear at which point the tax should be performance indicators. If an insti- levied, at the trading desk or at pay- tute falls behind the average per- ment. There are further complications formance of the branch, it risks an through the recording of netting opera- increase of disbursements, i.e. tions (in house, and bilaterally), and these institutes have to be con- there are problems relating to the in- tinuously solvent. Therefore they clusion of foreign exchange opera- concentrate on securities that are tions by producer companies. This all short-term market favorites, and renders it difficult to define the tax they rely on frequent changes in base. the compos ition of their portfolio. It Finally the possible reactions of mar- also implies a frequent change of ket participants are all but clear. currencies. These will severely be affected by If the change of securities de- decisions as to the level of taxation, nominated in different currencies the tax rate, the tax base, and the is more costly through the tax than dosage of the when introducing trading securities of one single the tax. currency, portfolio investors will As I have argued before I plead for a focus on the latter and avoid for- very small tax rate in the range of one eign exchange trading as far as half to one basis point at both ends of possible. It implies that shifting the a currency trade, but only on the “euro tax burden onto this group of mar- leg” that is settled through TARGET ket participants is more difficult and corresponding clearing operations than for longer-term investors 4 before settlement. such as insurance companies. A rough estimate on the basis of in- But even within the investment formation by the BIS is given in Table fund branch there are significant 7. On the one hand it is likely to be on differences. For instance those the safe side as far as the tax base is funds that specialize in securities concerned because foreign exchange of industrialized countries will have transactions by producing firms are no difficulty to change their strate- not included. One can also expect the gies, because there are deep and tax base to increase over time again liquid markets within the respec- once the consolidation of the financial tive currency areas that do not ne- industry has come to an end. cessitate frequent changes in cur- rency positions. This is less com- On the other hand it could be ques- pelling for fund that specialize in tioned whether the corrections securities of developing and necessary to cope with possible emerging economies. reactions of market participants have been assessed appropriately. There is

of course no information on the price- » What could be the revenue of a PFTT? It should have become clear 4 I have also argued that Switzerland from this study that any attempt to would have to cooperate in collecting the estimate the potential revenue of a tax. This does not necessarily imply the PFTT is fraught with severe difficulties inclusion of the Swiss franc into tax obli- and risks. The process of consolida- gation as long as it is traded against other tion of international foreign exchange currencies than euros. This is not without trading is far from being completed, problems since the franc could then play and further structural developments an increasing role as a euro substitute, can be expected to take place in these especially if the Swiss central bank pegs markets (for instance the continuous its currency to the euro. The same is true for pound sterling.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 5 ® Page 56 course no information on the price- salers. If one follows the method pro- elasticity of foreign exchange trading posed by Kenen according to which as a whole, let alone for market seg- the tax rate is 2 basis points for all ments. It is also unclear whether the transactions, and wholesalers would tax base is sufficiently protected carry only half the rate, the yearly tax against loopholes. revenue could amount to 20.8 bill. euros. In this calculation I only count As a whole a PFTT with a tax rate of that part of the trade whose leg is set- one basis point could eventually yield tled in euros. Neither pound sterling a yearly tax revenue of 16.6 bill. eu- nor the Swiss franc is included unless ros. This assumes that the tax to be traded against euros. paid at both ends of the trade, imply- ing a tax of 2 basis points for whole-

Table 7: Rough estimates of the revenue of a PFTT Daily values Total Euro leg Revenue (except for the last two rows) in bill. US $ in bill. US $ in mill. US $

Total turnover (in 2001 US dollars) 1,210.0

Minus estimates by the BIS -36.0

Preliminary basis 1,174.0 440.0

Non-taxed instruments -20.0

Contraction of trading volume -173.1

Total taxable trading volume 980.9 367.6 58.34

Trader-trader transactions 575.7 215.8 43.15

With other financial institutions 274.9 103.0 10.30

With non-financial institutions 130.3 48.9 4.89

Yearly amount in mill. US $ 91,907.2 14,585

Yearly amount in mill. euros (Tax rate 1 basis point) 16,573

(Tax rate 2 basis points Yearly amount (alternative) 20,800 with half the rate for wholesalers)

» Summary. The introduction of a The strongest impact will undoubtedly PFTT will provoke very different reac- be on arbitrage trading, but in this tions by actors in foreign exchange segment of the market the spreads markets. are so low that the tax must be passed on to other sectors, i.e. the tax Generally one may expect that the has to be carried by producing firms trading volume will decline, and the and households (private and public). spreads will widen. But at a reason- To what extent this will occur is again able rate of one basis point the an open question. spreads for liquidity trading would still be lower than several years ago. Nev- The proper speculators in the market, ertheless the question is open as to for instance hedge funds, are held who would finally carry the additional back by the tax only little, because costs. they operate with significantly larger margins than liquidity traders.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 5 ® Page 57 Of institutional investors, insurance Cautious estimates of the potential companies are likely to carry a rela- revenue of a PFTT indicate that the tively higher tax burden, because of tax could yield some annual 17 to 20 their lower turnover rotation and their bill. euros with a tax rate of one basis longer-term perspective, than invest- point for the area of the European ment funds. Among the latter, those Union plus Switzerland. This estimate groups that specialize in trading secu- does not include transactions that are rities of industrialized countries can carried out in British pounds or Swiss avoid the tax more easily. francs against non-euro currencies.

Paul Bernd Spahn ® [email protected] On the feasibility of a tax on foreign exchange transactions ® Chapter 5 ® Page 54