Financial Transaction Tax

Testo dell’intervento all’incontro “Financial Transaction Tax” organizzato a Milano da FeBAF e Assosim nella giornata del 15 luglio 2013

***

The recent financial crisis has caused public finance economists to rethink how the financial sector should be taxed. Before the crisis, it was generally assumed that no special tax regime for financial services was necessary, and that divergences from the general tax regime were likely to be distortive. For example, the IMF generally recommended against having a higher rate of corporate income tax on the financial sector, which some countries impose¹, because it can distort the allocation of capital and deter financial sector development. It was recognized, however, that imposing a standard credit-invoice VAT on the financial sector is difficult, because payment for financial services is often bundled into a financial margin, such as loan or deposit interest, which obscures the VAT tax base. For this reason, financial services have generally been exempted from the VAT.²

Things changed with the financial crisis: First, the crisis produced a desire to recoup revenues lost due to the crisis by taxing the companies that precipitated it. In the years leading up to the crisis, the financial sector logged exceptionally high levels of profit and compensation, which in retrospect reflect its assumption of large tail risks. Realization of these risks during the crisis, coupled with implicit public guarantees for too-big-to-fail institutions, imposed a heavy burden on public coffers: The direct costs of bailing out financial institutions in the most affected countries averaged about 7 percent of GDP through 2012, slightly more than half of which has been recovered for a net cost of 3.3 percent of GDP.³

Second, the apparent inadequacy of existing financial regulations to curb excess leverage and risk-taking in the financial sector raises the question of whether tax policy can be used to help achieve that goal. The crisis made apparent that excessive risk taking had caused severe externalities, and raised the issue of whether a Pigouvian tax should be introduced to internalize these externalities. But, even leaving this aside, it has long been known that the standard corporate income tax, which gives a deduction for interest payments but none for dividends, encourages non-financial companies to prefer debt over equity finance. A recent study by Keen and de Mooij (2012)⁴ shows that this also applies to banks, which usually carry more than minimal regulatory capital and are thus influenced at the margin by the tax benefits of interest deductibility.

It was in this context that the G-20 charged the IMF with designing a plan for a “how the financial sector could make a fair and substantial contribution toward paying for any burden associated with government interventions to repair the banking system.” In response, IMF (2010) proposed two new tax instruments to help accomplish this goal: a financial stability charge (FSC) on bank leverage and a financial activity tax (FAT) on financial sector profits and compensation. The 2010 report also examines the effects of a third tax widely considered to raise revenue and regulate financial markets in the wake of the crisis: a financial transaction tax (FTT), but, as we noted in our report, we regard an FTT as a much weaker option.

A broad-based FTT has been widely promoted in the wake of the crisis as a means of raising revenue and reducing financial sector risks. FTTs, which are quite common through both developed and developing countries, are imposed on a wide variety of transactions ranging from real property transfers to bank deposits/withdrawals to securities trading. Post-crisis debate has focused on security transaction taxes (STTs), with numerous governments and civil society organizations supporting introduction of a multilateral transaction tax on securities and derivatives trading to prevent future crises and help pay for the past one. Imposition of new STTs reverses the trend of the past two decades toward reducing financial transaction taxes: Since the 1990s, most major European countries have eliminated their FTTs on equity trading in an effort to develop their financial markets. (A notable exception to this is the UK, which maintains its stamp duty at the fairly high rate of 50 basis points; however, it has a fairly narrow base insofar as all market-makers and equity derivatives are exempt.).

Belief that STTs can reduce risk is a major reason for their promotion following the crisis, but the evidence for this is at best mixed. Numerous studies confirm that imposition of an STT, like any increase in transaction costs, reduces asset prices and trading volume or liquidity. The more controversial question is whether they can reduce price volatility, and hence one form of financial risk. Numerous studies relate trading volume positively to price volatility, so in theory a tax that reduces trading volume could reduce volatility as well. However, reduced trading volume is also associated with reduced liquidity and wider bid-ask spreads, which can also result in higher price volatility. So the relationship between an STT and price volatility is unclear, and it may be non-linear: a small STT in a highly liquid market may reduce short-term price volatility, while a large STT may reduce liquidity sufficiently to increase volatility. Major price swings, or financial bubbles, are believed to be driven by excessive leverage, not by trading activity. Since STTs do not in general reduce leverage—and depending on their design may even increase it—it is unlikely that they would reduce the risk of bubbles.

The revenue-raising capability of an STT depends not only on trading volume, but also on the availability of substitute assets and trading platforms. Imposition of an STT eliminates trades that do not yield at least the increase in transaction costs related to the introduction of an STT, especially short-term trading. If close substitutes for the taxed security (such as derivatives) are available, or the security also trades on untaxed platforms (such as offshore exchanges), then some of the trading volume will be displaced into those activities. For example, the 50 basis point U.K. stamp duty on share trading has encouraged the growth of the market for “contracts for difference” (CFDs), or daily-settled equity swaps; and Sweden’s imposition of transaction taxes on stock and bond trading in the early 1990s displaced stock trading activity to London. Thus, base elasticity can undermine the anticipated revenues from an STT.

The EU’s FTT proposal seeks to limit this form of displacement by design. The proposed tax would be imposed on a wide array of financial products—both equity and fixed income securities as well as their derivatives—which would limit displacement between instruments. The EU FTT would also seek to limit geographical displacement of trading by taxing all trading in securities issued by EU-headquartered corporations, regardless of where in the world it takes place. While this would in theory prevent transactions from migrating outside the EU to escape tax, it is likely that the tax on extraterritorial transactions would be very difficult to enforce. Concern that a European FTT would drive trading outside of Europe is a major reason for the UK’s opposition to a European FTT.

Indeed, opposition by some EU members, Sweden in addition to the UK, led an 11-country coalition to pursue a reduced version of the proposed FTT under the EU framework for “enhanced cooperation”. This narrowing of geographic scope lowered the estimated revenue from the FTT from approximately EUR 57 billion to EUR 30-35 billion. Although the official proposal is still for a 10 basis point tax on stocks and bonds and 1 basis point on derivatives (levied on both buyer and seller), more limited versions are reportedly being discussed, which could reduce the expected revenue by as much as an order of magnitude.⁵

The new STTs introduced by France and Italy could reportedly serve as models for this reduced-form STT, so it is worth examining their design and impact in greater detail. They have many similarities: Both tax transactions in the shares of domestically headquartered companies and their derivatives, regardless of where in the world they are traded. The tax charged on equity trades is generally much higher than the tax charged on derivatives.⁶ New share issues and market maker trading are exempted. Innovatively, both FTTs also levy a very low-rate transaction tax on high-frequency trading in the domestic market.⁷

Despite their fairly modest rates, the French and Italian FTTs can be expected to depress trading activity. Early empirical studies of the effect of the French FTT on the market for French equities show that it reduces trading volume by about 15 percent and decreases market depth; however, no appreciable effect on share market volatility was found.⁸

The French and Italian FTTs increase to some extent existing tax incentives for leverage. Taxing equity but not debt trading increases the relative cost of equity finance, thus compounding the debt bias of the corporate income tax. Levying a substantially higher tax rate on equities than on their derivatives encourages trading in the latter; and since derivatives carry inherent leverage, this may increase financial market risk. A uniform tax rate based on notional value would discourage use of leveraged instruments, but would disproportionately raise transaction costs in derivatives markets, which are generally much lower than those in securities markets.

One major difference between the French and Italian FTTs and the proposed EU FTT is their treatment of market makers: While the French and Italian taxes, like the UK stamp duty, provide a broad exemption for market makers (except in the case of HFT), the EU proposal would fall in particular on taxable transactions executed by financial institutions, regardless of whether they were proprietary or on behalf of a second party. The EU proposal would potentially produce significant “cascading”, or multiple taxation of a single economic transaction, since some financial arrangements such as unit trusts can introduce intermediate entities between final transactors. The unusual design of the EU FTT appears to be aimed at reducing the size of the financial sector and discouraging financial complexity, whereas the more conventional exemption for market-makers allows the FTT to function more as a realization-based wealth tax on securities holders.

These new FTTs offer a couple of innovative features with quasi-regulatory impact. Both the French and the Italian FTT target high-frequency trading in particular. Despite their very low-rate, these taxes should be sufficient to eliminate most HFT due to its high speed and ultra-thin margins. Although high speed and automation are not inherently pernicious—indeed the majority of algorithmic trading is used to improve execution for third parties—proprietary HFT is frequently associated with practices that can distort markets. Although it can improve liquidity, it is also thought to produce higher short-term volatility and sudden cascades (such as the “flash crash” of May 2010). However, since the HFT taxes are territorial, they will likely just displace HFT outside of France and Italy.

Another innovative feature of Italy’s FTT is its heavier tax rate on OTC trades. This may have quasi-regulatory benefits of channeling equity trading to organized exchanges, which offer greater transparency and control. This would likely offer the greatest benefit to securities other than equities, since equities are most likely already to be traded on organized exchanges.

However, the Italian derivatives tax could arguably be better designed. The series of flat rates charged according to the level of the notional value of the underlying security leads to sharp discontinuities in the effective tax rate. And as the underlying value gets larger, the effective rate goes to zero. A flat rate, such as the two basis points that Italy levies on HFT, would arguably make more sense for derivatives as well.

Altogether, the specific design of FTT does affect their effectiveness and distortionary effects, as this discussion has shown. Not all FTTs are equally good or bad (depending on the standpoint. This said, the IMF remains of the view that, if the goal of these new financial taxes is to raise revenues and reduce systemic risks, there are better options than FTTs. In particular, in our report to the G20 we gave preference to two different taxes, which we called the financial stability contribution (FSC) and the financial activities tax (FAT).

An FSC is a Pigouvian tax on bank balance sheet debt aimed at internalizing bank incentives to use excessive leverage and at raising revenue to offset the costs of potential bailouts. If deposits, a relatively stable source of funding compared to interbank loans, are adequately insured then they should be excluded from the base. If larger institutions are more likely to rely on excessive debt due to market perception of an implicit government guarantee, the rate of the FSC can be progressive to offset this effect. IMF (2010) estimates that too-big-to-fail institutions have a funding advantage of 20-60 basis points over smaller institutions, which can serve to indicate an appropriate top tax rate for an FSC on larger banks.

FSCs, or bank levies, have been widely adopted since the financial crisis, particularly across Europe. The most common base for these taxes is balance sheet liabilities net of equity and insured deposits, although there is significant variation: Portugal and Cyprus include deposits in the tax base; and France, Hungary and Slovenia tax different types of assets rather than liabilities, and thus do not alter financing incentives. Korea’s bank levy is based specifically on cross-border short-term funding in order to deter foreign exchange risk. Several countries (Austria, Germany, Hungary, Netherlands, and the UK) levy progressive rates, imposing higher burdens on larger banks, and the UK and Korea also offer reduced rates for longer-term debt, reflecting its reduced refunding risk.

However, existing bank levies appear modest in terms of incentives and revenue yields: The tax range for these levies runs as high as 53 basis points, but is typically much lower: Only three countries in Europe (France, Hungary and Slovakia) have a top rate above 10 basis points. On average, they thus appear too low to internalize the implicit government guarantee of the large banks. Expected revenues from the bank levies are quite modest: In Europe, median yield should be around 0.14 percent of GDP.

Nonetheless, preliminary analysis shows that bank levies have been successful in increasing bank reliance on equity and deposits as funding sources. Analyzing data for European banks, Devereux et al. (2013) show that each basis point increase of the levy rate increases bank equity by approximately one quarter of a percentage point. Similarly, each basis point increase in the levy rate raises the ratio of customer deposits to total assets by about one half of a percentage point. However, the results in this paper also shows that, in tandem with this reduction in funding risk, banks increased the riskiness of their assets: A one basis point increase in the levy rate increases the ratio of risk-weighted to total assets by one third of a percentage point.⁹

As I noted, our report also proposes a “financial activities tax” (FAT), which comes in three different versions: The most comprehensive version (FAT-1) would be levied on total cash-flow profits and compensation in the financial sector, and thus would be equivalent to an addition-method VAT. This version of the FAT would be most useful for addressing the potential under-taxation of financial services due to their exemption under current VATs. Because this tax would probably not be credited on a per-transaction basis, it would contribute to VAT cascading on business purchases of financial services; however, it would correct for the under-taxation of consumer financial services under VAT exemption. Keen and others (2012) estimate that the tax base for this version of the FAT averages just under 5 percent of GDP among developed countries, but varies substantially depending on financial sector development, from as much as 23 percent in Luxembourg to less than 2 percent in Finland.

Iceland introduced the first FAT-1 in 2012, imposing a 5.45 percent tax on payroll and a 6 percent tax on profits above one billion Icelandic krona in the financial sector. In contrast to an ideal FAT, the base of the profit tax is accounting rather than cash-flow profits—that is, investment is depreciated rather than expensed—so it taxes the normal return to capital. Moreover, the FAT is imposed on top of Iceland’s payroll tax for VAT-exempt businesses, so it serves more to raise revenues than to correct for VAT exemption. It was expected to raise 0.28 percent of GDP in 2012.

Depending on how it is structured, an FAT can also serve as a Pigouvian tax aimed at correcting financial sector externalities. As noted, the implicit bailout option encourages excessive risk-taking that produces abnormally high profits and compensation for large financial firms in good years, and large losses in bad years, which are put to the public sector. A surtax on abnormal profits and compensation in the financial sector could reduce this incentive, as well as generate a fiscal buffer to offset abnormal losses in bad years. This is the idea behind the “FAT-2” and “FAT-3” versions of the FAT. A FAT-2 is levied on financial sector cash-flow profits and extraordinary compensation above a certain level, however defined. Like a FAT-2, a FAT-3 would also tax extraordinary compensation, but would also exempt a certain level of cash-flow profits, taxing only the extraordinary profits associated with excessive risk assumption in the financial sector. Keen and others (forthcoming) estimate that the average base of a FAT-2 among developed economies is about 2.5 percent of GDP, while that of a FAT-3 is about half as large.

No countries thus far have introduced a full-fledged FAT-2 or FAT-3, but the bonus taxes enacted after the crisis can be viewed as partial FATs on compensation. The UK and French taxes, at 50 percent of variable compensation, were sizeable but temporary, levied for a year or less beginning in 2009. The Italian bonus tax is permanent, but much more modest in scope: It imposes a 10 percent tax rate on bonuses that exceed three times fixed remuneration. In contrast to an FAT, which would tax extraordinary compensation regardless of form, a bonus tax affects only incentive compensation (and sometimes only bonuses paid in cash or options, rather than stock). Financial institutions could thus avoid it by increasing regular compensation, which is fact what was observed in the UK.¹⁰

Let me conclude by noting that achieving the right balance of taxes on the financial sector requires that the role of taxes in determining leverage or risk-taking must be coordinated with the role of financial regulation. With full information, and where revenue is irrelevant, either instrument could be used, but these restrictions are clearly unrealistic. The focus of the international community has been so far on the use of regulation to address financial stability issues. Broadly speaking this is appropriate. But I think that not enough attention has been paid to looking at the implications of taxation for financial sector decisions. Much more work isneeded in this area.