The Empirical Case for a Financial Transaction Tax
Discussions of common-sense financial regulation inevitably evoke dire warnings of impending catastrophe. The argument that a financial transaction tax (FTT) will impoverish us all is a classic of the genre. Never mind that the United Kingdom and Hong Kong, two of the world’s pre-eminent financial centers, levy such taxes, as do many other nations. Pay no attention to Jack Bogle, founder of the Vanguard Group and legendary advocate for low-fee retirement investing, responding to the idea with: “I love it.” Finance, the financiers insist, must be left alone. For our own good.
FTTs vary in their specifics but generally entail a small fee—say, 10 basis points (i.e., 0.10%)—imposed on the buyer or seller of a security such as a stock, bond, or derivative. Why consider an FTT on securities trades in the secondary market?
First, short-term speculation in financial markets has negative economic value. It increases volatility, distorts price signals, and draws capital and talent away from more productive pursuits. An FTT “throws sand in the gears” of the financial system’s least useful contraptions, to steal a phrase from Nobel laureate economist James Tobin, and skims speculative froth from the market. A suitably low FTT discourages trades that contribute little to the long-term prosperity of the nation, while having little effect on the value-creating investments that the financial sector exists to facilitate.
Second, securities trades provide an ideal tax base within the financial system. All else being equal, a new tax imposed on securities trades could conceivably have some negative effect on investment levels. However, this concern ignores not only the reality that financial markets are currently doing an inadequate job directing capital to productive investment, but also the opportunity to use an FTT’s revenue to improve tax policy elsewhere or otherwise support investment and growth. The net effect of shifting the incidence of taxation from investment and profit to relatively useless securities transactions represents a boost to the rewards for more productive economic activities.
Opponents of an FTT offer a number of arguments. Some of them seem on balance less intended to persuade than to provide cover for the generic impulse that financial markets shall not be impinged upon. That impulse should be resisted.
Claim #1: An FTT will reduce market liquidity.
In Reality: The marginal cost of any liquidity reduction would be small, and the policy is as likely to benefit financial markets by suppressing costly latency arbitrage.
FTT opponents warn that a transaction tax would cause the “bid-ask spread” between what buyers offer and sellers demand to widen. Traders known as “market makers” who buy from sellers and then sell to buyers, lubricating the market by making desirable prices quickly available, would find their job more difficult and costly to perform. The result would be less efficient prices and, ultimately, lower asset values.
Certainly, investors are right to value the ability to buy and sell assets at any time, and the broader economy does benefit from such transactions occurring cheaply and quickly. This is why many nations and proposals exempt legitimate market makers from FTTs or refund their taxes paid. Beyond that, the question must then be asked: what constitutes a useful amount of liquidity?
Like most things, market liquidity has declining marginal utility. The advent of decimalization and internet trading has driven the average bid-ask spread to almost zero—for high-volume stocks, it declined from 0.2% in the late 1990s to 0.02% in 2004. As high-frequency trading (HFT) took off in the mid-2000s, this fell even further. In the UK, bid-ask spreads went from around 0.02% in 2004 to 0.002% in 2011.
Would the American financial market suffer if bid-ask spreads and liquidity returned to, say, their late 1990s level, when the Dow Jones Industrial Average delivered its greatest annualized returns since the Coolidge Administration? That seems unlikely. Indeed, the question brings to mind economist Thomas Phillipon’s argument that the unit cost of financial intermediation has not appreciably decreased since the late 1800s. J.P. Morgan reports no decline in the cost of capital for corporations between the 1990s and the 2010s. To the contrary, the advent of HFT and the explosion of trading volumes has coincided with a clear trend of disinvestment in the real economy.
The better case holds that an FTT would improve matters by curtailing HFT. While the HFT industry is nothing if not extraordinarily opaque (hardly a point in its favor), evidence suggests that HFT firms are often users rather than providers of liquidity. HFT’s technologically driven use of algorithmic trading to conduct extremely high numbers of trades in fractions of a second now accounts for at least half of total trading volume. The goal is often “latency arbitrage,” technologically exploiting the advantage of obtaining information or executing trades just millionths of a second faster than another trader. The UK’s Financial Conduct Authority has concluded that these practices reduce liquidity and introduce a “latency arbitrage tax” on other investors, elimination of which would reduce the cost of trading by 17%.
Claim #2: An FTT would reduce investment.
In Reality: An FTT whose revenue reduced other taxes or subsidized investment would have a net positive effect on investment and growth.
Assessment of any tax proposal’s net effect requires an assumption about how the revenue will be used—to reduce some other tax, facilitate some other spending or investment, or reduce debt. All of these activities have the potential to increase investment, and a tax’s effectiveness depends in part on these choices. For instance, the American Enterprise Institute’s analysis of the $4 trillion in tax increases proposed by then-candidate Joe Biden in 2020 found no effect on growth because of “two offsetting effects in that model: the reduction in output from higher taxes on work, saving, and investment and the increase in output from the lower debt.” Nor did the 2017 tax reform law cause the desired spike in new capital investment, as even potentially pro-investment measures were limited or sacrificed in the hunt for an ever-lower corporate tax rate.
Policymakers could use an FTT’s revenue to correct such mistakes and move towards full immediate expensing for capital investment or better tax treatment for research and development. They could provide equity for a domestic development bank. They could create a new category of very-long-term capital gains with a very low tax rate—potentially the largest capital-gains tax cut in American history. The goal should be to raise tax revenue from the financial system in a way that promotes well-functioning markets and orients activity toward actual investment in the real economy—both by taxing activity that is not productive and rewarding activity that is. Specific options are described in more detail below.
Claim #3: An FTT will harm saving for retirement.
In Reality: An FTT would have little effect on the wealth accumulation of long-term investors and may even improve their returns.
The most politically potent argument against an FTT is that it is a “tax on retirement.” Industry interest groups warn that that average investors would see their retirement investments reduced by about $6,000 over the long term. Or was that $20,000? Or $64,232?
Such estimates tend to rely on two faulty assumptions: that the investor is buying and selling at a very high rate (known as “turnover”), and that an FTT would not appreciably diminish that turnover. These assumptions are both out of step with reality. An FTT is only paid when a trade occurs. If trading frequency declines in response to the cost of the tax—and evidence suggests that trading is indeed responsive to costs—the actual cost to retail investors is likely to be low. When confronted on its analysis, Vanguard quietly revised its estimated impact down by a factor of 15. An alternative estimate, not financed by the industry, found that an FTT might cost the average middle-income family a grand total of $13 to $35 per year.
Even that may overstate the true cost. As Vanguard founder John Bogle observed, by reducing turnover an FTT would lower costs for long-term investors:
So we’ve taken the frictional costs out and that helps explain why we’ve had this orgy of speculation. No question about that. So I like the idea of a transaction cost. …
In 1929, the turnover was about 145 percent in the stock exchange. It was about 25 percent, believe it or not, my first 15 or 20 years in this business. Last year, it was 350 percent. That’s an orgy of speculation we’ve never seen before. If the idea of a transaction cost or a tax on short term capital gains is to cut back on that transaction volume, then it wouldn’t produce much revenue, but it would succeed in its primary goal of reducing those costs. …
By pitting one investor against another and having that croupier in the middle, which is apparently necessary for the transaction to take place, you ensure that investing is a loser’s game. If investors acted in the community interest, that is, by owning the market, which they own anyway, and not trading, it would be a winner’s game. So by doing what is best for society, our investors would end up being winners rather than losers.
The boogeyman of a tax helps the financial industry divert focus from the larger driver of costs to investors: the industry’s own asset management and performance fees, which dominate investment costs. Slowing the pace of excessive churn and encouraging a return to low-turnover, lower-cost strategies would benefit average investors. By making it harder for high-frequency traders to “front-run,” an FTT may also allow slower large investors (pension funds, for example) to participate in markets on a more level playing field and enjoy access to better prices.
Claim #4: FTTs have failed elsewhere.
In Reality: Some poorly designed FTTs have failed or been repealed. Well-designed ones have, well, worked well.
Some FTTs have worked poorly or have been repealed. Opponents generalize from these cases, whereas the real lesson is that, as with all policy, the devil is in the design details.
The policy enacted by Sweden in 1984 presents a clear example of an FTT gone wrong. Sweden set rates far too high (100 basis points increased to 200 in 1986), and traders avoided the tax by moving their activity elsewhere, which the tax’s poor design made easy. Predictably, the Swedish stock market weakened and trading in Sweden declined precipitously until the tax was repealed in 1991. Other nations have repealed FTTs as well. Like Sweden, Germany saw trading move to London and repealed its FTT in 1991—though it has supported a European FTT more recently. Japan, meanwhile, had an FTT between 1954 and 1999—a period roughly overlapping with its celebrated economic boom.
Conversely, the United Kingdom has had a Stamp Duty on securities transfers since 1694. Its design is far from ideal: the tax is relatively porous and does not apply to derivatives, so the UK “contract for difference” (CFD) market has grown meaningfully as traders substitute trading in derivatives for trading in other securities. Even so, the UK FTT generates about £3.5 billion in revenue per year, or about 0.2% of UK GDP. The lesson learned by many British financial experts and political leaders is not that an FTT is futile, but that theirs should be strengthened. If a similarly designed FTT captured an equivalent share of U.S. GDP, annual revenue would be approximately $40 billion. Hong Kong annually raises well over 1% of its GDP from its long-standing FTT and enjoys low levels of high-frequency trading. Shanghai is subject to China’s 0.1% duty on stock sales. After New York, these are the world’s next three largest financial centers.
Major economies like Taiwan, South Korea, South Africa, Brazil, and India impose some form of tax on securities trades. Numerous European countries in addition to the UK impose an FTT. European efforts to enact a common, multistate FTT are active, with Germany playing a leading role. Decisions by Italy, France, and Spain to repeal FTTs garner frequent mention, but all three have in fact re-established FTTs since. Spain’s new tax went into effect in 2021.
Claim #5: The United States already rejected an FTT.
In Reality: A 1965 law repealed a series of tax provisions that happened to include a stamp tax on securities trades.
The United States did have an FTT, in the form of a Documentary Stamp Tax on the issuance and transfer of stocks and other securities, from 1914 until 1965. Its repeal was not a result of concern that an FTT was an existential threat to America’s then-robust financial markets and investment levels. It was simply swept into the sprawling Excise Tax Reduction Bill of 1965, an effort to repeal “almost all of the hodge-podge of federal retail and manufacturers’ excise taxes which had accumulated over the course of three decades.” The repeal of the Documentary Stamp Taxes on securities trades and real estate conveyances was included under the Miscellaneous Taxes title, right next to the repeal of excise taxes on playing cards and coin-operated amusement devices. The congressional report on the bill made little comment on the tax, beyond its administrative complexities.
On signing the law, President Johnson made no mention of financial markets. Instead, he expounded at length on the ongoing economic boom—for which he credited tax policy that incentivized capital investment in the real economy.
To this day, the United States collects a form of FTT: the SEC’s Section 31 fee. This nominal fee has been imposed for decades on securities exchanges based on the volume of their transactions. The fee rate, adjusted regularly by the SEC, is vanishingly small ($5.10 per million as of early 2021, or roughly one-twentieth of one basis point). A full-fledged FTT would be much larger. But if the question is whether American financial markets can bear the administrative burden of transaction fees, the answer is yes.
* * *
Pro-Investment Tax Reform
An FTT’s revenue depends not only on its rate, but also on the extent to which it reduces trading activity, which can be difficult to predict. Thus, FTT revenue estimates range widely, depending on the proposed tax rate, the proposed scope of coverage, and the methodological assumptions made about the response by traders. When considering possibilities, a good starting point is the Congressional Budget Office and Joint Committee on Taxation’s 2020 revenue estimates for a 10-basis-point transaction tax on secondary market sales of stocks, bonds, and derivatives. They estimated that such a tax would raise $751.9 billion over 10 years, or an average of $75 billion annually.
What would new revenue of that magnitude allow?
- Moving Toward Full Expensing for Capital Investment. Permanently allowing immediate expensing of all capital investment, including structures, would cost $809 billion over 10 years (scored dynamically). More modestly, enacting permanent immediate expensing for all capital assets other than long-life structures, and simultaneously creating a neutral cost recovery system (NCRS) for structures, would cost $386 billion (dynamically scored). This more modest option could increase the nation’s capital stock by 13% and, after initial transitional costs in the first 10 years, settle into a long-run annual cost of around $60 billion.
- Canceling the Amortization of Research and Development Costs. The 2017 Tax Cuts and Jobs Act ended the ability of businesses to immediately expense research and development costs beginning in 2022, reducing the attractiveness of such investments and damaging U.S. competitiveness. There is bipartisan interest in repealing this provision. Doing so would cost between $108 billion and $131 billion over 10 years. The cost could be lower if policymakers privileged research and development conducted in the United States.
- Cutting Capital Gains for Very-Long-Term Investment. To encourage longer holding periods, policymakers could create a new category of very-long-term capital investment, taxed at a very low top rate (for example, a rate of 5% for assets held for over five years). Annual long-term capital gains revenue averaged $75.2 billion from 2000 to 2014.
Policymakers should think broadly about the tools at their disposal. If they were to impose an FTT and also eliminate the deductibility of business interest, as Oren Cass proposes in “Confronting Coin-Flip Capitalism: A Pro-Market Agenda for Financial Reform,” they would have even more revenue available to support an economic rebalancing away from speculation and toward real investment. Beyond offsetting tax cuts, they could use revenue to provide equity to a domestic development bank, which could directly facilitate increased investment in the American economy, especially infrastructure. They could provide further tax relief or fund research partnerships to support domestic production in industries of critical national importance.
A well-designed FTT could both improve the functioning of the markets in which it operates and raise revenue for other policies that would advance complementary goals. Policymakers interested in restoring well-functioning financial markets that play their needed role should consider an FTT.