1. No. 702 July 9, 2012
Would a Financial Transaction Tax
Affect Financial Market Activity?
Insights from Futures Markets
by George H. K. Wang and Jot Yau
Executive Summary
In the wake of the recent financial crisis, such computation. We show that a transaction
several commentators have suggested a trans- tax on futures trading will not only fail to gener-
action tax on financial markets. The potential ate the expected tax revenue, it will likely drive
consequences of such a tax could be hazardous business away from U.S. exchanges and toward
to the financial markets affected as well as to the untaxed foreign markets.
economy. In this paper, we review the relevant A review of the literature and estimates con-
theoretical and empirical literature and apply tained here indicates that there is an inverse
our findings to estimate the possible impact of relationship between transaction cost (bid-ask
a transaction tax on U.S. futures market activity spread) and trading volume; to the extent that
as well as its utility as potential tax revenue. a transaction tax increases costs, trading vol-
We find that the impact of a transaction tax umes will likely fall. There is also a positive re-
on market activity (trading volume, bid-ask lationship between transaction cost and price
spread, and price volatility) will determine the volatility, suggesting that the imposition of a
potential of such a tax as a source of government transaction tax could actually increase financial
revenue. We also find that the current estimat- market fragility, increasing the likelihood of a
ed elasticity of trading volume with respect to financial crisis rather than reducing it. Perverse-
a transaction tax in the U.S. futures markets is ly, the imposition of a financial transaction tax
much higher than those reported in the extant could have results that are exactly the opposite
literature and those used by the government in of those hoped for by its proponents.
George H. K. Wang is research professor of finance at the School of Management, George Mason University, and
former deputy chief economist at the U.S. Commodity Futures Trading Commission. Jot Yau is Dr. Khalil Dibee
Endowed Chair in Finance at the Albers School of Business and Economics, Seattle University.
2. Financial Introduction forms, dating from Great Britain’s 1694
transaction stamp duty9 to recent Tobin taxes on cur-
The financial crisis of 2007–2008 has rency transactions and the latest rejection
and securities raised many concerns and questions about of a proposed bank tax in Europe.10 In 1978
transaction financial regulation and policymaking. One James Tobin first proposed a tax on foreign
of the more popular proposals in the wake of exchange transactions. It aimed to curb ex-
tax proposals the crisis has been to impose financial trans- cessive speculative volatility as exchange
have been action taxes (FTTs).1 For example, the Re- rates freely fluctuated after the collapse of
brought up for public of Korea pushed for an international the Bretton Woods Agreement, which had
levy on bank transactions at the G-20 meet- established a fixed exchange rate system fea-
consideration by ing in 2010, while the International Mon- turing many other countries pegging their
several American etary Fund had presented its own bank-tax currencies to the U.S. dollar.11
administrations proposal at the same meeting.2 The Euro- FTTs have come under severe criticism by
pean Union (EU) has also considered various legislators, regulators, and the public—espe-
and Congresses. FTTs, though a proposal for a bank transac- cially the financial services industry and in-
tion tax was rejected by the EU in 2010.3 In vestors—because the effects of the taxes are
September 2011 the European Commission so wide-ranging. Those who oppose FTTs
proposed a new plan for a pan-EU Tobin tax argue that a tax on securities transactions
taking effect in 2014. It had the backing of would increase price volatility, reduce mar-
France and Germany but outright opposi- ket liquidity (trading volume), and decrease
tion from Britain.4 Proponents of the finan- price efficiency, thus increasing the cost of
cial transaction tax suggest that it can be capital and lowering security values.12 In ad-
used effectively as a means to curb excessive dition, an STT could drive trading in some
financial market volatility, stabilize the mar- securities to overseas markets not burdened
kets, and raise revenues for various purposes. by tax. Thus, the tax may affect the relative
Financial transaction and securities competitiveness of taxing countries in global
transaction tax (STT) proposals have been financial markets and would naturally be
brought up for consideration by several of particular concern for the U.S. futures
American administrations and Congresses.5 industry.13 Some pundits warn that FTTs
During the fiscal year 1990 budget negotia- are easily avoidable and likely to drive finan-
tions, the Bush administration proposed a cial activity underground, beyond regula-
broad-based 0.5 percent tax on transactions tory oversight.14 FTT proponents argue that
in stocks, bonds, and exchange-traded deriv- FTTs would increase government revenues
atives.6 In 1993 the Clinton administration that could be used for various purposes, in-
proposed a fixed 14-cent tax on transactions cluding funding regulatory agencies (e.g., the
in futures and options on futures.7 The U.S. Commodity Futures Trading Commis-
Obama administration has proposed a user sion or Securities Exchange Commission),
fee in the 2012 federal budget on all futures pay back the bailout money to the govern-
trading to fund the U.S. Commodities Fu- ment, fund a country’s future budget, or ex-
tures Trading Commission, while 28 mem- tract a larger contribution from the financial
bers of the House of Representatives have sectors toward funding public goods.15
co-sponsored legislation that would im- Transaction tax rates vary with the type
pose a transaction tax on regulated futures of financial instruments in question (e.g.,
transactions. The proposed tax is 0.02 per- equities are typically taxed at higher rates
cent of the notional amount of each futures than debt instruments or derivatives), the lo-
transaction, to be charged to each party of cation of trade (i.e., on- or off-exchange, on
the transaction, with a projected revenue of domestic or foreign markets), and the status
hundreds of billions of dollars per year.8 of the buyer or seller (domestic or foreign
FTTs have a long history in various resident, market maker, or general trader). As
2
3. financial markets have become more global- an FTT/STT with regard to trading volume,
ized with advances in information and trad- price volatility, pricing efficiency, and esti-
ing technologies, exchanges can now attract mated revenue. Following that, we discuss a
more business by lowering trading costs. For methodology for estimating potential trans-
example, countries such as Sweden and Fin- action tax revenue with an example illustrat-
land removed all STTs, while others such as ing the application for estimating potential
Australia, Japan, the United Kingdom, and tax revenues that can be raised from U.S.
Taiwan lowered their tax rates. John Camp- futures markets. Finally, we present our con-
bell and Kenneth Froot in 1994 referred to clusion.
this shift in taxation as an important func-
tion of STTs, that they “reveal the nature
and scope of powerful underlying changes What Does Theory
in international capital markets, and offer a Say about the FTT?
glimpse into a future in which government
policy not so much disciplines, but is instead The theoretical arguments for an FTT are
disciplined by, competition in modern capi- based on rational economic theories, which
tal markets.”16 That STTs are not common- assume participants in the markets are all ra-
place in most countries lends credence to the tional, having complete information about
Extant literature
hypothesis that implementing an STT will future prospects. Participants make deci- presents myriad
have a negative impact on a given market’s sions based on the maximization of their theoretical
relative competitiveness. utility functions given the assumed (cost-
Before one can properly evaluate the pros less) complete information they have. Trans- arguments
and cons of a transaction tax on financial action costs in financial transactions, in- in support of
markets, one needs to know what potential cluding all kinds of taxes and levies charged
impact an increase in transaction cost would by the authorities, are considered by traders
and against a
have on trading volume, market liquidity, as they optimize their welfare in these mod- transaction tax.
price volatility, potential revenue, and the els. However, with different assumptions on
general welfare of market participants. the rationality and composition of market
The objective of this paper is to review the participants and the degree of market effi-
relevant literature on the theoretical ratio- ciency, arguments for and against the FTT
nale for a financial transaction tax as well as both have reasonable theoretical appeal.
empirical evidence that measures outcomes Extant literature presents myriad theo-
from the imposition of such a tax. In this re- retical arguments in support of and against
view, we concentrate on STTs because (1) this a transaction tax. The arguments all address
type of transaction tax has been proposed in the following basic questions: (1) Does the
Europe, and a proposal is still under consid- FTT reduce price volatility? (2) Does the FTT
eration in the United States, and (2) STTs reduce trading volume and market liquidity?
are the most common FTT that has been ap- (3) Does the FTT affect cost of capital and
plied in other parts of the world, allowing us stock prices? Also, does the FTT cause cor-
to analyze the impact of the tax on trading porate management to emphasize long-term
volume, volatility, and price efficiency.17 or short-term results relatively more? (4) Will
The rest of the paper is organized as fol- the FTT cause trading to shift to overseas
lows. In the following section, we review the untaxed markets and make domestic mar-
theoretical arguments for and against a FTT. kets less competitive? (5) Will the FTT raise
Specifically, we review the literature on the substantial tax revenue?
analysis of the imposition of a FTT on trad-
ing activities (measured in terms of trading Reduced Excess Speculation and Price
volume) and price volatility. Next, we review Volatility
the empirical evidence on the imposition of Proponents of an FTT have suggested
3
4. that a transaction tax may reduce specula- This line of argument for a transac-
tive trading and excess market volatility. tion tax can be attributed to John Maynard
They believe short-term speculative trading Keynes. In light of excessive speculation
is the source of excess volatility, so by impos- and volatility during the Great Depression,
ing a transaction tax, short-term speculative Keynes proposed a securities transaction tax
trading and price volatility will be reduced. as a means of mitigating the predominance
The argument rests on the assumption of speculation in the stock market during
that there is a positive relationship between the Great Depression.18 Witnessing excess
short-term speculation and excess price vol- volatility in the foreign exchange markets
atility. To FTT proponents, there appear to after the dissolution of the Bretton Woods
be two types of traders in a financial mar- Agreement, James Tobin proposed an inter-
ket: value investors and noise traders. Value national transfer tax on currencies in 1978.
investors, also known as fundamentalists, Keynes’s and Tobin’s proposals, although
purchase stocks on the basis of comparison made decades apart, were based on the
of security price with estimates of funda- same assumption that short-term trades are
mental values. That is, they buy stocks when likely to be more destabilizing to financial
market price is below the fundamental value markets than longer-term trades.19 In sub-
and sell stocks when market price is above sequent work, Tobin stated that a financial
the fundamental value. This value-based transaction tax is “fundamental valuation
trading is assumed to reduce stock price efficient” since it lowers excess volatility. 20
volatility by pushing stock prices back to- In 1953, however, Milton Friedman ar-
ward estimates of the worth of the compa- gued that speculation cannot be destabiliz-
ny. Conversely, short-term noise traders act ing in general; if it were, the participants
on the basis of past price movements or the would lose money.21 Other advocates of the
results of technical analysis of stock market Efficient Market Hypothesis argue that spec-
data itself. They purchase stocks when mar- ulators—by rationally arbitraging the unex-
ket prices rise and sell the stocks when they ploited profit opportunities when a market
fall. This type of trading, based on positive becomes inefficient—help to clear markets,
feedbacks from market prices, is assumed to stabilize prices, and bring the assets and se-
destabilize markets because it often drives curities back to their fundamental values.22
market price away from estimates of funda- This suggests that the impact of a transaction
mental values and thus creates excess price tax on price volatility may hinge on whether
volatility. Value investors who trade on the markets are dominated by speculators, ar-
basis of market price deviations from the bitrageurs, or long-term investors. In other
fundamental values are long-term inves- words, any FTT’s success is dependent upon
tors and have no need to trade frequently. the composition of traders in the market.
Keynes proposed On the other hand, short-term speculative Joseph Stiglitz suggests that a desirable
a securities traders do need to trade frequently because FTT should not impede the functioning of
transaction tax their strategy is to follow recent past price the capital market as an allocator of scarce
behavior. Because the trading frequency resources. As such, an FTT based on the
as a means of of short term traders is much greater than value of the transaction (e.g., a turnover tax
mitigating the that of long-term investors, the imposition on trades) should be broad-based in order to
predominance of a transaction tax will increase the trad- avoid the frequent introduction of unneces-
ing cost for short-term speculative traders sary distortions, set at a low rate, and be eq-
of speculation in but will have less impact on the trading cost uitable. Stiglitz further suggests that a suffi-
the stock market of long-term value investors. As a result, a ciently small transaction tax (~ 0.5 percent to
transaction tax will curb the frequency of 1 percent) is negligible and would not affect
during the Great short-term speculative trading and thus, exchange efficiency, but would have different
Depression. theoretically, curb excess volatility. impacts on the welfare of different groups of
4
5. traders. He asserts that the uninformed trad- have irrationally or that waste too many re- Proponents of
ers are hardly likely to be affected by a tax at sources for this speculative zero-sum game.24 transaction taxes
a moderate rate of less than 1 percent. Like- Short-term technical traders are not neces-
wise, the informed traders (insiders) would sarily amateurs or low-volume traders. Port- argue that they
not be discouraged from trading with that folio managers, for example, are often evalu- will discourage
amount of transaction tax. This is because ated on the basis of quarterly performance.
they operate more often on long-term bases Thus, portfolio managers are incentivized
short-term
for valuation and thus are unlikely to have to maximize performance in the near term. trading.
their behavior greatly affected.23 This leads them to give short-term prospects
Stiglitz posits that the turnover tax pri- a disproportionate weight in determining
marily affects short-term market partici- stock purchases. As a consequence, corpo-
pants—noise traders and speculators—who rate managers are forced to slight long-term
buy and sell within the trading day and within investment in favor of delivering short-term
days or weeks. As such, a transaction tax may earnings.
represent a significant fraction of the returns Proponents of transaction taxes argue
they hope to achieve on each transaction. He that they will discourage short-term trading
argues that the large number of noise trad- and reduce the number of speculative short-
ers and liquidity providers (those who trade term traders due to higher trading costs
with the noise traders) bear the lion’s share in the markets. More market participants
of the transaction tax and may actually ex- would, theoretically, look beyond quarterly
perience a welfare gain from impeding these earnings reports and short-run prospects,
exchanges. There may be greater volatility if resulting in more stable prices. In this mod-
the FTT is too big (barring arbitrage trades), el, corporate managers would pursue more
but this is unlikely if the tax is small. Based long-term investment projects.
on this logic, Stiglitz establishes that the
upper bound on the volatility increase will Reduced Cost of Capital
not be greater than that without the trans- Stock markets allow firms to raise new
action tax. He therefore believes that if the capital from shareholders by way of exchange.
tax is small, there will be significant reduc- Thus, a transaction tax that impedes the ex-
tion in volatility as noise traders drop out change function of the stock market might
of the market. Thus, he argues that such a interfere with the capital raising function of
tax may actually be beneficial because it dis- the market, ironically forcing management
courages short-term speculative trading. The and investors to focus on short-term returns
tax won’t affect the long-term investors too rather than long-term concerns.25 Stiglitz ar-
much because a transaction tax, on average, gues that this potential impact is negligible
has the property of automatically phasing for a small transfer tax and, to the contrary,
itself out for long-term investments; that is, would enhance the capital-raising function
as a proportion of returns, it becomes negli- of the stock market if the tax reduces stock
gible as the holding period increases. Thus, market volatility.26 Reducing market volatil-
the tax will not have a significant effect on ity will make it easier for firms to raise eq-
long-term investors. He suggests this feature uity capital at a lower cost, thus increasing
makes a turnover tax more desirable than a efficiency. If true, management would focus
capital gains tax because a capital gains tax their orientation toward a longer-term strat-
subsidizes noise traders and penalizes arbi- egy.27
trageurs, leading to increased price volatility.
Lawrence Summers and Victoria Summers Increased Tax Revenue
also agree that a securities transactions tax Transaction tax proponents suggest that
improves the efficiency of financial markets the revenue potential of a transaction tax
by crowding out market participants that be- is formidable.28 The Congressional Budget
5
6. Office (CBO), in its publication Reducing Franklin Edwards argues that transac-
the Deficit: Spending and Revenue Options, esti- tion taxes increase trading costs, making
mated the revenue from a broadly based 0.5 U.S. futures markets less competitive be-
percent securities transaction tax to be about cause of the impact on price efficiency and
$12 billion per year based on a five-year aver- on the cost of hedging. He argues that a tax-
age. Based on the same tax rate used by the induced reduction in trading may decrease
CBO, Summers and Summers suggest a sim- informational efficiency by discouraging
ilar figure, estimating government revenues “information” trades by informed specula-
of at least $10 billion a year.29 Another esti- tors and hedgers. He admits that it is not
mate indicates that revenue from a securities easy to determine the impact on price effi-
transaction tax could be as large as $70–$100 ciency because the tax also discourages noise
billion per year.30 Outside the United States, trading.35
it was estimated that a securities transac- Evidence from a pair of studies suggests
tion tax in Japan would bring in $12 billion that reducing the transaction tax in the Tai-
a year.31 The European Commission in a wan futures market greatly improved the
June 2011 budget proposal calculated that efficiencies of price execution36 and price
a financial transaction tax would contribute discovery.37 Likewise, a study by Shinhua
A small fixed €50 billion per year to the European budget, Liu examined the impact of a 1989 change
transaction cost or €350 billion over a seven-year period.32 in tax rates on securities in Japan. Liu found
significantly significant decreases in estimates of the first
Effects on Trading Volume, Market auto-correlations in returns for Japanese
reduces trading Liquidity, and Information Efficiency stocks listed in Japan, but no changes for
volume. Some literature suggests that there is a Japanese stocks dually listed in the United
negative relationship between trading vol- States as American Depository Receipts
ume and trading costs. Increases in trading (ADRs), which were not subject to the tax
costs lower the profitability of trading, lead- law change. Liu also found a lower price ba-
ing traders to trade less frequently or extend sis between the ADRs and their underlying
their hold period in order to minimize their Japanese stocks, concluding that these re-
trading costs over time. As trading volume is sults are consistent with the hypothesis that
reduced, traders will take more time to off- a reduction in transaction costs (transaction
set their trades and face a larger price impact tax) improves the efficiency of the price dis-
of a given trade, thus diminishing market li- covery process.38
quidity as well. When the market is illiquid,
information will be more slowly incorporat- FTTs Do Not Necessarily Reduce Price
ed into equity or futures prices, impairing Volatility
overall market efficiency. Donald Kiefer argued that a transaction
Andrew Lo, Harry Mamaysky, and Ji- tax can theoretically increase or decrease
ang Wang proposed a dynamic equilibrium volatility.39 Paul Kupiec demonstrated that
model of asset prices and trading volume. a transaction tax has ambiguous effects on
It shows that a small fixed transaction cost price volatility in a general equilibrium mod-
significantly reduces trading volume.33 Even el framework. In the context of his model, he
Stiglitz, a supporter of the transaction tax, shows that a transaction tax can reduce the
agrees that a sizable transaction cost can re- price volatility of risky assets.40 However, the
duce trading, thinning market liquidity if the reduction in price volatility is accompanied
buy and sell sides are symmetric, although by a fall of the taxed asset’s price, while con-
“for widely traded stocks, on both theoretical versely the volatility of risky asset returns will
and empirical grounds, it is hard to believe increase with the transaction tax. Thus, the
that this effect [larger bid-ask spread due to a net effect of a transaction tax on price volatil-
transaction tax] would be significant.”34 ity could be to increase it, decrease it, or leave
6
7. it unchanged, depending on other factors in volatility is caused by changes in the relative
the scenario. share of noise traders and fundamentalists.
In a general equilibrium model, Frank In their general equilibrium model, Shi and
Song and Junxi Zhang examined the effects Xu analyzed entry costs for both informed
of a transaction tax on a set of noise trad- and noise traders after an introduction of
ers and the resulting market volatility. They a transaction tax. The model assumed in-
showed that a transaction tax may not only formed traders’ unconditional expectation
discourage the trading activity of noise trad- of excess return depends on the ratio of noise
ers but also discourage rational and stabi- entrants to informed entrants, but this does
lizing value investors from trading.41 The not influence noise traders’ expectations. An
net effect of a transaction tax on volatility increase in the noise component increases
depends on the change of trader composi- market volatility. In analyzing three equilib-
tion that results from the implementation ria with different entry costs to the market,
of the tax. They referred to this as the “trader their major finding was that the imposition
composition effect.”42 Furthermore, a trans- of a Tobin tax did not reduce volatility and
action tax may decrease trading volume and may, in fact, increase it, depending on the ra-
increase the bid-ask spread. This potential tio of noise to informed entrants. An increase
effect of a transaction tax on liquidity is la- in market volatility was also an important as-
beled as the “liquidity effect.” The net im- sumption for the impact of a transaction tax,
pact of a transaction tax could decrease or demonstrated in models assuming stochas-
increase market price volatility. The final re- tic interaction between agents, who are as-
sults depend on the relative magnitude and sumed not to be able to influence aggregate
interaction of the trader composition and variables.45 In these models, exchange rate
liquidity effects. volatility will be low if the market is domi-
Paolo Pellizzari and Frank Westerhoff an- nated by fundamentalists and will be high
alyzed the effect of a transaction tax on mar- if the market is dominated by noise trad-
ket price volatility in a number of computa- ers. These models reflect the stylized facts
tional experiments. They showed that the of financial markets, most notably, “volatil-
effectiveness of transaction taxes depends ity clustering”—in which the exchange rate
on the types of trading markets: specifi- switches irregularly between phases of high
cally, they compared a continuous double- and low volatility.
auction market versus a dealership market. Overall, the implications of theoretical
In a continuous double auction market, the models on the price volatility effects of an
imposition of a transaction tax is not likely FTT are mixed. Conclusions about the im-
to reduce market volatility since a reduction pact of a transaction tax on price volatility
in market liquidity amplifies the average depend on the assumptions of the theoreti-
price impact of a given trade order. Liquid- cal models and assumed mechanisms of in-
ity is endogenously generated in this type of formation transmission. We will examine
market. Their model predicts that in a deal- some of these in section III.
ership market, a transaction tax may reduce
market volatility because abundant liquidity Increased Costs of Capital and of
is exogenously provided, prompting special- Hedging The net effect
ists and some traders to retreat from the Trading costs affect stock prices because of a transaction
market, causing volume to decline.43 trading costs reduce the expected return of
Kang Shi and Juanyi Xu examined the im- stocks. Investors demand higher expected tax on volatility
pact of a transaction tax on foreign currency return when paying increased costs. depends on the
transactions, which was designed to limit the Yakov Amihud and Haim Mendelson
impact of noise traders in order to reduce found that the expected rate of return on
change of trader
volatility.44 They also believe exchange rate equities (i.e., the cost of equity) is an increas- composition.
7
8. The imposition ing concave function of the bid-ask spread, trading in 1989, they admitted that a trans-
of transaction a proxy measure of liquidity.46 Since other action tax could have damaging impacts on
transaction costs of equity trading (e.g., the the industry as evidenced in the demise of
taxes will brokerage fee) are positively related to the the Sweden Options and Futures Exchange
increase the bid-ask spread, the estimates obtained by following implementation of a tax on op-
Amihud and Mendelson imply that, for a tions.51
trading costs on given increase in the bid-ask spread, expect- Joseph Grundfest and John Shoven sug-
stocks, and thus ed returns increase at a larger amount as the gest that an STT would cause distortions in
investors will equity issue becomes more liquid. Likewise, the financial markets and could cause many
a securities tax that is analogous to a bigger investors—particularly institutions—to shift
demand a higher bid-ask spread will raise the expected return their equity trading away from organized
expected return of equity (i.e., the cost of capital). Later stud- domestic exchanges toward foreign coun-
commensurate ies have also documented two similar re- tries. They believe even a small STT can have
sults: (1) The greater the liquidity of a given major adverse consequences for the value of
with the added stock, the lower its expected return,47 and instruments subject to the tax and for the
cost. (2) lower trading costs are associated with cost of capital in the U.S. economy. They
lower expected return of the stock.48 also criticize the CBO’s static model because
In 2002 Amihud investigated the effects it does not consider the STT’s effect on trad-
of changing overall market liquidity on stock ing volume or market prices, nor does it
prices over the period from 1963 to 1996. He consider the possibility of substitution away
observed that a decline in market liquidity from taxable instruments and transactions.
was accompanied by a significant decline in Thus, they contend, the CBO overstates the
stock prices and subsequent increase in ex- actual tax revenue that can be collected.52
pected return (i.e., the cost of capital).49 Pre- Franklin Edwards has argued that even
vious studies clearly indicate that the impo- a very small transaction tax would be suf-
sition of transaction taxes will increase the ficient to drive all U.S. futures trading to
trading costs on stocks, and thus investors untaxed overseas markets.53 He considered
will demand a higher expected return com- a tax of 0.5 percent on the value of the con-
mensurate with the added cost. As a conse- tract to be prohibitively high as a percentage
quence, firms’ cost of equity would rise and of total transaction cost on trading as com-
their stock prices will decrease. pared to stocks.54 Should a lower rate be set
Franklin Edwards argued that a declin- on futures trading vis-à-vis stocks, the differ-
ing trading volume due to a transaction tax ence in the transaction tax may cause traders
would likely increase the risk premiums that to pursue transactions that bear a lower tax.
hedgers would have to pay to speculators Substitution may take place domestically or
who provide liquidity. This makes futures across international borders. This is why Sti-
less efficient risk management instruments glitz suggested in 1989 that transaction tax
and thus the FTT undermines one of the rates should be uniform within the United
primary economic functions of futures mar- States on substitutable assets.55
kets.50 Edwards also pointed out that a critical
feature of futures markets across the globe is
Migration of Trading and Relative Com- low transaction costs. “If U.S. markets were
petitiveness to have higher trading costs . . . it would
Previous literature on transaction taxes be a relatively simple matter for trading to
shed light on the potential adverse effects of shift to foreign markets.”56 The shift will
FTTs on the international competitiveness take place because: (1) there are restrictions
of the U.S. financial services industry. While that require foreign exchanges to trade the
Summers and Summers did not believe a same contract in order to compete for the
transaction tax would cripple U.S. equities same business (e.g., TAIFEX and Nikkei 225
8
9. indexes were traded on the Singapore Ex- the actual transaction tax. Some caveats are
change) and commodity futures are traded in order here. First, although test results on
all over the world; and (2) there are no re- the hypothesized impact of a transaction tax
strictions on Americans shifting their trad- are useful in explaining its possible impact,
ing to foreign exchanges. This has been the the actual impact may differ in practice. An
main reason why the futures industry op- explicit tax charged on a transaction may
poses a transaction tax on futures trading. not be perceived as an implicit transaction
John Campbell and Kenneth Froot ex- cost embedded in the bid-ask spread or a
plained that investors in Sweden moved eq- reduction in brokerage fee. Second, despite
uity trading offshore and fixed income trad- controlling for variables that could affect the
ing to untaxed local substitutes in response empirical results, the results obtained from
to the country’s imposition of an STT. In the foreign countries where market environ-
United Kingdom, the stamp duty STT stim- ments are different may vary considerably
ulated trading in untaxed substitute assets when and if an FTT is applied to U.S. mar-
and seemed to reduce total trading volume kets. Third, because most studies are done in
to some degree. They concluded that if an securities and foreign exchange markets and
STT is aimed at reducing overall trading vol- very few in futures markets, observers should
ume or raising revenue, most likely it won’t recognize the limits of similarities between Investors in
achieve its goal because investors would the results and implications obtained from Sweden moved
move trading to offshore markets or un- those markets and what would happen if equity trading
taxed assets. They believe the British type of they were applied to a futures market.
STT would be more workable for the United offshore and
States than the Swedish type.57 Effects on Trading Volume and Market fixed income
Liquidity
One major argument against transac-
trading to
What Does the tion taxes is that a transaction tax would untaxed local
Empirical Evidence increase trading cost, which would reduce substitutes.
trading volume and market liquidity. A nar-
Say about the FTT? rowly based transaction tax would provide a
strong incentive for traders to migrate to an
There are ample empirical studies on the alternative domestic instrument or to un-
impact of FTTs on various aspects of finan- taxed foreign markets that have lower costs.
cial market quality, including trading vol- Furthermore, a reduction in trading volume
ume, volatility, liquidity, and price discov- would increase trading costs (e.g., a wider
ery. In general, the literature can be placed bid-ask spread) and decrease market liquid-
into two groups. The first group of studies ity. Market and price efficiency would be im-
has used direct ex post tests on the impact paired when market liquidity deteriorated.
of transaction taxes on market quality in Several studies provide estimates of the
countries where actual direct taxes, whether elasticity of trading volume in equity mar-
STTs or Tobin taxes, had been charged on kets. In 1976 Thomas Epps estimated the
financial transactions. share transaction cost turnover elasticity
The second group of studies has used to be about -0.26 in U.S. equity markets,59
ex ante analysis of the impact of a transac- whereas Patricia Jackson and Gus O’Donnell
tion tax on the quality of financial markets estimated the transaction cost turnover
where there have been no actual STT or elasticity of equities traded in London to be
Tobin taxes. The studies use different mea- -0.70 nine years later.60 Put simply, previ-
sures and proxies of transaction costs (e.g., ous studies find that transaction costs and
changes in bid-ask spreads, brokerage com- trading volume have a negative relationship.
mission, and tick size58) but not necessarily Likewise, empirical studies find that the
9
10. SST had a negative effect on local trading. fect on the volume of trade in Swedish eq-
For example, Steven Umlauf documented in uities by foreign institutions. There is little
1993 that 60 percent of the trading volume evidence that total trading volume in Swed-
of the 11 most actively traded Swedish share ish stocks responded strongly to changes in
classes, amounting to 30 percent of the total taxation of trades in Stockholm. This lends
trading volume, shifted to the London stock additional support to the view that interna-
exchange when the Swedish transaction tional investors easily evaded Swedish turn-
tax on equity increased from 1 percent to 2 over taxes. They find that the transaction tax
percent in 1986.61 Two econometric studies had a larger impact on local fixed-income
on equity turnover, one in the United King- trading volume than on stocks. Campbell
dom62 and one in Sweden,63 found that the and Froot offer several observations: (1) The
long-run elasticity of turnover with respect effect of the tax seems to be quite large; (2)
to overall transactions costs is in the range much of the volume decline in futures oc-
of -1 and -1.7. Their best estimate is -1 (i.e., curred in anticipation of the tax; (3) these ef-
for each reduction or removal of 1 percent fects ran in reverse once the tax was removed
round trip transaction tax (or 0.5 percent on in April 1990. In short, the turnover tax in
buy and 0.5 on sell), trading volume would fixed income securities raised little revenue
increase by 100 percent.64 since substitution toward other Swedish do-
Taking into consideration the margins mestic securities was easy, with little need
of substitution, Campbell and Froot esti- for migration abroad given the existence of
mated the elasticity of trading volume af- less costly domestic substitutes.
ter changes in transaction taxes in Sweden Campbell and Froot proposed two prin-
and found evidence that foreign investors ciples that might be used to rationalize
tended to move toward more trading abroad transaction tax rates across securities. The
and domestic investors became less likely to first principle is that the transactions that
engage in any trading at all. They reported give rise to the same pattern of payoffs
that when the SST was in place from 1988– should pay the same tax, though they admit
91, the fraction of trading taking place in that it is conceptually impossible to apply
Stockholm was much lower for unrestricted this principle consistently. The second prin-
shares. This is corroborated by further evi- ciple is that transactions that use the same
dence that commissions paid by large U.S. resources should pay the same tax. For ex-
institutional investors when trading Swed- ample, they point out that Sweden used to
ish equities remained constant but the share tax domestic brokerage services, whereas the
Sixty percent of their taxes paid fell from 68 percent in United Kingdom taxes registration (i.e., the
of the trading 1987 to 13 percent by 1990. That is, foreign stamp duty).
investors such as U.S. institutions (and their Previous evidence of STT impacts in eq-
volume of the brokers) were increasingly able to evade the uity markets shows that a transaction tax re-
11 most actively tax by eliminating the use of Swedish bro- duces trading volume and market liquidity.
traded Swedish kers when trading in Sweden or by exchang- For highly elastic instruments, substitution
ing Swedish securities in London and New will take place, driving some or all trading
share classes York. They reported that by 1990, 50 percent to overseas markets where the tax rates are
shifted to the of trading volume was shifted to the London lower, or out of the market entirely.
London stock equity exchange. The authors also found Likewise, previous studies in futures mar-
that the Swedish tax shifted fixed-income kets demonstrated a statistically significant
exchange when trading activity from income-securities and negative relationship between trading vol-
the Swedish futures markets to untaxed markets such ume and trading costs in U.S., Taiwanese,
as variable-rate notes, corporate loans, and and Indian futures markets.65 For example,
transaction tax on forward rate agreements. They contended Johan Bjursell, George Wang, and Jot Yau ex-
equity increased. that the Swedish tax had only a marginal ef- amined the relations among trading volume,
10
11. bid-ask spread, and price volatility in 11 U.S. The first group of literature includes the Transaction
futures markets from 2005 to 2010. They work of Harold Mulherin, who examined taxes do not
found that a transaction tax would have the the relationship between trading costs in
same effect as a wider bid-ask spread, reduc- the New York Stock Exchange and the daily necessarily cause
ing trading volume, increasing price volatil- volatility of the Dow Jones Industrial Aver- volatility to
ity, and generating a modest amount of tax age returns from 1897 to 1987. He conclud-
revenue.66 ed that the imposition of a transaction tax
decrease.
Robert Aliber, Bhagwan Chowdhry, and may not necessarily reduce volatility.69 Like-
Shu Yan studied the impact of a small trans- wise, Charles Jones and Paul Seguin used the
action cost (averaged around 0.05 percent) elimination of the minimal brokerage com-
on the trading volume and price volatility of missions in the U.S. stock market in 1975 as
four currency futures traded on the Chicago a proxy for a one-time reduction in the trans-
Mercantile Exchange (CME) over the period action tax. They found that volatility fell the
of 1977–1999. They found that an increase year after the abolishment of the minimum
of 0.02 percent in transaction costs leads to a commission rates in NYSE and AMEX mar-
reduction in trading volume as well as an in- kets, but the decline in volatility was also ob-
crease of volatility of 0.5 percentage points.67 served in the NASDAQ market.70
Robin Chou and George Wang found Using the cross-sectional data of 23 coun-
that before Taiwan cut the tax on the Taiwan tries for the period up to, during, and after
Index (TIX) futures trading by 50 percent in the October market crash (1987–1989),
2000, futures trading volume was smaller Richard Roll found no significant evidence
than that in the Singapore futures exchange. that volatility is negatively related to trans-
However, since July 2002, the trading volume action taxes.71 In other words, transaction
for TIX exceeded that of the same contract taxes do not necessarily cause volatility to
on the Singapore futures exchange.68 This decrease across countries, and it is question-
evidence indicates that lower transaction able whether transaction taxes should be
costs change the relative competitiveness used with confidence as an effective policy
of exchanges and significant migration of instrument. Shing-yang Hu analyzed nu-
trade may take place because of lower trad- merous changes in STT rates in Hong Kong,
ing costs. Japan, Korea, and Taiwan during the period
In summary, evidence in extant futures 1975–1994 but concluded that, on average, a
literature is consistent with the anti-tax change in STT rates had no effect on volatil-
hypothesis that increasing trading costs ity and market turnover.72
through a transaction tax would reduce In one study, Ragnar Lindgren and An-
trading volume and market liquidity, and ders Westlund found no significant effect of
increase price inefficiency of taxed financial an STT on price volatility of Swedish stocks,
instruments. but in later work they found a weak positive
relationship between STT and price volatil-
Effects on Price Volatility ity.73 Steven Umlauf examined the impact
The empirical studies of transaction of increasing the transaction tax on market
taxes’ impact on price volatility can be clas- price volatility. He reported several inter-
sified into two groups. The first group of pa- esting results: (1) there was no significant
pers that examine the relationship between difference in volatility across the three tax
price volatility and trading costs does not regimes (i.e., before 1984 [0 percent], 1984–
find any definitive pattern or relationship, 1986 [1 percent ], after 1986 [2 percent]);74
whereas the second group of papers finds (2) there was a statistically significant in-
evidence of either an increase or a decrease crease in daily volatility of returns, which
in volatility. Putting them together, empiri- was higher during the 2 percent tax regime
cal results are inconclusive. than in other regimes, but there was no sys-
11
12. tematic relationship between transaction tax trading volume for selected U.S. futures
regimes and volatility;75 and (3) the volatil- contracts. Pravakar Sahoo and Rajiv Kumar
ity of London-traded shares of 11 companies analyzed the relations for five most traded
was lower than the volatility of these com- commodity futures contracts in India. These
panies’ Stockholm-traded classes.76 Robin studies showed that there is a significantly
Chou and George Wang found that there positive relation between price volatility and
were no significant changes in the daily price bid-ask spread (transaction costs) for each
volatility of Taiwan index futures after the futures contract examined.83
tax reduction.77 Robert Aliber, Bhagwan Chowdhry, and
Among studies that belong to the second Shu Yan studied the impact of a small trans-
group, Shinhua Liu and Zhen Zhu studied action cost (averaged around 0.05 percent)
the deregulation of fixed brokerage commis- on four currency futures traded on CME in
sions and the removal of an STT in Japan in the period of 1977–1999 on their trading vol-
October 1999, and found results contrary to ume and price volatility. They found that an
those of Jones and Seguin. They found that increase of 0.02 percent in transaction costs
the reduction in transaction costs (in which on the four currency futures traded on CME
an STT was included) increased volatility in leads to an increase of volatility of 0.5 per-
Most of the the Tokyo Stock Exchange.78 Liu and Zhu cent points, coupled with a decline in asset
previous offered a possible explanation for contradic- prices due to the decline in the demand be-
empirical tory results: commission rates were drasti- cause of higher transaction costs.84 Markku
cally reduced in Japan, whereas they were Lanne and Timo Vesala found larger trans-
evidence does not not in the United States. Hendrik Bessem- action costs impact on foreign exchange rate
support the use binder found that larger tick sizes were as- volatility between 1992–1993.85 Badi Baltagi,
sociated with higher transaction costs and Dong Li, and Qi Li investigated the effect of
of a transaction also with higher volatility.79 Bessembinder an increase in the stamp tax on price volatil-
tax as an effective and Subhrendu Rath found that stocks ity in the two Chinese stock exchanges using
regulatory policy that had moved from NASDAQ to NYSE, an event study methodology. They found
where trading costs were lower, saw a reduc- market price volatility significantly increased
tool to reduce tion in volatility.80 Harald Hau also found after the increase in the stamp tax rate.86
market price a positive relationship between transaction Overall, most of the previous empirical
volatility. costs and price volatility in the French stock evidence does not support the use of a trans-
market, where significant volatility increases action tax as an effective regulatory policy
were observed when there was an increase in tool to reduce market price volatility.
the cost of trading stocks due to an increase
in the tick size.81 Effects on Information Efficiency and
Franklin Edwards examined the relation Price Discovery
between trading volume and volatility for Robin Chou and Jie-Haun Lee provided
16 U.S. commodity markets during 1989 interesting empirical evidence of the effect of
and found no significant relationship be- a transaction tax cut on the price efficiency
tween the two.82 He concluded that even if a of the Taiwan Futures Exchange (TAIFEX)
transaction tax were to succeed in reducing and the Singapore Stock Exchange (SGX).87
speculative or short-term trading in futures They demonstrated that after the tax reduc-
markets, there was no evidence that it would tion in 1986, the TAIFEX assumed a lead-
reduce price volatility in either futures or ing role over the SGX in the price discovery
the underlying spot markets. Based on the process for index futures contracts. They
three-equation structural model used by showed that a reduction in the transaction
George Wang and Jot Yau in a 2000 study, tax greatly improves the efficiency of price
Bjursell, Wang, and Yau examined the rela- execution. Wen-Liang Hsieh also noted that
tions among volatility, bid-ask spread, and the information advantage of the SGX di-
12
13. minished as the TAIFEX lowered its transac- magnitude of 0.5 percent would probably
tion tax.88 Badi Baltagi, Dong Li, and Qi Li eliminate all futures trading in the United
found that volatility shocks were not quickly States, driving all of those transactions
incorporated into stock prices in the Shang- overseas. Of course, no revenue would be
hai and Shenzhen exchanges once China collected in that case. According to his con-
had increased its STT from 0.3 percent to 0.5 servative estimate, only negligible revenue (~
percent in July 1997, suggesting that the in- $287 million) could be raised from futures
crease adversely affected the price discovery trading even if the lowest tax rate (0.0001
function of the stock market.89 In contrast, percent) and a low demand elasticity (-0.26)
Shinhua Liu showed a reduction in the first- were assumed.97 He also computed esti-
order autocorrelation of Japanese stock pric- mated potential tax revenue with a range
es after the STT reduction in 1989, implying of assumed elasticities (from -1 to -20) and
an improvement in the efficiency of the price concluded that a transaction tax on futures
discovery process. trading would not generate substantial rev-
enue.98
Effects on Potential Tax Revenue George Wang, Jot Yau, and Tony Baptiste
Transaction tax proponents argue that provided the first empirical estimates of the
the revenue potential of a transaction tax is elasticity of trading volume for several U.S.
formidable.90 The U.S. Congressional Budget futures contracts. They documented that
Office estimates the revenue from a broad- estimates of trading volume elasticity with
based 0.5 percent securities transaction tax to respect to trading costs were in the range
be about $12 billion per year over five years.91 of -0.116 to -2.72, which were less than the
Based on the same tax rate used by the CBO, elasticities (-5 to -20) used by Edwards, but
Lawrence Summers and Victoria Summers higher than the elasticity of -0.26 used in
provided a similar estimate of at least $10 bil- the CBO report.99 Bjursell, Wang, and Yau
lion a year.92 Robert Pollin, Dean Baker, and provided empirical estimates of the elastic-
Marc Schaberg suggested that revenue from ity volume for 11 U.S. futures for the years
an STT could be as large as $70–$100 billion 2005–2010 in a three-equation structural
per year.93 Outside of the United States, a model.100 They estimated the potential post-
Japanese STT was estimated to bring in $12 tax trading volume and tax revenue with up-
billion a year in the late 1990s,94 and the dated cost estimates of trading volume elas-
European Commission expected an FTT to ticity under alternative tax rates. For a 0.02
provide €50 billion per year over a seven-year percent tax rate, they found that the trading
period as recently as 2011.95 volume for six futures (S&P 500, E-mini S&P
Edwards doubts that a tax on futures 500, 10-year T-Note, British pound, soy-
transactions would potentially generate sig- beans, and gold) would be totally eliminat-
nificant tax revenue. He argues that the elas- ed, resulting in zero post-tax revenue from
ticity of trading volume in futures markets these six futures.101 They concluded that a
is much more than that of equities because sizable transaction tax could have signifi-
close substitutes are easily available in inter- cant adverse impacts on market quality and
national futures markets, due to the growth would not raise substantial revenue for the A sizable
in trading and information technologies.96 government as suggested in other studies. transaction
Thus, he contends that the CBO overesti- Moreover, the tax could potentially hurt the
mated the revenue from the STT because international competitiveness of U.S. futures
tax could have
the elasticity of demand (-0.26) used in the markets. A study on Indian futures trading significant
estimate was based on an assumption of no also concluded that implementation of a adverse impacts
suitable international substitutes. Given a transaction tax would not bring in substan-
more elastic trading volume in futures, Ed- tial tax revenue.102 on market
wards argues that a transaction tax of the Chou and Wang found that the 50 per- quality.
13
14. Trading cent reduction in the TAIFEX transaction Estimation of Potential
volume may tax rate reduced tax revenue, but the pro- Transaction Tax Revenue
portional decrease in the tax revenue (30
precipitously percent) was less than the 50 percent reduc- One major argument for implementing
decline in tion in the tax rate. Interestingly, tax revenue a transaction tax in security and futures
increased in the second and third year after markets is to raise substantial tax revenue
response to the tax reduction when compared to the for the government. However, proponents
increased year before the tax reduction.103 This sug- of the transaction tax often employ a naïve
tax-induced gests that tax reduction has no permanent method to calculate transaction tax revenue,
negative impact on tax revenue. multiplying the tax fee by current aggregate
trading costs. Finally, some of the literature suggests trading volume in the given markets assum-
that the burden of transaction taxes on mar- ing a static model.105 In other words, their
ket participants depends on the availability models assume the imposition of a tax will
of no-tax substitutes for their instruments. not affect the trading volume in the market.
For instance, the elasticity of financial and This assumption can vastly overestimate
metals futures are much higher than those the potential tax revenue because it does
of agriculture futures. Thus, the traders of not take into account the relation between
agriculture futures would have a larger tax transaction costs and trading volume (see
burden relative to traders of financial fu- no. 1, above). Trading volume may precipi-
tures.104 A transaction tax would raise the tously decline in response to increased tax-
relative cost of hedgers using agriculture fu- induced trading costs.
tures to hedge against their underlying asset William Schwert and Paul Seguin pre-
price risk. sented a model to estimate tax revenues that
In sum, the review presented above sug- accounts for the impact of the transaction
gests the following: tax on trading volume and price volatility.106
They assumed a flat tax rate, τ, for all finan-
1. There is an inverse relationship be- cial transactions. Revenues can be estimated
tween transaction cost (bid-ask spread) as follows:
and trading volume; R = τ (P + ΔP) (Q + ΔQ) + ΔOR, where R is
2. There is a positive relationship that the revenue, P the volume-weighted average
may or may not be statistically signifi- price level, Q the quantity of transactions, ΔP
cant between transaction cost and price and ΔQ the change in P and Q, respectively,
volatility; and ΔOR the change in other government
3. There is a positive relationship between revenues associated with the tax. The magni-
trading volume and price volatility; tude of the decline in trading volume (ΔQ)
4. Relationships among trading volume, depends on the elasticity of trading volume,
bid-ask spread, and price volatility are which requires estimation with respect to the
jointly determined; percentage increase in trading costs due to
5. Demand for U.S. futures trading is very the transaction tax for each financial instru-
sensitive (i.e., very elastic) with a strong ment (see no. 5, above). Likewise, the impact
substitution effect between domestic of the transaction tax on prices (ΔP) needs
and untaxed overseas markets; and to be estimated. More importantly, previous
6. Although estimated potential tax rev- studies have shown that the elasticity of de-
enue is formidable in the equities mar- mand in the futures market is not likely to be
kets, tax revenue raised by a transaction the same as that in the equities market.107 As
tax in the U.S. futures markets would such, the potential revenue that can be raised
not be as much as many would believe from a transaction tax in these two markets
because the demand for U.S. futures is can be substantially different depending on
found to be very elastic. the differing elasticities (see no. 6, above).
14
15. It can be inferred that the estimation of estimating the elasticity of trading volume
the elasticity of trading volume is critical to for the purpose of estimating the potential
the accurate estimation of the potential tax tax revenue, one needs to have a model that
revenue. Previous studies found a strong allows relevant variables to be jointly deter-
and significant positive relation between mined in estimating the parameters of the
trading volume/liquidity and price volatil- model. However, in Wang, Yau, and Bap-
ity (see no. 2, above), and an inverse rela- tiste’s model, price volatility was omitted.
tion between transaction costs and trading It is imperative to estimate the elasticity of
volume/liquidity (see no. 1, above). The em- trading volume with respect to the transac-
pirical relation between price volatility and tion tax (bid-ask spread) in a structural mod-
transaction costs depends on how transac- el that jointly determines the relationships
tion costs affect trading volume, which, in between price volatility, bid-ask spread, and
turn, affects the price volatility as theory trading volume.111 In 2000 George Wang
suggested. Model specifications of these and Jot Yau proposed a three-equation struc-
previous studies, however, are incomplete tural model that allows trading volume and
because trading volume is assumed to be a price volatility to be jointly determined to-
function of transaction costs and/or price gether with transaction costs in the estima-
per share only.108 Likewise, one study found tion of tax revenues, which has been used in
The empirical
that trading volume was not only a function other studies discussed above.112 relation between
of volatility, but also one of open interest, In 2011, Bjursell, Wang, and Yau provid- price volatility
interest rates, exchange rates, and other vari- ed updated estimates of the trading volume
ables in futures markets.109 Unfortunately, it elasticity to bid-ask spread on the 11 select- and transaction
does not include any measure of transaction ed U.S. futures based on Wang and Yau’s costs depends on
costs (such as the bid-ask spread) or transac- methodology.113
tion taxes/fees as an explanatory variable in In Table 1, the estimates of the elasticity
how transaction
the model. In other words, the estimation of trading volume with respect to transac- costs affect
of the relationship between trading volume tion costs (proxied by the bid-ask spread) for trading volume.
and other explanatory variables is done sepa- 11 U.S. futures are presented. The estimates
rately instead of jointly in a structural model range from -2.6 (E-mini S&P 500 index fu-
(i.e., ignoring no. 4, above). tures) to -0.81 (heating oil), suggesting that
In light of the deficiencies in the empiri- the trading volume of a futures contract will
cal estimation of the relation of trading vol- decline if transaction costs increase, as by the
ume and price volatility, and the relation of imposition of an FTT. For example, the elas-
bid-ask spread and price volatility, Wang, ticity of -0.81 for the S&P 500 index futures
Yau, and Baptiste proposed a two-equation indicates that the trading volume for these
structural model to examine the relations be- futures will decrease 0.81 percent for each
tween trading volume and transaction costs one percent increase in the bid-ask spread
in seven financial, agricultural, and metals or financial transaction tax. The lower-end
futures.110 By estimating the elasticities in a of the corresponding interval estimates with
simultaneous-equation system, they explicit- a 95 percent confidence level are all greater
ly formalize the jointly determined relation- than one, except for 30-year T-bond (-0.972)
ship between trading volume and bid-ask and heating oil (-0.923) futures. These results
spread. Their study confirms that trading suggest that the elasticity of trading volume
volume and bid-ask spread are jointly de- with respect to transaction costs had been
termined in the U.S. futures markets. It also very high during the period 2005–2010 for
shows that the differences in the estimates most futures examined. Bjursell, Wang, and
for four U.S. futures markets underestimate Yau pointed out the important implication
the elasticities and overestimate the poten- that an increase in the bid-ask spread due to
tial tax revenues in these markets. Thus, in a new transaction tax would substantially re-
15
16. Table 1
Elasticity of Trading Volume with Respect to Transaction Costs in Selected U.S.
Futures Markets
January 2005–December 2010 January 1990–April 1994
Contract Elasticity Estimatesa Elasticity Estimatesb
S&P500 -0.81 -0.78
E-mini S&P500 -2.60
30-Year T-Bond -0.87
10-Year T-Note -1.36
British Pound -0.97
Wheat -0.98
Soybean -1.66
Copper -1.44
Gold -2.02 -1.31
Crude Oil -1.00
Heating Oil -0.80
Deutschemark -1.30
Silver -0.90
Source: aC. Johan Bjursell, George H.K. Wang, and Jot Yau, “Transaction tax and market quality of U.S. futures
market: An ex ante analysis,” Review of Futures Markets (2012), forthcoming. bGeorge H.K. Wang, and Jot Yau,
“Trading volume, bid-ask spread, and price volatility in futures markets,” Journal of Futures Markets 20, no. 10
(2000): 943–70.
Note: Numbers in parentheses denote standard errors for the corresponding point estimates.
duce trading volume and decrease liquidity proximated by the average yearly price in
for the U.S. futures exchanges. The elasticity 2010. The transaction tax revenue is then
used in the CBO’s 1990 report, -0.26, esti- expressed as a percentage of the total fixed
mated by Thomas Epps based on U.S. stock transaction costs (TFC), which is $14.8.
data, seriously understates current elasticity Thus, for S&P 500 futures, the transaction
in the futures markets. Hence, the CBO over- tax revenue as a percentage of the total fixed
estimated the potential revenue of a transac- transaction costs (TR%TC) is
The elasticity tion tax in futures markets.
used in the We use the following example to illustrate $283,981 * 0.0002
= 3.837581 or 383.7581%
CBO’s 1990 how Bjursell, Wang, and Yau computed the $14.8
estimated tax revenue for S&P 500 index
report seriously futures, using a transaction tax of 0.02 per- Second, the post-tax volume (PTV), i.e.,
understates cent.114 the estimated trading volume after a trans-
current elasticity First, we calculate the 0.02 percent trans- action tax is imposed, is calculated based
action tax revenue on the S&P 500 futures on the current elasticity of trading volume
in the futures transactions based on the notional value (TV) with respect to transaction costs/taxes
markets. of the futures contract, or $283,981 as ap- (-0.81 for the S&P 500 futures, Table 1) and
16
17. the total trading volume (the number of transaction tax (e.g., 0.02 percent) is big Even a small
contracts traded in 2010). Thus, enough to wipe out all S&P 500 index fu- transaction
tures transactions, leaving no tax revenues to
be collected by the government. This result tax (e.g., 0.02
PTV = Total TV*[1 + (current elasticity of suggests that the impact of a transaction tax percent) is big
TV*(TR%TC/TFC))] on trading costs and trading volume can vary
PTV =7,689,961*[1 + (-0.81*3.837581)] significantly with different types of futures
enough to wipe
= -16,213,825 since they have different degrees of trading out all S&P 500
volume elasticity. index futures
Third, the change in trading volume Finally, the estimated potential tax rev-
(ΔTV) is computed to be equal to enues to be collected on various futures con- transactions,
tracts is calculated based on the post tax vol- leaving no tax
ΔTV = PTV − TV ume (column 3, Table 2) estimated with the revenues to be
ΔTV = -16,213,825 – 7,689,961 recent estimates of trading volume elasticity
= -23,903,786 (from Table 1). Column 5 in Table 2 presents collected by the
the estimates of the potential post-tax reve- government.
This result shows that the trading volume nue for the eleven futures computed by Bjur-
of S&P 500 index futures is very sensitive to sell, Wang, and Yau with recent estimates of
changes in transaction costs. Even a small trading volume elasticity. The potential tax
Table 2
Estimates of Post-Tax Revenue in Selected U.S. Futures Markets
(4) Post Tax (5) Post Tax
(2) Average Yearly (3) Post Tax Revenue Naïve Revenue Elasticity
(1) Contract Price (2010) ($) Trading Volume Method ($)a Adjusted ($)b
S&P 500 283,981 0 436,760,532 0
E-mini S&P 500 56,776 0 6,305,896,991 0
30-Year T-Bond 124,069 29,208,455 2,072,187,486 724,770,376
10-Year T-Note 121,174 0 7,118,223,079 0
British Pound 96,522 0 583,383,582 0
Wheat 29,512 14,747,726 136,289,676 87,048,101
Soybean 52,434 0 387,315,432 0
Copper 8,572 8,340,933 17,668,989 14,300,463
Gold 122,616 0 1,096,935,043 0
Crude Oil 79,621 0 2,685,649,749 0
Heating Oil 9,033 21,518,357 48,725,003 38,875,710
Total 20,889,035,562 864,994,651
Source: C. Johan Bjursell, George H.K. Wang, and Jot Yau, “Transaction Tax and Market Quality of U.S. Futures
Market: An ex ante Analysis,” Review of Futures Markets (2012), forthcoming.
Notes: a Estimated potential revenue under this method is computed as Trading volume 2010 x Average Yearly
Price (2010) x Tax rate (0.02%). bEstimated potential revenue under this method is computed as: Post-tax trading
volume x Average Yearly Price (2010) x Tax rate (0.02%).
17
18. revenues (PTR) collected from each futures Conclusion
contract is computed as follows:
In this paper, we reviewed the theoreti-
PTR = PTV*Average Yearly Price (2010)* cal and empirical studies on the impact of a
Tax rate (0.02%) transaction tax.116 Specifically, we reviewed
the empirical evidence on the imposition
For comparison purposes, presented in col- of a FTT in futures markets with regard to
umn 4 in Table 2 are the post-tax revenues trading volume, price volatility, pricing effi-
calculated by the naïve method, that is, as- ciency, and estimated revenue. We discussed
suming trading volume will stay the same a methodology for estimating the potential
and not be affected by the transaction tax. transaction tax revenue that can be raised
The tax revenue generated by the naïve from U.S. futures markets. The empirical
method (column 4) is often used by propo- model proposed by the authors and used in
nents of transaction tax as the basis for ar- several previous studies accounts for the en-
guing that transaction taxes would generate dogenous relationships among trading vol-
substantial revenue.115 For the 0.02 percent ume, bid-ask spread (transaction cost), and
tax rate, six futures (S&P 500, E-mini S&P volatility.117 We explained the estimation of
The transaction 500, 10-year T-Note, British pound, soy- the empirical elasticity of trading volume
tax revenue bean, and gold) would cease to be traded at and post-tax adjusted trading volume us-
estimated by the all in U.S. markets, and would therefore gen- ing Bjursell, Wang, and Yau’s estimates on
erate zero tax revenue (column 5). The other 11 futures traded in the United States. We
pre-tax trading five futures (30-year T-Bond, wheat, copper, showed that current estimates of the elastic-
volume or with crude oil, and heating oil), given their trad- ity of trading volume with respect to a trans-
ing volume elasticity, generate tax revenues action tax in U.S. futures markets are much
an unrealistically that are less than the corresponding esti- higher than those reported in the extant lit-
low elasticity mated tax revenues from the naïve method. erature and those used by the government in
can seriously There are three noteworthy findings from transaction tax revenue estimation. As such,
the study by Bjursell, Wang and Yau. First, a transaction tax would reduce trading vol-
overestimate the magnitude of the decline in the post-tax ume significantly, may not reduce price vola-
the potential tax volume depends on the relative importance tility, and might only raise a modest amount
revenue. of the transaction tax to the total fixed cost of tax revenue, much smaller than expected.
and/or the elasticity of trading volume with More importantly, results indicate that
respect to transaction costs on each future. with such high estimates of trading volume
For example, the post-tax trading volume of elasticity, it is very likely that futures trad-
the S&P 500 index futures is reduced to zero ing activities would be shifted to untaxed
when the transaction tax is 383.76 percent foreign markets should a transaction tax be
of the total fixed transaction cost with an imposed. We conclude that a transaction tax
elasticity of -0.81. In the soybean case, the on futures trading will not only fail to gen-
elasticity is high (i.e., -1.66) but the post-tax erate the expected tax revenue, it will likely
trading volume still drops to zero even if the drive business away from U.S. exchanges
transaction tax is only 65.75 percent of the and toward untaxed foreign markets.
total fixed transaction cost.
Second, the impact of a transaction tax
on transaction costs and trading volumes Notes
varies significantly with different types of We would like to thank Mark A. Calabria, direc-
futures. Third, the transaction tax revenue tor of financial regulation studies at the Cato
estimated by the pre-tax trading volume or Institute, for encouraging us to write this paper.
with an unrealistically low elasticity can seri- The views expressed here do not necessarily re-
flect those of our current or former employers.
ously overestimate the potential tax revenue.
18
19. 1. Financial transaction taxes (FTTs) can be Chicago Press, 1994), pp. 277–308.
classified into (1) securities transaction taxes
(STT); (2) currency transaction taxes (CTT or To- 10. The European Commission rejected a pro-
bin tax); (3) capital levy or registration taxes; (4) posal for a bank tax to pay for the bailout of
bank transaction taxes (BTT); and (5) real estate failed European banks due to the 2007–2008 fi-
transaction taxes. Thornton Matheson, “Taxing nancial crisis.
Financial Transactions: Issues and Evidence,”
International Monetary Fund working paper, 11. James E. Tobin, “A Proposal for Internation-
WP11/54, 2011. The term “Tobin tax” originally al Monetary Reform,” Eastern Economic Journal 4,
referred to the tax on currency transactions (i.e., no. 3 (1978): 153–9.
CTT). It is now used interchangeably with finan-
cial transaction taxes, as in this paper. 12. For a review of arguments, see Schwert and
Seguin, pp. 27–35; Paul H. Kupiec, Patricia A.
2. Evan Ramstad, “World News: Seoul Will White, and Gregory Duffee, “A Securities Trans-
Seek Support for Levy,” The Wall Street Journal, action Tax: Beyond the Rhetoric,” Research in Fi-
June 4, 2010, p. A14. nancial Services Private and Public Policy 5 (1993):
55–76; R. Glenn Hubbard, “Securities Transac-
3. Jason Zweig, “Flash Tax: Why Levies on tion Taxes: Can They Raise Revenue?” MidAm-
High-Speed Trading Won’t Work,” The Wall Street erica Institute Research Project on Transaction
Journal, September 3–4, 2011, p. B11. Tax, July 1993; Summers and Summers; Steven
R. Umlauf, “Transaction Taxes and the Behavior
4. Alex Barker, George Parker, and Brooke Mas- of the Swedish Stock Market,” Journal of Financial
ters, “Fresh Clashes Brew Over Tobin Tax,” Finan- Economics 33 (1993): 227–240; and Neil McCull-
cial Times, January 5, 2012, p. 4. och and Grazia Pacillo, “The Tobin Tax–A Re-
view of the Evidence,” working paper, Institute
5. Joseph A. Grundfest and John B. Shoven, of Development Studies, University of Sussex,
“Adverse Implications of a Security Transaction 2010.
Excise Tax,” Journal of Accounting, Auditing, and Fi-
nance 6 (1991): 409–42. Grundfest and Shoven 13. See Franklin R. Edwards, “Taxing Transac-
believe the STT is “a resilient proposal,” p. 410. G. tions in Futures Markets: Objectives and Effects,”
William Schwert and Paul J. Seguin, “Securities Journal of Financial Services Research 7 (1992): 75–91;
Transaction Taxes: An Overview of Costs, Bene- and George H. K. Wang, Jot Yau, and Tony Bap-
fits and Unresolved Questions,” Financial Analysts tiste, “Trading Volume and Transaction Costs in
Journal 49 (September–October 1993): p. 28. Futures Markets,” Journal of Futures Markets 17, no.
7 (1997): 757–80.
6. Donald W. Kiefer, “The Securities Transac-
tion Tax: An Overview of the Issues,” CRS Report 14. “Charlemagne: The Seven-Yearly War,” Econ-
for Congress (1990). omist, June 30, 2011.
7. General Accounting Office, GGd-93-108- 15. Kiefer.
Futures Transaction Fee, p. 1.
16. Campbell and Froot, p. 277.
8. Kathleen Cronin, “A Transaction Tax’s Unin-
tended Consequences,” 2010, http://archive.opn 17. Previous studies present summaries for vari-
mkts.com/regulatory/a-transaction-taxs-unin ous types of financial transaction taxes around
tended-consequences/; E. Noll, “The Perils of a the world over different time periods. See Robert
Financial Services Transaction Tax,” 2010, http:// Pollin, Dean Baker, and Marc Schaberg, “Secu-
www.rollcall.com/news/-42770-1.html. rities Transaction Taxes for U.S. Financial Mar-
kets,” Eastern Economic Journal 29, no. 4 (2003):
9. The stamp duty in the U.K. is a levy on the 527–58; McCulloch and Pacillo; Schwert and Se-
transfer of assets and securities. Richard Roll, guin; Roll; and Matheson.
“Price Volatility, International Market Links, and
Their Implications for Regulatory Policies,” Jour- 18. John Maynard Keynes, The General Theory of
nal of Financial Services Research 3 (1989): 211–46; Employment, Interest, and Money (New York: Har-
Lawrence H. Summers and Victoria P. Summers, court Brace, 1936).
“When Financial Markets Work Too Well: A Cau-
tious Case for a Security Transaction Tax,” Jour- 19. James Tobin, “A Proposal for International
nal of Financial Services Research 3 (1989): 261–86; Monetary Reform.”
John Y. Campbell and Kenneth A. Froot, “Inter-
national Experiences with Securities Transaction 20. James E. Tobin, “On the Efficiency of Fi-
Taxes,” in Jeffrey A. Frankel, ed. The Internation- nancial Systems,” Lloyds Bank Review 153 (1984):
alization of Equity Markets (Chicago: University of 1–15.
19