Abstract
We consider the impact of transaction taxes on financial markets in the context of four questions. How important is trading? What causes price volatility? How are prices formed? How valuable is the volume of transactions? Drawing on the literature on market microstructure, asset pricing, rational expectations, and international finance, we argue that securities transaction taxes “throw sand” not in the wheels, but into the engine of financial markets. We conclude that transaction taxes can have negative effects on price discovery, volatility, and liquidity and lead to a reduction in the informational efficiency of markets. JEL G13, G14, H10
KARL HABERMEIER and ANDREI A. KIRILENKO*
This paper argues that transaction taxes can have negative effects on price discovery, volatility, and market liquidity in securities markets. These effects can lead to a reduction in market efficiency and may contribute to increased asset price volatility.
Financial markets transform latent demands of investors into realized financial transactions. Securities transaction taxes (STTs) alter this transformation. Proponents of STTs argue that such taxes can reduce market volatility, help to prevent financial crises, and reduce excessive trading.1 Opponents believe that STTs are difficult to implement and enforce and that they can do great damage to financial markets.
This paper considers the impact of transaction taxes on financial markets in the context of four broad questions. How important is trading? What causes price volatility? How are prices formed? How valuable is the volume of transactions? These questions are at the core of the debate on the role of transaction taxes. Our arguments draw on research on market microstructure, asset pricing, rational expectations, and international finance.
Market microstructure studies suggest that trading is essential for price discovery—the process of finding market clearing prices. A large number of markets rely on dealers to provide price discovery as well as liquidity and price stabilization. Levying STTs on the dealers inhibits their ability to assist investors with the transformation of latent demands into realized transactions. The literature also finds that much of the volatility is caused by informed traders as their information is aggregated into transaction prices. Taxing financial transactions does not reduce the volatility due to “noise” trading. Rather, it introduces additional frictions into the price discovery process.
The literature on option pricing under transaction costs shows how frictions on the trading in one asset affects prices and volumes of that and other assets. Using a simple framework based on this literature, we demonstrate how volume can migrate to the assets that are not subject to the tax. We also argue that it is very difficult to design and implement a tax that does not favor one portfolio of assets over another portfolio with exactly the same payoff.
Recent studies on rational expectations question the traditional view that volume is just an outcome of the trading process and is not valuable per se. These studies find that volume can play an informational role. Consequently, if transaction taxes cause volume to migrate, then they can hamper the informational efficiency of markets.
International finance provides other interesting examples of volume fragmentation and market segmentation. Volume fragmentation can occur due to restrictions on trading of substitutable securities such as different classes of company shares. This leads to market segmentation and inefficient price discovery.
Many of the issues that arise in the debate over STTs are also relevant to the debate on controls on international capital flows. Indeed, a number of controls on capital flows have taken the form of STTs—for example, the Chilean unremunerated reserve requirement on capital inflows.
I. Literature on Securities Transaction Taxes
Opinion is divided on the merits of securities transaction taxes. Many proponents of STTs advance the following propositions:2
The contribution of financial markets to economic welfare does not justify the resources they command. During a given time period, the resources that change hands in financial markets far exceed the value of the underlying or “real” transactions.
Many financial transactions are highly speculative in nature and may contribute to financial or economic instability.
Market instability, including crashes, enriches insiders and speculators, while the costs are borne by the general public.
Financial market activity increases inequalities in the distribution of income and wealth.
STTs can be an important and innovative source of revenue for the financing of development.
From this perspective, it is argued by some that governments ought to tax financial transactions in order to discourage destabilizing speculation that can threaten high employment and price stability, as well as to raise revenue. Higher rates—they argue—should be levied on short-term transactions, since these seem to benefit primarily market intermediaries and not “real” users. The massive volume of financial transactions in well-developed modern markets would—they reason—allow substantial revenue to be raised by imposing very low tax rates on a broad range of transactions. It is not surprising that a number of governments around the world have succumbed to this temptation, all the more so as such taxes have a certain popular appeal.
Opponents of STTs have more faith in the ability of markets to allocate resources efficiently without direct intervention from public policy. However, the opponents also lack a convincing argument to justify the volume of resources flowing through financial markets. In addition, numerous documented anomalies, as well as a history of market crashes, do not lend themselves easily to the idea that financial markets are fully efficient. Neither does the fact that market participants devote considerable resources to analyzing previous transaction prices and volumes. Thus, instead of showing that the allocation of resources to the financial sector is justified on efficiency grounds, or that observed market volatility is optimal, the opponents of STTs have focused on practical shortcomings of the taxes themselves.3
There are two dimensions to the difficulties in implementing STTs. First, if an STT is applied in one financial market but not in others, the volume of transactions tends to migrate from the market that is taxed to markets that are not. Effective enforcement of STTs thus requires either a cross-market and perhaps even a global reach or measures to segregate markets. For example, tax authorities in one country may attempt to require payment of the tax on transactions made by their residents not only in financial markets within their own borders, but in other markets as well. Alternatively, they may impose controls on cross-border financial transactions, for example, the Chilean tax on capital inflows.
Second, since the composition of the assets used in financial transactions matters less than the distribution of payoffs over time and in uncertain states of the world, the tax base must be defined as a function of the final payoff rather than the assets employed. A securities transaction tax would be considered neutral if it did not favor one portfolio of assets over another portfolio with exactly the same payoff. Since payoffs can be replicated by portfolios consisting of different types of assets, the imposition of an STT can create a greater distortion than it is trying to mitigate.4 Instead of trading less because of the tax, investors may transact more in assets that are taxed at a lower rate or not taxed at all. As a result, real resources devoted to financial transactions may in fact increase rather than diminish following the imposition of an STT.
Given the lack of a consensus on the theory, there have been many attempts to resolve the debate empirically. However, empirical studies undertaken so far have not been able to decisively resolve the debate on the effects of transaction taxes on financial markets.5
Empirical research has encountered three major problems. First, the effects of taxes on prices and volume are hard to disentangle from other structural and policy changes taking place at the same time. Therefore, estimates based on the assumption that everything else in the economy is held constant are potentially biased.
Second, it is difficult to separate transaction volume into stable (or “fundamental”) and destabilizing (or “noise”) components. Thus, it is hard to say which part of the volume is more affected by the tax.
Third, it is hard to differentiate among multiple ways in which transaction taxes can affect asset prices. These ways include changes in expectations about the impact of the taxes, the cost of creating and trading in close substitutes not covered by the tax, and changes in market liquidity.
Empirical studies seek answers to three main questions. The first question is whether transaction taxes have an effect on price volatility. Roll (1989) studies stock return volatility in 23 countries from 1987 to 1989. He finds no evidence that volatility is reliably related to transaction taxes. Umlauf (1993) studies equity returns in Sweden during 1980-87, before and during the imposition of transaction taxes on brokerage service providers. He finds that the volatility did not decline in response to the introduction of taxes. Saporta and Kan (1997) study the impact of the U.K. stamp duty on volatility of securities’ prices. They also find no evidence of a relationship between the stamp duty and volatility. Jones and Seguin (1997) examine the effect on volatility of the introduction of negotiated commissions on U.S. national stock exchanges in 1975, which resulted in a permanent decline in commissions. They argue that this event is analogous to a one-time reduction of a tax on equity transactions. They reject the hypothesis that the lowering of commissions increases volatility. Hu (1998) examines the effects on volatility of changes in transaction taxes that occurred in Hong Kong, Japan, Korea, and Taiwan from 1975 to 1994, and does not find significant effects. Finally, Hau and Chevalier (2000) examine the effect on volatility of minimum price variation rules in the French stock market. They argue that minimum price variation rules result in a doubling of transaction costs for stocks priced above a certain threshold (500 francs). They argue that this is analogous to the application of a transaction tax on the stocks above the threshold. They find that the increase in transaction costs results in “a statistically significant, but economically insignificant” reduction in the daily, weekly, and monthly return volatility.
The second question is whether transaction taxes affect trading volume. Umlauf (1993) reports that after Sweden increased its transaction tax from 1 percent to 2 percent in 1986, 60 percent of the volume of the 11 most actively traded Swedish stocks migrated to London. The migrated volume represented over 30 percent of all trading volume in Swedish equities. By 1990, that share increased to around 50 percent. According to Campbell and Froot (1995), only 27 percent of the trading volume in Ericsson, the most actively traded Swedish stock, took place in Stockholm in 1988. Hu (1998) examines 14 tax changes in four Asian markets and finds that differences in turnover before and after changes in the tax level are not statistically significant.
Thirdly, empirical studies seek to find out whether transaction taxes have an impact on securities’ prices. Umlauf (1993) reports that the Swedish All-Equity Index fell by 2.2 percent on the day a 1 percent transaction tax was introduced and again by 0.8 percent on the day it was increased to 2 percent. Saporta and Kan (1997) find that on the day stamp duty in the United Kingdom was increased from 1 to 2 percent, the stock market index declined by 3.3 percent. Hu (1998) finds that on average the return on the announcement date is -0.6 percent in Korea and -1.6 percent in Taiwan, with the result for Taiwan being highly statistically significant.
One of the main reasons for the dispersion and inconclusiveness of results is the lack of appropriate data. Since the questions are essentially of the market microstructure-type, an ideal dataset would consist of transaction frequency data for individual financial instruments. In order to take revisions in expectations into account, the data should start well before the announcement of the transactions tax and include a sufficient number of observations following its imposition. Furthermore, in order to separate volume into meaningful categories, the data should be broken down according to the type of investor, for example, institutional investors, hedge funds, and mutual funds. In contrast, most empirical studies rely on weekly equity index returns.
II. The Swedish Experience
In order to illustrate the subsequent arguments, we devote this section to a brief description of the Swedish experience with STTs. The Swedish experiment lasted for more than eight years. The first measure was announced in October 1983 and the last one was abolished in December 1991. The analysis in this section is based on the studies by Umlauf (1993) and Campbell and Froot (1995).
The initiative to impose financial transaction taxes came from the Swedish labor sector in 1983. The labor sector did not claim that trading in financial markets led to inefficient outcomes. Rather, according to Umlauf (1993), in the opinion of the labor sector, “the salaries earned by young finance professionals were unjustifiable … in a society giving high priority to income equality,” especially given the seemingly unproductive tasks that they performed. On this basis, the Swedish labor sector proposed to levy taxes directly on domestic brokerage service providers.
Despite the objections of the Swedish Finance Ministry and the business sector, popular support led to the adoption of taxes by Parliament. The taxes became effective on January 1, 1984. They were levied on domestic stock and derivative transactions. Purchases and sales of domestic equities were taxed at 0.5 percent each, resulting in a 1 percent tax per round trip. Round-trip transactions in stock options were taxed at 2 percent. In addition, exercise of an option was treated as a transaction in the underlying stock and, thus, was subject to an additional 1 percent round-trip charge. The tax coverage and rates reflected a popular perception about the “usefulness” of transactions in different financial instruments, with those involving equity options being the least “useful.”
Continuing pressure from the labor sector compelled the Parliament to double the tax rates in July 1986 and broaden its coverage in 1987. Furthermore, following large losses in interest futures and options (most notably by the city of Stockholm, which lost SEK 450 million), the tax was extended to transactions in fixed-income securities, including government debt and the corresponding derivatives in 1989.6 The maximum tax rate for fixed-income instruments was set at 0.15 percent of the underlying notional or cash amount. In addition, the tax was designed to be “yield-neutral,” with longer maturities instruments being taxed at progressively higher rates.
The revenue performance of the tax was disappointing. According to the Finance Ministry of Sweden, the government collected SEK 820 million in 1984, SEK 1.17 billion in 1985, and SEK 2.63 billion in 1986. This accounted for 0.37, 0.45, and 0.96, respectively, percent of the total revenue for the corresponding years. After doubling the tax rates, the government was able to collect SEK 3.74 billion in 1987 and SEK 4.01 billion in 1988. This accounted for 1.17 and 1.21 percent of the total revenue.7 Thus, a 100 percent increase in the tax rate resulted in a 22 percent increase in revenue.
Widespread avoidance was one reason for the weak performance of the tax. Foreign investors avoided the tax by placing their orders with brokers in London or New York. Domestic investors avoided it by first establishing offshore accounts (and paying the tax equal to three times the round-trip tax on equity for funds moved offshore) and then using foreign brokers.
The scale of avoidance was manifested by a massive migration of stock trading volume from Stockholm to other financial centers. Since the brokerage business is very competitive, finding a close substitute for brokerage services offshore was not very costly. According to Umlauf (1993), following the doubling of the tax, 60 percent of the volume of the 11 most actively traded Swedish stocks migrated to London. The migrated volume represented over 30 percent of all trading volume in Swedish equities. By 1990, that share increased to around 50 percent. According to Campbell and Froot (1995), only 27 percent of the trading volume in Ericsson, the most actively traded Swedish stock, took place in Stockholm in 1988.
Broadening the tax to fixed-income instruments resulted in a sharp drop in trading volume in Swedish government bills and bonds and in fixed-income derivatives contracts. Campbell and Froot (1995) estimate that during the first week of the tax, bond trading volume dropped by about 85 percent from its average during the summer of 1987 and trading in fixed-income derivatives essentially disappeared. This significantly undermined the ability of the Bank of Sweden to conduct monetary policy, made government borrowing more expensive, and eroded both popular and political support for the tax. Taxes on fixed-income instruments were abolished in April 1990. Taxes on other instruments were cut in half in January 1991 and abolished altogether in December 1991.
Following the abolition of the tax, some trading volume came back to Sweden. According to Campbell and Froot (1995), 41 percent of the trades in Ericsson took place in Stockholm in 1992. Overall, the proportion of the trading volume in Sweden increased for almost all equities in 1992. That year, 56 percent of all trading volume in Swedish equities took place in Stockholm.
The Swedish experience highlights the following points. First, investors avoid the tax by finding or creating close substitutes. Since the brokerage business is very competitive, finding a close substitute for brokerage services offshore was not very costly. However, the markets do not necessarily move offshore, if close substitutes are available domestically. For example, trading in bonds did not move offshore, but shifted to debentures, forward contracts, and swaps. Second, markets suffer greatly following the imposition of the tax. Even very low tax rates on fixed-income instruments led to an 85 percent decline in volume in the first week after the tax was imposed compared to its pre-tax average. The fixed-income options market virtually disappeared. Third, after the removal of the tax, the trading volume gradually comes back across all previously taxed assets.
III. How Important Is Trading?
The Swedish labor sector believed that trading in financial markets is an essentially unproductive task. Just how important is trading? The answer to this question depends on how the trading is conducted. In Sweden, investors had to carry out financial transactions mostly through dealers.
However, trading does not have be conducted exclusively through dealers. It can be done through other mechanisms. For example, in continuous electronic auctions, buyers and sellers trade directly with each other, bypassing the dealers. Why didn’t such an auction develop in Sweden? In fact, under the law, transactions executed without dealers were exempt from taxes.
According to the market microstructure literature, under some circumstances, dealers offer services that cannot be provided by other types of market designs at lower cost. It is especially true for infrequently traded assets such as most of the Swedish stocks. Perhaps for that reason the order flow migrated not to another trading design but to dealers in London and New York.
Dealers provide several important services. They provide liquidity and assume substantial risks by contributing their own capital. Accordingly, they demand adequate compensation for the provision of liquidity and the capital that they put at risk. The dealer’s compensation is higher for illiquid assets.
In addition, dealers who act as market makers in particular securities must furnish competitive bid and offer quotations on demand and be ready, willing, and able to effect transactions in reasonable quantities at the quoted prices. In other words, a buyer does not have to wait or look for a seller, but can simply buy from a dealer who sells from his inventory. According to Pagano and Roell (1990), “this implies that, in contrast with what happens on auction markets, traders are insured against execution risk, i.e., the risk of finding few or no counterparties to trade.” The dealer’s compensation is higher for assets with a higher execution risk.
This highlights another important function that dealers play, namely, the provision of price stability. According to Madhavan (2000), “the presence of market makers who can carry inventories imparts stability to price movements through their actions relative to an automated system that simply clears the market at each auction without accumulating inventory.”
The provision of liquidity, price discovery, and price stabilization requires inventory management. Inventory management is achieved through the buying and selling of securities. Hasbrouck and Sofianos (1993) examine a set of quote, trade, and inventory data for market makers (specialists) on the New York Stock Exchange. According to their data, the market maker’s activity (both purchases and sales) averages to about 26 percent of the total transaction flow (also both purchases and sales). For the most frequently traded stocks, this number is 20 percent, while for the least frequently traded stocks, it rises to 38 percent.8 Thus, dealers become much more important as liquidity providers in less frequently traded stocks.
Inventory management can involve both customer and interdealer trading. When a competitive interdealer market is available, dealers can adjust their inventory without waiting for a public order flow to arrive. According to the empirical evidence, dealers trade in the interdealer market when they want to manage large inventory positions. Lyons (2001) suggests that interdealer trading in the foreign exchange market currently accounts for about two-thirds of the total volume. Hansch, Naik, and Viswanathan (1998) show that the average size of an interdealer trade on the London Stock Exchange is much larger than the average size of a trade with the general public. They also show that inventory levels at which dealers trade among themselves are about twice as large as those at which they trade with the general public. They find that 38 percent of the variation in interdealer trading is explained by variation in inventory levels. They conclude that “interdealer trading is an important mechanism for managing inventory risks in dealership markets.”
Thus, trading is important. It helps manage risks. Dealers demand compensation for the services that they provide and the risks that they take. If trading becomes costly as a result of transaction taxes, dealers cannot manage their risks effectively. Accordingly, they become less willing to put their own capital at risk in order to provide liquidity. Investors cannot carry out their desired trades, their latent demands are not fully satisfied, and resources are not allocated to their best uses.
IV. What Causes Volatility?
In the previous section we argue that trading is important. But can it also be the cause of volatility?
French and Roll (1986) conduct an empirical study of the variability of stock returns over trading and nontrading periods. Using data for all stocks listed on the NYSE and AMEX for the period 1963 to 1982, they find that on an hourly basis, the variance of stock returns is between 13 and 100 times larger when markets are open for trading than the variance when the markets are closed, depending on the definition of nontrading period.
They investigate three possible causes for the higher volatility during trading hours. First, higher volatility may be caused by the arrival of more public information during trading hours. Second, it may be caused by informed investors as their private information is incorporated into prices. Finally, higher volatility may be caused by the process of trading itself as prices fluctuate due to market frictions and transaction costs.
They also find that the process of trading accounts for at most 12 percent of the daily return variance. The rest of the variance is attributable to the arrival of public and private information during trading hours. While they cannot directly decompose the effects of public and private information on volatility, they conduct a test that suggests that most of the variability in stock returns can be attributed to the arrival of private information during trading hours.
Later studies relied on much more refined transaction-level data to further decompose transaction price volatility. Madhavan, Richardson, and Roomans (1997) develop a stylized, reduced-form model of price volatility and use transaction-level, intraday data on 274 NYSE-listed stocks during 1990 to estimate it.
They argue that price volatility can be explained by the variability of four components: public information, private information, transaction costs, and other market frictions (price discreteness). They estimate that the impact of public information accounts for 46 percent of volatility in the beginning of the trading day and 35 percent at the end. The impact of private information (including the interaction between cost and private information effects) drops from 31 percent in the morning to 26 percent at the closing of trading. Variability in transaction costs increases from 22 percent at the opening to 35 percent at the end of the trading day. Finally, price discreteness accounts for the remaining 1 to 4 percent at the beginning and the end of the trading day, respectively.
Transaction costs in the Madhavan, Richardson, and Roomans (1997) model capture dealers’ costs for supplying liquidity on demand. They include compensation for inventory costs, putting their capital at risk, and other transaction costs. The model implies that other things being equal, higher transaction costs increase volatility. If transaction costs also include transaction taxes, then introduction of STTs can result in higher rather than lower volatility of transaction prices.
V. How Are Prices Formed?
In perfect, frictionless markets, asset prices immediately reflect all available information. As the new information arrives, investors rebalance their portfolios of assets. The rebalancing results in an updated set of prices. In the absence of transaction costs, the rebalancing can be done continuously and price discrepancies are eliminated instantaneously. However, in real markets, agents face transaction costs. The presence of even very small transaction costs makes continuous rebalancing infinitely expensive. Therefore, valuable information can be held back from being incorporated into prices. As a result, prices can deviate from their full information values.
The dissatisfaction with the assumption of continuous portfolio rebalancing was the starting argument for the literature on the replication of assets under transaction costs. The literature recognizes that continuous rebalancing is not feasible and formulates discrete rebalancing under transaction costs.
In this section, we study the impact of STTs on portfolio rebalancing and price formation.
A Simple Example
Consider a simple two-period example (following Hull, 1985). There are three assets in the market: a risk-free bond yielding 12 percent per year, a nondividend paying stock, and a call option on the stock. The starting price of each share of stock is equal to $20. After a year, we assume that the stock price will either have increased to $22 or fallen to $18, with equal probability. The strike price of the option at the end of the year is taken to be $21.
Simple option pricing theory can be employed to compute in what proportions a call option and a risk-free bond must be held in order to be equivalent to 100 shares of stock. As shown in the Appendix, on the assumptions given, this portfolio requires exactly 400 options (worth $0.63 each) and $1,747 of the bond.
But a 1 percent transaction tax on buying or selling the stock greatly lowers the value of the option, as a tax of $0.22 will have to be incurred twice if the option is exercised and the stock then sold. Working through the arithmetic reveals that the option is only worth $0.39 and that now 694 options must be bought (along with $1,728 worth of bonds) to match 100 shares.
If the transaction tax is also levied on option transactions, or on bonds, there is a further change in the required number of options in the portfolio to replicate the shares, but in these cases the changes are very small. Thus extending the transaction tax to all three assets certainly does not restore neutrality.
Note that even in this simple example, it is quite difficult to design and even more difficult to implement a tax that does not favor one portfolio of assets over another portfolio with exactly the same payoff (e.g., a stock versus a bond and a call option). A uniform transaction tax is not payoff-neutral. For a tax to be payoff-neutral, the tax rates must be such that a change in the value of a replicating portfolio is exactly equal to the change in the price of the underlying asset. In other words, the tax rates must depend on the “delta” of the replicating portfolio. Since in practice, “delta” changes as more information is revealed about the (unknown) underlying stochastic process, a payoff-neutral tax would have to be frequently adjusted. This would make it very difficult to implement.
A Generalized Model
Boyle and Vorst (1992) have generalized the simple two-period example to a multiperiod case using a method proposed by Cox, Ross, and Rubinstein (1979), who assumed a dynamic price process according to which, during each subperiod of length Δt, the stock price increases by a factor
Boyle and Vorst show that, under simplifying assumptions, the call option can still be priced after the introduction of transaction costs by adding to the variance an amount fraction that is positively related to the rate of the transaction cost or tax and inversely related to the length of the rebalancing period.9 Specifically, if δ2 is the original variance, the modified variance,
where k is the rate of transaction cost.10
VI. How Valuable Is the Volume of Transactions?
According to the example presented in the previous section, demand for assets changes following the introduction of a transaction tax on a stock. The demand for derivatives goes up and the demand for both stocks and bonds decreases. Changes in demand translate into changes in the volume of realized transactions. Was anything lost as a result of this change in volume? Does it matter if transaction volume migrates to other instruments, markets, or countries? It does not, if the volume is not valuable. But how valuable is the volume of realized transactions?
According to standard rational expectations models with supply uncertainty, trading orders have both informational (or “signal”) and “noise” components. Without the noise, aggregate supply uncertainty is resolved, and prices adjust to their full information level. Otherwise, the informational component is aggregated into prices and the “noise” is left in volume. Consequently, volume is just an outcome of the trading process. It does not have any information about the fundamentals or the trading process and, therefore, lacks value.
According to this view, the migration of volume to other instruments, markets, or countries does not result in any loss of value or efficiency. It just means a reallocation of supply uncertainty. In other words, if transaction volume moves from Stockholm to London, investors in Stockholm become exposed to less uncertainty associated with “noise” trading and investors in London to more of it. Thus, if following the imposition of a transaction tax, volume migrates away from the taxed asset, the policymakers should perhaps just change their revenue projections and not worry about any fundamental market effects.
The long-held view that volume is not valuable per se has recently come under scrutiny. Blume, Easley, and O’Hara (1994) investigate the informational role of volume. In their model, the source of “noise” is not supply uncertainty, but the precision of private information about the signal. Prices aggregate information about the average level of private information. Trading volume contains information about the precision of individual private signals. Thus, volume does not just contain “noise,” but has a nontrivial informational role to play. Price-volume sequences are more informative than prices alone. This role becomes especially important for infrequently traded stocks that often do not get much analyst coverage.
In addition, Easley, O’Hara, and Srinivas (1998) investigate the informational role of transaction volume in options markets. They develop a model where informed traders can trade in stock or options markets. They empirically test the model and find that option volume data contain information about future stock prices. Thus, they conclude that “volume plays a role in the process by which markets become efficient.” Consequently, a migration of volume from the derivative market may also result in the loss of informational efficiency.
This new view represents a fundamentally different perspective on the role of volume. It can be summarized as saying that “volume matters.” The migration of volume results in lower informational efficiency of instruments and markets from which it migrated. If transaction taxes cause the volume to migrate, then they do affect the ability of markets to aggregate information and prevent a more efficient allocation of resources.
VII. Evidence from International Finance
The international finance literature provides examples of market segmentation and execution costs in different markets. Market segmentation can result from direct restrictions on foreign ownership, exchange and capital controls, and regulatory and accounting aspects including disclosure rules, settlement practices, and investor protection rights. Bekaert (1995) studies 19 emerging markets and finds that exchange and capital controls (and taxes that have a similar effect) as well as regulation and accounting practices are significant in explaining market segmentation. Restrictions on foreign ownership are apparently being circumvented by the closed-end country funds.
Domowitz, Glen, and Madhavan (2000) use a comprehensive database of execution costs (including transaction taxes) for 42 countries from September 1996 to December 1998. They use panel data techniques to study the interaction between cost, liquidity, and volatility across countries and through time.
They find that except for North America, explicit equity trading costs such as brokerage commissions, taxes, and fees account for about two-thirds of total execution costs. In the United States average explicit one-way trading costs are the smallest for the countries in their study, accounting for 8.3 basis points or a fraction of 2.2 percent of mean return (374 basis points) for the period 1990-98. In other words, a complete rebalancing of the portfolio once a year results in an average explicit cost of 2.2 percent of its annual mean return. The largest explicit cost of 106 basis points is in Ireland, which has a stamp duty of 1 percent. In Ireland, the explicit costs of turning over a portfolio of equities just once a year accounts for a full 25 percent of the annual mean return.
They also find that over time, with the exception of transition economies, costs have generally declined, and that higher trading costs are positively related to increased volatility and lower volume.
VIII. Securities Transaction Taxes and Controls on International Capital Flows
There is an important similarity between securities transaction taxes and controls on international capital flows. Such flows are the result of financial transactions that involve parties who happen to be on different sides of national borders. Such transactions can in principle be subject to general taxes on financial transactions or to taxes that specifically target cross-border transactions. Capital controls may differ substantially in the types of transactions they apply to: inflows versus outflows, short-term versus long-term, all markets or assets, or only a subset of them.11
The arguments in favor of capital controls overlap to some extent with those advanced in favor of domestic securities transaction taxes, notably in that they are often seen as a policy response to financial market imperfections arising from informational asymmetries and other microeconomic distortions. However, the debate on capital controls has more strongly emphasized their use in macroeconomic policy, as a means of reconciling conflicting monetary and exchange rate policy objectives, or in preventing and managing balance of payments and financial crises, either by seeking to discourage volatile short-term inflows that could later be reversed, or by seeking to stem outflows during a crisis.
Probably the best-known example of a tax on short-term international capital flows is the Chilean encaje, or unremunerated reserve requirement on capital inflows. As with securities transaction taxes more broadly, there is evidence of substantial avoidance of this and other types of capital controls, which notably reduced their effectiveness and created various distortions in financial markets. More recently, Forbes (2002) shows that the Chilean tax on capital inflows increased financial constraints for small firms.
IX. Summary and Conclusions
This paper examines finance research relevant to assessing the impact of securities transaction taxes on financial markets. This research includes work on market microstructure, asset pricing, rational expectations, and international finance. We conclude that in most circumstances, transaction taxes can have negative effects on price discovery, volatility, and liquidity and lead to a reduction in market efficiency.
The arguments made in this paper may be summarized as follows. First, in dealership markets, trading facilitates the provision of liquidity, price discovery, and price stabilization. Trading also helps to manage risks. If investors cannot carry out their desired trades, their latent demands are not fully satisfied and resources are not allocated to their best use.
Second, price volatility can be explained by the variability of four components: public information, private information, transaction costs, and other market frictions. Other things being equal, higher transaction costs increase volatility. Consequently, the introduction of STTs can increase the volatility of transaction prices.
Third, a simple theoretical framework based on the literature on option pricing with transaction costs shows that following the introduction of a transaction tax, the demand for derivatives can increase substantially. Moreover, it is difficult to design and implement a tax that does not favor one portfolio of assets over another portfolio with exactly the same payoff.
Fourth, if transaction volume has an informational content, then a migration of volume would result in lower informational efficiency of instruments and markets from which it migrated. If transaction taxes are the cause of volume migration, then they can inhibit the informational efficiency of markets.
Finally, the international finance evidence on market segmentation and execution costs in different markets suggests that except for North America, explicit equity trading costs such as brokerage commissions, taxes, and fees account for about two-thirds of total execution costs. The conclusion was that higher trading costs, some of which are due to STTs, are positively related to increased volatility and lower volume. The paper also briefly summarizes similarities between securities transaction taxes and controls on international capital flows.
Transaction taxes can thus have a substantial effect on the transformation of investor demands into transactions. STTs can obstruct price discovery and price stabilization, increase volatility, reduce market liquidity, and inhibit the informational efficiency of financial markets.
APPENDIX
Working Through the Numerical Example
In order to compute the portfolios in Section V’s example, we begin by choosing a number of shares δ so that holding that number of shares and selling 100 call options provides a risk-free portfolio, that is, one that has the same value whether the share goes up or down. Since the value of option at maturity when it is “in the money” is exactly 1 (since then the option allows the share to be bought at the strike price of 21 and sold at 22), δ must satisfy:
The solution to this equation is δ= 25. The value of this portfolio at the end of the year will be 18 δ = 450, which equals $437 discounted to the present at 3 percent per year. This, then must be the value of the risk-free portfolio at the outset. Therefore, since the 25 shares will then cost $500, we can conclude that the price of the 100 options is $500 – $437 = $63.
Rearranging, we conclude that a portfolio consisting of 100 call options and $437 worth of bonds will exactly replicate the payoff on 25 shares. Equivalently, to replicate 100 shares requires exactly 400 options and $1,747 worth of bonds.
Suppose now that a transaction tax of 1 percent is introduced on all period-one transactions in the stock. Once more, when the stock price is equal to 22, the option gives a right to buy the stock at 21 and sell it at 22. But now this round-trip transaction is subject to transaction taxes. To buy the stock, the option’s holder must pay an additional $0.21 when buying the stock and $0.22 when selling it. Accordingly, the net terminal value of no call options is now just 0.57.
Let δ* be the amount of stock in the risk-neutral valuation portfolio adjusted for the transaction tax. Then, subtracting 1 percent transaction tax from the price of the share in each case,
The solution is now δ* = 14.4 and the value of the portfolio at the end of the year will be 17.82
Rearranging, we conclude that a portfolio consisting of 100 call options and $249 worth of bonds replicates just 14.4 shares. In order to replicate 100 shares requires 694 options (plus $1728 worth of bonds).
REFERENCES
Ariyoshi, Akira, Karl Habermeier, Bernard Laurens, Inci Otker-Robe, Jorge Ivan Canales-Kriljenko, and Andrei Kirilenko, 2000, Capital Controls: Country Experiences with Their Use and Liberalization, IMF Occasional Paper No. 190 (Washington: International Monetary Fund).
Bekaert, Geert, 1995, “Market Integration and Investment Barriers in Emerging Equity Markets,” World Bank Economic Review, Vol. 9, pp. 75–107.
Blume, Lawrence, David Easley, and Maureen O’Hara, 1994, “Market Statistics and Technical Analysis: The Role of Volume,” Journal of Finance, Vol. 69, pp. 153–81.
Boyle, Phelim P., and Ton Vorst, 1992, “Option Replication in Discrete Time with Transaction Costs,” Journal of Finance, Vol. 47, pp. 271–93.
Campbell, John Y., and Kenneth A. Froot, 1995, “Securities Transaction Taxes: What About International Experiences and Migrating Markets,” in Securities Transaction Taxes: False Hopes and Unintended Consequences, ed. by Suzanne Hammond (Chicago: Catalyst Institute).
Cox, John C., Stephen A. Ross, and Mark Rubinstein, 1979, “Option Pricing: A Simplified Approach,” Journal of Financial Economics, Vol. 7, pp. 229–63.
Domowitz, Ian, Jack Glen, and Ananth Madhavan, 2000, “Liquidity, Volatility, and Equity Trading Costs Across Countries and over Time” (unpublished; Washington: International Finance Corporation).
Easley, David, Maureen O’Hara, and P.S. Srinivas, 1998, “Option Volume and Stock Prices: Evidence on Where Informed Traders Trade,” Journal of Finance, Vol. 53, pp. 431–65.
Eichengreen, Barry, James Tobin, and Charles Wyplosz, 1995, “Two Cases for Sand in the Wheels of International Finance,” Economic Journal, Vol. 105, pp. 162–72.
Forbes, Kristin, 2002, “One Cost of the Chilean Capital Controls: Increased Financial Constraints for Smaller Traded Firms” (unpublished; Cambridge, Massachusetts: Massachusetts Institute of Technology).
French, Kenneth, and Richard Roll, 1986, “Stock Return Variances: The Arrival of Information and the Reaction of Traders,” Journal of Financial Economics, Vol. 17, pp. 5–26.
Hansch, Oliver, Narayan Y. Naik, and S. Viswanathan, 1998, “Do Inventories Matter in Dealership Markets? Evidence from the London Stock Exchange,” Journal of Finance, Vol. 53, pp. 1623–56.
Hasbrouck, Joel, and George Sofianos, 1993, “The Trades of Market Makers: An Empirical Analysis of NYSE Specialists,” Journal of Finance, Vol. 68, pp. 1565–93.
Hau, Harald, and Anne Chevallier, 2000, “Evidence on the Volatility Effect of a Security Transaction Tax,” (unpublished; Fontainebleau: INSEAD).
Honohan, Patrick, 2003, “Avoiding the Pitfalls in Taxing Financial Intermediation” in Taxation of Financial Intermediation: Theory and Practice for Emerging Markets, ed. by Patrick Honohan (New York: Oxford University Press).
Hu, Shing-yang, 1998, “The Effects of the Stock Transaction Tax on the Stock Market—Experiences from Asian Markets,” Pacific-Basin Finance Journal, Vol. 6, pp. 347–64.
Hull, John C., 1985, Options, Futures, and Other Derivatives (Upper Saddle River, New Jersey: Prentice Hall).
Jones, Charles M., and Paul J. Seguin, 1997, “Transaction Costs and Price Volatility: Evidence from Commission Deregulation,” American Economic Review, Vol. 87, pp. 728–37.
Leland, Hayne E., 1985, “Option Pricing and Replication with Transaction Taxes,” Journal of Finance, Vol. 60, pp. 1283–301.
Lyons, Richard K., 2001, The Micro structure Approach to Exchange Rates (Cambridge, Massachusetts: MIT Press).
Madhavan, Ananth, 2000, “Market Microstructure: A Survey,” Journal of Financial Markets, Vol. 3, No. 3, pp. 205–58.
Madhavan, Ananth, Matthew Richardson, and Mark Roomans, 1997, “Why Do Securities Prices Change? A Transaction-Level Analysis of NYSE Stocks,” Review of Financial Studies, Vol. 10, pp. 1035–64.
Pagano, Marco, and Alisa Roell, 1990, “Auction Markets, Dealership Markets and Execution Risk,” CEPR Financial Market Paper Series (London: Centre for Economic Policy Research).
Reinhart, Vincent R., 2000, “How the Machinery of International Finance Runs with Sand in its Wheels,” Review of International Economics, Vol. 8, pp. 74–85.
Roll, Richard, 1989, “Price Volatility, International Market Links, and Their Implications for Regulatory Policies,” Journal of Financial Services Research, Vol. 3, pp. 211–46.
Saporta, Victoria, and Kamhon Kan, 1997, “The Effects of Stamp Duty on the Level and Volatility of U.K. Equity Prices,” Working Paper (London: Bank of England).
Stiglitz, Joseph E., 1989, “Using Tax Policy to Curb Speculative Short-Term Trading,” Journal of Financial Services Research, Vol. 3, pp. 101–15.
Summers, Lawrence H., and Victoria P. Summers, 1989, “When Financial Markets Work Too Well: A Cautious Case for a Securities Transaction Tax,” Journal of Financial Services Research, Vol. 3, pp. 261–86.
Tobin, James, 1984, “On the Efficiency of the Financial System,” Lloyd’s Bank Review, Vol. 153, pp. 1–15.
Umlauf, Steven R., 1993, “Transaction Taxes and the Behavior of the Swedish Stock Market,” Journal of Financial Economics, Vol. 33, pp. 227–40.
Knrl Habermeier is an Advisor in the Monetary and Exchange Affairs Department of the International Monetary Fund, Andrei Kirilenko is an Economist in the Internatation a I Capital Markets Department of the International Monetary Fund. The authors thank Stefan Ingves, Kristin Forties. Patrick Honohan. Richard Lyons, and participants at the 2001 Australasian Finance and Banking conference, 2002 World Bank Workshop on the Taxation of Financial Intermediation, and the 2002 IMF Research Conference.
For example, Eichengreen, Tobin, and Wyplosz (1995) argue that “transaction taxes are one way to throw sand in the wheels of super-efficient financial vehicles,”
See, for example, Tobin (1984), Summers and Summers (1989), Stiglitz (1989), and Eichengreen, Tobin, and Wyplosz (1995).
See, for example, Campbell and Froot (1995).
Honohan (2003) gives a comprehensive overview of the difficulties in designing an optimal tax system for the financial sector.
A collective volume published by the Catalyst Institute in 1995 reviews most of the empirical research on financial transaction taxes. Empirical studies since 1995 have sought to address similar issues by using other datasets.
Officially, the extension of the tax to fixed-income instruments was supposed to achieve “neutrality” with the tax on equity transactions. See Campbell and Froot (1995).
By contrast, tobacco taxes accounted for 1.26 and 1.37 percent of the total revenue collected in 1987 and 1998, respectively.
The statistics are calculated by taking the participation rates reported in the paper as a fraction of 50 percent, the rate that implies that the market maker is a counterparty to all trades.
Reinhart (2000) argues that the introduction of STTs may also make asset prices more variable in the general equilibrium setting.
Leland (1985) develops an extension to the Black-Scholes continuous-time model and shows how to modify the variance in order to price call options in the presence of transaction costs. In Leland’s model, the variance increases in the presence of transaction costs, reflecting the discontinuous rebalancing of portfolios necessitated by transaction costs.
Ariyoshi and others (2000) provide a detailed review of the literature.