S ecurity trading differs dramatically from country to country and from security to security within a country. For example, the hurly-burly of the open outcry system in the trading pits at the Chicago Mercantile Exchange (CME) and the Chicago Board of Trade (CBOT) stands in marked contrast to the tranquillity of the empty trading floor at the Sydney Stock Exchange, where all trades are conducted electronically off the floor of the exchange. The New York Stock Exchange (NYSE) relies on a specialist system to trade stocks whereas its chief US competitor, Nasdaq (the National Association of Securities Dealers Automated Quotation system), uses competing market-makers.
Should investors be concerned about the technology and organizational details of trading when making investment decisions? Should investors treat the trading process as a black box not worth scrutinizing for any reason other than intellectual curiosity? Or are there important lessons to be learned from studying the minutiae of the trading process?
Investors should be concerned about the returns they expect to realize in a market, net of the cost of trading. Various features of markets, such as their reliance on electronic technology, may be highly visible but ultimately the investor’s real concern is with the economic cost of trading. These costs are more than the out-of-pocket costs associated with a trade. Some elements of the economic costs, however, are difficult to estimate.
Consequently, market features (such as low trading volume) are often used as indications of trading costs in lieu of precise measurements. Some short-term traders, especially those who buy and sell at least once during a day, base their strategies on institutional features of the market — which are not discussed in this article.
Designers of markets cannot treat the trading process as a black box because they need to understand how particular features of the trading process affect the cost of trading. Nevertheless, the cause and effect relationship is not well understood. In particular, why some markets are highly liquid and others are not remains a mystery. Academic inquiry into most aspects of the trading process is known as market microstructure (although our colleague Merton Miller argues that micromarket structure seems equally descriptive) and is in its relative infancy.
An organized securities market or exchange is a network of potential buyers and sellers or their agents. Although the term ‘market’ originally connoted a central physical location, improvements in communication have allowed geographically dispersed individuals to form an organized market. Organized markets have rules about who is allowed to trade, what securities can be traded, the types of trades that can be made, how trades are consummated and how the integrity of trades is ensured.
Exchanges compete both domestically and internationally for trading volume and for new listings of securities. Even though most organized securities markets are privately owned and operated, virtually all securities markets that allow public participation are subject to government regulation. For example, in the US the Securities and Exchange Commission (SEC) regulates the behavior of stock exchanges.
Stocks can be bought and sold privately. For example, you may be able to convince your (soon to be ex) brother-in-law to pay $5 per share for your 10,000 Bre-X shares, even though those shares are no longer traded on a major stock exchange. However, private sales are typically not cost effective because of the costs of searching for potential buyers and sellers, negotiating a price and ensuring the trade will still be consummated if one of the parties reneges on the deal.
The potential volume of trading in some securities (such as corporate bonds, shares in limited partnerships and small firms) does not warrant the overheads associated with a public market. Markets for those securities closely resemble private transactions and these markets are labeled generically as over-the-counter (OTC) markets.
OTC trades are arranged between a would-be buyer or seller and a large institution, typically an investment bank such as Morgan Stanley. Those trades can be tailored to meet the specifications of the buyer or seller, whereas in a public market trades are relatively homogeneous in order to enhance liquidity.
The OTC markets are more active in derivative securities, such as options, where some buyers or sellers seek highly specific contracts (particularly in terms of the time to maturity). However, the foreign currency market involves homogeneous trades and is dominated by the interbank market, a principal-to-principal market akin to the OTC market.
Electronic technology has reduced search costs dramatically and electronic bulletin boards provide efficient ways to advertize willingness to buy or sell particular securities. However, the integrity of those trades is more difficult to guarantee.
Dealer markets are familiar to lay people. The dealer (or market-maker) is essentially a middle man who buys at one price (the bid) and sells at a slightly higher price (the ask, or offer).
Dealers buy stocks into inventory, sell stocks from that inventory and may even sell stocks short. Dealers attempt to make money from the spread (the difference between the bid and the ask) and bear the risk of adverse price moves in their inventory. They can manage their inventory by adjusting their quotes, by transacting with other dealers or by hedging in derivatives markets. Nasdaq and the London Stock Exchange (LSE) provide examples of important dealer markets for stocks; and most bond markets are dealer markets.
In an auction market, all orders to buy or sell are channeled to a central location (even if that location is an address in a computer) and a market-clearing price is determined by means of a set of rules (or an algorithm) that determines, among other things, the priority of different offers to buy and sell. These rules are more complex than those of auction houses such as Sotheby’s because in securities markets the supply of the item is not known when potential buyers submit their bids. Moreover, auction markets for securities can be discrete or continuous.
In a discrete auction, orders to buy or sell a stock are accumulated and at a particular time a single price is set to clear the market, again according to a prespecified set of rules. Opening prices on large markets such as the NYSE and Tokyo are set using this process. They are known as a call market. Some smaller markets, especially nascent emerging markets, operate a call market once or several times a day and in some a call market is operated for a sub-set of securities.
Call markets are seldom pure auction markets because a dealer (usually known as a specialist) is often introduced to supply liquidity. Occasionally, discrete auctions are designed to produce multiple prices, such as the auction for newly issued US government securities. In these auctions, the fixed supply is allocated first to the highest bidder at the bidder’s price, then to the next highest bidder at that bidder’s price and so on until the supply is exhausted.
A continuous auction market is a hybrid of an auction market and a dealer market because of the specialist’s role. Customers submit market, or limit, orders to a retail or discount broker. Those orders are transmitted to the specialist either electronically or by a floor broker. The specialist for a particular stock matches (crosses) some customers’ orders with each other where possible and provides additional liquidity by selling from inventory or buying into inventory to accommodate customers’ orders.
Some continuous auction markets do not employ specialists although a subset of traders (variously known as ‘locals’, ‘scalpers’ or ‘jobbers’) assumes that de facto role by being prepared to take the opposite side of trades initiated by brokers on behalf of customers. During normal trading hours, the NYSE and the Tokyo Stock Exchange are continuous auction markets for stocks, and the CME and the CBOT for futures and options contracts.
One of the chief differences between auction and dealer markets (seen as an advantage by those who favor auction markets) is the greater extent to which customers trade with customers in an auction market with a limit order book. If there is a limit order book in an auction market, the current quote reflects either a limit order from a customer or the specialist’s quote. Thus, an incoming market order can be executed at a price established by another investor.
In contrast, in many dealer markets (in particular Nasdaq) the current quote reflects the best quote (highest bid, lowest ask) by a dealer but not necessarily customers limit orders. Consequently, in many dealer markets, customers market orders and limit orders are executed at dealers’ quotes, not against each other. For example, if the best dealer quotes are bid $40 and ask $40.75, a limit order to buy at $40.50 will not execute even if a market order to sell arrives and is executed at the dealer s bid of $40. In fact, the limit order will not execute until the quoted ask falls to $40.50.
Recent rule changes by Nasdaq allow customer limit orders for some stocks to be executed against market orders so this distinction is not as sharp as it once was. Proponents of dealer markets point out that competition among dealers should provide more favorable trading conditions for investors than the auction market with its monopoly market-maker (the specialist). This issue is hotly debated and remains an open question.
Many who witness the frenzy of trading floors are convinced electronic trading would be more efficient. This is a controversial topic, particularly in options and futures markets relying on trading pits with open outcry. Three points are crucial.
First, there is no unique electronic trading system. All of the protocols and rules necessary for trading must be programmed into the electronic market-clearing algorithm. Although those rules may be easier to monitor and enforce electronically, there is not a universally accepted set of protocols and rules.
Second, unless dealers monitor their posted quotes constantly, those quotes become stale when new information arrives and dealers risk being ‘picked off’ (taken advantage of by a trader with up-to-date information). It is not feasible to program electronic systems to replicate human judgment in revising or suspending quotes to accommodate flows of information.
For example, traders know that quotes and limit orders need to be revised dramatically when Intel warns of substantially lower earnings due to reduced demand for its products in Europe. But the algorithm or artificial intelligence that would produce the desired result is beyond the current capabilities of electronic systems. Manual revision of quotes on electronic systems is slower than human reaction time, especially if the electronic system becomes congested as traders try to exploit stale quotes.
Third, proponents of floor trading argue that the physical proximity of traders on the floor provides clues and strategic information to facilitate trading (especially to enhance liquidity) and those intangibles cannot be reproduced in an electronic system. No one has yet devised a scientific test of this proposition nor of a compelling alternative explanation, namely that those who have a comparative advantage at floor trading are reluctant to see it eroded. Some exchanges have converted to electronic trading (for example, London, Toronto, Paris and Sydney) without apparent harm; newly created exchanges are almost universally electronic; and some exchanges that have hard core ‘eyeball-to-eyeball’ trading systems during normal trading hours resort to electronic trading systems for after-hours trading.
Even exchanges with floor trading have made extensive use of electronics for transmitting or routing orders from customers to the trading floor. This creates interesting contrasts with orders being transcribed from computer screens to trade tickets by hand. One prominent US academic compared the process to the use of a telephone if it had been invented in the days of the Pony Express delivery system: the sender of the message could have used the telephone to warn the recipient that the rider had just left with an important message.
There are also electronic trading systems that are independent of any big exchange. Instinet, a continuous auction system now owned by Reuters, is the best known of these and is the only stand-alone electronic system with significant trading volume. The Arizona Stock Exchange, despite its name, is a computer (located in Arizona) and is a single-price auction system that, once a day, establishes a price that will produce the largest trading volume from a list of buy and sell orders.
Some systems (such as Posit) are little more than bulletin boards and are called crossing systems because they are simply a mechanism to match a buyer and seller at a price determined in another market, typically the closing price on a major exchange. Virtually all of the customers who trade on these electronic systems are institutions such as pension or mutual funds and market-makers.
The economic cost of trading in a particular market comprises more than out-of-pocket transactions costs such as fees, commissions and transactions, or transfer taxes, that
Reuters' Instinet: a stand-alone continuous action system
are common outside the US. Investors’ returns are reduced by the normal bid-ask spread and the price impact (if any) of the trade; counterparty risk can be prohibitive in some markets.
The bid-ask spread represents a cost of trading because if an investor pays the ask (say $10/4) and sells at the bid (say $10), the investor has lost the spread (12.5 cents, or 1.23 per cent). Equivalently, the stock price has to increase by the spread if the investor is to break even. For some stocks, the spread presents a considerable obstacle to earning returns. For example, in May 1997, Sharper Image (a catalog retailer of high-end electronic gadgets and toys) was quoted as $3/4 (bid) and $3% (ask). An investor who buys this stock requires a 25 cent price move to break even, a rate of return of 7.4 per cent. This is 7.4 per cent per transaction not 7.4 per cent per year, so if the investor is an active trader with a short holding period the spread looms large.
Consider an analogy. Instead of renting a car on your next three-day business trip you could buy a new car and sell it at the end of your trip. Car rental rates do not appear high compared with the cost of this round-trip transaction.
The bid-ask spread provides income to the dealer to compensate for the costs of carrying an inventory or of bearing the risks of short-selling if the dealer is willing to sell more than the inventory. Spreads are wider for more volatile stocks because the risk of holding inventory is greater and spreads widen when uncertainty about a particular stock increases. Actual transactions often occur at prices within the posted (quoted) spread (estimates range from 20 per cent to 40 per cent of the time), that is, at prices higher than the bid for sales and lower than the ask for purchases. Thus effective spreads are narrower than quoted or posted spreads. How this occurs depends on the trading system.
In a dealer market, trades can be negotiated within the spread, especially by institutions. In a continuous auction market, trades occur within the quoted spread when the specialist or a broker in the crowd (essentially floor brokers who congregate at the trading post) fills an order at a better price than the quoted spread.
Transactions prices (and quotes) change by discrete amounts (the tick) and the smallest change is the minimum tick size, again set by exchange rules. In US markets, typical minimum tick sizes have been eighths of a dollar (12.5 cents) historically, although 32nds (3.25 cents), 16ths (‘teenies’, 6.25 cents) and quarters (25 cents) apply for some securities and for some markets. The bid-ask spread must be at least the minimum tick size.
In markets outside the US, prices are typically stated in decimals (for example, in London, £10.26). The NYSE has announced plans to move to decimals (for example $40.30) by the year 2000. Consequently, the minimum tick size will be 1 cent. There is considerable speculation about the effect of the proposed change on the size of the bid- ask spread. It is widely believed that the spread will be reduced. Unless there is a commensurate increase in volume, profits to those who post the spreads (specialists and dealers) must decline.
A market for a particular security is highly liquid (or very deep) if large quantities of that security can be bought or sold without affecting the price of the security. Price volatility is often taken as an indicator of illiquidity, but even a highly liquid security can have high volatility if there is considerable uncertainty about it. Similarly, large trades can move the price of even highly liquid securities if other traders believe that such a trade is a measure of the amount of information possessed by the trader.
Other common indicators of stock liquidity are the bid-ask spread, the float (the number of shares available for trading) and the typical trading volume of the stock. Liquidity is in great demand but its supply is something of a mystery. If a market is highly liquid, someone must be willing to take the opposite side of each trade. If that were not so, a buy order would languish until the arrival of a fortuitous sell order and a sell order would be in limbo until a buy order arrived.
Liquidity can be supplied in various ways. For example, a dealer or specialist might be prepared to buy into inventory any stock offered at the prevailing price and to sell from inventory (or sell short) at slightly above the prevailing price.
Additional liquidity can be provided by the limit order book. If there are offers to buy and sell large quantities of the stock at slightly below and above the current bid and ask, large quantities of the stock can be traded at close to prevailing market prices. Buying or selling large quantities of a stock relative to normal trading volume usually involves procedures that differ from those associated with a normal trade.
Quoted bids and asks apply to only a specific number of shares (ranging from 100 to several thousand) with the specific number set by the rules of the stock exchange. Larger trades must be negotiated with the dealer or the specialist and, if the trade is sufficiently large, the trade may occur outside the market in what is known as the third market or upstairs market.
Trades above a certain size (10,000 shares for the NYSE) are called ‘blocks’. For actively traded, highly liquid stocks such as AT&T, GE or British Petroleum, 10,000- share trades may be accommodated readily in the normal trading process whereas for other companies, such as Sharper Image, such trades represent several days’ normal trading volume. Would-be buyers and sellers of large blocks often rely on block houses (generally arms of investment banks) to trade the block off the exchange.
For a block sale, the block house can adopt one of two strategies: either ‘shop the block’ or buy the block and then attempt to resell it. To shop, or position, the block, the
broker contacts large institutions to determine their interest in the block or in pieces of the block. Such a strategy exposes the customer to front running, that is, to the risk that when an institution is offered the block it will decline but then use its knowledge of the impending trade strategically in the market. Successful block houses mitigate front running by establishing working relationships with large institutional investors.
Alternatively, the block house can purchase the block outright and then attempt to resell it at a profit. For example, when the Kuwaiti government decided to sell its 170m shares in British Petroleum, it invited big investment banks to bid for the block. Goldman Sachs paid almost $2bn for the block when the closing price for BP in London was $12.25 per share. Goldman resold those shares to institutional clients around the world within 36 hours for a price of $11.75, a discount of four per cent from the closing price. It is estimated that Goldman’s profit (before expenses) on the deal was $15m.
Similarly, in a $1.4bn transaction when Time Warner was trading at $48,875 per share, Merrill Lynch paid $46.33 per share for a block of 30 million Time Warner shares (formerly owned by Seagram), a discount of slightly over five per cent. Merrill resold the shares to institutional clients within a day for $46.75 per share, a profit (before expenses) of almost $13m. Time Warner then traded at $47,125, a discount of almost four per cent from its pre-block price.
Investors who want to buy or sell large quantities can accumulate or dispose of their position gradually but such a strategy has costs. First, it lacks immediacy so cash is not readily available if the investor is selling or cash sits idle if the investor is buying. Second, if the investor believes he or she is better informed than the market, delay makes it more likely others will become similarly informed or will detect the selling or buying activity and mimic it.
Professional investors attempt to measure the market impact of their trades and to adopt trading strategies to trade off market impact with their desire for immediacy. Market impact is measured by noting the price of the security when the buy or sell decision is made and comparing that with the price of the security after the trade has taken place.
Counterparty risk unambiguously increases the economic costs of trading but the magnitude of the increase is difficult to estimate. In some emerging markets, concern about counterparty risk may be sufficient to deter participation by rational investors. Every purchase and sale entails counterparty risk, that is, the risk that the other party to the transaction will not fulfil the terms of the agreed trade.
Stock exchanges establish clearing and settlement rules to eliminate or minimize counterparty risk. Essentially, the exchange (technically, the clearing house) becomes the counterparty to every trade. If either party defaults, the other party looks to the dealing house to fulfil the terms of the trade. The clearing house in turn protects itself by monitoring the creditworthiness and reputation of those who offer to buy and sell in the market. Although counterparty risk is negligible in well-established markets such as New York, London and Tokyo, it is a serious concern in some emerging markets.
Counterparty risk can manifest itself in various forms. For example, a buyer may not receive good title to the security or the seller may relinquish title to the security but may not be paid on time (or at all). Less obviously, the agreed-upon trade may not be honored under certain circumstances. If the stock price rises after the supposedly agreed trade, the seller experiences remorse and if the stock price falls the buyer
experiences remorse. To prevent that remorse from being translated into action, there must be a mechanism (such as a well-capitalized clearing house) to ensure that the terms of the trade will be honoured by both parties regardless of any subsequent movement in the price of the security.
Some institutional investors claim that, even in some established markets, the incidence of ‘misunderstandings’ about the terms of the trade seems to be a function of the subsequent performance of the stock. In those markets, an offer to buy a stock is akin to giving the putative seller a free option to deliver the stock if its price declines and to retain the stock if its price rises.
The language and the picayune technical characteristics of securities markets can be daunting, even to an otherwise savvy investor. Ultimately, however the technical aspects of trading are relevant to an investor only to the extent that they affect the economic costs of trading.
Conceptually, the economic costs are easy to identify: the bid-ask spread, liquidity or depth, and counterparty risk. Precise measurement of these costs is not as simple but investors should recognize that these costs, not simply out-of-pocket costs, such as brokerage fees and commissions, potentially reduce the returns they can achieve in securities markets. Moreover, the costs of trading can differ substantially from market to market and from security to security within a market.
This article is about the minutiae of the security trading process, different types of international stock exchange (and over-the-counter markets), and the costs of dealing.
Brad Barber and Richard Leftwich note the fundamental distinction between ‘dealer’ markets in which middlemen stand ready to buy at one price and sell at another, making money from the spread, and ‘auction’ markets in which all orders are channeled to a central location and a market-clearing price determined.
The merits of electronic trading are often advanced, but this is a controversial topic and there are arguments in favor of human reaction time and the physical proximity of traders.
Conceptually, the economic costs of the bid-ask (offer) spread, liquidity and counterparty risk are easy to identify. Precise measurement is trickier and investors should realize that these as well as brokerage, fees and commission potentially reduce returns.
Suggested further reading
Christie, W.G. and Schultz, P., (1994), ‘Why do NASDAQ market makers avoid Odd-Eighth Quotes?' Journal of Finance 49, 1813-1840.
Holthausen, R.W., Leftwich, R.W. and Mayers, D., (1990), ‘Large-block transactions, the speed of response, and temporary and permanent stock-price effects’, Journal of Financial Econom ics 26, July, 71—95.
Huang, R.D. and Stoll, H.R., (1996), 'Dealer versus auction markets: a paired comparison of execution costs on Nasdaq and the NYSE’, Journal of Financial Economics 41, July, 313-357.
Keim, D.B. and Madhavan, A., (1996), ‘Execution costs and investment performance: an empirical analysis of institutional equity trades’, Working paper, The Wharton School of the University of Pennsylvania.
Peterson, M. and Fialkowski, D., (1994), Posted versus effective spreads: good prices or bad quotes? 35, 269-292.
Stoll, H.R., (1993), ‘Equity trading costs in-the-large’, Journal of Portfolio Management, Summer, 41-50.