The Impact of Derivatives on Cash Markets: What Have We Learned? Stewart Mayhew Department of Banking and Finance Terry College of Business University of Georgia Athens, GA 30602-6253 October 27, 1999 Revised: February 3, 2000 The Impact of Derivatives on Cash Markets: What Have We Learned? Abstract This paper summarizes the theoretical and empirical research on how the introduction of derivative securities a? ects the underlying market. A wide array of theoretical approaches has been applied to the question of how speculative trading, the introduction of futures, or the introduction of options might a? ct the stability, liquidity and price informativeness of asset markets. In most cases, the resulting models predict that speculative trading and derivative markets stabilize the underlying market under certain restrictive conditions, but in general the predictions can go either way, depending on parameter values. The empirical evidence suggests that the introduction of derivatives does not destabilize the underlying market—either there is no e? ect or there is a decline in volatility—and that the introduction of derivatives tends to improve the liquidity and informativeness of markets. 1 Introduction
Writing in 1688, Joseph de la Vega describes various strategies used by a syndicate of bear traders to manipulate prices in the market for Dutch East India Shares at the Amsterdam Exchange. Some of these tricks involved trading options. For example, de la Vega reports that one strategy employed by the bears was to… … enter as many put contracts as possible, until the receivers of premiums [assumed to be bulls] do not dare buy more stock [on their own initiative]. [Their hands will be largely tied] because they are already obliged to take the stock [covered by the put premiums, if requested to do so].
Therefore the speculation for the decline has free course and is an almost sure success. Roger Lowenstein, writing in the Wall Street Journal, November 6, 1997, comments: When the Nobel Prize was awarded to the inventors of the formula for pricing stock options, the formula’s practical utility was widely noted. Last Monday, when the market cracked, we saw how practical indeed. To start at the conclusion, the vast growth in the market for options unleashed tremendous selling pressure in the stock market and accentuated Monday’s decline.
Previously, the very same options had encouraged stock prices to rise above the level justified by business valuations. The popular assertion that derivative securities tend to destabilize the underlying asset markets has persisted for more than three centuries. To what extent is this notion theoretically justi? ed? To what extent has it been supported by empirical evidence? This topic has been the focus of much academic scrutiny. The purpose of the present article is to provide a comprehensive review of the theoretical and empirical literature on this issue, and more generally on the relationships between 1 nderlying and derivative markets. 1 The debate on the e? ects of derivative trading is closely related to the more fundamental issue of the extent to which speculative trading in general in? uences market prices. Accordingly, we will begin by reviewing the theoretical literature on speculation and price stability. Much of the early literature in this area focused on the role of speculators in smoothing out seasonal price ? uctuations in commodity markets. As we shall see, traditional models of this ? avor generally conclude that under certain restrictive assumptions, speculative trading tends to stabilize prices.
When these assumptions are violated, it is often found that speculative trading can stabilize or destabilize prices, depending on parameter values. Next, we will discuss several models that explicitly incorporate forward contracting. Again, most of these models were developed in the context of a market for storable commodities. And again, most come to the ambiguous conclusion that under su? ciently restrictive assumptions, the introduction of futures stabilizes the cash market, but in general the result can go either way depending on parameter values.
Turning to the introduction of option contracts, the non-linear payo? structure of options adds an interesting dimension to the problem, that has inspired much insightful theoretical research. Still, no real consensus has emerged as to whether we should expect options to stabilize or destabilize the underlying market. As in the case for the introduction of futures, the range of theoretical possibilities is su? ciently broad as to accommodate nearly any conclusion, depending on what assumptions are made.
Nevertheless, it can be quite instructive to examine this literature carefully, for in doing so we may gain valuable insights to help us gain a richer appreciation of the complicated relationships between derivative and primary assets. The empirical literature on this issue is vast. Studies have been performed on data from commodity, ? xed-income, individual stock, stock-index, and currency futures, as well as individual For earlier surveys related to this topic, see Damodaran and Subrahmanyam (1992), Hodges (1992) and Sutcli? e (1997). 1 2 tock stock-index, and currency options, including markets in the United States and in many other nations. Researchers have studied the impact of derivatives by comparing underlying market characteristics before and after introduction dates, by studying the behavior of the underlying market around the expiration dates of the derivative contracts, and by examining lead-lag relationships between cash and derivative markets. As we shall see, most of the empirical evidence suggests that derivative markets do not increase volatility in cash markets, but do tend to make spot markets more liquid, and more informationally e? ient. There is very little evidence from any market that volatility increases with the introduction of derivatives. 2 Theory A variety of theoretical arguments have been advanced over the years to explain why speculative trading in general, or the existence of derivatives markets in particular, might a? ect the volatility of the underlying asset market. Let us ? rst summarize the long-running debate on whether speculative trading stabilizes prices. Next we will discuss several theoretical models that speci? cally address the e? ects of futures markets. After that, we will review models that deal speci? ally with the introduction of options. 2. 1 E? ects of Speculation Adam Smith (1776) observed that speculators help prevent extreme shortages (and by implication, extreme price movements) by buying and storing grain in periods when they forecast a shortage. 2 John Stuart Mill (1871) elaborated on this idea, explicitly observing that speculators play an important role in stabilizing prices. 3 Because they buy when prices are low and sell when prices are high, speculators improve the intertemporal allocation of resources and have a dampening e? ect on seasonal price ? uctuations.
Similarly, Mill observed, local price ? uctuations are reduced as 2 3 Book IV, Chapter 5. Book IV, Chapter II, sections 4-5. 3 speculators geographically reallocate goods, buying in regions where prices are low and selling them where prices are high. Mill recognized the possibility that speculators may attempt to manipulate prices, but argued that pro? table manipulation should not be possible. He argued that speculation will only be pro? table when speculators, as a group, buy low and sell high. The basic notion here is that destabilizing speculation cannot persist because it is not pro? table.
This argument was resurrected by Friedman (1953), who stated that “people who argue that speculation is generally destabilizing seldom realize that this is largely equivalent to saying that speculators lose money. ” Various authors over the years have attempted to demonstrate the fallacy of this proposition. As noted by Kaldor (1939), it is possible that speculators as a group may indeed lose money. Conceivably, the population of speculators may be composed of two groups, one of seasoned traders who, on average, make money and one a “? oating population” of novices, most of whom lose money and are driven out of the market.
Friedman’s (1953) implicit suggestion that pro? table speculation ought to stabilize prices, later dubbed “Friedman’s proposition,” inspired a new wave of attempts to demonstrate the contrary— that speculative trading can simultaneously be pro? table and destabilize markets. This line of research soon transformed into an formal investigation of the conditions under which Friedman’s proposition holds. This literature includes papers by Stein (1961), Baumol (1957), Telser (1959), Kemp (1963), Farrell (1966), Schimmler (1973), Jesse and Radcli? e (1981) and Hart and Kreps (1986).
Stein (1961) suggests that it is possible for foreign exchange traders to make positive pro? ts and destabilize exchange rates by attacking a currency. In spirit, this argument is similar to that of Kaldor (1939). Both arguments rely on a group of unpro? table speculators to o? set the gains of the pro? table ones. In Kaldor’s example, old speculators make pro? ts at the expense of new speculators. In Stein’s example, speculators make pro? ts at the expense of the monetary authority. An early attempt to construct a formal model in which speculative trading is destabilizing was made by Baumol (1957).
In this model, prices in the absence of speculators have predictable 4 seasonal ? uctuations. When speculators enter, they accelerate price movements by buying when prices are rising, and selling while prices are falling. The model captures the popular intuition that speculators are momentum traders. By modern standards, Baumol’s (1957) model seems awkward—there is no random component to prices, the speculative demand function is not derived from utility maximization, but is speci? ed ad hoc, and stability is measured by the frequency and amplitude of oscillations of sinusoidal functions, not by variance.
Moreover, as Telser (1959) pointed out, Baumol’s model is subject to a logical pitfall explicitly addressed by Mill (1871). In particular, the speculators in Baumol’s model generate only paper pro? ts; it is not clear that they could ever realize these pro? ts inasmuch as the trading activity necessary to unwind the position will a? ect prices. In response to Baumol (1957), Telser (1959) constructed a model with a pro? t-maximizing monopolist speculator. Roughly speaking, prices are mean reverting and the speculator, having a noisy estimate of the long-run mean, optimally trades on this information.
In this model, speculation unambiguously stabilizes prices, supporting the view espoused by Mill (1871) and Friedman (1953). Kemp (1963) supplied another counterexample, in which supply is inelastic and the asset is a Gi? en good, having a demand curve that slopes upward over a range of prices. Due to the perverse nature of the demand for the good, there are multiple equilibrium prices. In this framework, Kemp (1963) is able to construct an example where a relatively small change in speculative demand can cause a jump between equilibria, and this gives rise to the possibility of pro? able speculation. For a linear excess demand function, however, Kemp showed that positive speculative pro? ts imply price stabilization. Farrell (1966) investigates this proposition further, showing that if consumer demand is intertemporally independent, as indeed it is in the models of Telser (1959) and Kemp (1963), a linear excess demand function is necessary and su? cient for Friedman’s proposition to hold. In other words, Farrell (1966) demonstrated that if demand is intertemporally independent and linear, then whenever speculators are making pro? ts, they are stabilizing prices.
In addition, speculators may 5 stabilize prices even if they are making losses, as long as their losses are not too large. However, if demand is not linear, then pro? table speculation may be destabilizing. Schimmler (1973) showed that this result does not in fact require the independence assumption. In this framework, it turns out that Friedman’s proposition is equivalent to the proposition that consumer demand is linear. Jesse and Radcli? e (1981) show that the same result holds when consumer demand has a random component and speculators are risk-neutral.
Hart and Kreps (1986) consider a market where supply is constant and demand is subject to large but rare shocks. Each period, a noisy signal of next period’s demand is observed—the signal is always positive prior to a shock but a false positive is frequently observed. In the absence of speculative trading, prices are stable until the shock actually occurs. When speculators are added to this market, then the price increases in response to a positive signal as speculators buy the underlying asset in anticipation of a possible demand shock.
If the signal turns out to be false, then the price in the subsequent period is depressed as speculators dump their shares. Thus, the presence of speculators increases price ? uctuations, except in the rare case when the event actually occurs. In the various models discussed above, speculators trade on the basis of legitimate information about future demand. Other authors have addressed the possibility that speculators can destabilize prices through pure price manipulation. Among others, Hart (1977), Allen and Gale (1992), Allen and Gorton (1992), and Jarrow (1992) have investigated the conditions under which pro? able manipulation is possible. 2. 2 Futures Markets According to the traditional economic paradigm, prices are determined by the interaction of supply and demand functions, and prices change in response to shifts in these functions. In the absence of storage, supply is derived from the producers’ cost function, and demand is determined by utilitymaximizing consumers. If the product is storable, then the demand function for a given period 6 may be augmented by speculative demand by investors who anticipate price increases.
Likewise, supply is increased as speculators sell their inventories. To the extent that changes in production costs and consumer demand are predictable, the intertemporal reallocation made possible by storage generates trading activity that more properly may be viewed as inventory management, as opposed to speculation. If there are random shocks, however, then the act of buying and storing an asset is inherently risky. In the absence of a futures market, we would presume that risk averse speculators will hold inventories only if the expected return from doing so is su? ient to compensate them for bearing this risk. Thus, in equilibrium, the amount of intertemporal price smoothing will depend on the extent to which the commodity is storable, the relative magnitude of the predictable and random components of supply and demand changes, and the speculators’ level of risk aversion. When forwards or futures are introduced, a dealer can buy commodities when prices are low, and immediately lock in a selling price using a forward. To the extent that carrying costs are predictable, price smoothing through storage becomes an arbitrage activity.
If agents are risk averse, this should lead to increased intertemporal price smoothing. Futures markets may also in? uence spot prices if they have an e? ect on the behavior of producers. Since futures markets allow producers to hedge price risk, the existence of futures may a? ect a producer’s decision of what to produce, how much to produce, and what production techniques to use. In addition, the futures price may contain information about anticipated demand that can feed back into production decisions. As we shall see in the following sections, di? rent authors have modeled various dimensions of this complicated problem, not only in the traditional Marshallian paradigm, but also in a rational expectations equilibrium framework, and in a general equilibrium context. 7 2. 2. 1 Traditional Approach Various aspects of the relationship between futures markets, storage, and production are modeled by Peck (1976), Turnovsky (1979, 1983), Kawai (1983), Sarris (1984), Chari, Jagannathan and Jones (1990), and Chari and Jagannathan (1990). In Peck’s (1976) model, neither demand nor supply are random, and price movements are governed by an adaptive expectations model for the various agents.
The result is that when storage decisions are made on the basis of the futures price, futures have a stabilizing e? ect on prices. But when both production and storage decisions use the information in the futures price, the e? ect depends on the parameter values. In Turnovsky’s (1979) model, both supply and demand have stochastic components, and expectations about the future price are adaptive in the absence of a futures market, but rational when futures are introduced. Unlike Peck (1976), Turnovsky (1979) ? nds that futures markets may be destabilizing in he presence of storage alone, albeit for parameter values he considers to be implausible. Kawai (1983) formulates a model with rational expectations and three stochastic components: consumer demand, inventory demand, and supply. The resulting equilibrium is too complicated to analyze in general, but with additional restrictions, the model shows that futures are more likely to stabilize prices when the variance of consumer demand is high and the variance of inventory demand is low, with variance of supply having an ambiguous e? ect. Again, the net e? ect depends on the parameter values.
Turnovsky (1983) comes to a similar conclusion, except that in his model, the inventory demand e? ect is also ambiguous. In Turnovsky’s (1983) model, however, futures have a net stabilizing e? ect for a wide range of parameter values, as shown through simulations performed by Turnovsky and Campbell (1985). Another model worthy of note is that of Sarris (1984). This model focuses on the interaction between speculators in the cash and futures markets. In his model, futures markets unambiguously stabilize cash markets. The models of Peck (1976), Kawai (1983) and Turnovsky (1983) predict that futures are un- ambiguously stabilizing in the absence of storage and production uncertainty. Chari, Jagannathan and Jones (1990), however, present a model where this is not the case. In their model, if supply and demand are linear, the demand shock is additive, and producers have constant absolute risk aversion, then the introduction of futures stabilizes prices. However, the authors use counterexamples to demonstrate that when these conditions are violated, the introduction of futures may destabilize prices. More details of this model are discussed in a second paper, by Chari and Jagannathan (1990).
Chari and Jagannathan (1990) also suggest another way in which futures markets may in? uence spot markets. Recall that the introduction of futures may a? ect current production levels, by allowing producers to respond to new information about anticipated demand shocks. Chari and Jagannathan (1990) point out that this may a? ect tomorrow’s spot price, not only through storage activity, but also through intertemporal dependence in the production function. For example, suppose we have a non-storable good, for which the marginal cost function next year is an increasing function of the quantity produced this year.
And suppose that demand is low this year but, by observing the futures price, producers learn that demand will be unusually high next year. Then producers may ? nd it optimal to cut back on production this year in order to lower production costs for the following year. In this model, futures markets may stabilize or destabilize prices, with stabilization more likely when demand is inelastic and when supply is elastic. Even when prices are destabilized, however, in this model the introduction of futures improves welfare for all parties. 2. 2. 2 Rational Expectations Equilibrium Approach
Grossman (1977), Danthine (1978) and Bray (1981) develop fully-revealing rational expectations equilibrium models of futures markets. In these models, speculators observe a noisy signal of next period’s market demand, and trade in the futures market. As usual in rational expectations equilibrium models, all sources of uncertainty are assumed to be normally distributed, and all traders have utility functions characterized by constant absolute risk aversion. The rational expectations 9 equilibrium is fully revealing, meaning that all of the relevant information observed by speculators is impounded into market prices.
Danthine (1978) addresses the issue of price stability in this context. Because producers observe the futures price in making their production decisions, the equilibrium spot price will re? ect the speculators’ signals, including both the true information and the noise. The futures market in this model has a stabilizing in? uence insofar as the future price conveys information to farmers that help them adjust production to demand shocks. On the other hand, noise in the speculators’ signals has an o? setting, destabilizing e? ect.
This model is sometimes quoted as demonstrating that futures markets will stabilize prices. In fact, whether the futures market increases or decreases volatility depends on parameter values. Futures markets will stabilize spot markets more when the number of speculators is large, their signals are precise, and the variance of demand is high. If the speculators have poor information and there are not many demand ? uctuations to stabilize, then futures markets can destabilize prices in this model. Demers and Demers (1989) construct a partially-revealing rational expectations equilibrium model of the futures market.
In their model, there are two sources of uncertainty, production risk and demand risk. Information is costly, and producers have a comparative advantage at collecting information on production uncertainty, while speculators specialize in collecting information about demand uncertainty. When information collection is costly, speculators will not pay to acquire poor information. In this model, futures markets unambiguously stabilize prices, unlike Danthine’s (1978) model, where poorly informed speculators can destabilize prices. The Demers and Demers (1989) model also illustrates another interesting possibility.
Information about production is more useful to producers when used in conjunction with information about demand. Thus, it is possible for producers, in the absence of futures markets, to rationally refrain from gathering costly information about production, but to begin gathering that information once the futures market opens. 10 2. 2. 3 General Equilibrium Approach Weller and Yano (1987) analyze the e? ect of futures markets on spot prices in a general equilibrium model of an exchange economy with two agents, two goods, two periods and two states of nature.
In the second period, agent one receives a random endowment of good one, and agent two receives a non-random endowment of good two. The authors examine equilibrium both with and without forward contracting in period one. The stabilization question is addressed by comparing the variability of the time-two equilibrium price around its time-one expected value. In this context, it is shown that the introduction of a futures market has two e? ects. First, in the absence of a futures market, the two agents may have a di? erent marginal rate of substitution between wealth in state one and wealth in state two.
The introduction of a futures contract completes the market, allowing state prices to equilibrate across investors, and this, in general, leads to a more stable spot price. When the agents are risk-neutral, the period two spot price is perfectly predictable in period one. Second, if one or both of the agents are risk averse, then the futures market plays a role in allocating risk. In this case, the extent to which, futures markets stabilize or destabilize the spot price depends on the agents’ preferences with respect to the two goods, and their level of risk aversion.
The more risk averse the agents, the more important the risk transfer feature of the futures market becomes. If the agents are highly risk averse, and if the agent that produces the risk-free good generally prefers to consume the risky good, while the agent that produces the risky good prefers to consume the risk-free good, then the introduction of futures may increase the amount of uncertainty in spot prices. 2. 2. 4 Other Models Stein (1987) presents another information-based model, in which the increased speculation associated with the opening of a futures market may lead to price destabilization and decreased welfare.
In this model, there are two periods, and the supply of the asset is subject to both permanent and transitory shocks. There are two types of traders, hedgers and speculators. The hedgers have 11 an initial inventory position, can store the underlying asset costlessly, and observe the transitory shock. The speculators have no inventory, but observe information about the permanent shock. In order to assess the impact of futures markets, Stein compares the equilibrium with and without speculators. He ? nds two o? setting e? ects. As in other models, the in? x of new traders has a stabilizing, welfare-increasing e? ect through improved risk sharing. However, there is also an information externality arising from the fact that the speculators may have new information unavailable to the hedgers. In the special case where the speculators have perfect information about the permanent supply shock, the market price becomes perfectly informative, and the introduction of futures markets stabilizes prices. Likewise, when speculators have no information, futures markets also stabilize prices. But when speculators have a noisy signal, the hedgers eact to the possibility that the speculators may have additional information. Because the hedgers react, in part, to noise, there is a destabilizing e? ect. For some parameter values, this destabilizing e? ect outweighs the risk sharing bene? ts, prices are destabilized and welfare is reduced. Subrahmanyam (1991) presents an information-based model speci? c to “basket trading,” which may be applied to stock index futures. This model is based on that of Kyle (1985), in which informed and uninformed traders submit orders to a competitive market maker.
In Kyle’s model, the equilibrium pricing schedule is directly related to the level of informed trading—the greater the level of informed trading, the greater is the adverse selection problem faced by the market maker, the more sensitive will be the stock price to order ? ow, and, from the point of view of uninformed traders, the less liquid will be the market. Subrahmanyam (1991) extends Kyle’s framework to allow simultaneous trading in individual stocks and baskets of stocks. Under the assumption that most informed trading is based on ? m-speci? c information, it is shown that uninformed investors can reduce the deleterious e? ects of informed trading by trading the basket, rather than the individual stocks. Thus, this model predicts that with the introduction of stock index futures, uninformed investors will migrate from the stock market to the index futures market. This would leave a greater concentration of informed traders and poorer liquidity in the stock market. 12 Newbery (1987) analyzes another channel through which futures markets may increase spot price volatility.
Since futures markets allow producers to hedge production risk, they encourage producers to adopt more risky production technologies. Newbery illustrates this concept using a simple model where the producer may choose between two production modes, one in which output is certain and one in which output is uncertain but has a higher mean. In the absence of futures markets, producers, being risk averse, optimally select the safe production mode. When a futures market is introduced, producers move to the risky production mode and transfer the risk to speculators. The net e? ct is an increase in the variance of spot prices. It should be noted that even though futures markets destabilize prices in this model, welfare increases with futures introduction, both for the producers and the speculators. A similar intuition is modeled by Artus (1990). He constructs a simple model of the equilibrium interaction between producers and speculators, both with and without futures markets. In the basic model, futures markets stabilize prices, but the addition of production uncertainty may have a destabilizing e? ect, as in the model of Newberry (1987).
Artus (1990) also examines other extensions of his basic model under which the introduction of futures may destabilize prices. For example, destabilization may occur when market imperfections prevent arbitrageurs from stabilizing prices or when market participants have false expectations and learning is slow. In a subsequent paper, Artus (1996) examines the “market imperfections” model in more detail. Here, it is shown that whether futures markets stabilize or destabilize the underlying spot market may depend critically on the nature of the ? nancial constraints faced by market participants.
In particular, Artus (1996) considers two models. In one model, there are wealthy and poor investors, where poor investors face a binding wealth constraint on their desired investment position. In the second model, the two groups of traders have di? erent access to credit. In both cases, futures markets tend to have a stabilizing e? ect, but the e? ect of risk aversion and asymmetric information is di? erent in the two models. In the ? rst model, futures markets stabilize spot markets less when market participants are risk averse, while in the second model, the opposite result is true.
In both cases asymmetric 13 information causes futures markets to be less stabilizing. In the second model, but not the ? rst, asymmetric information can, for some parameter values, reverse the result, causing futures to destabilize the spot market. Zhou (1998) analyzes the interaction between producers, consumers and speculators in a continuoustime equilibrium model. He derives equilibrium conditions in an economy where producers may face liquidity constraints and consumers’ relative risk aversion is a decreasing function of the level of consumption. In this model, the e? ct of speculation is to induce a positive relationship between volatility and the futures price level. In particular, speculation makes futures prices more volatile in states of the world were futures prices are high, but less volatile in states where futures prices are low. Moreover, when prices and volatility are high, liquidity constraints will dampen volatility, but when prices are volatility are low, liquidity constraints lead to increased volatility. Finally, in Zhou’s (1998) model, the e? ect of agricultural price subsidies is to dampen volatility, especially in states when futures prices are low.
The Brady Commission (1988), in their analysis of the 1987 market crash, suggested that stock index futures markets may have contributed to the crash through yet another channel. Although they develop no formal model, they suggest that … the market’s break was exacerbated by the failure of institutions employing portfolio insurance strategies to understand that the markets in which the various instruments trade are economically linked into one equity market. Portfolio insurance theory assumes that it would be infeasible to sell huge volumes of stock on the exchange in short periods of time with only a small price impact.
These institutions came to believe that the futures market offered a separate haven of liquidity sufficient to allow them to liquidate huge positions over short periods of time with minimal price displacement. 14 2. 2. 5 Market Manipulation If the underlying asset market is not perfectly competitive, the introduction of futures may induce large producers to manipulate the cash prices through their production and storage decisions. Newbery (1984) analyzes a futures market on a commodity that is produced by one dominant producer with market power and multiple “fringe” producers.
A number of interesting results may arise. The dominant producer may ? nd it bene? cial to deliberately destabilize prices by randomizing production or storage decisions, in order to impose costs on smaller competitors. This is particularly likely if the large producer is less risk averse than the small producers. The dominant producer in this model will generally be harmed by the introduction of a futures market. It should be noted that in this type of model, various di? erent results may obtain, depending on what assumptions are made about isk aversion, and about the extent to which the producers have information about each others’ production plans. For further analysis of this issue, see Anderson and Sundaresan (1984). Other authors have advanced explicit models of market manipulation involving derivative markets. Most of these models ? nd that under certain assumptions, pro? table manipulation is possible. For a review of this literature, see Pirrong (1995). Papers in this literature include, among others, Kyle (1984), Kumar and Seppi (1992), Jarrow (1994), Pirrong (1995), and Cooper and Donaldson (1998). 2. 3
Option Markets To a large extent, the results described above on the introduction of futures markets may be extended to the introduction of options. If both calls and puts are introduced, then synthetic forward positions may be created, and in this sense listing options is like listing futures. There may be further e? ects of option introduction, however, arising from the special role of options in completing the market, as discussed by Ross (1976). To illustrate this concept, suppose that today the stock index is at 100, and tomorrow there are three possible states of the world, 15 n which the stock index goes to 99, 100, or 101. Two investments are available, an index fund, which, using vector notation, will be worth [99 100 101] and a risk-free bond, which will be worth [100 100 100]. There are various payo? pro? les that can be achieved using portfolios of the bond and the index fund, but there are some that can never be achieved. For example, one cannot create a portfolio with a payo? vector [1 0 0]. Suppose we introduce a call option with strike price of 100. This has payo? s [0 0 1]. Now, the payo? ectors of the available securities span the entire state space—any payo? function may be achieved using a portfolio of the bond, the index fund and the option. For example, to achieve the payo? [1 0 0], one would buy one unit of the bond, short the index fund, and buy one call option. If there are four states of the world, then the market can be completed by introducing two call options with di? erent strike prices. If there are n states of the world, the market can be completed by adding n ? 2 options. Alternatively, the market could be completed by adding put options or Arrow-Debreu securities.
Although moving from an incomplete market to a complete market is welfare-enhancing (in the sense that it represents a Pareto improvement), Hart (1975) demonstrated that the same cannot be said for a market expansion that fails to complete the market. 4 As illustrated by Hodges (1992), the volatility of the underlying asset may increase or decrease with the addition of new derivative securities that complete the market. If an option can be synthesized using a dynamic trading strategy in the stock and cash, then the option would appear to be redundant.
One might suspect that the introduction of a redundant option ought to have no impact on the underlying stock. Surprisingly, this is not necessarily the case. As argued by Grossman (1988), the price of a traded option can convey information that would otherwise be unobservable in an economy where the option can only be replicated. An option that appears redundant, in the sense that it can by dynamically replicated, might not actually be redundant, since introducing it might convey information that will change state prices. 4 Hakansson (1982) explores the relationship between market changes and welfare in more detail. 6 Grossman (1988) illustrates this point using a simple model where a random number of traders elect to follow a dynamic replicating strategy. The volatility of the stock price depends on the number of traders intending to follow the dynamic strategy. In the absence of a traded put option, the number of traders following the dynamic strategy, and therefore the volatility, is unknown. After the option is introduced, its equilibrium price conveys information about the anticipated volatility of the stock, and therefore about the fraction of traders intending to follow a dynamic trading strategy.
Grossman (1988) also suggests that derivative markets help curb excess volatility that might be induced by dynamic trading strategies, by lowering transaction costs and thereby allowing institutional traders to trade more smoothly over time. Back (1993) extends Kyle’s (1985) continuous-time model of informed trading in the presence of noise traders to include trading on an option. In Kyle’s model, the underlying stock in equilibrium follows a Brownian motion with constant volatility. Back (1993) shows that the introduction of options causes volatility to be stochastic.
It does not, however, change the expected average level of volatility. The basic intuition underlying Back’s model is the similar to that of Grossman’s (1988)—option trading conveys information not available in a similar market where options may be synthesized with dynamic trading strategies. Even though an option may be replicatable before it is introduced, that does not mean that introducing it has no e? ect on the spot process. A third model that captures the same intuition has been proposed by Kraus and Smith (1996).
This model highlights the fact that there is a distinction between market completeness and option replicability. Here, it is shown that it may be possible for each individual investor to dynamically replicate an option with portfolio of stock and cash, even if markets are not dynamically complete. In this case, di? erent investors may have di? erent information sets and di? erent state prices, and value the option di? erently. Once options are introduced, the option price aggregates information across investors, allowing di? erent investors’ state prices to equilibrate. In this context, the introduction of options may a? ct the equilibrium stock price even if each investor views it as redundant. 17 Detemple and Selden (1991) analyze the interaction between stocks and options when markets are incomplete. As a general result, they show that in incomplete markets, the equilibrium stock price depends on the characteristics of the option contracts available for trading. There are no comparative statics describing the direction of the e? ect for the most general model. However, examining a special case where the agents have quadratic utility and the heterogeneity of their beliefs is modeled in a simple way, the authors ? d that the introduction of options leads to an increase in the equilibrium stock price, and a reduction in the stock’s volatility. An alternative theoretical framework, based on the noisy rational expectations model of Hellwig (1980), is explored by Brennan and Cao (1996) and Cao (1999). In this framework, Brennan and Cao (1996) show that the introduction of an option with a quadratic payo? function can lead to a Pareto-e? cient allocation, and thus e? ectively complete the market. Option introduction in this model leads to increased market depth in the underlying market, and an indeterminate e? ct on trading volume in the underlying market, but has no e? ect on the price process for the underlying asset. Cao (1999), however, extends the model to show that this result changes when the acquisition of information is endogenized. He shows that the introduction of a “generalized straddle,”—of which the quadratic option considered in Brennan and Cao (1996) is a special case—increases the incentive to collect information, and leads to an increase in the price and a decrease in the volatility of the underlying stock.
It also leads to an increase in the price of other assets that are positively correlated with the underlying asset. John, Koticha and Subrahmanyam (1991) examine the simultaneous trading of a stock and option in the presence of an informed investor, using the sequential order arrival framework of Glosten and Milgrom (1985). In this model, there are risk-neutral market makers and noise traders in both markets, and the informed trader optimally trades in both markets.
Comparing the resulting equilibrium to the case without options trading, it is shown that the introduction of options leads to an improvement in liquidity and a reduction of volatility in the underlying stock market, but stock prices become less informative. Also, the insider’s optimal trading strategy and equilibrium 18 bid-ask spreads in both markets are shown to depend on the margin requirement in the option market. A similar model is considered by Biais and Hillion (1994). In their model, the market is incomplete without the option, and the option completes the market.
The liquidity traders are risk averse and trade to hedge endowment shocks. This endogeneity of liquidity trading turns out to have important implications. For some parameter values, the introduction of the option decreases the informativeness of the stock price, as in John, Koticha and Subrahmnayam (1991). But the authors show that the introduction of the option makes it less likely that the market will break down due to adverse selection. In cases where the option prevents the market from breaking down, the option improves the price informativeness of the stock.
Another result of this model is that the pro? tability of insider trading may increase or decrease with the introduction of options, depending on parameter values. A third sequential-trade model of simultaneous trading in option and stock markets is advanced by Easley, O’Hara and Srinivas (1998). In this model, it is shown that the propensity of the insider to trade in the option market depends on the leverage achievable through the option and the depth of the two markets. In some cases, there is a corner solution and the insider trades only in the stock market. 3 3. Empirical Evidence Commodity Futures Table 1 summarizes the results of several studies that have examined the e? ects of futures introduction on cash market stability in commodity markets. The papers summarized in this table employ various methods to test whether futures markets stabilize prices. The earliest papers simply compared price ranges or standard deviations between periods or across markets. For example, Emery (1896) examined the annual high-low range before and after the advent of corn and wheat futures in the United States, and Hooker (1901) compared 19
Table 1: Results of various studies on the volatility e? ect of the introduction of commodity futures. Author Emery (1896) Hooker Working Gray Powers Tomek Johnson Taylor and Leuthold Brorsen, Oellermann and Farris Weaver and Banerjee Antoniou and Foster Netz Kocagil (1901) (1960) (1963) (1970) (1971) (1973) (1974) (1989) (1990) (1992) (1995) (1997) Market Cotton Wheat Wheat Onions Onions Pork bellies Cattle Wheat Onions Cattle Cattle Cattle Crude Oil Wheat 4 Metals Volatility Lower Lower Lower Lower Lower Lower Lower Lower No e? ect Lower Higher No E? ect No E? ect Lower No E? ect he standard deviation of wheat price changes in Berlin to those in Chicago and Liverpool before and during a three-year trading suspension in the Berlin market. Subsequent authors have suggested various methodological re? nements, particularly with respect to how the data should be de-trended, and how “variance” should be de? ned. For example, Brorsen, Oellermann and Farris (1989) correct for seasonality by regressing daily price di? erences on sinusoidal functions with six- and twelve-month periods. Then, for each year, they calculate the standard deviation of the residuals from the ? st-stage regression. Finally, they subject these standard deviations to another regression analysis, with dummy variables for three subperiods. Antoniou and Foster (1992) estimate a Generalized Auto-Regressive Conditional Heteroskedasticity (GARCH) model for crude oil prices before and after the introduction of futures, and test for structural changes in the GARCH parameters using a Chow test. This approach allows for non-constant volatility in the pre- and post-event subsamples. Netz (1995) performed a more careful study of commodity futures introduction by directly 0 analyzing the e? ects of wheat futures markets on storage decisions from 1858 to 1890. Another interesting feature of Netz’s (1995) analysis is the inclusion of dummy variables corresponding to known attempts to corner the market. While she does ? nd that attempted corners do tend to raise volatility, markets are still more stable with futures markets than without. In addition to those mentioned in the table, other authors have examined the e? ects of futures introduction by testing whether the degree of price autocorrelation changed with the introduction of futures.
Cox (1976) studied the Onion, Potato, Pork Belly, Hog, Cattle and Orange Juice contracts. After the introduction of futures, he found less autocorrelation, and that the price forecast error is lower. This suggests that futures markets improve the information content of spot prices. Kodres and Schachter (1996) examine the impact of the introduction of wheat and corn futures options in January, 1926, and report a signi? cant decline in the volatility of the underlying futures prices. 3. 2 Fixed Income Futures Table 2 reports results from various studies testing for volatility e? cts associated with the introduction of GNMA and other ? xed-income futures. Most studies ? nd either no signi? cant e? ect, or else a decrease in volatility following futures introduction. Simpson and Ireland (1985) and Edwards (1988b) ? nd mixed results, depending on the size of the event window. Only Figlewski (1981) reports higher volatility following futures introduction. 3. 3 Stock Index Futures Results on the introduction of stock index futures are somewhat ambiguous. Many authors ? nd no signi? cant volatility e? ect associated with stock index futures listing.
Others, including Maberly, Allen and Gilbert (1989), Brorson (1991), Lee and Ohk (1992), Antoniou and Holmes (1995) and Gulen and Mayhew (2000) report a volatility increase in highly developed markets such as the United States, United Kingdom, and Japan. On the other hand, Antoniou, Holmes and Priestley (1998), Salih and Kurtas (1999) and Gulen and Mayhew (2000), ? nd evidence that volatility decreased with futures listing in many other countries. The results of these and other studies are 21 Table 2: Results of various studies on the volatility e? ect of the introduction of ? ed income futures. Author Froewiss (1978) Figlewski (1981) Dale and Workman (1981) Simpson and Ireland (1982) Bortz (1984) Corgel and Gay (1984) Moriarty and Tosini (1985) Simpson and Ireland (1985) Edwards (1988b) Ely (1991) Hegde (1994) Market GNMA GNMA T-Bill GNMA T-Bond GNMA GNMA T-Bill T-Bill Eurodollar Interest Rates T-Bond Volatility Lower Higher No E? ect No E? ect Lower Lower No E? ect Mixed Mixed Lower No E? ect No E? ect summarized in table 3. Many of these papers are reviewed by Sutcli? e (1997), so we will not go into too much detail here.
These papers use one of three basic approaches to analyze the e? ect of futures on the cash index. The majority of these papers compare the volatility of the index before and after the introduction of the futures contract, either using an unconditional measure of volatility, or using an ARCH/GARCH framework. Most of these papers look at only one market, making it somewhat di? cult to compare results across markets. However, there are a few exceptions, including papers by Lee and Ohk (1992), Antoniou, Priestley and Holmes (1998), Salih and Kurtas (1999) and Gulen and Mayhew (2000).
A second approach, exempli? ed by Harris (1989), Laatsch (1991), and Kumar, Sarin and Shastri (1995), is to compare the volatility of individual stocks within the index to a control sample of stocks that are not in the index. Harris (1989) reports that after futures listing, the volatility of stocks in the S&P 500 increased, relative to the volatility of stocks in a control sample. Laatsch (1991) performs a similar test for the introduction of futures on the Major Market Index (MMI), and ? nds no signi? cant e? ct. Kumar, Sarin and Shastri (1995) ? nd that in Japan, the 22 Table 3: Results of various studies on the volatility e? ect of stock index futures. Author Santoni (1987) Edwards (1988a, 1988b) Aggarwal (1988) Harris (1989) Mabery, Allen and Gilbert (1989) Fortune (1989) Becketti and Roberts (1990) Lockwood and Linn (1990) Brorson (1991) Chan and Karolyi (1991) Laatsch (1991) Gerety and Mulherin (1991) Hodgson and Nicholls (1991) Baldauf and Santoni (1991) Bessembinder and Seguin (1992) Board and Sutcli? (1992) Kamara, Miller, and Siegel (1992) Lee and Ohk (1992) Koch and Koch (1993) Bacha and Vila (1994) Brenner, Subrahmanyam and Uno (1994) Choi and Subrahmanyam (1994) Robinson (1994) Antoniou and Holmes (1995) Brown-Hruska and Kuserk (1995) Chen, Jarrett and Rhee (1995) Darrat and Rahman (1995) Kumar, Sarin and Shastri (1995) Kan (1996) Reyes (1996) Galloway and Miller (1997) Pericli and Koutmous (1997) Ragunathan and Peker (1997) Antoniou, Holmes & Priestley (1998) Chang, Cheng and Pinegar (1999) Gulen and Mayhew (2000)
Market S&P 500 S&P 500 Value Line S&P 500 S&P 500 S&P 500 S&P 500 S&P 500 DJIA S&P 500 Nikkei 225 MMI S&P 500 Australian AOI S&P 500 S&P 500 FT-SE 100 S&P 500 Australian AOI Hang Seng US, UK, Japan S&P 500/MMI Nikkei 225 Nikkei 225/TOPIX MMI FT-SE 100 FT-SE 100 S&P 500 TOPIX S&P 500 Nikkei 225 Hang Seng CAC 40 KFX (Denmark) MidCap 400 S&P 500 Australian AOI S&P 500, Nikkei225 FT-SE 100, IBEX 35 DAX 100, SWISS MI Nikkei 225 US and Japan 8 Countries 15 Countries Volatility No E? ect Lower No E? ect No E? ect Higher Higher No E? ect No E? ect Higher Higher No E? ect No E? ect No E? ct No E? ect No E? ect Lower No E? ect Mixed No E? ect Mixed Higher No E? ect No E? ect Mixed No E? ect Lower Higher Lower No E? ect No E? ect Lower No E? ect Lower No E? ect No E? ect No E? ect Mixed No E? ect Lower Mixed Higher No E? ect Lower 23 volatility of indexed stocks decreased relative to non-indexed stocks with the listing of index futures. This is interesting in light of the results reported by various other authors that Nikkei 225 index volatility either increased or remained unchanged at the time of futures listing. Chang, Cheng and Pinegar (1999) analyze the e? ct of index futures listing on the underlying stocks by decomposing portfolio volatility into the average volatility of component stocks and the cross sectional dispersion of returns. They ? nd that when Nikkei 225 futures were listed in Japan, the cross-sectional dispersion of returns across stocks in the index decreased, and index volatility increased proportionally more than the average volatility of the individual stocks. No such result was found for stocks outside the index, nor was any e? ect found at the time of o? -shore listing of Nikkei 225 futures in Singapore.
A third approach, used by Bessembinder and Seguin (1992, 1993) and others, is to test whether the introduction of stock index futures a? ects the volume-volatility relationship in the spot market, and whether spot market volatility is contemporaneously related to trading volume or open interest in the futures market. These authors ? nd that the unexpected component of futures trading activity (measured by volume or open interest) is positively related to spot market volatility, suggesting that futures market volume responds to unexpected volatility events.
The expected component of trading activity, however, was found to be negatively related to spot market volatility, suggesting that futures markets help stabilize cash markets. Gulen and Mayhew (2000) verify that open interest is negatively related to stock index volatility in many other countries, but with respect to trading volume, they do not ? nd any e? ect that is robust across countries. In addition, some e? ort has been made to test whether futures volume by di? erent types of traders has a di? erent e? ect on volatility.
For example, Hogan, Kroner and Sultan (1997), using a bivariate GARCH model and four years of data from the S&P 500 index futures market (1988-1991), ? nd that the positive relationship between volume and volatility is driven primarily by program trading. Chatrath, Ramchander and Song (1998) analyze the relationship between futures volume and volatility using the CFTC’s “Commitments of Traders” data, in which traders are classi? ed 24 as commercial traders (hedgers), non-commercial traders (speculators), small traders or spreaders. Using S&P 500 data from January 1986 through March 1995, they ? d that volatility is more sensitive to changes in non-speculative trading volume than speculative volume. They also ? nd that volatility is not sensitive to changes in the number of speculators or the size of their positions. Daigler and Wiley (1998) use the Chicago Board of Trade “Liquidity Data Bank” to di? erentiate the trading volume of ? oor traders and trading volume by the general public. Examining two years of daily data on MMI stock index futures and four other futures contracts, they ? nd that the positive volume-volatility relationship is driven by public volume, not by ? oor-trader volume.
Jegadeesh and Subrahmanyam (1993) examine whether the introduction of stock index futures had any e? ect on the liquidity of the component stocks. Although trading volume in the component stocks did increase, Jegadeesh and Subrahmanyam (1993) ? nd that after controlling for changes in price, volume, and volatility, bid-ask spreads on the component stocks increased signi? cantly with the introduction of S&P 500 index futures. The component of the bid-ask spread attributable to adverse-selection was also estimated to be higher after the introduction of index futures, but this result was not statistically signi? ant. 3. 4 Currency Futures Most research on derivative markets has concentrated on exchange-listed derivatives, for which data are readily available. In the FX market, exchange-traded futures represent only a small portion of the total forward trading activity. It is not surprising, therefore, that relatively few authors have studied the impact of currency futures on the volatility of exchange rates. Clifton (1985) documents a positive relationship between currency futures trading in Chicago and exchange rate volatility.
His study is based on the Japanese Yen, Swiss Franc, German Mark and Canadian Dollar futures markets between January 1980 and October 1983, and he uses the daily high-low spread as a measure of volatility. Chatrath, Ramchander and Song (1996) con? rm this result. Their study is based on the same four currencies plus the British Pound, they use a 25 much longer time period (ranging from twelve to seventeen years depending on the currency), and they employ a GARCH framework for modeling volatility. Adrangi and Chatrath (1998), examining eleven years of data on the Yen, Mark, Pound, and Canadian Dollar, ? d exchange rate volatility to be positively related to the temporary component of the speculators’ and small traders’ futures market positions. The volume-volatility relationship for currency futures is also examined by Bahr and Malliaris (1998). Jochum and Kodres (1998) study the relationship between futures trading and volatility for the Mexican Peso, Brazilian Real and Hungarian Forint, and ? nd no signi? cant e? ect. Shastri, Sultan and Tandon (1996) document a decrease in exchange rate volatility associated with the listing of currency options. 3. 5 Individual Equity Options
Many authors have examined the impact of option listing on the volatility or systematic risk of the underlying stocks. The results reported by various authors are summarized in table 4. In this table, a down arrow indicates that the authors report a signi? cant decrease in volatility following option listing, and a question mark indicates that the result was not statistically signi? cant, or that it went in opposite directions in di? erent subsamples. A number of comments are in order. First, it should be noted that nearly all of these papers, the only exceptions being Trennepohl and Dukes (1979) and Chaudhury and Elfakhani (1997), fail to ? d a signi? cant change in beta at the time of option introduction. Second, it should be noted that of all these papers, only one reports a signi? cant increase in volatility following option listing—that of Wei, Poon and Zee (1997), who studied the listing of options on OTC stocks. Most authors report a decrease in volatility. As demonstrated by Lamoureux and Panikkath (1994), Freund, McCann and Webb (1994) and Bollen (1998), however, this e? ect is probably due to market-wide phenomena—similar volatility declines may be observed in a control sample of ? rms that did not have options listed. While the results reported by these 6 Table 4: E? ect of Stock Option Listing on Volatility and Beta Author Nathan Associates (1974) CBOE (1975) Trennepohl and Dukes (1979) Klemkosky and Maness (1980) Whiteside, Dukes and Dunne (1981) Whiteside, Dukes and Dunne (1983) Nabar and Park (1988) Bansal, Pruitt and Wei (1989) Conrad (1989) Skinner (1989) Detemple and Jorion (1990) Haddad and Voorheis (1991) Rao, Tripathy and Dukes (1991) Damodaran and Lim (1991) Fedenia and Grammatikos (1992) Watt, Yadav and Draper (1992) Chamberlain, Cheung and Kwan (1993) Stucki and Wasserfallen (1994) Lamoureux and Panikkath (1994) Freund, McCann and Webb (1994) Long, Schinski and O? er (1994) Gjerde and S? ttem (1995) Elfakhani and Chaudhury (1995) Becchetti (1996) Chaudhury and Elfakhani Kabir Niendorf and Peterson Wei, Poon and Zee Alkeb? ck and Hagelin a Bollen St. Pierre (1997) (1997) (1997) (1997) (1998) (1998) (1998) Sample 16 40 32 103 35 71 390 175 96 293 300 327 45 200 438 39 37 11 527 685 111 7 119 174 30 35 110 144 32 1,010 140 Period 1973 1974-75 1973 1973-75 1973-75 1973-81 1973-86 1976-86 1974-80 1973-86 1973-86 1973-86 1985-87 1973-83 1973-87 1973-83 1979-87 1988 1973-88 1973-90 1985-90 1990-94 1979-87 1984-1989 1975-89 1978-93 1985-91 1985-90 1985-94 1973-92 1973-90
Market USA USA USA USA USA USA USA USA USA USA USA USA USA USA USA UK Canada Switzerland USA USA USA (OTC) Norway Canada Japan (warrants) Canada (puts) Netherlands USA USA (OTC) Sweden USA USA Volatility v v ? Beta v ? ? ? ? ? ? ? v v v v v v v v ? v ? v ? ? ? ? v v v ? ? ^ v ? ? ? ? ? ? ? v 27 authors do cast serious doubt on previous ? ndings, they do not necessarily imply that option listing has no e? ect. If the introduction of an option on one stock helps complete the market, this might, in equilibrium, a? ect all stocks, not just the stock underlying the option. Detemple and Jorion (1990) ave argued that the introduction of options on individual stocks has led to a decrease in market volatility. In this case it is unclear whether the control sample methodology is entirely appropriate. Proponents of the control-sample approach might argue that it is unreasonable to expect the introduction of options on one stock to a? ect the volatility of another stock. On the other hand, if a cross-stock listing e? ect does exist, one might expect it to be strongest within the same industry. In this case, selecting a control sample using stocks matched by industry would bias the researcher against ? ding a volatility e? ect. Third, one would think that if there is a robust volatility e? ect, it should go the same direction in di? erent subperiods, yet this does not appear to be the case. Bollen (1998) and Freund, McCann and Webb (1994) report di? erent volatility e? ects in di? erent subperiods. One interpretation is simply that the option listing e? ect is spurious. On the other hand, Detemple and Jorion (1990) argue that we should expect to see a di? erence between early option listings and late option listings, since the early option listings had more of a “market-completing” role than the later listings.
The earliest stock option listings occurred prior to the introduction of stock index futures and options. Options on large stocks may be viewed as substitutes for options on the market. The introduction of the ? rst stock options thus represented a signi? cant increase in the variety of investment and risk-management strategies available to investors. Now that index futures and options, industry sector options and a wide cross section of individual stock options are traded, the introduction of additional options on new stocks has only a marginal impact on market completeness. Fourth, many studies have veri? d that volume in the underlying stock tends to increase after stock options are listed. Among the authors who have addressed this issue are Hayes and Tennenbaum (1979), Branch and Finnerty (1981), Skinner (1989), Bansal, Pruitt and Wei (1989), Rao, Tripathy and Dukes (1991), Gjerde and S? ttem (1995), Wei, Poon and Zee (1997) and Kumar, Sarin 28 and Shastri (1998). It is possible that the apparent change in volatility may be related to a concurrent change in liquidity. In support of this hypothesis, Niendorf and Peterson (1996) have found evidence that changes in volume and volatility at the time of option listing are cross-sectionally elated. Fifth, it has been found that after the introduction of options, prices tend to re? ect new information more quickly, bid-ask spreads tend to narrow, and the adverse selection component of the bid-ask spread becomes smaller. 5 This e? ect could be related to the increased volume and decreased volatility. Fedenia and Grammatikos (1992) ? nd that bid-ask spreads become narrower for NYSE stocks but wider for OTC stocks, for which Wei, Poon and Zee (1997) reported a volatility increase. Using a vector augoregression approach, G? er (1996) analyzes the relationship between u price changes and trading volume for stocks with and without listed options. Among other results, she reports that stock trading volume is more informative for optioned stocks than for non-optioned stocks. G? ner (1996) also ? nds that for non-optioned stocks, long volume in the stock is more inu formative than short volume, but the same result does not hold for optioned stocks, suggesting that options markets signi? cantly lower the cost for informed traders to take short positions in the underlying asset.
Sixth, several authors have tested for abnormal stock returns associated with option listing. A positive price e? ect is documented by Branch and Finnerty (1981), Rao and Ma (1987), Conrad (1989), Detemple and Jorion (1990), Kim and Young (1991), Watt, Yadav and Draper (1992), Stucki and Wasserfallen (1994) and Gjerde and S? ttem (1995). These studies di? er with respect to whether they examine the announcement date or the event date, and disagree as to whether this price response is permanent or transitory. 6 Rao and Ma (1987) report a negative price e? ct on the announcement date and a positive price e? ect on the listing date, while several other authors ? nd no signi? cant announcement day e? ect. In some respects, this price e? ect is reminiscent of the See Jennings and Starks (1986), Rao, Tripathy and Dukes (1991), Fedenia and Grammatikos (1992), Gjerde and S? ttem (1995), Niendorf and Peterson (1997) and Kumar, Sarin and Shastri (1998) 6 See Broughton and Smith (1997). 5 29 S&P 500 listing e? ect. Indeed, Damodaran and Lim (1991) document an increase in institutional ownership following option listing.
On the other hand, recent evidence suggests that the positive listing e? ect is not nearly as robust as previously thought. Ho and Liu (1997) and Sorescu (1999) document that in the United States, option listing is associated with negative excess returns in the post-1981 period. Moreover, Kabir (1997) documents a negative price response associated with the listing of options in the Netherlands. Schinski and Long (1995) ? nd evidence from options on OTC ? rms suggesting that the price response to option listing is driven by changes in liquidity in the underlying asset.
Speci? cally, they found that smaller ? rms experience an increase in