The price effect of option introductions

MISSING IMAGE

Material Information

Title:
The price effect of option introductions 1973-1992
Physical Description:
viii, 131 leaves : ill. ; 29 cm.
Language:
English
Creator:
Sorescu, Sorin
Publication Date:

Subjects

Subjects / Keywords:
Finance, Insurance and Real Estate thesis, Ph. D
Dissertations, Academic -- Finance, Insurance and Real Estate -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1996.
Bibliography:
Includes bibliographical references (leaves 128-130).
Statement of Responsibility:
by Sorin Sorescu.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 023836156
oclc - 35790437
System ID:
AA00013595:00001


This item is only available as the following downloads:


Full Text









THE PRICE EFFECT OF OPTION INTRODUCTIONS
1973-1992














By

SORIN SORESCU


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1996


UNIVERSITY OF FLORIDA LIBRARIES





























Copyright 1996

by

Sorin Sorescu













To my wife Alina, living proof that Love is the only raison d'6tre.

To my parents Gabriela and Mihai whose courage and determination made all of

this possible.













ACKNOWLEDGMENTS


This dissertation was inspired by initial conversations with Mark Flannery,

David T. Brown and Jerome Detemple. I am truly indebted to Mark Flannery, who

provided guidance, support and inspiration throughout the entire process. I also wish to

express my sincere thanks to Mahendrarajah Nimalendran, Jonathan Hamilton, Andy

Naranjo, and in particular to David T. Brown, for many excellent comments received on

earlier versions

This research was supported in part by a Doctoral Fellowship from the Social

Sciences and Humanities Research Council of Canada, whose contribution is greatly

appreciated















TABLE OF CONTENTS


ACKNOWLEDGMENTS ........


ABSTRACT.

CHAPTERS


1 INTRODUCTION .............

2 REVIEW OF LITERATURE


Intro ductio n .......... .. .. ... ... .. .............
E m pirical P apers ............... .. ............ ..... ............
Theoretical Papers Based on Market Completion ........
Other Theoretical Papers .................. ................
Conclusions of the Literature Review .......................


3 THE PRICE EFFECT OF OPTION INTRODUCTIONS
REVISITED: 1973-1992 ... ....... ...


Introduction .. ............. ......
Data Sources and Methodology.
Empirical Results ......... ..........


4 THE REGULATORY ENVIRONMENT OF STANDARDIZED
OPTION TRADING: 1973-1992 ....................

Introduction ............... .... .. ...... .......... ....
The O options Study .......... ........ ........................................
Regulatory Changes Related to the Quality of Information
Available to Investors .......................................
Regulatory Changes Related to the Prevention of Stock
Price M manipulation ...... ......... .......................
M multiple Listings of O options ........ .... .................................

5 EVIDENCE OF STOCK PRICE MANIPULATION AROUND
OPTION INTRODUCTIONS: 1973-1980 ...................


. ....... 3 7


....... .. ..... .............. ...... iv


.. ...... .. .... ... .. v ii


............ .............. .... ... 4
.. ... .............. ............ ........... 4


...... ..... .. 4
.. ......... 4
. .. ........ 7
8
11.


13
......... 13
........ ... 14


Introduction











How Does Market Manipulation Work? ... ............................. ... 37
When Will Market Manipulation Work Better? .......... ........... 38
What Are the Empirical Effects of Market Manipulation ........... 41
M methodology .. .... .. ... ......... ........ .................. 42
Tests of Price Reversal ......... .............. ............ .... 43
Cross-Sectional Analysis of Price Effect and Share
Turnover as a Function of Firm Size and
Stock V olatility .. ... ........ ...... ........... ........ 43
Combined Cross-Sectional and Price Reversal Analysis ...... 43
Abnormal Returns Estimation ............... ........................ .... 44
Evidence of Stock Price Manipulation ...................................... 45
An Alternative Explanation .................... ........................... ... 55

6 THE SELECTION BIAS HYPOTHESIS .................. ..... ... 75

Introduction ... ... .. .. .. ... ...... ... ......... ............. 75
The "Price Threshold" Hypothesis .......................................... 76
The "Over-Performance" Hypothesis ............................ ........... 78
Stock Price M manipulation Revisited ........................................ 82
The Case Against the Selection Bias Hypotheses ..................... 83

7 THE INFORMATIONAL EFFECT OF OPTION LISTINGS ... 95

Introduction ....................................... ........... ............ ..... ..... 95
Theoretical Foundations .. ..... ....... ............ 95
M ethodology ...... ........... .. .. ..... ......... .. ................ ......... 99
Identifying the Appropriate Samples .............................. 99
Testing the Information Hypothesis .................................. 101
D escriptive Statistics ....... ......... ... ... ................... ......... ........ 102
Results of the Information Hypothesis Tests ............................... 105
The Case Against the Information Hypothesis ........................ 109

8 SUMMARY AND CONCLUSION. ..................................... 118

APPENDIX

MATHEMATICAL RELATION BETWEEN RETURNS STANDARD
DEVIATION AND INCENTIVES TO MANIPULATE
ST O C K PR IC E S ......... .. .................... ....... ...... .. ..... ... ...... 123

R EFE REN CE S ... .. .. .............. ... ...... ... ....... .. .... .................. 128

BIOGRAPHICAL SKETCH .... ... ......... ......................... ...... ........... 131













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE PRICE EFFECT OF OPTION INTRODUCTIONS:
1973-1992

By

Sorin Sorescu

August 1996


Chairman: Mark J. Flannery
Major Department: Finance, Insurance and Real Estate

I examine the effect of option listings on the price of underlying stocks for the

period 1973-1992 In accordance with previous studies, I find that options increase stock

prices during the 1973-1980 period. While some authors attribute this price effect to

market completion, I show that it is more likely caused by option dealers manipulating

underlying stock prices during that period.

Federal regulations introduced at the end of 1980 have eliminated many of the

opportunities for manipulating stock prices during the 1981-1992 period Accordingly, no

evidence of price manipulation is found for that period. Surprisingly however, after

manipulation ceases, the price effect of option introductions does not merely vanish, but

becomes negative.

Two explanations are proposed for the occurrence of this negative effect after

1980: The first is the selection bias hypothesis, according to which exchanges would only











introduce options on stocks that have over-performed the market in the near past. If this

hypothesis holds true, the negative price effect would therefore simply represent the return

to a "more normal" stock performance after the option becomes listed.

The second is based on the premise that option listings reveal new information,

which causes stock prices to adjust accordingly.

The data do not support either of these hypotheses, suggesting that the negative

price effect of option introductions is consistent with equilibrium changes in stock prices.

Understanding the cause of these equilibrium price changes remains an important issue for

further research.













CHAPTER 1
INTRODUCTION


An increasingly important issue in the study of financial markets concerns the

extent to which options interact with their underlying stocks. Although Black and Scholes

(1973) view options as redundant securities, a growing number of empirical studies have

established that stock prices are generally affected when traded options are first listed.

Conrad (1989) shows that between 1974 and 1980, underlying stocks experienced positive

abnormal returns around the date of their option listing, and concludes that "options are

not entirely redundant" (p. 488)'. Skinner (1989) reinforces Conrad's conclusion by

documenting a decline in stock volatility after listing.2

Several academic authors indicate that options' positive price impact and negative

volatility effect enhance shareholder wealth and contribute to economic growth.3 By

contrast, the popular press has long suggested that options destabilize markets by

encouraging speculative behavior. This concern was also shared by the Securities and

Exchange Commission (SEC) in 1977, when it suspended the introduction of new stock




' This evidence is also supported by Detemple and Jorion (1990), Kim and Young (1991),
Branch and Finnerty (1981). An interesting (and puzzling) finding in both Conrad and
Detemple and Jorion is that stock prices adjust only on the date of option introduction, not
on the date of the listing announcement.
2 See also Damodaran and Lim (1991), Bansal, Pruitt and Wei (1989), Conrad (1989).
SSee, among others, Allen and Gale (1994), Branch and Finnerty (1981), Klemkosky and
Maness (1980), Bansal, Pruitt and Wei (1989).








options and mandated a general study (the Options Study) to determine "whether

standardized options trading represents a threat to the integrity of the capital raising

function of the securities markets"4 (SEC Release No. 30-14056).

Concerns about the negative effects of options surfaced in the legal arena in 1983,

when Golden Nugget's stock price dropped 17.2% around the initial listing of its traded

option.5 Golden Nugget's management promptly sued the American Stock Exchange (on

which the option had been listed), contending that options are "competing investment

vehicles, that draw potential capital away from the company's issues of common stock."6

More recently, Netscape stock price dropped from $75 to $62 in the 10-day period

surrounding the listing of its first option contract.

While the collapse in Netscape and Golden Nugget stock prices appear to contrast

with the findings of Conrad, it is important to note that both cases occurred almost three

years after the last year in Conrad's study period. It is therefore interesting to determine if

these cases are a simple exception to Conrad's "rule," or whether they have become the

"new rule" at some point after the end of Conrad's study period. In other words, how have

stocks generally reacted to their option introduction after 1980?


4Unfortunately, the question could not be answered, due to limited data availability at the
time. (Exchange traded options had only existed in the US since 1973.) Instead, the study
recommended a number of policy changes, aimed at preventing reported instances of stock
price manipulation and unethical sale practices
SGolden Nugget's call options were first listed on the American Stock Exchange on
August 17, 1983. The price of the underlying stock dropped from $16.00 to $13.25 in the
ten-day window surrounding the listing event
6 Quoted in Regulatory and Legal Developments, DER No. 215, P. A-4, November 4,
1983, emphasis added. The law suit was dismissed in 1986 on grounds that the company's
stock belongs to its shareholders, and that the management is not entitled to sue on their
behalf.








In Chapters 3, 4 and 5 of this dissertation I show that the positive price effect

documented by Conrad (1989) is confined to the period prior to 1980, and is likely to

result from option dealers manipulating underlying stock prices. After late 1980, when

federal regulators limited the extent to which option dealers could trade in underlying

stocks, the positive price effect becomes negative, and no further evidence of market

manipulation is detected.

Chapter 6 investigates whether the post-1980, negative price effect of option

introductions is caused by a bias in the exchange selection process, and presents evidence

refuting this hypothesis.

Chapter 7 explores whether the price effect is a reflection of the private

information available to informed traders at the time of listing, which would be revealed to

the market as a consequence of these traders' migration from the primary to the derivative

security. The chapter, however, presents evidence which is strongly inconsistent with this

information hypothesis

The overall results from Chapters 3 through 7 leave wide open the possibility that

the observed negative price effect could in fact reflect a reduction in the demand for

underlying stocks, produced by the listing of options This hypothesis remains an

important topic for further research.

The dissertation concludes with a summary and a brief description of this study's

implications for regulators and researchers.













CHAPTER 2
REVIEW OF LITERATURE


Introduction

Empirical studies of the interaction between options and stocks are primarily

concerned with documenting the effect of option listings on the price, volatility and

bid/ask spread of underlying stocks. By contrast, theoretical studies address either (1) the

ability of options to "complete" the markets, or (2) the effect of options upon the stock

market information structure. With respect to the integration of theory and empirics into

one, unified body of literature, Damodaran and Subrahmanyam (1992) write.

Although the empirical hypotheses tested could be tied in with the
theoretical developments, they appear to have evolved in separate
directions. Hence, the empirical work has not been tied in with particular
model structures, but devoted to testing more general hypotheses. (p. 17,
emphasis added)

This chapter reviews a number of empirical and theoretical papers which are

relevant to the present study, and points to specific instances where the literature would

benefit from a better integration of theory and empirics.


Empirical Papers.

Branch and Finnerty (1981) provide the first study of the impact of option listings

upon stock prices. They evaluate excess returns for a 12-week period around the

announcement date, using a market model with betas obtained from Value Line.' They


STheir sample period extends from 1973 to 1977.








find positive excess returns in the second week following the announcement date, and

conclude that options increase demand for underlying stocks. The authors interpret this

post-announcement effect as 'a thing of value' to the affected company" (p. 12).2

Conrad (1989) examines the effect of option listing on stock prices and volatilities,

using all options introduced between 1974 and the end of 1980. She excludes the first

options listed in 1973 in order to eliminate learning effects associated with these securities.

She examines abnormal returns around both listing and announcement dates She finds

significant positive abnormal returns around listing dates, and no abnormal returns around

announcement dates.3 She concludes that (1) "options are not completely redundant

securities" (page 488), (2) the observed price effect is due to an increase in the demand

for stock caused by option dealers, and (3) the absence of an effect at announcement

date is due to transaction costs: "the price effect may be sufficiently small (2%) that

transaction costs make it unprofitable for any but those traders who anticipate acting as

dealers in the new option and using the security to hedge" (page 488).

Detemple and Jorion (1990) also examine the effect of option introduction on

stock prices for a sample of 300 stock options listed between April 1973 and December

1986. Since many of these options were listed simultaneously, their sample includes only



2 The presence of positive abnormal returns two weeks after the announcement date is
also consistent with abnormal returns occurring around option listing
3 The absence of abnormal returns around the announcement date could be explained by
the fact that the initial listing announcement is in fact only an "option" to list at that date.
Conrad in fact observes that some options do not get listed on the day in which they were
announced, but rather during a window of 125 trading days after the originally-announced
date. Accordingly, to the extent that announcements convey noisy information, one
should not expect to find any reaction in underlying stock prices on that date.








53 event dates. They, like Conrad find significant price increases (2%) around the listing

date, and no price impact at announcement.

While the absence of an announcement price effect seems puzzling in an efficient

market, Detemple and Jorion present two possible explanations for this idiosyncrasy.

First, an announcement is only an option to introduce a contract on a given date. In their

study, approximately one third of all option announcements did not materialize as

scheduled. In some cases, the contracts have been listed more than one year behind

schedule. Second, among the remaining two thirds of the sample for which the

announcement materialized as scheduled, the listing date followed the announcement date

by approximately one to two days, making it difficult to differentiate between listing and

announcement effects.

The main contribution of the Detemple and Jorion study is to examine the effects

of options on stocks other than the underlying one. In particular, they find that upon

option listing, a portfolio of stocks belonging to the same industry as the underlying stock

(but excluding that stock) also experiences an increase in its market value (1.5%). More

surprisingly, they also find that the entire market experiences an increase in its value

around listings of individual stock options (1 1%).

The authors also divide their sample period into three sub-samples, with June 1975

and March 1982 as cutoff dates. They find strong, positive abnormal returns in the first

two sub-periods, and no significant abnormal returns in the last sub-period. The authors

suggest that their results are consistent with options completing the markets. "Contrary to

the premises of classic arbitrage pricing models, option markets have a real effect on

equilibrium prices and allocations. We find evidence of significant price increases in the








optioned stock around listing dates" (p. 800). Moreover, they explain the absence of

positive abnormal returns after March 1982 by the fact that options on S&P 500 futures

were introduced in April 1982

The price effect seems to be localized in the first two sample periods, and is
much weaker in the 1982-1986 sub-period. Since options on stock indices
have been introduced in 1982, these results are consistent with the
numerical analysis in [a previous Detemple and Jorion working paper]: the
price effect of a new option listing is marginal when an option on the
market already exists. (p. 798)

I provide empirical evidence in Chapter 5 which is inconsistent with this

hypothesis.

More recently, Allen and Gale (1994) interpret the findings of Detemple and Jorion

as follows:

On each [one of the 53 listing days], their results indicate that the market
rises by approximately 1%. Based on this finding, one could argue that the
total increase in the value of the market due to innovations in the option
market may be as high as (1.01)" 1 = 69% or roughly two thirds. (p. 34)


Theoretical Papers Based on Market Completion

Ross (1976) shows that the introduction of options in an incomplete market

increases the span of the investor's consumption space: "in an uncertain world, options

written on existing assets can improve efficiency by permitting an expansion of

contingencies that are covered by the market." (p.75). Thus, to the extent that options do

complete the markets, they must have a real allocation impact contrary to the redundancy

assumption embedded in the Black and Scholes (1973) model.

Detemple (1990) demonstrates that in an economy with incomplete markets, the

introduction of a stock option generally changes asset prices and consumption bundles. He








presents one example in which options decrease stock prices, as a result of a general

portfolio re-balancing. Other examples show that stock prices may also rise, depending on

initial endowment bundles or state probabilities. Like Ross (1976), this model can only

predict that if options complete the markets, individual stock prices will be affected either

upwards or downwards. The model does not predict the general direction of such price

changes, or whether the aggregate effect is different from zero.

Detemple and Selden (1991) use a mean-variance economy to examine the impact

of options on underlying stock prices. Assuming agents have different beliefs about the

stock variance, but similar beliefs about its mean returns, they show that options increase

stock prices.

The only empirical implication of theoretical models based on market completion is

that options change stock prices if they complete the markets. The theory does not,

however, provide inferences about the direction or magnitude of these price changes.

While empirical studies confirm that stock prices generally react in response to option

introductions, they cannot be directly tied to the predictions of any given model.


Other Theoretical Papers

Stein (1987) is the first theoretical paper to recognize that options need not

necessarily complete the market in order to affect stock prices.4 That is, even if options

were totally redundant, their introduction may still affect the price of their underlying

stocks. The core logic of Stein's model is as follows: (1) options are only redundant to


4 Stein's main analysis centers around the opening of a futures market, but his model can
be easily extended to the opening of an options market, as indicated by the author himself
"A similar logic would apply to the opening of an options market (p. 131)."








option market makers, who are "endowed" with the Black and Scholes technology,5 (2)

the introduction of options attracts a new class of investors in the market: those who

would have wanted to use options before, but were not endowed with the Black and

Scholes technology,6 (3) in some cases, the arrival of these new investors modifies the

market information structure, leading to appropriate changes in underlying stock prices.

The author summarizes his view on this point as follows:

It is often claimed that "since options are redundant, they do not change
the equilibrium asset prices." This would hold true only if everybody in the
economy could borrow and lend costlessly so that introducing options
changes nobody's opportunity set. But if this were the case, why would
anybody trade in options at all? The Black-Scholes formula still holds
when only some agents can borrow and lend costlessly, although asset
prices cannot strictly be taken as exogenous: opening an option market
indirectly changes asset demand, hence their price process. (p. 1130,
emphasis added)

A second implication of Stein's model is that if the new class of investors consists

mainly of speculators, the introduction of options may be welfare-reducing, leading to a

drop in underlying stock prices.7

Back (1993) also demonstrates that the introduction of redundant options may

affect stock prices, by modifying the market information structure. In his model options

are ex-ante redundant, in the sense that the stock follows exactly the stochastic process

postulated in the Black and Scholes paper, and the option market makers may freely and



That is, only option market makers can execute costless hedges. The other traders
cannot.
6 This class includes, among others, information traders and speculators, with no access to
riskless hedges.
7A speculator in Stein's model is a trader who has a noisy observation of future stock
values.








costlessly trade any quantities of stocks and/or bonds required to form a riskless hedge.

That is, in Back's model options are totally redundant before they start trading.

However, in a market composed of both informed and uniformed traders, the

author shows that the introduction of any new asset with nonlinear payoffs (such as an

option) will change the portfolio allocation of the informed traders. Whereas in the stock-

only economy all trades are restricted to one asset, in a stock-option economy the

informed chooses an optimal trading strategy between stocks and options, to maximize his

total returns on private information. By observing the trades occurring in the options

market, the uninformed simply adjusts, in a Baysian fashion, his prior beliefs about the

future states of nature. Thus, in this economy, every option trade reveals a new (and

different) information structure. In a rational expectation framework, every option trade

should therefore change underlying stock prices, depending upon the type of information

being revealed. If there is sufficient information asymmetry in a given market, options

which are ex-ante redundant become ex-post non-redundant, and no perfect hedge is

available, even to option market makers with zero transaction costs.

Like models based on market completion, this model cannot predict the direction

or magnitude of price changes following an option introduction. The model nevertheless

has a very interesting, indirect empirical implication: if stock price changes around option

listings are caused by a the revelation of new information, they should be correlated with

unexpected earnings occurring immediately after listing.8 Moreover, this effect should be

stronger to the extent that informed traders switch their trades to the options market.


8 Unexpected earnings (or dividends) are defined as the actual earnings announced
immediately after listing, minus the latest market expectation which precedes the listing








Koticha (1993) uses a similar framework to focus on the effect of options on the

bid/ask spread of underlying stocks. In his model the informed makes a tradeoff between

the higher leverage and the lower depth of the option markets when selecting a trading

strategy. His main conclusion is that if option markets are sufficiently deep (i.e., the

number of option noise traders is sufficiently high), the informed traders will move out of

the stock market This then reduces bid-ask spreads in the stock market. In reviewing his

conclusion it is important to emphasize that the drop in the stock bid/ask spread occurs

only if the options market is sufficiently noisy. The model does not permit to determine

when this happens, since the relative number of noise traders in each market--a critical

model variable--is exogenous (in this model) and cannot be measured empirically.


Conclusions of the Literature Review

Theoretical papers dealing with the effect of option introductions are generally

based on two distinct premises. (1) options affect stock prices by completing the markets,

and (2) options affect stock prices by unveiling a new informational structure. Though

based on different economic foundations, both theories share essentially the same broad

empirical implication: stock prices react when new options are introduced.

While empirical studies confirm that stock prices generally react in response to

option introductions, the documented price effects are consistent with both market

completion and information-based models.

This dissertation improves upon the previous literature in three respects. First, it

analyses data from a longer time period, and shows that the price effect of option

introductions becomes negative starting in 1981. Second, it shows that the positive price





12


effect observed prior to 1980 is likely to be caused by stock price manipulation, not by

general price changes resulting from market completion. Third, it shows that the negative

price effect observed after 1980 may not be attributed to either a bias in the exchange

selection process, or to the information hypothesis postulated in Back (1993).













CHAPTER 3
THE PRICE EFFECT OF OPTIONS INTRODUCTIONS
REVISITED 1973-1992


Introduction

This chapter investigates the price effect of options listed during 1973-1992, using

the same methodology as Conrad (1989). The analysis repeats Conrad's study for a longer

time period, and using a larger number of listing events.' Like Conrad, I find significantly

positive price effects for options listed between 1973 and 1980. By contrast, I find

significantly native price effects for options listed between 1980 and 1992.


Data Sources and Methodology

The Chicago Board of Options Exchange provided listing dates for 1455 options

listed on all organized options exchanges in the United States, since 1973. Since the list

was accompanied by a disclaimer about possible errors in the data, each listing has been

checked in the Wall Street Journal Expiration dates for the first option contracts were

also obtained from the Journal at the same time. When the CBOE information was not

corroborated by the Journal, the observation was dropped from the dataset. More

observations were dropped when the CRSP tapes did not contain sufficient data for

underlying stock returns, either in the pre-listing or post-listing window.2 This resulted in a




'Conrad's study period ended at the end of 1980.
2 At least 40 trading days were required in each period to calculate abnormal returns.








1237 "clean" listing events, which have been categorized as either (1) call-only, (2) joint

put/calls, and (3) put-only. Of these, only 179 were call-only events, with options listed on

the CBOE

I investigate the price effect of option listings for 1973-1992, using the same

methodology as Conrad (1989). Like previous authors I form equally weighted portfolios

of stocks with identical option introduction dates, and treat them as single securities.

This eliminates possible biases coming from cross-sectional correlation in excess returns. I

then estimate the market model over the 100 trading days preceding the listing, and use it

to calculate abnormal returns for the 10-day window surrounding the listing date.


Empirical Results

When Conrad's analysis is repeated for all options listed on all major exchanges

between 1973 and 1992, the positive price effect documented prior to 1980 becomes

generally negative after 1980, as illustrated in Figure 1. To understand the reasons for this

shift, I first try to establish whether 1980 marks a significant change in the nature (or

characteristics) of the various option markets in the United States. Table 1 shows several

descriptive statistics for new option contracts listed on U.S. exchanges between 1973 to

1992.4 While the majority of option contracts listed prior to 1980 were calls only, there

were four instances where calls and puts were listed at the same time. Panel A indicates

that call-only listings continued to be most prevalent from 1981 to 1987, with only 5 out


SSee Conrad (1989) and Detemple and Jorion (1990)
4 To preserve comparison with the rest of the dissertation, this table includes only options
which meet the following two criteria: (a) the listing date has been verified in the Wall
Street Journal, and (b) data for the underlying stock is available on the CRSP tapes for at
least 40 trading days on either side of the listing event.








of 325 first-time listings being joint put/calls. After 1987, put/calls became the norm.

Because the change from a call- to a put/call-dominated market occurred in 1987, it does

not appear to be related to the 1980 price-effect shift illustrated in Figure 1.

The middle section of Panel A shows a breakdown of all first option contracts

according to the exchange where they were listed. While the number of option exchanges

increased through time, no significant change in the listing pattern is observed in 1980

The last two columns of Panel A show that prior to 1985 all optioned underlying

stocks were traded on either NYSE or AMEX. Starting in 1985, NASDAQ-traded stocks

also became optioned, in roughly equal proportions to NYSE/AMEX stocks. This change,

however, occurred five years after 1980, and does not appear to be related to the pattern

observed in Figure 1.

Panel B illustrates the change in the size of the "typical" optioned firm, by

comparing mean and median market capitalizations across time. Two measures of market

capitalization are used: the current-dollars measure, and an inflation-adjusted, constant-

dollars measure, using the 1983 CPI as a deflator. It becomes readily apparent that firms

which became optioned during 1973 or 1974 were substantially larger than firms optioned

in subsequent years, with median market capitalization in excess of $7 Billion (constant,

1983 dollars). Starting in 1975, and especially after 1976, the average size of the optioned

firm became much smaller, and remained relatively constant throughout the rest of the

sample period. Since the most significant change occurred in 1976, firm size does not

appear to be related to the Figure 1 pattern.








Table 2 presents a more formal analysis of the results illustrated in Figure 1. This

table shows--for various sub-samples--the mean cumulative abnormal returns (CARs)

experienced by underlying stocks during the ten-day window surrounding the listing date.

Panel A of Table 2 shows results for all stocks which became optioned for the first time,

either with call-only contracts or joint put/call contracts. In agreement with previous

studies, I find that for the first sub-period (1973-1980) CARs were strongly positive

(2.47%) and significantly different from zero at better than the 1% level (t=4.66). When

the same analysis is repeated to the second sub-period (1981-1992), the opposite obtains:

CARs become negative (-1.03%), and significantly different from zero at better than the

2% level (t=-2.49). In an attempt to explain the switch in the sign of the CARs, I repeat

the analysis from Panel A for different sub-samples of my original data.

Panel B shows the results for call-only events, whereas Panel C shows the results

for put/call events. By comparing the two panels, it appears that the post-1980 negative

CARs are especially driven by joint put/call listings, with abnormal returns of -1 34%

(t=-2.55), compared to -0.55% (t=-0.84) for call-only listings. However, this does not

seem to account for the reversal in the sign of the CARs shown in Panel A, which still

persists--in a milder form--in the call-only sub-sample: the mean price effect in the post-

1980 period is significantly lower than in the pre-1980 period (-2.94%, t=-3.46).

Panel D repeats the analysis for put-only listings, but finds no significant results in

either period. It should be noted that put-only listings always followed call-only listings for

the same stock. In other words, each time a put option was listed on a given stock, a call

option for the same stock was already trading.








Panels E and F show that the exchange on which the underlying trades does not

affect the way in which it reacts to its option introduction: while NASDAQ stocks were

not optioned in the first sub-period, when they do become optioned (in the second sub-

period) they react similarly to contemporaneous NYSE/AMEX stocks. Moreover, the

CAR sign reversal continues to persist for NYSE stocks in Panel E.

Panel G uses a sub-sample of all stocks, for which bid/ask quotes are available on

the NASDAQ-NMS tape, to determine whether the negative CARs computed from

transaction data are merely resulting from more trades occurring at the bid quote during

the event window. Even when bid/ask averages are used instead of transaction prices, the

negative CARs persist.

Panels H through L show that the CAR sign reversal documented in panel A is

driven mostly by options listed on CBOE, AMEX and PSE. PHLX option listings cause a

similar, but smaller (and insignificant) effect, whereas options listed on NYSE seem to

have no impact on the price of their underlying stocks

It therefore appears that the shift from a positive to a negative price effect

occurring at the end of 1980 is unrelated to either the type of option contracts being listed,

the relative size of the optioned firm, the exchange on which the stock trades, or the

method used to estimate abnormal returns. The rest of the dissertation argues the

following:

1. The pre-1980 positive price effect is not a permanent price change, but rather a
temporary price run-up, induced by option dealers manipulating stock prices around
the date of option introduction








2. Regulatory changes introduced at the end of 1980 successfully prevented stock price
manipulation from re-occurring However, instead of merely vanishing after 1980, the
price effect of option introductions becomes negative starting with 1981.

3. This negative, post-1980 price effect is not consistent with the selection bias
hypothesis, according to which exchanges only list options on stocks that have over-
performed the market in the past. Nor is it consistent with the information hypothesis,
which states that options reveal the content of privately held information at the time of
listing.

4. The post-1980 negative price effect is however permanent, and persistent. This
suggest that option listings are causing a drop in the equilibrium price of underlying
stocks.

5. A longer term negative price effect is discernible even prior to 1980, after the initial
(manipulation-driven) price run-up disappears. As in the post-1980 period, this longer
term price effect also appears to be permanent.

6 Overall, the dissertation asserts that the true price effect of option introductions is
negative. This price effect is readily observed around listing date in the post-1980
period, when the data are not affected by manipulation activities. Moreover, this price
effect is also discernible in the pre-1980 period, after the manipulation-driven price
run-up disappears at the end of the first contract expiration period.

7. Understanding the cause of the true price effect of option introductions remains an
important issue for further research.


















0-

1-
0
3
-2-

-3-


-4 ........... .. ... .......
1973 1975 1977 1979 1981 1983 1985 1987 1989 1991
Year

|0 Mean U Median














Figure 1:
Stock mean abnormal returns
around the first option listing date:
1973-1992


Ln~rA









Table 1: Descriptive statistics


Panel A: Summary of new stock option contracts: 1973-1992


Number of Options Number of Options Listed Underlying
Year Listed (by Type) (by Exchange) Stock
Calls and Put/Calls Only Exchange
C P/C P ALL CB AM NY PS PH MS NY NSD


26
-- -- 6
6
S 84
S 59
22 41



4 29 73
3 49 61
-- 1 80
28
12
1 -- 50
-- 1 40
1 -- 106
77 4 92
84 -- 88
124 124
131 -- 151
102 1 116


41 -- -- 9 --
15 -- 15 16 8
3 -- 8 6 4



18 -- 12 9
2 -- 3 3
18 -- 19 19
10 -- 5 7
5 -- -- 2
21 8 11 5
12 4 7 7
24 16 26 23
23 13 16 15
27 9 17 17 --
29 18 25 32 --
48 31 49 49
36 8 20 17 --


1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992









Table 1. Continued

Note:

This table only includes options which meet the following two conditions:

The listing date has been verified in the Wall Street Journal, and

-The CRSP tapes contain data for at least 40 trading days on either side of the listing
event.


Abbreviations:

C Call contracts only.

P/C: Joint put/call listings.

P. Put listings only.

ALL. The total of calls, put/calls, and put listings.

CB: Chicago Board of Options Exchange.

AM: American Stock Exchange

NY: New York Stock Exchange

PS: Philadelphia Stock Exchange

MS: Midwest Stock Exchange

- NSD: National Association of Security Dealers Automated Quote (NASDAQ)









Table 1. Continued

Panel B: Market capitalization of newly-optioned stocks: 1973-1992 (Calls and
Put/Calls only)


Year Market Capitalization in Current Market Capitalization in Constant
Dollars (1983) Dollars


Mean Std. Median I.Q. Mean Std. Median I.Q.
Dev. Range Dev. Range


6,549 9,672 3,422
4,361 3,871 3,633
1,587 1,526 1,031
877 868 537
677 695 469



1,063 1,349 535
796 459 649
2,633 15,403 588
653 556 477
2,493 2,462 1,428
1,959 2,345 965
1,298 801 1,198
2,116 2,893 1,437
988 776 819
1,224 3,165 528
2,200 6,990 767
670 1,226 405
422 448 289


5,439
6,725
1,702
754
316



647
724
687
572
3,392
1,744
1,000
1,747
888
751
1,126
435
285


14,701 21,441 7,715
8,403 7,458 7,001
2,989 2,895 1,927
1,542 1,532 956
1,135 1,161 793



1,286 1,634 656
867 497 704
2,763 16,214 616
653 554 479
2,387 2,365 1,384
1,811 2,163 891
1,183 731 1,095
1,864 2,518 1,274
827 648 684
978 2,517 427
1,678 5,362 585
490 895 295
300 318 204


Abbreviation:


- LI. Inter-quartile range


1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992


12,347
12,957
3,234
1,283
532



780
791
722
564
3,211
1,645
903
1,533
730
591
824
319
205


---





23


Table 1. Continued

Note:

In order to be included in these calculations, a firm must meet the following criteria:

- The firm's stock became optioned for the first time during the year shown, with either
call-only contracts, or joint put/call contracts. Put-only contracts are always listed after
calls, and are therefore excluded;

- The listing date of the call contract has been verified in the Wall Street Journal; and

- The CRSP tapes contain data for at least 40 trading days on either side of the listing
event









Table 2: Cumulative Abnormal Returns (CARs) of optioned stocks
during the ten-day window around listing date.


When more than one stock became optioned on a given day, a portfolio of all such stocks
is formed, and subsequently considered a single security. Abnormal returns are calculated
using the Brown and Warner (1985) methodology. The benchmark period extends from
day L-100 to L-6, where L represents the listing date. The event window extends from
L-5 to L+5. This 10-day window was selected to preserve comparison with previous
studies.


Panel A: All stocks, all call and put/call listings.

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 83 2.47% 4.66*** 2.54% 3.72***
1981-1992 427 -1.03% -2.49** -1.05% -2.38**


Panel B: All stocks, call options only

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret
Model

1973-1980 82 2.39% 4.41*** 2.45% 3.58***
1981-1992 170 -0.55% -0.84 -0.51 -0.72


Panel C: All stocks, put/calls only

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic -
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 -- -- -
1981-1992 261 -1 34% -2.55** -1.39% -2.56**









Table 2. Continued

Panel D: All stocks, put options only

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 9 -1.06% -0.66 2.84% 1.25
1981-1992 10 -0.03% -0.01 0.08% 0.03


Panel E: NYSE/AMEX stocks only, calls and put/calls

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 83 2.52% 4.73*** 2.63% 3.82***
1981-1992 275 -0.80% -2.16** -0.71% -1.64*


Panel F: NASDAQ stocks only, calls and put/calls

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980
1981-1992 208 -1.61% -2.16** -1.59% -1.99**


Panel G: NASDAQ-NMS stocks only, calls and put/calls, returns calculated using bid/ask
averages

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 -- --
1981-1992 203 -1.62% -2.26** -1.62% -2.06**









Table 2. Continued

Panel H: CBOE options only, calls and put/calls

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic -
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 23 3.85% 4.17*** 3.51% 2.98***
1981-1992 151 -1.38% -2.12** -1.36% -2.01**


Panel I: AMEX options only, calls and put/calls

Period Number of CAR(s) -- T-statistic CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 23 3.04% 2.61** 4.46% 3.06***
1981-1992 159 -1.79% -2.33** -1.79% -2.13**


Panel J: PSE options only, calls and put/calls

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret.
Model

1973-1980 19 1.95% 1.49 2.89% 1.74*
1981-1992 112 -1.92% -2.05** -1.97% -1.97*


Panel K: PLHX options only, calls and put/calls

Period Number of CAR(s) -- T-statistic-- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret
Model

1973-1980 25 1.05% 0.98 -0.54% -0.42
1981-1992 104 -0.31% -0.47 -0.71% -0.99








Table 2. Continued

Panel L: NYSE options only, calls and put/calls

Period Number of CAR(s) -- T-statistic -- CAR(s) -- T-statistic --
Events Market Market Constant Constant
Model Model Ret. Model Ret
Model

1973-1980 -- -- -- --
1985-1992 44 0.85% 0.75 0.99% 0.79


Significance Levels:

** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level
Insufficient data

Variable Definition.

-Number of Events: Number of days during which one or more option contracts get
listed.

-CARs (Market Model): Cumulative abnormal returns, from day L-5 to L+5,
calculated using the market model estimated from L-100 to L-6.

-CARs (Constant Ret. Model): Cumulative abnormal returns, from day L-5 to L+5,
based on the mean portfolio returns estimated from L-100 to L-6.













CHAPTER 4
THE REGULATORY ENVIRONMENT OF
STANDARDIZED OPTION TRADING: 1973-1992


Introduction

Stock options have been traded over the counter in the United States since the

early Twentieth Century. Rule 9b-1 of the Security and Exchange Act of 1934 prohibited

any transaction in puts, calls, straddles or other options, except in accordance with a plan

established by the exchange, that the SEC had determined to be "in the public interest and

for the protection of investors." In the early part of 1973 the Chicago Board Options

Exchange (CBOE) submitted under rule 9b-1, a plan for the initiation of standardized

options trading. The SEC was particularly concerned that standardized option trading

"may involve complex problems and special risks to investors and to the integrity of the

market place" (SEC Release No. 34-10552). Accordingly, it decided to approve the

CBOE plan, on an experimental basis, and only on condition that "the economic benefits

of options ownership outweigh the dangers of options trading." That is, the SEC

reserved the right to terminate the plan, if subsequent evidence indicated that options were

harmful. Throughout the mid 1970s, the SEC also authorized the American, Philadelphia,

Pacific and Midwest Stock Exchanges to implement similar plans, on similar experimental

conditions.




SSecurities Exchange Act Release No. 34-10552









The Options Study

The early years of standardized option trading were marked, on the one hand, by a

sharp rise in trading volume, and on the other, by numerous instances of unethical sales

practices, fraud, and stock price manipulation.

In addition, as option volume grew by a factor of 30 in less than three years, the

SEC became particularly worried that option trading might draw investor capital away

from equity securities: "[The Chairman of the SEC] is concerned that options trading is

drawing off investors' funds that otherwise might be going to smaller companies in need of

speculative capital." (WSJ, 8/3/76.) In July of 1977, as the overall effect of option trading

became questionable, the SEC felt it appropriate to review the experimental standardized

option trading programs. Accordingly, it ordered a comprehensive study on this subject

(the Options Study), and imposed a moratorium on any new option listings. (That is, the

SEC asked all option exchanges to refrain from initiating trading in new options until

further notice.)

At its inception, the Options Study had two main objectives. First, it was to

investigate alleged instances of fraud, price manipulation, unethical sale practices and

inadequate surveillance systems in all options exchanges. If sufficient evidence were found

in support of these allegations, the Study was to recommend specific regulatory changes

and disciplinary sanctions, in order to prevent re-occurrences.






2 See, among others, "J. Newmann cited for Manipulation of Stock Prices," in The Wall
Street Journal, Jan 26, 1978.








Second, the Study was to assess the impact of options upon the integrity of the

stock market and its capital raising function, in order to determine whether options are

"consistent with fair and orderly markets, the public interest [and] the protection of

investors" or whether they "represent a threat to the integrity of the capital raising

function of the securities markets." (Securities Exchange Act Release No. 34-14056

Emphasis added.) It quickly became clear that this second objective could not be

accomplished with only four years of trading data.

In 1978, the SEC directed the Study exclusively towards its first objective, leaving

the broader question of economic impact to others. Upon its completion in 1979, the

Study made a number of recommendations, which--for the most part--fell under two broad

categories: increase in the quality of information available to investors and prevention of

stock price manipulation.

Regulatory Changes Related to the Quality of Information Available to Investors

In order to protect inexperienced investors from the dangers of options trading: (i)

option brokers had to disclose complete information to all prospective traders, including

the average performance of all accounts, the realized rate of return net of transaction costs

and a disclaimer that past performance is not an indication of future performance; (ii)

during option seminars conducted by brokerage firms the financial interest of these firms

had to be clearly disclosed; (iii) option brokers had to complete an SEC-approved training

course; and (iv) option brokers had to ensure that potential customers have a thorough

understanding of options, before they could recommend options as an investment

instrument. The majority of these recommendations were successfully implemented








between February 1979 and March 1980. Later that month, the SEC lifted its three-year

long moratorium and allowed exchanges to once again initiate trading in new option

contracts.

Regulatory Changes Related to the Prevention of Stock Price Manipulation

The Options Study observed that between 1973 and 1980, the Federal Reserve

Board in essence permitted option dealers to purchase unlimited quantities of underlying

stock without providing any margin deposit. This practice was made possible by

inadequate regulations, and encouraged option dealers to manipulate underlying stock

prices in order to increase their gains from call writing. At the end of 1980 the Federal

Reserve severely restricted the availability of zero-margin credit to option dealers,

discouraging stock price manipulation from re-occurring. In the following paragraphs I

briefly describe the regulatory environment of option dealer margins between 1973 and

1980, and show that the two regulatory regimes prior to 1980 permitted stock price

manipulation. I then examine the regulatory changes implemented at the end of 1980, and

explain how the post-1980 regime substantially eliminated opportunities for manipulating

underlying stock prices.

The "Good Faith Credit" Regime: 1973-1977 When standardized option trading

begun, option dealers did not have to provide any margin deposits for the purchase of

underlying stocks, so long as the acquired stocks were used for hedging their option

positions.3 However, the Federal Reserve did not specify the precise amount of stock that

qualified for credit under this rule, leaving it to individual market makers to determine "in


* Other option dealer transactions were subject to the usual 50% margin requirements.








good faith" the appropriate quantity required for hedging. By 1977 the Federal Reserve

observed that this leniency gave way to regular abuses as many market makers acquired

substantially larger quantities of stock than those economically required for hedging, and

misrepresented these transactions as being hedge-related. In April of 1977 the Federal

Reserve issued a proposed rule change, according to which stock credit for option dealers

would be made available only under the following terms:

(i) For options in- or at-the-money, one share of stock per option
outstanding may be purchased with a reduced, 25% margin requirement
instead of the usual 50%. Zero-margin credit was no longer available.

(ii) In all other cases, underlying stock transactions were subject to
the regular 50% margin requirements.

Although the Federal Reserve never adopted this proposed rule, on June 20, 1977

the Securities and Exchange Commission asked the Option Clearing Corporation to

finance option dealer accounts in accordance with the April 1977 rule proposal as if it

were in effect. (Options Study, p. 679.) That is, starting in June of 1977, the "good faith,"

zero-margin requirement has been replaced with a 25% margin requirement for stocks

used to offset in- or at-the-money calls, and a 50% margin for all other stocks.

The "Free Riding" Regime. 1977-1980. The Options Study, however, observed

that the regulatory change implemented in June of 1977 was inadequate, and did not

prevent option dealers from trading excessive stock quantities without margin This is

because after the end of the "good faith credit" regime, option dealers quickly devised a

scheme (later called "free riding") which in essence allowed them to circumvent the newly

imposed margin requirements, and continue to control large quantities of stocks with no

marin deposits. Taking advantage of a regulatory inefficiency not corrected by the June








1977 rule change, the scheme consisted of acquiring the underlying stock position, selling

it back at the end of the five-day grace period, and buying it back immediately thereafter.

When this procedure was repeated every five days for an extended period, the option

dealer was in effect "acquiring" the underlying stock without having to maintain a margin

deposit. The Federal Reserve Board did not prohibit "free-riding" at the time, but stock

exchanges had adopted rules which prohibited free riding, except fo option dealers' stock

transactions which were executed as a "good faith" hedge for an options position:4

The practice of acquiring a stock position and liquidating it within five
business days without making a required margin deposit is called "free-
riding." The Federal Reserve Board does not prohibit free-riding but all
self-regulatory organizations have adopted rules which prohibit their
broker-dealer member firms from permitting a public customer to engage in
free riding The self regulatory organization regulations, however, have not
been applied to market maker stock transactions, For that reason, an
options market maker has five business days within which to liquidate a
stock position without making a margin deposit when the stock was
originally acquired as a bona fide hedge of an options position.

Some option market makers have made a practice of selling their stock
within this five day period and then immediately repurchasing the stock to
avoid the necessity of putting up a margin deposit. This practice permits
the options market maker to speculate in the stock underlying an option
without being required to maintain a margin deposit. (...) The Options
Study does not believe that this type of activity contributes to an orderly
market or to the financial integrity of the options market. (Options Study,
p. 681, emphasis added)



SThus, option dealers were given preferential treatment over all other stock investors. To
better appreciate this difference, consider the following example: If on day 0 an option
dealer purchases 100 shares of stock at $1 per share, he owes his creditor the amount of
$100 on day 5. If on day 4 he sells his position, he no longer has to pay the $100 on day 5.
At the end of day 4 the option dealer can repeat the procedure for another five days, and
can continue to do so every five days for an unlimited period. This in essence allows him
to control the stock with zero margin. The difference between an option dealer and an
"ordinary" investor is that the ordinary investor who sells the stock on day 4 must still
provide the margin deposit on day 5, and collect the proceeds on day 9.








The "Permitted Credit" Regime: Post-1980. Whether de jure (1973-1977) or de

facto (1977-1980), the availability of unlimited, zero-margin credit made it possible for

option dealers to control large quantities of underlying stocks with no up front investment.

Both the SEC and the Federal Reserve believed that a change in margin regulations was

necessary to discourage manipulation in underlying stock prices. On August 11, 1980, the

Federal Reserve officially prohibited "free riding," except for the quantities of stock

legitimately required for hedging, which it referred to as "permitted offset positions."

These quantities were defined as one share of stock for every call option which is in- or at-

the money, and zero shares for options which are out-of-the-money. When a stock

purchase did not qualify as a "permitted offset position," the option dealer was compelled

to provide the margin deposit by the end of the fifth business day after the purchase, even

if he had sold the stock in the mean time. Failure to do so will preclude him from obtaining

any credit for stock purchases during the following 15 business days.5 This new regime in

essence created two separate classes of underlying stock transactions, subject to separate

margin requirements and differential "free riding" treatments

(i) Permitted offset stock positions, consisting of one share of
stock for each outstanding call option in- or at-the-money. This class was
subject to a 25% margin requirement. However, "free riding" continued to
be permissible, which resulted in an effective zero margin requirements for
stocks belonging to this class

(ii) "Ordinary" stock positions, consisting of all other stock
transactions which did not qualify as "permitted offset positions." This
second class was subject to a 50% margin requirement, and "free riding"
was no longer allowed. If convicted of "free riding," the option dealer



SFederal Register, 6/17/1980, pp. 40967-8. The CBOE has characterized this 15-day
penalty as "unduly harsh".








would no longer be able to obtain stock credit for a period of 15 days, for
both "ordinary" and "permitted offset" stock purchases.

Individual exchanges were allowed to make minor modifications to the new policy,

so long as they did not contradict the essence of the Fed's ruling. The Philadelphia Stock

Exchange decided to follow the ruling without further revisions. The other three

exchanges proposed minor modifications, which were subsequently accepted by both the

SEC and the Federal Reserve. The effective enforcement date of the new policy therefore

varied according to the following schedule

Philadelphia Exchange: August 11, 1980
Chicago Board Options Exchange: September 4, 1980
Pacific Stock Exchange: December 26, 1980
American Stock Exchange March 16, 1981.

The Options Study thus observed that between 1973 and 1980 the availability of

"good faith credit," followed by the practice of "free riding" permitted option dealers to

control large quantities of stock without providing any margin deposit. This made it

possible for them to manipulate underlying stock prices in order to increase profits from

call writing. 1 test this assertion in Chapter 5, and show that the positive price effect

observed prior to 1980 is consistent with stock price manipulation: option dealers

acquiring excessive stock quantities around option introductions, in order to sell

"overpriced" call options to the inexperienced investor. When excessive "free riding" was

no longer permissible at the end of 1980, the abilities of option dealers to purchase

underlying stocks with zero margin deposit were limited to at most one share of stock per

outstanding call option. Since substantially larger stock quantities are required to

influence prices in the desired direction, the profitability of stock price manipulation

considerably diminished after the end of 1980.









Multiple Listings of Options

In a world with perfect competition among option dealers, manipulation of

underlying stock prices requires a collusion which may not always be sustainable in

equilibrium, since at any time an individual dealer may have the incentive to deviate from

the collusion, for example by short-selling the stock which is being manipulated. Thus, if

multiple listings of options were allowed prior to 1980, the market manipulation argument

would be substantially weakened.

A brief examination of the regulatory environment reveals that between 1973 and

1980 only 22 call contracts were traded on more than one exchange. None of these

contracts, however, has been listed simultaneously on more than one exchange: the listing

on the second (and third) exchanges always lagged by at least three months the listing on

the first exchange (usually the CBOE). Rule changes implemented in 1980 completely

prohibited multiple listings until 1989, when the SEC decided to allow unrestricted

multiple listings. The regulatory regime pertaining to multiple listings may therefore be

summarized as follows:

1973-1980: Sequential multiple listing only occurred for 22 stocks. First time
option contracts were always listed on only one exchange

1980-1989: No multiple listings allowed. Exchanges had to select optionable
stocks according to an agreed-upon rotation system.

1989-1992- Simultaneous and sequential multiple listing allowed without
restrictions.

Since no simultaneous multiple listings occurred between 1973 and 1980, market

manipulation could have persisted as a sustainable equilibrium during that period.













CHAPTER 5
EVIDENCE OF STOCK PRICE MANIPULATION
AROUND OPTION INTRODUCTIONS: 1973-1980


Introduction

In this chapter I investigate whether the pre-1980 positive price effect is consistent

with option dealers manipulating stock prices. As a first step, I show that under the pre-

1980 margin regulations, option dealers had both the incentives and opportunities to

manipulate underlying stock prices. Subsequently, I identify the empirical effects of market

manipulation, and show that they are supported by the data.


How Does Market Manipulation Work?

Market manipulation is the process by which option dealers attempt to control

stock prices, to benefit from their call option positions. In particular, option dealers would

like to artificially inflate stock prices just before the first option contract begins trading, in

order to sell over-priced calls. As this first contract approaches expiration, option dealers

would like to depress stock prices, to reduce the call's terminal value (which corresponds

to the option dealer's liability). The profit derived from market manipulation therefore

corresponds to the difference between the "artificially inflated" call price, and its "true"

price that would have obtained in a non-manipulated economy.

To the extent that stocks have a down-sloping demand (Shleifer, 1986), stock

prices can be manipulated as follows: First, option dealers acquire a large inventory of








underlying stocks before the option contract begins trading. (This pushes call prices above

their "true" values.) Since option dealers must also purchase underlying stocks for

legitimate hedging activities, this acquired inventory represents shares purchased in excess

of those required for hedging. As the option contract approaches expiration, option

dealers liquidate their excessive stock inventory, placing downward pressure on underlying

prices. This lowers both stock and call prices towards their "true" levels.' Clearly, the

availability of free credit for underlying stock purchases prior to 1980 provides an unique

opportunity for option dealers to manipulate stock prices during that period.


When Will Market Manipulation Work Better?

Since market manipulation involves a round trip stock transaction, it will only be

profitable when the gains realized from call writing exceed the losses incurred in stock

trading. The option market maker's expected gains resulting from manipulation can be

calculated as follows:

E, Profit}= (C, P m-N-r Ak price, + N[E, Bidprice,}-Askprice,] (1)

where:

C is the initial number of call contracts written;
P, is the effective sale price of each call contract, based upon the
artificially inflated stock price in effect at time t;
P, is the "true," arbitrage-based price of a call contract, which would
have obtained in the absence of stock price manipulation;
N is the number of underlying shares transacted, in excess of the number
otherwise required for hedging2;


In some cases option market makers have allegedly caused options to expire worthless.
2 As discussed in Chapter 4, up to one share of stock per outstanding call option can be
purchased without margin deposit throughout the entire sample period, from 1973 to
1992. To the extent that manipulating stock prices requires the purchase of substantially









m is the effective margin requirement applicable to the stock purchase at
time t;
r is the opportunity rate of return on an alternative investment,
Askprice is the mean purchase price of the stock position at time t;
Bidprice is the mean sale price of the stock position at time t 1,
t is the time of the stock purchase and initial call writing; and
t+I is the time when the stock is sold and the call contract expires.

The first term of equation (1) represents the profits realized from selling

overpriced calls. It is equal to the difference between the net proceeds from call writing

under stock price manipulation and the net proceeds from call writing when no attempts

are made to influence stock prices.

The second term of equation (1) represents the opportunity cost of investing

$(moN.AskPrice,) at time t into the manipulation "technology": when option dealers are

required to maintain a positive margin deposit m, they are foregoing alternative investment

opportunities for the time period ranging from t to t1 1. The analysis presented in Chapter

4 shows that the value ofm through time varies as follows:

from 1973 to 1977, m=0, de jure
from 1977 to 1980, m=0, de facto;
starting in 1981 (both deure and de facto:
for options out-of-the money, m-0.5, for any quantity traded;
for options in- or at-the money, m-O.5, for all shares traded in
excess of one per outstanding option contract

The third term in equation (1) represents the actual loss resulting from the round-

trip stock transaction. It is equal to the net loss incurred for each stock share transacted

for manipulative purposes, times the total number of shares transacted.





more shares, the margin ratio of interest in this chapter is the one that applies to trades in
underlying stock in excess of those quantities which are exempt from margin deposits.









Equation (1) therefore predicts that market manipulation is likely to be more

effective under the following circumstances:

First, manipulative costs are substantially reduced when m=O, as the second term in

equation (1) becomes zero. Accordingly, market manipulation should be more apparent

between 1973 and 1980, and less apparent starting in 1981, when the value of m jumps

from 0% to 50%.

Second, the incentives to manipulate are substantially diminished in the post-1980

regime, where option dealers are subject to differential margin requirements, depending on

whether the call option is in- or out-of-the-money. In the pre-1980 regime option dealers

find it lucrative to write calls when stock prices are artificially inflated, because most calls

would finish out-of-the-money once stock prices revert to their normal level. With the

post-1980 differential margin requirements, forcing options out-of-the-money is no longer

profitable, since the margin ratio increases from 0 to 0.5, making the hedging operations

more costly. I therefore expect, once again, to find more evidence of market manipulation

after 1980, when differential margin requirements are in effect.

Third, assuming that the net proceeds from the stock transactions are negative, the

third term in equation (1) shows that manipulation is more profitable when N is small, i.e.

when only small stock quantities are required to move stock prices in the desired direction.

To the extent that market capitalization is inversely related to firm size (Shleifer, 1986), I

expect market manipulation to be more apparent for firms with small market

capitalization.








Fourth, if option dealers make money by forcing calls to expire worthless, they

should be manipulating mostly stocks for which the term [P*c Pc] is larger. An

examination of a simple two-state, two-period option pricing model reveals that the term

[P', P,] is greater for stocks with large volatilities.' Accordingly, I expect market

manipulation to be especially discernible for more volatile stocks.4

Lastly, market manipulation is not likely to be profitable when both calls and puts

are introduced simultaneously, since by artificially inflating call prices, the market maker

necessarily deflates put prices. Accordingly, this chapter considers call-only introductions

throughout the entire sample period.


What Are the Empirical Effects of Market Manipulation?

The first empirical effect consistent with market manipulation is that if the "good

faith credit" and "free riding" regimes encouraged manipulative activities, their

replacement with the "permitted credit" regime at the end of 1980 should have prevented

market manipulation from reoccurring This suggests that option listings should cause a

positive price effect before 1980, and no price effect afterwards.

In addition, if stock prices are manipulated, they should exhibit only temporary

positive cumulative abnormal returns (CARs) around the date of their option listing. That

is, if stock prices are artificially inflated during the listing window (in order to increase

profits from option writing), they should subsequently drop to their normal "equilibrium"


The proof is presented in the appendix.
4 One can argue that the more volatile stocks also present the greatest risk for the
manipulator. However, in a risk neutral environment, the only relevant losses resulting
from the stock transaction are those shown in equation (I), and do not directly depend on
stock volatilities.








levels. However, as discussed in sub-section B above, some stocks are more likely to be

manipulated than others. Thus, a second empirical implication is that small stocks with

large historical stock return standard deviations should exhibit a stronger price effect

during listing, and a stronger price reversal after listing. By contrast, large stocks with

low historical stock return standard deviations should exhibit little listing effect or post-

listing reversal.

Lastly, when option dealers attempt to manipulate stock prices, they must acquire

an abnormally large number of outstanding shares The third empirical implication is

therefore that higher share turnover should obtain when stock prices are manipulated.


Methodology.

To detect market manipulation I first form a sample composed of all 179 CBOE

call options listed between 1973 and 1992. I then divide it into two sub-samples,

according to the regulatory environment in effect at call listing date: (a) pre-moratorium

listings (1973-1980), and (b) post-moratorium listings (1981-1992). For each sub-sample

I use the listing price effect and abnormal share turnover as proxies for the occurrence of

stock price manipulation.5 Lastly, I perform the market manipulation tests, which

generally fall into three broad categories






5 While high abnormal turnover is definitely consistent with the occurrence of stock price
manipulation, it is also possible that high turnover may reflect the endogeneity of the
listing decision (i.e. exchanges listing options when stock turnover is high). Abnormal
turnover is therefore used only in conjunction with abnormal returns in all market
manipulation tests presented in this chapter









Tests of Price Reversal

These tests determine whether the positive price effect of option introductions is

followed by a corresponding negative effect, prior to the expiration date of the first option

contract Two types of price reversal tests are performed:

i) An "overall" price reversal test, verifying whether the average listing is followed
by an average negative price effect in the post-listing period. (The results of this
test are shown in Table 3 )

ii) Cross-sectional tests of price reversal, verifying if stocks with the strongest
positive listing price effect also experience the strongest post-listing price reversal.
(Results for these tests are shown in Tables 4 and 8.)


Cross-Sectional Analysis of Price Effect and Share Turnover as a Function of Firm Size
and Stock Volatility

Cross-sectional analysis is used to test the following joint hypothesis: (a) the pre-

1980 positive listing price effect is due to stock price manipulation, and (b) stock price

manipulation is more likely to occur for small stocks with large volatilities. Two types of

cross-sectional tests are performed:

i) Regression analysis, where either the listing price effect, or the listing abnormal
share turnover are regressed on firm size and stock volatility (Table 5).

ii) Quartile analysis, where the sample data is divided into quartiles, according to
firm size and stock volatility. For each quartile, the average price effect and
abnormal share turnover are computed, in order to test the hypothesis that market
manipulation (as detected by price effect and turnover) is more likely to occur for
small firms with large volatilities. (The results of the quartile analysis are presented
in Table 7.)


Combined Cross-Sectional and Price Reversal Analysis

These tests allow for the interaction of the price reversal and cross-sectional tests

described above. Two main hypotheses are tested:









i) Price reversal is more likely to occur for small stocks with large volatilities
(Table 8); and

ii) The cross-sectional analysis described in the previous paragraph is more
significant for stocks which experience price reversal (Table 6).


Abnormal Returns Estimation

To estimate the price effect of option introduction, I use both Cumulative

Abnormal Returns (CAR) and Cumulative Raw Returns (CRR), calculated over both

listing and post-listing windows. The following variables are used throughout Tables

3 to 8:

CAR: The cumulative abnormal returns computed using the Brown and Warner
(1985) market model, with parameters estimated over the 95 trading days
preceding the beginning of the listing window. CARs are calculated for
both the listing window, and the post-listing window prior to the expiration
date of the first option contract

CRR: The cumulative raw returns calculated for the listing and post-listing
windows.

Listing window: The 10-day period extending from L-5 to L+5, where L is the
option listing date.

Post-listing window: Alternatively, the period extending from day L+6 to day E,
and the period extending from day E-40 to day E, where E is the expiration
date of the first option contract.

Firm size The market capitalization of a firm's outstanding equity, adjusted for
inflation using the 1983 price level. Market capitalization is computed as
the product of share price and outstanding shares, averaged over the event
window.

Stock volatility: The standard deviation of raw returns, computed over the period
ranging from L-100 to L-6

Abnormal share turnover: The difference between the mean share turnover
computed over the [L-5 to L+5] period, and the mean share turnover
computed over the [L-100 to L-6] period. Share turnover is estimated as








the mean dollar volume of trade divided by the stock's market
capitalization (in $).


Evidence of Stock Price Manipulation.

Figures 2, 3 and 4 present preliminary, prima facie evidence of stock price

manipulation, by showing that the positive post-1980 price effect does not persist over the

50 trading days following listing dates. In Figure 2, cumulative abnormal returns are

plotted for the 200 trading days around the listing event (100 on each side), using all

stocks which became optioned between 1973 and 1980. While impressive CARs are

evident in the narrow event window, they quickly vanish over the following 50 trading

days, the average lifetime of the first option contract (during that time period). A formal

statistical test confirms these conclusions: whereas the cumulative price event at the end of

the 10-day listing window is strongly positive (CAR=2.47%, t-4.66), the price effect

drops to virtually zero by the 50'h trading day after listing (CAR=0.02%, 1-O. 01).6

Figure 3 depicts cumulative raw returns for the same period. Again, the price

pattern is consistent with an impressive price run-up during the event window, followed by

a period of stagnation leading to the expiration of the first option contract.

By contrast, no evidence of market manipulation is observed when post-1980

events are used: in Figure 4 cumulative abnormal returns are plotted for all stocks which


6 This finding contrasts with the implicit conclusion deriving from Conrad (1989). Using
the results of her Table I, one could conclude that the positive price effect is permanent.
However, Conrad's Table I only presents results for up to 30 days after listing, during
which the price effect remains positive. I find similar positive persistence over the same
30-day interval (CAR=1.86%, 1-2.27). However, as reported above, the CARs totally
vanish during the following 20 days. This 50-day price reversal is consistent with the
conjecture that stock prices return to their normal level prior to the expiration date of their
first option contract (which, on average, equals 50 days during that period)..








became optioned between 1981 and 1992. Unlike Figure 2, event-window CARs are

negative, and persistent over at the 100-day period after listing (CAR=-11.50%, 1--8.27).

This finding suggests that a change in equilibrium stock prices occurs after option

introductions, during the 1981-1992 period.

A first formal test of price reversal is presented in Table 3, which compares pre-

1980 stock performance across three different windows: (i) listing event, (ii) post-listing

event, extending to the expiration date of the first option contract, and (iii) cumulative

"holding period," composed of(i) and (ii)

The first row in Table 3 confirms that substantial price increases occur during the

listing event window as illustrated in Figures 1, 2 and 3. The second row shows that the

average listing is followed by a negative abnormal performance, for the period extending

to the expiration date of the option contract. For each row, the conclusion holds when

both raw and abnormal returns are used to estimate price effects.

More interesting are the results shown in the last row: between 1973 and 1980, if

an investor chose to purchase an optioned stock five trading days before listing, and held it

until the expiration date of the first option contract, it would not have realized any gains.

This is consistent with the event-window price effect being only temporary, reflecting

manipulative maneuvers by option dealers.






7The graph in Figure 4 shows a slightly different cumulative price effect at day L+100
approximately -14%, compared to the -11.5% reported. The difference is that Figure 4
uses abnormal returns of individual stocks, whereas the statistical result reported in the
text is based on portfolios of stock grouped by listing date.








Table 4 relates listing period returns and abnormal turnovers to stock price

reversals. One implication of the market manipulation hypothesis is that stocks which are

being manipulated experience negative returns close to the expiration date of their first

option contract. Consequently, a good proxy for whether or not a particular stock is

being manipulated is the sign of the stock's cumulative returns, close to the expiration date

of the first option contract. That is, stocks with negative near-expiration returns are more

likely to have been manipulated around option listing. Table 4 therefore splits the stocks

into two sub-samples, according to the sign of cumulative returns during the 40 days

preceding the option expiration date. The target sub-sample of interest is the one where

stocks have negative near-expiration returns. If the market manipulation hypothesis is

correct, I expect listing-window price and turnover effects to be significantly higher in the

target sub-sample compared to its counterpart. Table 4 presents therefore tests of the

joint hypothesis that (1) listing period price and turnover effects are due to stock price

manipulation, and (2) manipulated stocks have negative near-expiration returns.

Panel A of Table 4 shows that between 1973 and 1980, price effects and abnormal

turnovers are generally higher for stocks experiencing subsequent price reversals, although

the difference is statistically significant only in the case of cumulative raw returns. The

first row in Panel A shows that CARs in the target sub-sample are 4.07%, compared to

2.60% in its counterpart. While target CARs are higher (as expected), the difference

between the two groups is not statistically significant: ([CAR(target)-CAR(counterpart)]

= 1.47%, t=1.03). A similar pattern is discernible for abnormal share turnover, which is

larger in the target sub-sample without being statistically different: ([TURNOVER(target)

- TURNOVER(counterpart)] = 1.79%, 1=1.26). By contrast, cumulative raw returns are








significantly higher in the target sub-sample: ([CRR(target)-CRR(counterpart)]=5.39%,

t=3.33) Overall, the data in Panel A are moderately consistent with the market

manipulation hypothesis.

Panel B repeats the analysis for options listed between 1981 and 1992. Unlike

Panel A, no clear relationship is discernible between post listing performance and either

(1) the listing price effect or (2) abnormal turnover, consistent with the conjecture that

stock price manipulation substantially diminished during that period.

Table 5 uses cross-sectional regressions to test the joint hypothesis that (a) the

pre-1980 positive price effect and abnormal share turnover are due to stock price

manipulation, and (b) option dealers manipulate mostly small stocks with large volatilities.

Panel A estimates the following six equations for the pre-moratorium period (1973-1980):

( -i A*i,7wl i-. + /, LOG(C(AP)+ c (2)

( i'l 'l, mi.i =, +fL2 LOG(CAP)+ e, (3)

Abnormal Turnover = a, + f, LOG(CAP) + E3 (4)

CAR(listing)= a4 +,4 S)DR + e4 (5)

( h''lir ms'le -, +yI SDR + e, (6)

Abnormal Turnover = a + y, SDR + e (7)

A significantly negative p, or -, indicates that the price effect of option

introduction is strongest for smaller firms; likewise a significantly negative ,3 indicates that

smaller stocks experience higher abnormal turnover during the event window. In sum,

significantly negative ['s are consistent with the hypothesis that manipulation is easier to

achieve when only small stock quantities are required to move stock prices in the desired








direction. The reverse is true for the y coefficients, which must be significantly positive to

be consistent with price manipulation, since more volatile stocks command higher call

premia, making manipulation more attractive.

The results in Panel A are overwhelmingly consistent with market manipulation

occurring between 1973 and 1980: the p coefficients are always significantly negative (at

better than the 1% level in two of the three cases), while two of the three y coefficients are

significant at better than the 10% level. (The third coefficient--y, --is marginally significant,

with p=. 105.)8

Panel B of Table 5 estimates equations (2) through (7) for the period 1981-1992,

when "good faith credit" and "free riding" were no longer permitted. None of the

coefficients in the six cross-sectional regressions is significant, indicating that stock price

manipulation substantially declined in the post-moratorium period.

Table 6 repeats the analysis of Table 5 for two sub-samples of the original data,

selected according to the sign of the stock's post-listing performance. As in Table 4, I

conjecture that cross-sectional regressions will be significant only in the case of the target

sub-sample of stocks with negative near-expiration returns. This table in effect represents

an interactive test of price reversal and cross-sectional analysis, used to strengthen the

previous conclusions about stock price manipulation.




8 I have also estimated multivariate versions of equations (2) to (7), with LOG(CAP) and
SDR as joint independent variables. While individual coefficients always carry the correct
sign, they are generally insignificant due to the presence of multi-collinearity (the
correlation coefficient between LOG(CAP) and SDR is -0 588) The overall F-statistic
however, remains highly significant, suggesting that the two variables have a strong joint
explanatory power








Panel A of Table 6 presents the results for stocks experiencing negative near-

expiration returns during the 1973-1980 period. As conjectured, the coefficient of the

independent variable is strongly significant at better that the 1% level in the first five cases,

and at the 5% level in the case of equation (7). These coefficients are generally more

significant than the corresponding coefficients in Table 5. In Panel B of Table 6 the same

six equations are estimated for stocks with positive post-listing performance. In sharp

contrast with models estimated using the subsample of stocks with negative near-

expiration returns, none of the regression coefficients are significant for this data subset.9

Overall, this evidence suggests that the results in Table 5 are driven exclusively by the sub-

sample of stocks with negative post-listing performance. That is, between 1973 and 1980,

manipulation is only detected for stocks experiencing a post-listing price reversal.

Panels C and D of Table 6 present the results for the 1981-1992 period. As with

previous tests, none of the regression coefficients are significant during that period,

reinforcing the previous conclusion that federal regulators successfully prevented stock

price manipulation from re-occurring after 1980.

Throughout Tables 5 and 6, standard errors are corrected using a White

heteroskedasticity-consistent variance-covariance matrix. An alternative way to adjust for

heteroskedasticity would be to use the inverse of the abnormal returns standard deviation

as a weighting variable in an Ordinary Least Square (OLS) model. To assess the


9 A statistical comparison of the independent variable coefficients across the two panels
reveals that these coefficients are statistically different only for the second (t11.89) and
fifth (t=2.43) equations estimated in Panels A and B. Nonetheless, the data in Panel A
presents more significant regressions than that of Panel B, which is consistent with the
conjecture that stocks with negative near-expiration returns are the ones that are being
manipulated








robustness of my results, I have repeated the analysis in Tables 5 and 6 by substituting the

weighted least square (WLS) model for the White OLS model presented. When Tables 5

and 6 are repeated using WLS estimates, the market manipulation conclusions remain

qualitatively identical: manipulation occurs mostly for small stocks with high volatility.1o

Table 7 repeats the analysis of Table 5 to test the joint hypothesis that that (a) the

pre-1980 positive price effect and abnormal share turnover are due to stock price

manipulation, and (b) option dealers manipulate mostly small stocks with large volatilities

Unlike Table 5 which imposes a linear relationship between price effects and independent

variables, the analysis presented in Table 7 consists of comparing price effects across the

four different quartiles of each dependent variable. The advantage of this type of analysis

is that no specific relationship is imposed between the dependent and independent

variables. The cross-sectional regressions estimated in equations (2) through (7)

necessarily impose a linear functional form for the relationships between (a) price effects

or share turnover on the one hand, and (b) volatility or the logarithm of firm size on the

other When the exact functional form is unknown, the use of quartile analysis provides an

alternative method to test the market manipulation hypothesis, and to assess the

robustness of the results obtained in the cross-sectional regressions.

Panel A of Table 7 shows the results for all CBOE calls listed prior to the end of

1980. In the top part of the Panel, firm size is considered as a potential explanatory


10 In addition, WLS results tend to be more significant than the White-adjusted OLS
results shown in Tables 5 (Panel A) and Table 6 (Panel A) For example, in the fifth
regression of Panel A in Table 5, the coefficient of SDR is only marginally significant at
the 10% level. In the WLS regression, that coefficient is significant at better than the 10%
level (t=1.80).








variable for listing price and turnover effects. In accordance with previous findings, the

explanatory power of firm size is very substantial: firms in the lowest quartile experience

the highest positive price effect and abnormal share turnover. Moreover, the relation

between price effect and firm size is monotonic, as predicted by the previously presented

theoretical discussion. Of particular interest is the magnitude of the price effect for firms

in this lowest size quartile: 7.5% using raw returns, and 6.7% using abnormal returns! By

contrast, firms in the highest size quartile experience no significant price effect around the

date of their option introduction, consistent with the hypothesis that option dealers do not

attempt to manipulate the stock of some of the largest firms on the market."

The middle part of Panel A considers the effects of return volatility on the listing

price effect and abnormal share turnover. In general, the results are qualitatively consistent

with the conclusions of Table 5: there is an increasing monotonic relation between price

effects and volatilities, and in the case of cumulative abnormal returns, the price effect in

the most volatile group (6.095%) is statistically different from the effect in the less volatile

group (1.114%), with a t-statistic of 2.24.12 Unlike price effects, abnormal turnovers do

not appear to be monotonically related to volatilities. Nonetheless, the difference between

the first and fourth groups (3.847%) is marginally significant at the 15% level (t-1.61).

Overall, the last row of the middle section indicates that the average price effect is driven


Using statistical tests of mean differences, I find that all three parameters estimated in
the first group (small CAP) are significantly higher than those estimated in the fourth
group (large CAP):
CRR(group 1) CRR(group 4)= 6.242%, t=2.18;
CAR(group 1) CAR(group 4) = 6.029%, 1-3.19; and
ABNTURN(group 1) ABNTURN(group 4) = 6.779%, 1-2.77.








mostly by firms with large return volatility, for which price manipulation is more

profitable.

The bottom part of Panel A considers the interactive effects between firm size and

return volatilities on the one hand, and price and turnover effects on the other. The usual

pattern is once again discernible: most of the listing price and turnover effects occur for

small stocks with large volatilities.

Panel B repeats the quartile analysis for options listed after 1980. As with previous

tests, no clear pattern is distinguishable during this period.

In short, Table 7 shows that the qualitative conclusions of Table 5 are robust to a

plausible change in the model specification.

Table 8 presents a different analysis of the interaction between price reversal, firm

size and stock volatilities. The following general model is estimated throughout the

Table"3

CRR(list window) = yo Y CRR(exp40) (8)



12 In the case of cumulative raw returns, this difference is insignificant.

"3 This table uses exclusively the cumulative raw returns as a measure of stock price
performance. Comparing cumulative abnormal returns (CARs) estimated during the event
window with the CARs estimated near the expiration date poses a serious econometrical
problem: since both measures share the same estimated a (from the market model), they
also share the same measurement error. Consequently, CAR(list window) and
CAR(exp40) will always be positively correlated (by construction), regardless of the true,
underlying economic relationship. That is, given any random sample of stocks and event
dates, the two estimates will be positively correlated. As an alternative to using
cumulative raw returns, I have considered estimating CAR(exp40) using the market model
estimated in the L+100/L+200 period, and comparing it to CAR(list window) estimated
over L-100/L-6. However, as explained in Chapter 6, stock returns during the
L+100/L+200 period are affected by the introduction of options, which is likely to
introduce considerable noise in the cross-sectional regression.








where a significantly negative y, suggests that price reversal holds for every firm, in cross-

section. That is, firms with highly positive listing price effects also experience highly

negative post-listing performances. I estimate equation (8) for six sub-samples of my

original data, selected according to listing date, optioned stock size, and optioned stock

volatility.

Panel A presents the results for options listed prior to the end of 1980. The first

row of Panel A shows that on average, the optioned stock experiences a price reversal in

the post-listing period (the y, coefficient is significantly negative at the 5% level). In the

second row, equation (8) is estimated only for the subset of stocks with above-median

market capitalization, and below-median return volatility. As conjectured, no price

reversal is observed in this case. By contrast, the last row of Panel A presents strong

evidence of price reversal when equation (8) is estimated only for the smaller stocks with

higher volatilities.

The analysis is repeated in Panel B for options listed between 1981 and 1992, but

no evidence of price reversal is found for any of the three samples examined.

Overall, the totality of the evidence presented in Tables 3 through 8 is strongly

consistent with the following conclusions.

1. The positive price effect of option introduction documented between 1973 and
1980 may be explained by stock price manipulation.

2. During that period, option dealers have mostly manipulated smaller stocks
(which required lesser capital "input" to the manipulation "technology"), and
stocks with large volatilities (which commanded higher call premia).

3 The typically manipulated stock experienced a strong price increase during the
listing window and a corresponding strong decrease during the post-listing
window.








4. The price effect documented in Conrad (1989) is only temporary, and does not
reflect a change in the asset's equilibrium price level.

5. The repeal of "free riding" at the end of 1980 has considerably increased the
option dealers' costs of manipulating stock prices, substantially diminishing the
appeal of stock price manipulation in recent years.


An Alternative Explanation

The market manipulation hypothesis is clearly consistent with the disappearance of

the positive price effect after 1980. The only competing explanation encountered in the

literature is found in Detemple and Jorion (1990), and is based on market completion. The

authors attribute the disappearance of the positive price effect to the introduction of the

index options in April 1982, which would have effectively "completed" the markets. If this

hypothesis is correct, the positive price effect should have persisted through March 1982.

However, when I analyze event-window abnormal returns for all listings occurring

between January 1981 and March 1982, I find evidence of negative (rather than positive)

CARs for that period (CAR = -2.03%, t= -2.062). The data are therefore inconsistent with

this alternative explanation.


































o 0 0 0 0 0 0 0 0 0 0 0


Days from listing date


















Figure 2:
Stock abnormal returns around option listings:
1973-1980


Event window


o
E
U
i
uB













Event Window


Typical lifetime of first
Call contract listed


Days from listing date


Figure 3:
Stock cumulative raw returns around opyion listings:
1973-1980














Event Window


So Days f om listing date
Days from listing date


Figure 4:
Stock abnormal returns around option listings:
1981-1992


SC
0




,E
0
1
rt









Table 3. Stock price behavior around listing and expiration dates of first
Call option contracts listed on CBOE between 1973 and 1980.

Cumulative Abnormal Returns computed using the market model, estimated during the
pre-listing (L-100/L-5) period.


Window


Listing window (N=78)



Post-listing window, leading
to the expiration date of the
first option contract (N=78)

Holding period, comprised
of the previous two periods
(N=78)


Cumulative Raw Returns


3.77%
(1-4.37)***


-5.23%
(t=-2.57)**


-1.46%
(- -0.72)


Cumulative Abnormal
Returns

3.34%
(t-3.47)***


-4.74%
(t=-2.08)**


-1.40%
(t -0.55)


Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level



Variable Definition:

N : Number of stocks for which Call options were first listed on CBOE, during the
time period indicated.

-Listing Window: Period extending from L-5 to L+5, where L represents the listing
date.

-Post Listing Window: Period extending from L+6 to E, where E represents the
expiration date.

-Holding Period: Period extending from L-5 to E









Table 4: Listing returns, abnormal turnover, and price reversal.


This table relates listing period returns and abnormal turnover to stock price reversal. The
table tests the joint hypothesis that (1) listing period price effects and abnormal turnovers
are due to stock price manipulation and (2) manipulated stocks experience price reversal
in the post-listing period.


Panel A: CBOE call options listed between 1973 and 1980


Stocks With Positive
Expiration-Period Returns


CAR (list 0.0407 3.514***
window)


CRR (list 0.0646
window)

Abnormal 0.0243
turnover


6.053***


2.635**


CAR (list
window)

CRR (list
window

Abnormal
turnover


0.0260 3.126***


0.0107


0.0064


0.879


0.599


Panel B: CBOE call options listed between 1981 and 1992

Stocks With Negative Stocks With Positive
Expiration-Period Returns Expiration-Period Returns
( CRR(exp40)<0 ) N=47 ( CRR(exp40)>0) N=53
Variable Mean Value T-statistic Variable Mean Value T-statistic
Name Name

CAR (list -0.010 -0.917 CAR (list -0.007 -0.621
window) window)

CRR (list 0.011 0.916 CRR (list 0.013 1.156
window) window)

Abnormal 0.021 1.606 Abnormal 0.004 0.203
turnover turnover









Table 4. Continued

Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
Significant at the 10% level



Variable Definitions:

CAR: Cumulative Abnormal Returns, computed for the period indicated in brackets.
Abnormal returns are calculated using the Brown and Warner (1985) market
model, estimated from L-100 to L-6, where L represents the listing date

- CRR Cumulative Raw Returns, computed for the period indicated in brackets.

- Abnormal Turnover: The difference between the mean share turnover computed
during L-5 to L+5, and the mean share turnover computed during L-100 to L-6.
Share turnover is estimated as the volume of trade divided by the stock's market
capitalization.

- (list window): Period extending from L-5 to L+5.

- (exD40: Period extending from E-40 to E, where E represents the expiration date.









Table 5: Cross-sectional regressions between listing period returns and abnormal share
turnover (dependent variables) and two proxies for stock price manipulation: firm size and
stock volatility (independent variables).


This table is a test of the following joint hypothesis: (1) the listing period price effect and
abnormal turnover are due to stock price manipulation, and (2) option dealers manipulate
mostly small stocks, with large standard deviation. Standard errors are computed using a
White heteroskedasticity consistent variance-covariance matrix.



Panel A: CBOE call options listed between 1973 and 1980

Dep. Indep. Constant Constant Indep. Indep. N Adj.
Variable Variable Coeff T-stat. Var. Var. Obs.R2
Coeff T-stat.

CAR LOG(CAP) 0.189 3.114*** -0015 -2.699*** 78 0.0855
(list
window)
CRR LOG(CAP) 0.178 2.154** -0.014 -1.820* 78 0.0421
(list
window)
Abn. LOG(CAP) 0.190 2.684*** -0.017 -2.681*** 77 0.1152
Turnover


CAR SDR -0.032 -1.499 2.844 2.706*** 78 0.0907
(list
window)
CRR SDR -0008 -0.356 2.011 1.644A 78 0.0224
(list
window)
Abn. SDR -0.033 -1.570 2.123 1.861* 77 0.0451
Turnover









Table 5. Continued

Panel B: CBOE call options listed between 1981 and 1992

Dep. Indep Constant Constant Indep. Indep. N. Adj.
Variable Variable Coeff T-stat. Var. Var. Obs. R
Coeff T-stat.

CAR LOG(CAP) -0.035 -0.537 0.003 0.435 101 -0.0081


window)
CRR LOG(CAP) 0.0727 0.957
(list
window)
Abn. LOG(CAP) 0.0403 0.201
Turnover


-0.006 -0.814 101 -0.0013


-0.0031 -0.144 100 -0.0096


CAR SDR -0.0189 -0.971 0.377 0.504 101 -00075
(list
window)
CRR SDR 0.0047 0.249 0.290 0.387 101 -0.0086
(list
window)
Abn. SDR 0.0310 0.781 -0.643 -0.374 100 -0.0072
Turnover


Significance Levels:

** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level
A Marginally significant at the 10% level (p=0.105, two-tail test)








Table 5. Continued

Variable Definitions:

-CAR: Cumulative Abnormal Returns, computed for the period indicated in brackets.
Abnormal returns are calculated using the Brown and Warner (1985) market
model, estimated from L-100 to L-6, where L represents the listing date.

-CRR: Cumulative Raw Returns, computed for the period indicated in brackets.

-CAP: The inflation-adjusted market capitalization, equal to the number of outstanding
shares times the average share price during the event window. Figures are
converted in 1983 constant dollars.

-LOG(CAP): The natural logarithm of CAP.

-SDR: The standard deviation of raw returns, computed over the pre-listing period.
The pre-listing period extends from L-100 to L-6.

Abn. Turnover: The difference between the mean share turnover computed during
L-5 to L+5, and the mean share turnover computed during L-100 to L-6. Share
turnover is estimated as the volume of trade divided by the stock's market
capitalization.


- (list window): Period extending from L-5 to L+5.









Table 6: The relation between abnormal returns, turnover, firm size, return volatility, and
price reversal

This table presents a test of the following joint hypothesis: (1) the listing period price
effect and abnormal turnover are due to stock price manipulation, and (2) option dealers
manipulate mostly small stocks, with large standard deviation, and (3) stocks whose prices
are manipulated experience price reversal in the post listing period. Standard errors are
computed using a White heteroskedasticity consistent variance-covariance matrix.


Panel A: CBOE call options listed between 1973 and 1980
Stocks with negative expiration period returns (CRR(exp40)<0)

Dep. Indep. Constant Constant Indep Indep N. Adj.
Variable Variable Coeff T-stat Var. Var. Obs. R2
Coeff T-stat.

CAR LOG(CAP) 0.237 2.704** -0.019 -2.303** 39 0.1057
(list
window)
CRR LOG(CAP) 0.317 3.881*** -0.025 -3.098*** 39 0.2291
(list
window)
Abn. LOG(CAP) 0.214 3.296*** -0.018 -3.174*** 39 0.1675
Turnover


CAR SDR -0.062 -1.951* 4.333 2.804*** 39 0.1538
(list
window)
CRR SDR -0.045 -1.734* 4.628 4.451*** 39 02158
(list
window)
Abn. SDR -0.037 -1.469 2.621 2.166** 39 0.0770
Turnover









Table 6. Continued

Panel B: CBOE call options listed between 1973 and 1980
Stocks with positive expiration period returns (CRR(exp40)>0)


Dep. Indep. Constant Constant Indep. Indep. N. Adj.
Variable Variable Coeff. T-stat. Var. Var. Obs. R
Coeff T-stat.

CAR LOG(CAP) 0.120 1.204 -0.009 -0.986 39 0.0196
(list
window)
CRR LOG(CAP) -0.057 -0.363 0.006 0.457 39 -0.0155
(list
window)
Abn. LOG(CAP) 0.147 0.929 -0.013 -0.949 38 0.0379
Turnover


CAR SDR -0003 -0.129 1.138 0.969 39 0.0073
(list
window)
CRR SDR 0.0340 0.845 -1.037 -0.497 39 -00170
(list
window)
Abn. SDR -0.0273 -0.773 1.481 0.745 38 -0.0014
Turnover









Table 6. Continued

Panel C CBOE call options listed between 1981 and 1992
Stocks with negative expiration period returns (CRR(exp40)<0)


Dep. Indep. Constant Constant Indep. Indep. N. Adj
Variable Variable Coeff T-stat. Var. Var. Obs. R
Coeff. T-stat.

CAR LOG(CAP) -0.061 -0.587 0.005 0.486 47 -0.0160
(list
window)
CRR LOG(CAP) 0.088 0.608 -0.008 -0.526 47 -0.0100
(list
window)
Abn. LOG(CAP) 0.0557 0.431 -0.004 -0.287 46 -0.0212
Turnover


CAR SDR 0.007 0.031 -0.414 -0.502 47 -0.0198
(list
window)
CRR SDR 0.006 0.185 0.177 0.122 47 -0.0218
(list
window)
Abn. SDR 0.016 0.547 0.139 0.106 46 -0.0225
Turnover









Table 6. Continued

Panel D: CBOE call options listed between 1981 and 1992
Stocks with positive expiration period returns (CRR(exp40)>0)


Dep. Indep. Constant Constant Indep. Indep N. Adj.
Variable Variable Coeff T-stat. Var. Var. Obs. R
Coeff. T-stat.


CAR LOG(CAP) -0.016 -0.182 0.001 0.114 53 -0.0193
(list
window)
CRR LOG(CAP) 0.069 0.721 -0.006 -0.632 53 -0.0108
(list
window)
Abn. LOG(CAP) 0.0539 0.187 -0005 -0.184 53 -00177
Turnover


CAR SDR -0.0290 -1.111 -0.769 -0.716 53 -0.0063
(list
window)
CRR SDR -0.0028 0.124 0.352 0.388 53 -0.0167
(list
window)
Abn SDR 0.0321 0.592 -0.954 -0.392 53 -0.0140
Turnover


Significance Levels:

*** Significant at the I% level
** Significant at the 5% level
* Significant at the 10% level









Table 6. Continued

Variable Definitions:

- CAR: Cumulative Abnormal Returns, computed for the period indicated in brackets.
Abnormal returns are calculated using the Brown and Warner (1985) market
model, estimated from L-100 to L-6, where L represents the listing date

-CRR: Cumulative Raw Returns, computed for the period indicated in brackets.

- CAP: The inflation-adjusted market capitalization, equal to the number of outstanding
shares times the average share price during the event window. Figures are
converted in 1983 constant dollars.

- LOG(CAP): The natural logarithm of CAP

SDR: The standard deviation of raw returns, computed over the pre-listing period.
The pre-listing period extends from L-100 to L-6

- Abn. Turnover: The difference between the mean share turnover computed during
L-5 to L+5, and the mean share turnover computed during L-100 to L-6. Share
turnover is estimated as the volume of trade divided by the stock's market
capitalization.

- (ist window: Period extending from L-5 to L+5.

- (exp40) Period extending from E-40 to E, where E represents the expiration date.








Table 7: Quartile analysis of the relations between (1) listing period returns and share
turnover and (2) firm size and stock volatility.

Like Table 4, this is a test of the following joint hypothesis: (1) the listing period price
effect and abnormal turnover are due to stock price manipulation, and (2) option dealers
manipulate mostly small stocks, with large standard deviation. T-statistics are shown in
brackets below each entry.

Panel A: CBOE call options listed between 1973 and 1980

Independent Quartile CRR (list CAR (list Abn. Turnover
Variable window) (%) window) (%) (%)

CAP 1 (lowest) 7.503 6.727 6.751
(N=19) (2.810)** (4.064)*** (2.758)**
2 3.202 3.573 -0.573
(N=20) (2.094)** (2.057)* (-0.734)
3 3.158 2.393 0.162
(N=20) (3.229)*** (2.605)** (0.868)
4 (highest) 1.261 0.698 -0,028
(N=19) (1.222) (0.762) (-0.508)

SDR 1 (lowest) 2.188 1.114 0.427
(N=19) (2.103)** (1.337) (2.538)**
2 3.380 2.433 0.329
(N=20) (2.108)** (1.853)* (0.444)
3 3.763 3.714 1.193
(N=20) (2.988)*** (3.435)*** (0.954)
4 (highest) 5.573 6.095 4.274
(N=19) (2.182)** (2.952)*** (1.797)*

High CAP and low SDR 2.449 1.421 0.123
(N=29) (2.882)*** (1.864)* (0.996)
Low CAP and high SDR 5.809 5.868 3.655
(N=29) (3.173)*** (4.075)*** (2.076)**








Table 7. Continued

Panel B: CBOE call options listed between 1981 and 1992


Independent Quartile CRR (list
Variable window) (%)


CAR (list Abn. Turnover
window) (%) (%1


CAP 1 (lowest) 2.784 0.744 3.348
(N=25) (1.215) (-0.322) (0.733)
2 3.903 1 129 -0.112
(N=26) (2.990)*** (0.938) (-0.059)
3 -1 802 -3.419 1.939
(N=25) (-1.021) (-2.072) (1.270)
4 (highest) 0.199 -0.349 -0.0389
(N=25) (0.230) (-0.386) (-0.045)

SDR 1 (lowest) 1 317 -0.318 -0.278
(N=25) (1.384) (-0.372) (-0.352)
2 -0.328 -2.691 00527
(N=26) (-0.254) (-1.876)* (0.071)
3 2.722 -0.283 4.880
(N=25) (1.337) (-0.148) (1.995)*
4 (highest) 1.543 0.062 0.471
(N=25) (0.706) (0.031) (0.105)

High CAP and low SDR -0.085 -1.788 -0.222
(N=36) (-0.093) (-1.749)* (-0.351)
Low CAP and high SDR 3.990 0.675 2.185
(N=36) (2.345)** (0.396) (0.641)


Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level









Table 7. Continued

Variable Definitions:

CAR: Cumulative Abnormal Returns, computed for the period indicated in brackets.
Abnormal returns are calculated using the Brown and Warner (1985) market
model, estimated from L-100 to L-6, where L represents the listing date.

-CRR: Cumulative Raw Returns, computed for the period indicated in brackets.

- CAP: The inflation-adjusted market capitalization, equal to the number of outstanding
shares times the average share price during the event window. Figures are
converted in 1983 constant dollars.

- LOGCAP: The natural logarithm of CAP.

-SDR: The standard deviation of raw returns, computed over the pre-listing period.
The pre-listing period extends from L-100 to L-6

- Abn. Turnover: The difference between the mean share turnover computed during
L-5 to L+5, and the mean share turnover computed during L-100 to L-6. Share
turnover is estimated as the volume of trade divided by the stock's market
capitalization.

- list window): Period extending from L-5 to L+5.



Note:

In the bottom part of Panels A and B, the first group consists of stocks with above-median
market capitalization and below-median volatility. The second group consists of stocks
with below-median market capitalization and above-median volatilities.








Table 8: Evidence of price reversal

This table presents cross-sectional regressions of listing period returns on post-listing
returns, to determine whether price reversal holds in cross-section. The model estimated is
of the following form:

CRR(list window) a + f CRR(exp40) +

The values of a and p are shown in the table. Standard errors are computed using a
White heteroskedasticity consistent variance-covariance matrix.


Panel A: CBOE call options listed between 1973 and 1980

Sample Selection T-stat. of p T-stat. of N. Obs Adjusted
p__ ^R2

All call listings 0.0358 4.201*** -6.210 -2.523** 78 0.0889


Large stocks with 0.0244 2.824*** 0 036 0.0125 29 -0.0370
low volatility

Small stocks with 0.0548 3.089*** -9.368 -2.155** 29 0.1700
high volatility




Panel B: CBOE call options listed between 1981 and 1992

Sample Selection a -stat, of p T-stat. of N Obs Adjusted
I I pt P IIR2

All call listings 0.0129 1.540 0.236 0.153 101 -0.0100


Large stocks with -0.0009 -0.102 1 088 0.498 36 -0.0235
low volatility

Small stocks with 0.0409 2.435** -0.866 -0.394 36 -0.0270
high volatility








Table 8. Continued

Significance Levels.

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level




Variable Definitions:

-CAR: Cumulative Abnormal Returns, computed for the period indicated in brackets.
Abnormal returns are calculated using the Brown and Warner (1985) market
model, estimated from L-100 to L-6, where L represents the listing date.

-CRR: Cumulative Raw Returns, computed for the period indicated in brackets.

-CAP: The inflation-adjusted market capitalization, equal to the number of outstanding
shares times the average share price during the event window. Figures are
converted in 1983 constant dollars.

-LOG(CAP): The natural logarithm of CAP.

SDR: The standard deviation of raw returns, computed over the pre-listing period.
The pre-listing period extends from L-100 to L-6.

-(list window): Period extending from L-5 to L+5

-(exp40: Period extending from E-40 to E, where E represents the expiration date.













CHAPTER 6
THE SELECTION BIAS HYPOTHESIS


Introduction

While the market manipulation hypothesis presented in Chapters 4 and 5 is

consistent with the pre-1980 positive price effect, it does not account for the negative

price effect observed after 1980 That is, after manipulation ceases, it is not clear why the

price effect of option introductions becomes negative after 1980, and appears to persist

over the 100-day period following the listing

In an attempt to identify the cause of this negative price effect, I must first

determine whether the observed abnormal returns reflect a "true" change in equilibrium

stock prices, or whether they merely reflect a bias in the process by which exchanges

select optionable stocks.

This chapter identifies two selection bias hypotheses. The first is a "price

threshold" hypothesis, according to which exchanges select mostly stocks which have just

met the minimum price-per-share criterion required in order to become optionable. The

second is an "over performance" hypothesis, which states that exchanges select mostly

stocks which have over-performed the market in the immediate period preceding the

option listing.








The data do not however support either one of these two hypotheses, leaving wide

open the possibility that the negative price effect may indeed represent a permanent

change in the equilibrium prices of underlying stocks.


The "Price Threshold" Hypothesis

One possible explanation for the post-1980 negative price effect lies in the way in

which stocks are selected for option listings. While the ultimate listing decision belongs to

the options exchange (and is entirely driven by profit considerations), SEC regulations

impose minimal criteria which must be met by all stocks before they become eligible for

options trading. Among these criteria is the share price of the underlying stock, which

must remain above a certain threshold level during the three calendar months prior to the

listing of its first option contract For most of my sample period (1973-1988), this

threshold price was $10 per share, during each of the 60 days prior to the listing date.

SEC regulations introduced in 1988 lowered this threshold to an average price of $7.50

per share, computed over the 60-day period prior to listing. In either cases the option may

remain listed as long as the stock trades above $5 per share.

Under these circumstances, one must inquire whether the post-1980 negative price

effect is not mostly driven by stocks which became optioned as soon as they met the price

threshold criterion. In such case, the sample of stocks used in this dissertation would have

an upwards-biased pre-listing performance, resulting in an over-estimate of the constant

term (a) in the market model. With an over-estimated c, abnormal returns calculated

during an otherwise normal period would appear to be negative. This would then create








the appearance of negative abnormal returns in the post-listing period, even if the stock

performance during that period were normal.

One way to test the "price threshold" hypothesis is to identify a target sub-sample

of underlying stocks, which had met the price threshold criterion long before they became

optioned. (For example, Netscape stock was trading at $70 when its option became listed,

well above the $7.50 minimum threshold level.) Members of this target sub-sample are

therefore stocks for which the option listing decision was driven by considerations other

than the minimum threshold criterion. Then, if members of the target sub-sample also

experience a negative price effect in the post-1980 period, the "price threshold" hypothesis

is refuted.

I have included in the target sub-sample all stocks which have constantly traded at

over $12 per share during the 120 days preceding the listing. Thus, all members of the

target sub-sample became eligible for listing at least 60 days before their actual listing

date Clearly, if the target sub-sample represents a large proportion of the overall sample,

the "price threshold" hypothesis is not likely to be supported.

Table 9 presents descriptive statistics about the relative size of the target sub-

sample: between 1981 and 1992, 91.3% of all options listed were related to stocks which

had been trading at a price greater than $12 per share for at least 120 days. That is, like

Netscape, 91.3% of the stocks did not become optioned immediately after they met the

minimum criterion, but rather at some distant time in the future.

I have also repeated the analysis in Table 2, Panels A, B and C, using only stocks

which are member of the target sub-sample. The results are qualitatively identical to those








of Table 2-' the price effect of the target sub-sample is positive prior to 1980, and

becomes negative thereafter. Considering that a relatively large number of stocks became

optioned long after they first met the price-threshold criterion, the negative price effect of

option introductions appears to be unrelated to the price threshold hypothesis.


The "Over-Performance" Hypothesis

Even if no evidence of price-threshold selection bias is detected, a different type of

selection bias is also possible, and warrants further consideration. Some early authors (e.g.

Branch and Finnerty (1981) suggest that exchanges select mostly stocks which have over-

performed the market in the 60 calendar days prior to the option introduction. Given this

type of selection mechanism, one could argue that the negative event-window stock

performance merely reflects the return to a more "normal" state of nature.

There are three factors, however, which considerably reduce the power of this

argument: First, Skinner (1989) finds no evidence of any selection bias, for all options

listed between 1973 and 1986. Second, this hypothesis does not fairly represent the

exchange's selection incentives. The exchange profits rise with option volume, not with

past stock performance. Option volume itself is closely dependent on stock volatility (not

price) and to some extent the industry in which the firm trades.2 Thus, if selection bias


Repeating the analysis in Table 2 for the target sub-sample yields results which are
identical to the same degree of significance of the t-statistics. The few observations which
account for the difference between the full and the target sample are evenly distributed in
time, across listing events. When forming portfolios of stocks grouped by listing dates, the
number of listing events remains unchanged, and the value of the estimated coefficients
differs only by one significant digit in most cases.
2 CBOE traders have indicated that whereas volatility is a major factor in the listing
decision, many low-volatility firms also get listed, in order to offer a better cross-section
of industries to the general investor.









existed, it would have to be related to stock volatility as opposed to mean returns, but

Skinner also dismisses the presence of a volatility bias.

Third, even if exchanges really were to select better-than-average past performers,

there is no particular reason for which these firms will cease to be better-than-average

precisely on the day on which they become listed. Indeed, a close examination of Figure 4

shows that the change in the underlying stock price occurs inside the event window, rather

than at some point in the future.'

I therefore conjecture that the observed post-1980 negative price effect is not

consistent with the "Over-Performance" selection bias hypothesis. To test this assertion, I

estimate the market model over two periods different from the one used in Chapters

3 and 5:4

L-200 to L-100 (Model A); and

L+100 to L+200 (Model B).

I then repeat the relevant tests from Chapters 3 and 5 using each of the two newly

estimated market models.

If my conjecture is correct, I expect results estimated using Model A to be very

similar to those obtained in Chapters 3 and 5: positive price-effects prior to 1980, negative

price effects thereafter.


3 One explanation for the return to a "more normal" performance after listing would be to
assume that stock prices follow a mean reversion process, with an approximately 100-day
cycle. Thus, if a stock has over-performed for the past 100 days, it will more likely under-
perform for the next 100-day period. Academic evidence on mean reversion processes is
however inconclusive, thereby substantially reducing the appeal of this plausible
explanation.
4 In these models, L denotes the listing date.








More difficult is the interpretation of results obtained from model B. If this

model's estimation period (L+ 100/L+200) is itself affected by the introduction of options,

failure to find a negative listing price effect does not necessarily support the selection bias

hypothesis. To illustrate this point, consider the following scenario: stock XYZ yields an

average return of 10% per year for the 200 trading days prior to listing. Subsequently, its

average rate of return drops to 7% and remains at that level for 200 trading days after the

listing. The change occurs on day L. Then, by estimating the market model over either the

L-200/L-100 period, or the L-100/L-5 period, one would find a negative price effect

during the listing window. By contrast, when the market model is estimated over the

L+100/L+200 period, the price effect inside the listing window will no longer be negative,

since the estimation period is itself affected by the introduction of options

Consequently, in order to better interpret the post-1980 results obtained from

Model B, I must first determine if the post-listing period (L+100/L+200) has been

significantly affected by the introduction of options. This can be achieved by calculating

abnormal returns for the entire L+100/L+200 period, with respect to either one of the two

pre-listing market models. If these abnormal returns are significantly negative, it is

plausible to conclude that options have affected the rate of return of underlying stocks for

at least 200 trading days after listing In this case the absence of event-window abnormal

returns obtained using model B cannot be interpreted as an indication of a selection bias.

In Table 10 below, I repeat the relevant part of the analysis presented in Chapter 3,

with a market model estimated during the L-200 to L-100 period. As with the analysis

presented in Table 2, I first form portfolios of stocks listed at the same date, and treat

them as single securities. Similar to the results obtained in Table 2, I find that the price








effect of option introductions remains significantly positive during the 1973-1980 period,

and becomes significantly negative after 1981.

In the first part of Table 10 I show that pre-1980 results are qualitatively identical

to those obtained in Table 2: the positive price effect of option introduction is about 2%,

with a t-statistic significant at better than the 1% level. The second part of Table 10

shows that the post-1980 price effect is even more negative than the one computed in

Table 2: the entire sample experiences listing abnormal returns equal to -1.18%, with a t-

statistic of -3.204. This compares to the more modest -1.03% (t=-2.38) negative

abnormal return computed using the market model estimated in Table 2. Lastly, the

analysis in Table 10 reveals that the negative post-1980 abnormal returns are driven

especially by joint put/call listings, and are more pronounced for stocks listed on the

NASDAQ exchange.

Table 11 repeats the analysis presented in Table 10, for a market model estimated

over the L+100/L+200 period. As with results from Tables 2 and 10, I find significant

positive listing abnormal returns for stocks optioned during 1973-1980, which are

consistent with market manipulation occurring during that period.

As earlier conjectured, no evidence of negative price effect is detected for the

1981-1992 period: all estimated abnormal returns are not significantly different from zero.

In order to better interpret this finding, I must also determine whether stock returns over

the L+100/L+200 estimation period have also been negatively affected by the option

listing.








Table 12 shows abnormal returns for two different extended post-listing periods,

using market models computed in two different pre-listing periods. The bottom half of

Table 12 shows the results for the 1981-1992 period, as conjectured, abnormal returns in

the extended post-listing periods are substantially negative, both statistically and

economically: cumulative abnormal returns range from -9.82% to -13.33%, with t-

statistics significant at better than the 0.1% level. These results suggest that the listing of

an option significantly reduces the rate of return on the underlying stock for at least 200

trading days after listing occurs. Consequently, since returns over the L+100/L+200

period are also affected by the introduction of options, they cannot be reliably used to

estimate abnormal returns around the date of option introduction.

Equally interesting are the results shown in the top half of Table 12: long-term

post-listing performance is negative even during the pre-1980 period, when listing-window

abnormal returns are significantly positive. This further strengthens the conclusion that

the positive pre-1980 listing price effect of option introductions is only temporary.


Stock Price Manipulation Revisited

Since the upper half of Table 12 shows that the longer-term abnormal performance

of optioned stocks is negative even during 1973-1980, it is interesting to assess the

robustness of the market manipulation analysis presented in Chapter 5, using the abnormal

returns estimated over the L-200/L-100 period. Table 13 repeats the analysis of Table 5,

panel A, by analyzing the cross-sectional relation between listing abnormal returns and

either market capitalization or stock volatility (computed during the L-200/L-100

estimation period). As in Table 5, I find significant positive correlation between the listing








price effect and stock volatility (t=3.409), and significant negative correlation between

listing price effect and firm size (t=-3.202). These results are once again consistent with

the hypothesis that option dealers manipulate mostly small stocks with large volatilities.

As in Table 5, repeating the analysis in Table 13 for the 1981-1992 period (not shown)

reveals no statistical relation between price effects and either one of the two explanatory

variables.

The market manipulation hypothesis is further supported by Table 14, which like

Table 7, makes use of quartile analysis instead of cross-sectional regressions to detect any

relation between price effect and explanatory variables. As shown in Table 14, using the

L-200/L-100 estimation period does not substantially alter the conclusions already reached

in Table 7. The positive price effect of option introductions occurs mostly for smaller

stocks (4.71% for the lowest size quartile, t=2.52), and for stocks with high standard

deviation (4.95% for the highest volatility quartile, t=3.45). The interactive analysis

reveals that the positive price effect of option introduction is almost 6% for the smallest,

most volatile stocks, which are more likely to be manipulated

Using the L-200/L-100 estimation period does not therefore alter the conclusion of

the market manipulation analysis presented in Chapter 5.


The Case Against the Selection Bias Hypotheses

Overall, the evidence presented in this chapter appears to refute the selection bias

hypothesis: all of the results thus far presented in this dissertation are robust to a plausible

change in the market model estimation period, from L-100/L-5 to L-200/L-100.








In addition, this evidence also suggests that the "true" price effect of option

introduction is negative throughout the entire sample period, ranging from 1973 to 1992.

During the 1973-1980 period this negative effect is however eclipsed by the short-term

positive price run-up caused by manipulatory trading. However, when the longer-term

stock performance is examined, the negative price effect becomes readily apparent, as

indicated in the top half of Table 12. After market manipulation ceases at the end of 1980,

the negative price effect becomes discernible starting with the very first day of the post-

listing period, and appears to last for at least 200 trading days after listing.

In light of the findings presented in this chapter, it becomes important to

understand the reason for which options negatively affect the performance of their

underlying stocks. The following chapter presents one plausible explanation, based upon

the effect of options on the stock market's informational structure









Table 9: Descriptive Statistics for the Full and Target Samples

This table presents descriptive statistics about the size of the following two samples: the
full sample, consisting of all stocks optioned between 1981 and 1992, and the target
sample, consisting of stocks which were trading over $12 per share during the 120 days
prior to the option listing event.



Period Type of Overall Sample: Target Sample: Size of the Target
Listing Sample vs. the
Overall Sample (%)

1973-1980 Calls only 240 235 97.9

1981-1992 Entire 893 815 91 3
sample
1981-1992 Joint puts 523 465 88.9
and calls
1981-1992 Calls only 370 350 94.6



Sample Definition:

- Overall Sample: Total number of call and put/calls listed during 1981-1992.

- Target Sample: Total number of call and put/calls listed during 1981-1992, for which
the stock price was over $12 during the entire pre-listing period.









Table 10: The price effect of options introduction computed using a market model


estimated during the L-200/L-100 period


This table repeats the analysis of Table 2, with a market model estimated during the
L-200/L-100 period, where L represents the listing date. As with table 2, abnormal returns
are calculated using the Brown and Warner (1985) methodology. When more than one
stock became optioned on a given day, a portfolio of all such stocks is formed, and
subsequently treated as a single security. The event window extends from L-5 to L+5.


Period Sample


Number of CAR(s) -- T-stat. --
Events Market Model Market Model


1973-1980 All stocks (calls and
put/call listings)
Call listings only


Put /Call listings only


Put listings only


Calls and put/calls, NYSE
stocks only


81 2.02%


80 1.94%


9 -0.02 %


81 2.02%


3.037***


2.950***


-0.011

3.037***


Calls and put/calls, -- --
NASDAQ stocks only


1981-1992 All stocks (calls and
put/call listings)
Call listings only


Put /Call listings only

Put listings only


Calls and put/calls, NYSE
stocks only
Calls and put/calls,
NASDAQ stocks only


377 -1.18%


156 -0.52%


226 -1.64%


9 1.44%

244 -0.63%


179 -1.91%


-3.204***


-1.648*


-2.958***


0.700

-1.469


-2.811***





87


Table 10. Continued

Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level
Insufficient data



Variable Definition:

Number of Events: Number of days during which one or more option contracts get
listed.

-CARs (Market Model): Cumulative abnormal returns, from day L-5 to L+5, calculated
using the market model estimated from L-100 to L-6.









Table 11 The price effect of options introduction computed using a market model
estimated during the L+100/L+200 period

This table repeats the analysis of Table 2, with a market model estimated during the
L+100/L+200 period, where L represents the listing date. As with table 2, abnormal
returns are calculated using the Brown and Warner (1985) methodology. When more than
one stock became optioned on a given day, a portfolio of all such stocks is formed, and
subsequently treated as a single security The event window extends from L-5 to L+5.


Period Sample


Number of CAR(s) -- T-stat. --
Events Market Model Market Model


1973-1980 All stocks (calls and
put/call listings)
Call listings only


Put /Call listings only


Put listings only


Calls and put/calls, NYSE
stocks only


82 3.15 %


81 3.11%


9 0.25 %


82 3.15 %


4.624***


4.588***


0.192


4.624***


Calls and put/calls, -
NASDAQ stocks only


1981-1992 All stocks (calls and
put/call listings)
Call listings only


Put /Call listings only


Put listings only


Calls and put/calls, NYSE
stocks only
Calls and put/calls,
NASDAQ stocks only


377 0.07 %


155 0.29%


226 -0,06 %


9 0.96 %

244 0.53 %


179 -0.56%


0.190


0.526


-0.115


0.459

1.188


-0.820





89


Table 11. Continued

Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level
Insufficient data



Variable Definition:

SNumber of Events. Number of days during which one or more option contracts get
listed.

- CARs (Market Model): Cumulative abnormal returns, from day L-5 to L+5,
calculated using the market model estimated from L-100 to L-6.









Table 12: Cumulative Abnormal Returns (CARs) of
post-listing extended periods, estimated using ore-listing models


This table shows the cumulative abnormal returns of extended post-listing periods,
calculated using market models estimated in two different pre-listing periods, using the
Brown and Warner (1985) methodology. When more than one stock became optioned on
a given day, a portfolio of all such stocks is formed, and subsequently treated as a single
security. The event window extends from L-5 to L+5.


Listing Estimation Post-listing
Period Period Period


Number of CAR T-statistic
Events (Market)


1973-1980 L-194/L-100 L+6/L+100 82 10.46% -5.58***

L+100 /L+194 82 -11.17% -5.79***

L-100 / L-6 L+6 / L+100 83 -2.24% -1.20

L+100 / L+194 83 -3.98% -2.36**


1981-1992 L-194/L-100 L+6 / L+100 377 -13.33 % -10.42***

L+100 / L+194 377 -11.49% -9.07***

L-100 / L-6 L+6 / L+100 427 -10.48% -8.03***

L+100 / L+194 427 9.82% -7.85***


Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level





91


Table 12. Continued

Variable Definition:

Number of Events: Number of days during which one or more option contracts get
listed.

-CAR (Market): Cumulative abnormal returns, from day L-5 to L+5, calculated using
the market model estimated from L-100 to L-6.

L The listing date.








Table 13: Cross-sectional regressions between listing period returns
and two proxies for stock price manipulation: firm size and stock volatility.
Cumulative Abnormal Returns computed using the market model
estimated in the L-200/L-100 period.


This table is a test of the following joint hypothesis: (1) the listing period price effect and
abnormal turnover are due to stock price manipulation, and (2) option dealers manipulate
mostly small stocks, with large standard deviation. Standard errors are computed using a
White heteroskedasticity consistent variance-covariance matrix.

This table covers listing events occurring during 1973-1980.


Dep. Indep. Constant Constant Indep. Indep. N. Adj.
Variable Variable Coeff T-statistic Var. Var. Obs R
Coeff T-statistic

CAR LOG(CAP) 0.229 3.046*** -0.0210 -2.963*** 64 0.120

CAR SDR -0.078 -3.202*** 3.991 3.409*** 64 0.145



Significance Levels:

*** Significant at the 1% level
** Significant at the 5% level
* Significant at the 10% level

Variable Definitions:

- CAR: Cumulative Abnormal Returns, computed for the period ranging from L-5 to
L+5. Abnormal returns are calculated using the Brown and Warner (1985) market
model, estimated from L-200 to L-100, where L represents the listing date.

- CAP: The inflation-adjusted market capitalization, equal to the number of outstanding
shares times the average share price during the event window. Figures are
converted in 1983 constant dollars.

- LOG(CAP): The natural logarithm of CAP.

- SDR: The standard deviation of raw returns, computed over the pre-listing period. The
pre-listing period extends from L-100 to L-6.




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EKRQME271_31ZSU8 INGEST_TIME 2013-03-25T14:07:57Z PACKAGE AA00013595_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES