Journal of
Journal of Banking & Finance 19 (1995) 117-129
Futures trading, information and spot price
volatility: evidence for the FTSE-100 Stock
Index Futures contract using GARCH
Antonios Antoniou
a,*, Phil Holmes aab
a Centre for Empirical Research in Finance (CERF), Department of Economics, Brunei,
The University of West London, London, UK
b Department of Economics, University of Durham, Durham, UK
Received October 1992; final version receivedApril 1993
Abstract
This paper examines the impact of trading in the FTSE-100 Stock Index Futures on the
volatility of the underlying spot market. To examine the relationship between information
and volatility (as subject neglected in previous studies) the GARCH family of techniques is
used. The results suggest that futures trading has led to increased volatility, but that the
nature of volatility has not changed post-futures. The finding of price changes being
integrated pre-futures, but being stationary post-futures, implies that the introduction of
futures has improved the speed and quality of information flowing to the spot market.
Keywords:
Futures; Information;
JEL classification:
Volatility;
Speculation;
GARCH
G14; Gl; G15
1. Introduction
Almost
has been
since futures
concern
about
trading
began
the impact
at the Chicago
of futures
Board
of Trade
on the underlying
in 1865 there
spot
market.
A
* Corresponding
author. The authors gratefully acknowledge A. Damell, A. Foster, I. Garrett, J.
Hunter and J Rougier and two anonymous referees for helpful comments. We would also like to thank
G. Constantinides and A. Malliaris for fruitful discussions. The usual disclaimer applies.
0378-4266/95/%09.50
0 1995 Elsevier Science B.V. All rights reserved
SSDI 0378-4266(94)00059-X
118
A. Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) 117-129
major reason for this is the belief among market participants that speculators in
futures markets destabilise spot prices. In addition to generating much theoretical
discussion, the issue has attracted considerable empirical analysis and received the
repeated attention of policymakers.
This close scrutiny led to futures markets in the USA being subject to
substantial regulation (including, for example, the prohibition of trading in onion
futures). In spite of the volume of research and the long history of conflict
concerning whether futures stabilise or destabilise cash markets, futures trading is
still viewed with suspicion by spot market participants and policymakers alike. For
example, program and futures trading were blamed by some for the stock market
crash of October 1987, leading to suggestions that futures trading should be further
regulated, including, for example, higher margins. Further regulation may impact
negatively on the operation of financial markets and on economic welfare. Hence,
it is important to consider whether such action is desirable.
While much research has been undertaken on the stabilisation-destabilisation
question, there are problems associated with this work. Firstly, most studies relate
to commodity rather than financial futures. Clearly this is expected given the
relatively short time for which financial futures have been traded. However,
financial futures are being questioned in relation to the 1987 crash. Secondly, that
work undertaken on financial futures relates to the USA, rather than the UK.
Finally, the implications of previous studies for regulation are not as clear-cut as
some authors suggest. In particular, we shall argue that both the theoretical debate
and empirical testing fail to recognise the link between information and volatility,
leading to inappropriate policy implications.
This paper addresses these shortcomings by examining the impact of futures
trading on the FTSE-100 stock index, following its introduction
in May 1984,
utilising the Generalised Autoregressive
Conditional
Heteroscedastic
(GARCH)
family of statistical techniques. These techniques avoid methodological
problems
encountered in previous studies (to be discussed in Section 3) and enable the link
between information and volatility to be examined. The paper proceeds as follows.
The next section examines the theoretical debate concerning the impact of futures
trading. This is followed by a review of previous empirical studies, highlighting
the methodologies
used and results obtained. Empirical analysis relating to the
impact of futures trading in the FISE-100 stock index follows: the methodology
adopted, data used and results are set out. The final section provides conclusions
and a summary.
2. The theoretical debate
Concern over the impact of speculators on price volatility predates futures
trading. For example, John Stuart Mill, in discussing the progress of society argues
that:
A. Antonioy P. Holmes/Journal of Banking & Finance 19 (1995) 117-129
119
“The safety and cheapness of communications,
which enable a deficiency in one
place to be supplied from the surplus of another . . . render the fluctuations of
prices much less extreme than formerly.. . This effect is much promoted by the
existence of . . . speculative merchants. . . [T]he tendency of this operation [by
speculators]
is to equalise price, or at least to moderate its inequalities. . .
Speculators, therefore, have a highly useful office in the economy of society”
(1871, pp. 276-277).
Discussion of the impact of speculators intensified with the arrival of futures
trading, due to futures encouraging speculation. Indeed, speculators are required to
enable hedgers to transfer risks. Further, futures and spot prices are closely related,
yet futures trading imposes fewer costs than spot trading, making futures attractive
to speculators.
The opposing views on the impact of speculators are discussed by Kaldor
(1960, chapter l), and Friedman (1953). Kaldor, in line with Mill, points out that
traditionally speculation was viewed as a process which evens out price fluctuations. The idea that speculation might increase price fluctuations was not considered in the traditional view, since this requires that speculative activity results in
losses: selling when prices are low and buying when high. However, Kaldor
argues that this implies that speculative demand or supply is only a small part of
total demand or supply. If this is not true, it may be more profitable for individual
speculators to forecast the psychology of other speculators, rather than the trend of
the non-speculative
elements. Thus, argues Kaldor, speculation may produce a net
loss, with some speculators gaining, and at the same time destabilise the market.
While Friedman (1953) accepts this, he argues that while speculators may lose
money, it is hard to see why there is any presumption that they will. For Friedman
the presumption should be the opposite.
The debate about speculators and the impact of futures on spot price volatility
suggests that increased volatility is undesirable and reductions in volatility are
desirable. However, this is misleading as it fails to recognise the link between
information and volatility. Prices depend on the information currently available in
a market. As Cox (1976) argues, futures trading can alter the available information
for two reasons. First, futures attract additional traders to a market. Second, as
transaction costs in the futures market are lower than those in the spot market, new
information may be transmitted to the spot market more quickly. ’ The issue to be
addressed, then, is one of how the rate of information flow relates to spot price
volatility. This issue is addressed at the theoretical level by Ross (1989).
Ross assumes that there exists an economy that is devoid of arbitrage and
proceeds to provide a condition under which the no-arbitrage situation will be
’ Evidence on lead-lag relationships between stock and stock index futures prices certainly suggests
that new information is incorporated in futures prices first. See, inter alia, Kawaller et al. (1987) and
Chan (1992).
120
A. Antonioy
P. Holmes /Journal
of Banking & Finance 19 (I 995) I 17-129
sustained. He begins by assuming that asset prices are a martingale
represented by the following differential equation:
and can be
dp
- = ppdt + mPdzp
P
where p is the asset price with mean pP, standard deviation up and z N N(O,l).
Further, by assuming that prices are determined by a pricing standard, q, as in an
asset pricing framework for example, Ross (1989, p. 5, theorem 1) demonstrates
that expected returns will satisfy the following security market line expression:
-cov(p,q)
PJJ-”
(2)
where r is the risk free rate of interest.
information evolves according to:
In a similar fashion,
Ross assumes
that
ds
= psdt + usdz,
S
By imposing a terminal condition that at some future point in time, call this T,
the asset price will be such that p(T) = s(T), the following pricing relationship
holds:
from which Ross obtains the following
dp
_=__
P
Substituting
differential
equation:
ds
[ us - r + cov( q,s)]dt
s
Eqs. (1) and (3) into (5) yields:
Cl.,dt + apdzp = [r - cov( q,s)]dt
Using Eq. (2) in (6) and rearranging,
+ u,dz,
(6)
we obtain:
uPdz, = u,dz,
(7)
and therefore (Ross, 1989, p. 8, theorem 2):
UP = u s
(8)
Eq. (8) is Ross’s condition for no arbitrage, and implies that the variance of
price change will be equal to the rate (or variance) of information flow. The
implication of this is that the volatility of the asset price will increase as the rate of
information flow increases. If this is not the case, arbitrage opportunities will be
available. It follows, therefore, that if futures increase the flow of information,
then in the absence of arbitrage opportunities the volatility of the spot price must
change. Whether or not this actually happens is ultimately an empirical question
and we address this issue in the subsequent sections of the paper.
A. Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) I 17-129
121
3. Previous empirical studies ’
The general approach adopted in the literature is to examine spot price volatility
prior to the onset of futures trading and to compare this with spot price volatility
post-futures.
In analysing the behaviour of volatility pre- and post-futures it is
necessary to isolate influences not due to futures trading so that the impact futures
trading makes can be assessed more readily. This is typically achieved through the
inclusion of a proxy variable for which there is no related futures contract. Having
isolated “market-wide”
movements, the impact of futures trading is then captured
by the introduction of a dummy variable. Such a model is represented by the
following regression equation:
y, = a0+ qX, + a2DF + u,
where Y, is a constructed
measure of volatility for the spot market under
investigation,
X, is a proxy variable to capture “market-wide”
volatility, that is,
volatility unrelated to the onset of trading in a futures contract on the spot asset
under investigation, and DF is a dummy variable taking on the value 0 pre-futures
and 1 post-futures. If the futures dummy is statistically significant, then futures
trading has an impact on the volatility of the underlying spot market. Research on
the impact of financial futures has tended to be concentrated on two financial
instruments: the Government National Mortgage Association (GNMA) certificates
in the US, and stock index futures (again, predominantly
in the US>.
3.1. GNMA futures
Froewiss (1978) uses regression analysis to examine the variability of GNMA
prices relative to that of bond prices and finds no change after the introduction of
futures. Froewiss concludes that weekly spot price volatility has not been altered
by the introduction of futures. In contrast, Figlewski (1981) concludes that futures
trading in GNMA securities led to increased monthly price volatility.
Simpson and Ireland (1982) use regression analysis and a multivariate time
series model with an intervention term to analyse the impact of futures for daily
and weekly price changes. Their results suggest that futures did not affect spot
price volatility either on a daily or a weekly basis.
Intervention analysis is appropriate for examining the impact of futures according to Corgel and Gay (1984) as it allows a direct focus on the dynamic
characteristics of the response to the introduction of futures. Their results are in
line with Froewiss (1978).
’ This review is not meant to be exhaustive. Rather, it seeks to identify the most important work in
this area to date, together with the main techniques used to address the issue. The review, therefore, is
confined to empirical work relating to financial futures.
122
A. Antoniou, P. Holmes/Journal of Banking & Finance I9 (1995) I1 7-129
Moriarty and Tosini (1985) use the same volatility measure and regression
model employed by Figlewski (1981) to examine the validity of his results. They
extend the period analysed and in contrast to Figlewski (whose findings they refer
to as “unique in the futures literature”) their results suggest that the introduction
of GNMA futures did not cause cash market volatility to increase. They conclude
that the strength and significance of price relationships between cash and futures
markets depend critically on the sub-period analysed.
The influence of futures volatility on spot volatility is examined by Bhattacharya et al. (1986) using Granger’s methodology for testing for causality. While
they suggest that futures volatility has some causal influence on cash volatility,
they do not say whether futures have stabilised or destabilised the spot market.
3.2. Stock index futures
Edwards (1988a, b) examines stock market volatility before and after the
introduction of futures, and finds that it decreased post-futures for the S&P 500,
but finds no significant difference for the Value Line index.
Aggarwal(1988)
finds that while the post-futures period is more volatile, this is
true for all markets and hence stock index futures may not be the primary cause of
this increase in volatility. Similarly, Harris (1989) argues that support for the
hypothesis that trade in index futures increases cash market volatility is circumstantial. Harris believes that other index-related phenomena, such as the growth in
foreign ownership of American equities and the growth in index funds, could
account for the changes.
The frequency of jumps in daily stock returns is analysed by Becketti and
Roberts (1990) who conclude that stock market volatility is not related to either
the existence of, or the level of activity in, the stock index futures market.
Brorsen (1991) tests for homogeneity of variance for time periods before and
after futures and finds that while the variances of daily price changes are
significantly different, the variances of 5- and 20-day price changes are not.
To conclude this section, it is evident that the majority of studies discussed
above conclude that futures have had no discernible impact on spot price volatility.
For those studies that do find an impact, however, there is no decisive evidence to
suggest that futures either stabilise or destabilise the underlying spot market.
4. Methodology, data and results
4.1. Methodology
As discussed in the previous section, there is disagreement first on whether
futures actually have any impact on spot market volatility and second on whether
this impact (if any) stabilises or destabilises the underlying
spot market. The
A.
Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) 117-129
123
source of this disagreement lies mainly in the constructed measures of volatility
used, for as Board and Sutcliffe (1991) show, studies of volatility are sensitive to
the measure of volatility that is used. In addition, studies based on constructed
volatility measures implicitly
assume that price changes in spot markets are
serially uncorrelated and homoskedastic.
However, findings of heteroskedasticity
in stock returns are well documented (see inter alia Mandelbrot,
1963; Fama,
1965; and the review in Bollerslev et al., 1992). Inferences drawn from studies
failing to control for such dependence
are therefore unreliable.
Thus, while
observed differences in volatility may be due to the introduction of futures, they
may simply be the result of return dependence. Thus, the time period investigated
may significantly alter the results (see Moriarty and Tosini, 1985).
More importantly,
however, previous studies fail to make the connection
between information and volatility explicit. This connection is an important one
for, as demonstrated in Eq. (8), any change in the rate of information flow will
change the volatility of the price of the spot asset. Therefore, unless information
remains constant, volatility must be time varying, euen on a daily basis. A natural
way to capture the time varying nature of volatility is to model the conditional
variance as a GARCH process (Engle, 1982; Bollerslev, 1986; Engle and Bollerslev, 1986). In contrast to the estimation of regression equations such as Eq. (9) by
OLS, which requires the error term to be homoscedastic,
GARCH models the
conditional variance of the error term as a linear function of the lagged squared
residuals and the lagged residual conditional
variance. The advantage of a
GARCH model is that it captures the tendency in financial data for volatility
clustering. A model with errors that follow a GARCH ( p,q) process is represented
as:
Y,=a,+a,X,+e,,
i=
E,I!P-:_l-N(O,h,)
( IOa)
(lob)
I
j=l
where (lOa) is the conditional mean equation and (lob) is the conditional variance
equation. !Pt ~ I is the information set.
For a GARCH process to be well-defined it is necessary that both (Y~and p,
are non-negative.
Engle and Bollerslev (1986) put forward the integrated GARCH
(I-GARCH) as an extension of the GARCH model. With I-GARCH the model
specification is characterised by non-stationary
variables, such that any shock to
the variance of a process is permanent. For a process to be identified as I-GARCH
the parameters (Y; and pi in Eq. (lob) must together sum to unity. This implies
the presence of an approximate unit root in the autoregressive polynomial. Where
an approximate unit root is present, current information remains important for
forecasts of the conditional variances for all horizons.
In analysing the relationship between information, spot price volatility and the
impact of futures trading, there are two issues that need to be addressed. First,
124
A. Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) 117-129
does the existence of futures trading in itself have any effect on volatility? Second,
and perhaps more important,
if the existence of futures trading does affect
volatility, how does it, that is, what is the relationship between information and
volatility following the onset of futures trading? To address the first issue, we
augment the conditional variance equation with a dummy variable taking on the
value zero pre-futures and one post-futures. Thus, (lob) becomes:
h, = CY”+ i
i=l
cxi&
+ i
Pjhlpj + yDF
(11)
j=l
where DF is the dummy variable. If the dummy is statistically significant then the
existence of futures trading has had an impact on spot market volatility. To
address the second issue, the period under investigation
is partitioned into two
sub-periods relating to before and after futures trading began. GARCH models of
the form (lOa) and (lob) are estimated for both sub-periods, thereby allowing a
comparison of the nature of volatility before and after the onset of futures trading.
4.2. Data
Daily closing price indices for the period November 1980 to October 1991 are
used. The FTSE-100 stock index was introduced in 1984 to support the futures
contract on its introduction. Hence, a proxy for the FTSE-100 index had to be used
to allow comparison of pre- and post-futures periods. The candidates are the FT
All Share index, the FT 500 index and the FT 30 index. The construction of the
FT 30 is different from that of the FTSE-100. Hence, its use is inappropriate.
Since the non-synchronous
trading problem is most severe for the FT All Share
index and the correlation coefficient between it and the FTSE-100 is lower than
that for the FT 500 and the FTSE-100, the FT 500 is used.
As stated in the previous section it is necessary to remove market-wide
influences on spot price changes by incorporating a proxy variable in the mean
equation. None of the FT share indexes are suitable for this purpose since they are
all highly correlated with the FTSE-100 and it is necessary to have a proxy which
is not associated with a futures contract. Therefore, to capture market-wide
influences on price volatility the index on the Unlisted Securities Market (USM),
provided by Datastream, is used. After excluding non-trading days the daily time
series consists of 2709 observations:
883 relate to the period prior to the
introduction of futures and 1826 to the period post-futures.
4.3. Results
As shown in Table 1 the standard deviation of daily price changes for the FT
500 is higher for the post-futures period. In contrast, that for the USM index is
lower for this period. Hence, while volatility in the market without futures is lower
A. Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) I1 7-129
Table 1
Means and standard deviations of first differences
November 1980-October
1991
125
of the log of the FT 500 and the USM indexes,
Period a
n
FT500
Mean
Standard Deviation
Mean
Standard Deviation
1980-1991
1980-May 1984
May 1984-1991
2709
883
1826
0.00054
0.00072
0.00045
0.00950
0.00935
0.00958
-0.00011
0.00008
-0.00019
0.01129
0.01376
0.00987
a Excluding
Bank Holidays
USM
and other non-trading
days.
in the later period, the volatility of the spot market underlying the futures contract
has increased on the basis of this measure. However, inferences cannot be drawn
from these figures and further investigation is required. GARCH (p,q)
equations,
as shown below, are estimated for all combinations of p = 1, 2, 3, 4, 5 and q = 1,
2, 3, 4, 5.
R;=a,+a,R;+r,,
h, = (Y,, + i
i=l
E, I9,_1 -N(O,h,)
c&i
+
i
fijh,_j
+
yDF
( =a)
( 12b)
j=l
where Rf is the daily change in log prices for the FT 500 index, Ry is the daily
change in log prices for the USM index and DF is a dummy with value 0 for the
pre-futures period and 1 for the post-futures period.
Log likelihood ratio tests indicate that GARCH (1,l) is the most parsimonious
representation of the variance for all periods considered. ’
Table 2 shows the equations estimated and results for the whole period. The
model is estimated with and without a dummy for the October 1987 crash in the
mean equation (12a). In addition, a dummy for Big Bang was included, but found
to be insignificantly
different from zero. It was therefore excluded from the final
estimations. All parameters included are statistically significant at the 5% level.
The positive coefficient on the futures dummy suggests that the onset of futures
trading resulted in increased spot price volatility. While spot price volatility may
have increased as a result of the onset of futures trading, the analysis thus far does
not enable us to examine the reason for this change.
Table 3 reports results for the sub-periods. The post-futures model is estimated
with and without a dummy relating to the crash. For both pre- and post-futures
trading the GARCH parameters are significantly
different from zero at the 5%
level, with the exception of the constant term pre-futures. The increased volatility
suggested in Table 2 is investigated further by examining the behaviour of the
parameters in the GARCH equations for the two sub-periods.
3 In the interests of brevity, these results are not reported here. They are available
upon request.
from the authors
126
A. Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) 117-129
Table 2
R: = a0 + a,R: + a2 DC + E,
h,=(Yg+(Y1e:_,+Plh,-l+yDF
where DC is a dummy taking on the value 1 for the period around the 1987 stock market crash, and
zero otherwise.
a0
01
a2
a0
aI
PI
Y
0.7750 a
(5.32)
0.963 a
(6.52)
0.3384
(25.70)
0.3276
(25.30)
_
0.3969 b
(5.49)
0.4350 b
(5.56)
0.0862
0.8423
(54.97)
0.8190
(50.57)
0.1070 b
(2.70)
0.1171 b
(2.39)
- 0.0504
(-43.15)
(9.44)
0.0991
(13.13)
n = 2709
a Coefficients multiplied by lo3 for readability.
b Coefficients multiplied by lo5 for readability.
All parameters are statistically significant at the 5% level. Figures in parentheses
are t-statistics.
The first point to note in comparing results for before and after the onset of
futures trading is that the onset of futures has not led to a change in the nature of
volatility. For the periods before and after the onset of futures trading a GARCH
(1,l) representation is the most appropriate form of the model. The large increase
in LX,-,post-futures (indeed (Ye is not significantly different from zero pre-futures)
together with the changes in cxl and PI indicate that there has been an increase in
the unconditional
variance. The unconditional
variance, given by cu,/(l - (YePI), is 0.0000543 pre-futures and 0.0000787 post-futures (0.0000732 with the
crash dummy). This is consistent with more information being transmitted to the
market as a result of the onset of futures trading.
Similarly, the value of (Ye has increased post-futures,
again suggesting an
increase in volatility. (Ye is the coefficient relating to the lagged squared error
Table 3
Ri = a, + a,R: + a,D,
+ E,
h,=crO+cxIe:_,+&h,-l
Pre-futures
n = 883
Post-futures
n = 1826
Post-futures
n = 1826
a0
aI
0.642 a
(2.58)
0.842 a
(4.53)
0.778 a
(4.01)
0.4082
(23.62)
0.2107
(7.83)
0.1839
(7.46)
Augmented Dickey-Fuller
statistics:
Pre-futures - 3.33.
Post-futures (without 0,) - 10.56.
Post-futures (with 0,) - 7.75.
Definitions and footnotes as in Table 2.
a2
-0.0337
(- 23.57)
%
a/
PI
0.1151 a
(1.48)
0.6481 b
(6.73)
0.5662 a
(5.65)
0.0404
(2.73)
0.1178
(7.49)
0.1177
(9.07)
0.9384
(38.31)
0.7999
(39.34)
0.8050
(38.46)
A. Antoniou, P. Holmes/Journal
of Banking & Finance 19 (1995) 117-129
121
term. In the context of this analysis the lagged error term relates to changes in the
spot price on the previous day which are attributable to market-specific factors, i.e.
non-market-wide
factors. Assuming that markets are efficient, then these price
changes are due to the airival in the market of items of information which are
specific to the pricing of the FT 500. Hence, crl relates to the impact of
yesterday’s market-specific
price changes on price changes today. Given that this
relates to the arrival of information yesterday, cri can thus be viewed as a “news”
coefficient, with a higher value implying that recent news has a greater impact on
price changes. Thus the increase in (Ye post futures suggests that information is
being impounded in prices more quickly due to the introduction of futures trading.
Just as (Y, reflects the impact of recent news, p, can be thought of as
reflecting the impact of “old news”. p1 is the coefficient on the lagged variance
term and as such is picking up the impact of price changes relating to days prior to
the previous day and thus to news which arrived before yesterday. The increase in
the rate of information flows to be anticipated from the onset of futures trading is
expected to lead to a reduction in uncertainty regarding previous news. This in
turn will lead to a fall in the persistence of information.
In other words, “old
news” will have less impact on today’s price changes. This view is confirmed by
the fall in the value of p1 post-futures.
This conclusion is given further support by the fact that the pre-futures model is
a candidate for I-GARCH, whereas the post-futures model is not obviously so.
Pre-futures
(Y, and p, sum to 0.98, compared to 0.92 post-futures. ADF tests
were carried out to test for I-GARCH and reveal that while the pre-futures sample
is integrated at the 1% level, the post-futures
model is stationary. Thus the
persistence of shocks decreased since the onset of derivative trading. To illustrate
the impact of futures trading on the rate at which information is incorporated into
spot prices, consider a shock of 3 standard deviations to the volatility of the spot
market, where the standard deviation used is that of the pre-futures series. Fig. 1
plots the effect of the shock to both the pre- and post-futures periods. It is clear
that the immediate impact of the shock is much greater and less persistent in the
post-futures period than in the pre-futures period. This reaction is entirely consistent with the results discussed above. All the findings regarding changes in
GARCH parameters suggest that the spot market is more volatile post-futures, but
that this is the result of an increase in the rate of flow of information to the market.
These results are consistent with the theoretical arguments of Ross (1989) and the
view that futures trading increases the flow of information to the spot market.
5. Summary
and conclusions
There has long been a debate on the impact of speculation on price volatility.
While this debate preceded the introduction
of futures, trading in derivative
securities intensified concern over the role of speculators. Previous studies sought
to examine the impact of futures by modelling the volatility of prices for periods
128
A. Antoniou, P. Holmes/Journal
0 .QO
1
Before
.
,
5
.
of Banking & Finance 19 (I 995) I I 7-129
.
. ,
10
Time (Trading
fifter
Fig. 1. Effect of a 3 SD. shock on spot price volatility
“‘7 .-.::... .......i .......~_ ._......_
___._
15
20
Days)
i
before and after the onset of futures trading.
before and after the introduction of futures. However, these studies largely ignored
the interdependence
of the time series of returns in speculative markets, i.e. large
changes in prices are followed by large changes, and small by small. For this
reason it is more appropriate to analyse volatility using GARCH which allows for
time varying variance in a process.
More importantly,
previous studies largely ignored the relationship between
information
and volatility. Thus increasing volatility has been seen as a “bad
thing” and the fact that it may be a direct result of an increase in the rate of flow
of information has not previously been acknowledged.
Hence, previous studies
have failed to distinguish between message and messenger.
The results presented here for the impact of trading in the FISE-100
index
futures contract suggest there has been an impact on spot price volatility. In
particular, the variance of price changes pre-futures was integrated, suggesting
shocks (i.e. items of news> have a permanent effect on price changes, whereas the
post-futures sample is stationary. The results suggest that futures trading improves
the quality and speed of information flowing to spot markets. This is confirmed by
the increase in the news coefficient
(a,) of the GARCH equation and the
reduction in the persistence coefficient ( &).
Hence the evidence suggests that there has been an increase in spot price
volatility on a daily basis, but that this is due to increased information in the
market and not to speculators having adverse destabilising effects. Indeed, this
increased volatility appears to be the result of futures trading expanding the routes
over which information can be conveyed to the market.
A. Antoniou, P. Holmes /Journal
of Banking & Finance 19 (I 995) I1 7-129
12Y
References
R.. 1988, Stock index futures and cash market volatility,
Aggarwal,
Review of Futures Markets
7.
290-299.
Baldauf, B. and G.J. Santoni,
1991, Stock price volatility:
Some evidence from an ARCH
model.
Journal of Futures Markets 11, 191-200.
Becketti.
S. and D.J. Roberts, 1990, Will
increased regulation of stock index futures reduce stock
market volatility? Federal Reserve Bank of Kansas City Economic Review Novcmber/Decembcr,
33-46.
Bhattacharya, A.K..
A. Ramjee and B. Ramjee,
1986, The causal relationship between futures price
volatility and the cash price volatility of GNMA
securities, Journal of Futures Markets 6. 2Y-3Y.
Board, J. and C. Sutcliffc, 1991, Information, volatility, volume and maturity: An investigation of stock
index futures, University of Southampton Discussion Paper No. 91-S.
Bollerslev, T.,
1986, Generalised autoregressive conditional heteroscedasticity. Journal of Economct-
rics 33, 307-327.
Bollerslev. T., 1987. A conditional heteroscedastic time series model for speculative prices and rates of
return, The Review of Economics and Statistics August, 542-547.
Bollerslev. T., R.Y. Chou and K.F. Kroner, 1992, ARCH
modeling in finance: A review of the theory
and empirical evidence, Journal of Econometrics 52, S-59.
Brorsen, B.W.,
1991, Futures trading, transactions costs, and stock market volatility, Journal of Future\
Markets 11, 153-163.
Chan. K., 1992, A further analysis of the lead-lag relationship between the cash market and stock index
futures market, Review of Financial Studies 5, 123-152.
Corgcl, J.B. and G.D.
AREUEA
Gay.
1984, The impact of GNMA
futures trading on cash market volatility,
Journal 12, 176-190.
Cox, C.C., 1976. Futures trading and market information, Journal of Political Economy 84, 1215~1237.
Edwards, F.R., 1988a. Does futures trading increase stock market volatility? Financial Analysts Journal
44, 63-h’).
Edwards, F.R., 1988b. Futures trading and cash market volatility: Stoch index and intcrcst rate futures.
Journal of Futures Markets 8, 421-439.
Engle,
R.F..
1982, Autoregressive
conditional
heterosccdasticity
with cstimatcs of the variance of
United Kingdom inflation, Econometrica 50. 987-1008.
Englc, R.F. and T. Bollerslev,
1986. Modelling
the persistence of conditional variances. Econometric
Reviews 5, I -SO.
Fama, E.F.,
lY65, The behaviour of stock market prices, Journal of Business 38, 34-105.
Figlewski.
S..
1981.
Futures trading and volatility
in the GNMA
market.
Journal of Finance 36.
445-456.
Friedman, M., 1953, Essays in positive economics (University
Froewiss.
K.C.,
1978,
GNMA
futures: Stabilizing
of Chicago Press, Chicago).
or destabilizing?
Federal Reserve Bank of San
Francisco Economic Review, Spring, 20-29.
Harris, L.. 198’). S&P 500 cash stock price volatilities, Journal of Finance 44, 1155~1 175.
Kaldor, N.. 1960, Essays on economic stability and growth (Duckworth
Kawaller.
and Co.).
I.. P. Koch, and T. Koch. 1987. The temporal price relationship between S&P 500 future\
and S&P 500 index, Journal of Finance 42, 1309-1329.
Mandelbrot,
B., 1963, The variation of certain spcculativc prices. Journal of Business 36. 304-41(J).
Mill, J.S., 1871. Principles of political economy II, 7th ed. (Longmans. Green, Reader and Dyer).
Moriarty.
E.J. and P.A. Tosini, 1985. Futures trading and the price volatility of GNMA
certificates -
Further evidence, Journal of Futures Markets 5. 633-641.
Ross, S.A..
1989,
Information
and volatility:
The no-arbitrage
martingale
approach to timing and
resolution irrelevancy, Journal of Finance 44, t-17.
Simpson, W.G. and T.C. Ireland, 1982, The cffcct of futures trading on the price volatility of GNMA
sccuritics, Journal of Futures Markets 2, 357-366.
Stall.
H.R.
and R.E. Whaley,
Portfolio Management,
1988.
Volatility
Winter, 20-22.
and futures: Message versus messenger, Journal of