International Research Journal of Finance and Economics

ISSN 1450-2887 Issue 10 (2007)

EuroJournals Publishing, Inc. 2007
http://www.eurojournals.com/finance.htm
Testing the General Equilibrium Model of Stock Index Futures:
Evidence from the Asian Crisiss
Janchung Wang
Department of Financial Operations
National Kaohsiung First Universityof Science and Technology
1 University Road, Yanchao,Kaohsiung 824, Taiwan, ROC
Tel: 886-7-6011000ext3112; Fax: 886-7-6011039
E-mail: janchung@ccms.nkfust.edu.tw
Abstract
Hemler and Longstaff (1991) followed the CIR (Cox, Ingersoll and Ross, 1985a&b)
framework and developed a closed-form general equilibrium model of stock index futures
with both stochastic market volatility and interest rates. Using the Nikkei 225, the Hang
Seng, and the KOSPI 200 (South Korean composite stock index 200) index futures
contracts, this study represents the first attempt to apply the Hemler and Longstaff (1991)
model to the case of the Asian crisis, and compares the power of the cost of carry and
Hemler-Longstaff models in predicting stock index futures prices. The empirical results
from the Hang Seng and the KOSPI 200 futures markets imply that in highly volatile
periods, such as the Asian crisis period, incorporating stochastic market volatility into the
pricing model helps in predicting stock index futures prices. Therefore, when selecting a
pricing model to estimate the theoretical values of stock index futures, practitioners should
identify the degree of market volatility for the markets they are participating in.
Keywords: Asian Financial Crisis, Pricing of Stock Index Futures, Stochastic Volatility,
EGARCH
JEL Classifications: C00, G13.
1. Introduction
Until now, the cost of carry model has been the most widely used model for pricing stock index
futures. This model was based on the assumption of perfect markets with non-stochastic interest rates.
However, if interest rates are stochastic, Cox, Ingersoll, and Ross (1981) showed that futures and
forward prices cannot be equal. Some studies [e.g., Cornell and French (1983a&b), Modest and
Sundaresan (1983), Figlewski (1984), Eytan and Harpaz (1986), and Gay and Jung (1999)] found
significant deviations between actual futures prices and theoretical prices estimated by the cost of carry
model. Additionally, the cost of carry model implies that market volatility should not have explanatory
power for futures prices. However, Resnick and Hennigar (1983), Kamara (1988), and Fung and
Draper (1999) found that pricing deviations from the cost of carry model are a function of the volatility
of the underlying security price. Motivated by these considerations, Hemler and Longstaff (1991)
followed the CIR (Cox, Ingersoll and Ross, 1985a&b) framework and developed a closed-form general
equilibrium model of stock index futures prices with both stochastic market volatility and interest rates
(hereafter the Hemler-Longstaff model).
International Research Journal of Finance and Economics - Issue 10 (2007) 108
Moreover, the empirical tests in the study of Hemler and Longstaff also indicated that their
model is superior to the cost of carry model for the NYSE futures contract when the October 1987
observation is included in the sample. However, when October 1987 is excluded, the cost of carry
model has lower pricing errors. Hemler and Longstaff (1991) pointed out that the difference in results
occurs because the October 1987 market variance estimate is much larger than the variance estimates
for the other months. This implies that the Hemler-Longstaff model appears particularly suitable for
stock markets with high price volatility. Tuluca, Zwick, and Seiler (2003) pointed out that the Asian
crisis variances of daily returns are larger than the pre-crisis daily variances for Asian stock markets.
This study also finds that for the Hong Kong and the Korean stock markets, the Asian crisis period has
clearly higher stock price volatility than the other periods (see Table 2). Thus the question is whether
the Hemler-Longstaff model with stochastic market volatility can effectively predict stock index
futures prices during the Asian crisis. Using the Nikkei 225, the Hang Seng, and the KOSPI 200 (South
Korean composite stock index 200) index futures contracts, this study represents the first attempt to
apply the Hemler-Longstaff model to the case of the Asian crisis, and compares the predictive power of
the cost of carry and Hemler-Longstaff models.
1
2. Pricing Models of Stock Index Futures
One common way of calculating futures prices is to use the cost of carry model. If interest rates and
continuous dividend yields are non-stochastic, then in the absence of taxes and other market
imperfections, the cost of carry model can be written as
F
t
= S
t
e
(r - q)(T - t)
(1)
where F
t
is the theoretical futures price at time t; S
t
denotes the current stock index; r represents the
risk-free interest rate; q is the dividend yield; and T-t denotes the time to expiration.
If the underlying stock index pays irregular lumpy dividends, under the concept of continuous
compounding, the cost of carry model will be
F
t
(S
t
D
t
)e
r(T-t)
(2)
D
t
n
i t i
i i t
p
w d S
1 ,
) ((
) ( t t r
i
e )
where D
t
is the sum of the present values of all cash dividends distributed by the underlying
component stocks at time t during the life of the futures contract; d
i
is the cash dividend per share for
stock i during the life of the futures contract; w
i
is the weight of stock i in the index; t
i
is the time that
stock i pays the cash dividend; and p
i,t
is the price of stock i at time t.
For the Hang Seng index futures, following Fung and Draper (1999), the theoretical futures
prices can be estimated by the cost of carry model with continuous dividend yields (that is, Equation
(1)). Furthermore, in Japan and Korea, the cash dividend payouts are relatively lumpy.
2
Thus,
substituting the known risk-free rate r and the lumpiness of cash dividend D
t
, together with the current
index price S
t
and time to maturity Tt, into the cost of carry model (2), the theoretical prices of the
Nikkei 225 and the KOSPI 200 stock index futures can then be obtained.
The cost of carry model was developed under the assumption of non-stochastic interest rates.
Moreover, the cost of carry model implies that market volatility should not have explanatory power for
futures prices. Hemler and Longstaff (1991) followed the framework of CIR and developed a closed-
form general equilibrium model of stock index futures with both stochastic market volatility and
interest rates. The general equilibrium model of stock index futures is given by Equation (15) in
Hemler and Longstaff (1991). The natural logarithm of the general equilibrium model yields the
following regression equation
1
Although the Hemler-Longstaff model with both stochastic market volatility and interest rates has been developed, empirical evidence remains scarce.
To our knowledge, this is the first such study of the Asian crisis.
2
In Japan, cash dividends are clustered at the end of March and September. As for the Korean stock market, most of the companies pay dividends at the
end of the year.
109 International Research Journal of Finance and Economics - Issue 10 (2007)
L
t
r
t
V
t t
(3)
where L
t
= ln(
q
t
e F S
t
); F
t
is the theoretical futures price; denotes the time to expiration (i.e., T-
t); and V
t
represents the variance of stock index returns. With a lumpiness of cash dividends, L
t
will be
ln(F
t
/(S
t
-D
t
)). Substituting the data r
t
, V
t
and the coefficient estimates , , and into the regression
model (3) to generate the L
t
estimate, the theoretical futures price can then be inferred from L
t
.
Additionally, if 0, T-t, and 0 (that is, market volatility should not have explanatory
power) hold, Equation (3) reduces to
L
t
r
t
(4)
Equation (4) can be rearranged to demonstrate (1), the cost of carry model. Thus, the cost of
carry model can be regarded as a special case of the Hemler-Longstaff model.
3. Comparison of the Cost of Carry and the Hemler-Longstaff Models
3.1. Data and Basic Statistics
The Nikkei 225 index futures based on the Nikkei 225 stock index began trading on the Osaka
Securities Exchange (OSE) on September 3, 1988. The futures contract is a major stock index futures
contact in Japan. The Hang Seng index futures contract is based on the Hang Seng index, which
comprises 33 representative stocks listed on the Stock Exchange of Hong Kong. The KOSPI 200 index
futures contract is based on the KOSPI 200 stock index. The KOSPI 200 index is a value-weighted
index comprising 200 leading stocks which represent over 70% of the total market capitalization of the
Korean Stock Exchange. Table 1 describes the main features of the three futures contracts and their
underlying indexes.
For the three stock index futures, the nearest maturity contracts all have significant trading
volume. To reduce thin trading problems, this study only considers the near-month contracts. The daily
data cover the period from July 2, 1997 through December 31, 2005. To capture different market
conditions during the sample period, two approximately equal length sub-samples were considered.
The first and second sub-periods are from July 2, 1997 to December 31, 2001 and from January 1,
2002 to December 31, 2005, respectively. Additionally, the Asian financial crisis period is defined as
being from July 2, 1997 to September 30, 1998.
3
For the Nikkei 225 futures, the three-month Gensaki rates are used as the risk-free interest
rates. The dividend data of the underlying component stocks in the index were obtained from the
QUICK Research Institute and the Tokyo Stock Exchange. For the Hang Seng stock index futures, the
middle quotes of one-month Hong Kong inter-bank rates are
3
Although pinpointing the start of the Asian financial crisis is difficult, analysts often point to the 17% devaluation of the Thai baht on July 2, 1997 as the
triggering event. From July of 1997 to September of 1998, stock markets in the Asia-Pacific region displayed dramatic declines. For example, the
Nikkei 225 index dropped from 20575 on July 28, 1997 to 13406 on September 30, 1998. Meanwhile, the KOSPI 200 index dropped from 82.38 on July
4, 1997 to 33.17 on September 22, 1998. It was not until October 1998 that most Asian governments made effort to restore the confidence of investors.
This led to a strong improvement in the stock markets. Thus, following the wide definition of the Asian crisis of Baig and Goldfajn (1998), this study
defines the Asian crisis period as the period from July 2, 1997 to September 30, 1998.
International Research Journal of Finance and Economics - Issue 10 (2007) 110
Table 1: Main features of the Nikkei 225, the Hang Seng, and the KOSPI 200 index futures contracts
Nikkei 225 futures Hang Seng futures KOSPI 200 futures
1.Opening date September 3, 1988 May 6, 1986 May 3, 1996
2.Underlying index Nikkei Stock Average
(Nikkei 225)
Hang Seng ifndex KOSPI 200 index
KRW$500,000
4.Contract months March, June, September,
December cycle (five contract
months open at a time)
the spot month, the next calendar
month, and the next two calendar
quarterly months
March, June, September and
December
5.Minimum price
change
HK$50 per contract 0.05 index point
KRW$25,000 per contract
6.Last trading day the business day prior to the
second Friday of each
contract month
The business day immediately
preceding the last business day of
the contract month
the second Thursday of the
contract month
7.Settlement cash cash cash
Source: Osaka Securities Exchange (OSE), Hong Kong Exchanges and Clearing Limited (HKEx), and Korea Exchange (KRX).
used as the substitute of risk-free interest rates. The annualized month-end dividend yields of Hang
Seng index for the same period were obtained from the Hang Seng Index Services Ltd. In the case of
the KOSPI 200 futures, the three-month inter-bank rates are used as the substitute of risk-free interest
rates. The dividend data of the underlying component stocks in the index come from the Korea Stock
Exchange.
Table 2 lists descriptive statistics for the spot and futures return series. During the Asian crisis,
the daily standard deviations are 1.77%, 3.08%, and 3.52% for the Nikkei 225, the Hang Seng, and the
KOSPI 200 indexes, respectively. Furthermore, Table 2 also indicates that for the Nikkei 225 index,
the price volatility for the Asian crisis period is only slightly larger than for the other periods.
However, for the Hang Seng and the KOSPI 200 indexes, the Asian crisis period has clearly higher
price volatility than the other periods. Additionally, in Table 2, the Jarque-Bera statistics for all return
series are statistically significant at the 1% level. Thus, as is commonly found for daily equity returns,
normality is significantly rejected.
Table 2: Descriptive statistics
Spot Returns Futures Returns
Mean Std. Dev. Skewness Kurtosis Jarque-
Bera
Mean Std. Dev. Skewnes
s
Kurtosis Jarque-Bera
Nikkei 225
Full sample -0.00011 0.0149 -0.0272 4.8464 294.72*** -0.00011 0.0154 -0.0925 4.9084 317.52***
Period 1 -0.00059 0.0164 0.0669 4.8720 161.44*** -0.00059 0.0171 0.0226 4.7792 145.18***
Period 2 0.00044 0.0131 -0.1639 3.9965 44.61*** 0.00044 0.0132 -0.2833 4.2090 72.28***
Crisis period -0.00133 0.0177 0.1483 4.6700 36.92*** -0.00133 0.0188 0.0046 4.4508 27.01***
Hang Seng
Full sample 0.00001 0.0180 0.1773 13.3513 9341.88*** 0.00001 0.0204 0.3805 14.6782 11926.97***
Period 1 -0.00026 0.0227 0.1863 9.7355 2097.07*** -0.00027 0.0259 0.3845 10.6159 2700.15***
Period 2 0.00028 0.0103 0.0771 4.0481 46.02*** 0.00029 0.0115 -0.0908 4.2796 68.48***
Crisis period -0.00212 0.0308 0.4266 8.7388 433.39*** -0.00220 0.0351 0.6085 9.8717 627.02***
KOSPI 200
Full sample 0.00039 0.0243 0.0619 5.9516 760.01*** 0.00037 0.0284 0.3532 7.2761 1635.76***
Period 1 0.00008 0.0296 0.1225 4.7535 144.06*** 0.00005 0.0354 0.4076 5.4819 313.64***
Period 2 0.00072 0.0165 -0.2352 4.4549 96.15*** 0.00073 0.0176 -0.3679 4.8448 162.23***
Crisis period -0.00263 0.0352 0.4911 4.5741 43.89*** -0.00284 0.0477 0.7532 4.7087 66.16***
Note: *** denotes significance at the 1% level.
111 International Research Journal of Finance and Economics - Issue 10 (2007)
3.2. Methodology
One major approach to evaluating model pricing performance is via the most popular accuracy
measures, mean percentage errors (MPE) and mean absolute percentage errors (MAPE). MPE and
MAPE are defined as follows:
MPE =
n
t t
t t
AF
F AF
n
1
1
(5)
MAPE =
n
t t
t t
AF
F AF
n
1
1
(6)
where AF
t
is the actual futures price at time t. F
t
is the theoretical futures price estimated by the cost of
carry model or the Hemler-Longstaff model.
Additionally, for the Hemler-Longstaff model the only variable that cannot be directly observed
is the volatility of the underlying index (V
t
). To accommodate time-varying volatility in index returns,
estimators based on the moving average method or the GARCH model are commonly employed. This
study employs the exponential GARCH model (EGARCH) modified by Nelson (1991) to estimate V
t
.
4
The exponential GARCH model of Nelson (1991), which allows for asymmetric volatility
response to stock price shocks, is specified as follows:
t t
R (7)
t t-1
,
t-2
, N(0, h
t
)
2
) ln( ) ln(
1
1
3
1
1
2 1 1 0
t
t
t
t
t t
h h
h h
(8)
where R
t
is the spot index return on day t;
t
denotes an error term; and h
t
is the conditional variance on
day t. In the EGARCH (1,1) Model, since 0
2
, negative stock price shocks (i.e., 0
1 t
) will have
a greater impact on the conditional variance,
t
h , than positive stock price shocks (i.e., 0
1 t
).
Furthermore, if 0
3
and
2
1
1
t
t
h
, then the lagged residuals will have a positive impact on the
conditional variance.
Finally, to compare the relative performance between the cost of carry and the Hemler-
Longstaff models, the t-test was used to test whether the MAPE statistics generated from each model
were significantly different.
4. Empirical Results
Table 3 summarizes the pricing errors of the two models. First, the percentage error column shows
that the cost of carry model tends to overprice all three futures contracts. This finding is consistent with
the findings of previous studies, for example Bailey (1989) for the Nikkei 225 contract. Next, Table 3
shows that for all of the markets and periods the magnitude of MPE of the cost of carry model clearly
exceeds that of the Hemler-Longstaff model. Particularly in the Asian crisis period, the MPE measures
of the cost of carry model are extremely high, ranging from -0.0675% (Nikkei 225) to as high as -
2.2014% (KOSPI 200). However, the MPE values of the Hemler-Longstaff model are 0.0053% and -
0.0692% for the Nikkei 225 and the KOSPI 200 index futures, respectively.
Table 4 further classifies the percentage errors as either premiums or discounts for the Asian
crisis period.
5
The Hemler-Longstaff model displays no clear difference between the number and

4
This study also uses the GARCH(1,1) model to estimate Vt. The EGARCH (1,1) Model has slightly better pricing performance than the GARCH(1,1)
model.
5
If actual futures prices are higher than theoretical futures prices, the percentage errors are classified as premiums. Conversely, percentage errors are
classified as discounts.
International Research Journal of Finance and Economics - Issue 10 (2007) 112
magnitude of premiums and discounts for all three markets. However, as for the cost of carry model, in
terms of magnitude, discounts are about 195.40 percent and 384.25 percent of premiums for the Hang
Seng and the KOSPI 200 markets, respectively. Overall, for the cost of carry model, the Asian crisis
period shows that the magnitude of the discounts obviously exceeds that of the premiums for the Hang
Seng and the KOSPI 200 markets with high stock price volatility.
Table 3: Summary statistics for the pricing errors of the cost of carry and Hemler-Longstaff models
Cost of carry model Hemler-Longstaff model
Percentage error Absolute percentage
error
Percentage error Absolute percentage
error
Mean (%) Std (%) Mean (%) Std (%) Mean (%) Std (%) Mean (%) Std (%)
Nikkei 225
Full sample -0.0421 0.3552 0.2579 0.2477 0.0001 0.3550 0.2565 0.2453
Period 1 -0.0390 0.4013 0.2920 0.2779 0.0010 0.4008 0.2901 0.2764
Period 2 -0.0456 0.2945 0.2193 0.2016 -0.0006 0.2938 0.2177 0.1972
Crisis period -0.0675 0.4767 0.3304 0.3497 0.0053 0.4746 0.3225 0.3478
Hang Seng
Full sample -0.1310 0.6309 0.4056 0.5006 -0.0051 0.6091 0.4001 0.4592
Period 1 -0.1912 0.7894 0.5169 0.6263 -0.0092 0.7599 0.5048 0.5679
Period 2 -0.0633 0.3695 0.2805 0.2485 -0.0037 0.3743 0.2846 0.2429
Crisis period -0.4690 1.0972 0.8053 0.8797 -0.0060 1.0504 0.6943 0.6753
KOSPI 200
Full sample -0.3877 1.7763 1.0014 1.5173 -0.0206 1.5733 0.9481 1.2444
Period 1 -0.6535 2.3546 1.5448 1.8929 -0.0646 2.0875 1.4071 1.4620
Period 2 -0.0904 0.5642 0.3937 0.4139 0.0071 0.5511 0.3751 0.4037
Crisis period -2.2014 3.4802 3.0221 2.7949 -0.0692 2.9917 2.4158 1.7605
Table 4: Summary statistics for the premiums and discounts for the Asian crisis period
Cost of carry model Hemler-Longstaff model
Number
of
premiums
Number
of
discounts
Mean
premium
(%)
Mean
discount
(%)
Number
of
premiums
Number
of
discounts
Mean
premium
(%)
Mean
discount
(%)
Nikkei 225 123 185 0.3301 -0.3298 153 155 0.3299 -0.3151
Hang Seng 105 204 0.4955 -0.9682 177 132 0.6008 -0.8191
KOSPI 200 114 192 1.0868 -4.1761 157 149 2.2869 -2.5517
Figures 1 to 3 further plot the premiums and discounts for two alternative models. Clearly,
during the Asian crisis, the cost of carry model tends to seriously overprice the Hang Seng and the
KOSPI 200 futures contracts relative to the Hemler-Longstaff model. To summarize, as can be seen
from Tables 3 to 4 and Figures 1 to 3, in highly volatile periods, such as the Asian crisis period, the
magnitude of MPE of the cost of carry model is much larger than that of the Hemler-Longstaff model.
113 International Research Journal of Finance and Economics - Issue 10 (2007)
Figure 1: Percentage Errors of Two Alternative Pricing Models during the Asian Crisis for the Nikkei 225
Futures
-0.020
-0.015
-0.010
-0.005
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
07/02/97 08/02/97 09/02/97 10/02/97 11/02/97 12/02/97 01/02/98 02/02/98 03/02/98 04/02/98 05/02/98 06/02/98 07/02/98 08/02/98 09/02/98
Time
cost of carry model
Hemler-Longstaff model
Figure 2: Percentage Errors of Two Alternative Pricing Models during the Asian Crisis for the Hang Seng
Futures
-0.070
-0.060
-0.050
-0.040
-0.030
-0.020
-0.010
0.000
0.010
0.020
0.030
0.040
07/03/97 08/03/97 09/03/97 10/03/97 11/03/97 12/03/97 01/03/98 02/03/98 03/03/98 04/03/98 05/03/98 06/03/98 07/03/98 08/03/98 09/03/98
Time
cost of carry model
Hemler-Longstaff model
Figure 3: Percentage Errors of Two Alternative Pricing Models during the Asian Crisis for the KOSPI 200
Futures
-0.160
-0.140
-0.120
-0.100
-0.080
-0.060
-0.040
-0.020
0.000
0.020
0.040
0.060
0.080
0.100
07/02/97 08/02/97 09/02/97 10/02/97 11/02/97 12/02/97 01/02/98 02/02/98 03/02/98 04/02/98 05/02/98 06/02/98 07/02/98 08/02/98 09/02/98
Time
cost of carry model
Hemler-Longstaff model
International Research Journal of Finance and Economics - Issue 10 (2007) 114
To robustly test the relative pricing performance of the cost of carry and Hemler-Longstaff
models, this study also calculates the MAPE measures of these two models. Table 3 lists the empirical
results. Table 5 summarizes the results of statistical tests for differences in MAPE between the two
pricing models. First, in the case of the Nikkei 225 futures, Tables 3 and 5 indicate that the MAPE
values are similar across the two models for the Asian crisis period, with the Hemler-Longstaff model
being marginally preferred. Thus, overall, for the Nikkei 225 market with low price volatility, the
Hemler-Longstaff model provides no significant improvement over the cost of carry model in
estimating stock index futures prices. One reason is that, as indicated in Table 2, the Asian crisis period
only has slightly higher price volatility than the other periods for the Nikkei 225 index. Another reason
is that this comparison involves just one particular method (that is, the EGARCH(1,1) model) for
estimating the volatility of the underlying index in the Hemler-Longstaff model. Hence, the Hemler-
Longstaff model does not significantly outperform the cost of carry model during the Asian crisis.
Next, for the Hang Seng and the KOSPI 200 futures, from Tables 3 and 5, the MAPE value of
the Hemler-Longstaff model is significantly smaller than that of the cost of carry model during the
Asian crisis. For example, for the KOSPI 200 futures, the MAPE value of the cost of carry model
reached as high as 3.0221%, with a standard deviation of 2.7949%, whereas the MAPE value when the
Hemler-Longstaff model is used is substantially reduced to 2.4158%, with a standard deviation of
1.7605%. The reduction in MAPE (relative to the cost of carry model), shown in Table 5, is
statistically significant at the 1% level, based on a t-test of the mean difference. This result occurs
because the Hemler-Longstaff model appears particularly suitable for stock markets with high price
volatility and the Asian crisis period indeed has clearly higher price volatility than
Table 5: Statistical tests for differences in MAPE between the two pricing models
Cost of carry model (MAPE) vs. Hemler-Longstaff model (MAPE)
Full sample Period 1 Period 2 Crisis period
Nikkei 225 0.308 0.162 0.177 0.281
Hang Seng 0.371 0.476 -0.368 1.766*
KOSPI 200 1.245 1.914* 1.010 3.220***
Notes: If the MAPE of the cost of carry model is greater than that of the Hemler-Longstaff model, t-value is positive. Conversely, t-value is negative. *
and *** denote statistical significance at the 10% and 1% levels, respectively.
the other periods for the Hang Seng and the KOSPI 200 indexes (see Table 2). Thus, as one would
expect, the Hemler-Longstaff model with stochastic market volatility yields significantly better pricing
performance than the cost of carry model during the Asian crisis. The empirical results from the Hang
Seng and the KOSPI 200 futures markets imply that in highly volatile periods, such as the Asian crisis
period, incorporating stochastic market volatility into the pricing model helps in predicting the prices
of these two futures.
Table 6 summarizes the results of statistical tests for differences in MAPE between period 1,
period 2, and the Asian Crisis period. The Asian Crisis period has a significantly higher MAPE value
than periods 1 and 2 (except for crisis period vs. period 1 for the Nikkei 225). Additionally, Table 6
also indicates that the MAPE value of the second sub-period is significantly smaller than that of the
first sub-period for all the index futures contracts and for all the models. This implies that the
efficiency of these three markets appears to increase over time.
115 International Research Journal of Finance and Economics - Issue 10 (2007)
Table 6: Results of statistical tests for differences in MAPE between the sub-periods
Nikkei 225 Hang Seng KOSPI 200
Cost of carry model
Crisis period vs. Period 1 1.778* 5.402*** 8.722***
Crisis period vs. Period 2 5.307*** 10.373*** 16.422***
period 1 vs. Period 2 6.865*** 11.571*** 19.686***
Hemler-Longstaff model
Crisis period vs. Period 1 1.508 4.503*** 9.175***
Crisis period vs. Period 2 5.035*** 10.451*** 20.104***
period 1 vs. Period 2 6.913*** 11.741*** 22.503***
Notes: If the MAPE of the former period is greater than that of the latter period in column 1, t-value is positive. Conversely, t-value is negative. * and ***
denote statistical significance at the 10% and 1% levels, respectively.
5. Conclusions
Hemler and Longstaff (1991) followed the CIR framework and developed a closed-form general
equilibrium model of stock index futures, characterized by both stochastic market volatility and interest
rates. Furthermore, the empirical implication of the Hemler-Longstaff model indicates that this model
appears especially suitable when stock markets have high price volatility. Notably, the Asian crisis
period has higher stock price volatility than the other periods. Thus, by using the Nikkei 225, the Hang
Seng, and the KOSPI 200 futures contracts, this study represents the first attempt to apply the Hemler-
Longstaff model to the case of the Asian crisis, and examines the power of the cost of carry and
Hemler-Longstaff models in predicting stock index futures prices.
The results of MPE indicate that in highly volatile periods, such as the Asian crisis period, the
magnitude of MPE is much larger for the cost of carry model than for the Hemler-Longstaff model.
Based on MAPE, the results from the Hang Seng and the KOSPI 200 futures markets also indicate that
the Hemler-Longstaff model with stochastic market volatility provides more accurate pricing
performance than the cost of carry model during the Asian crisis. This finding implies that
incorporating market volatility into pricing models is beneficial for predicting stock index futures
prices for periods with high stock price volatility. Therefore, when selecting a pricing model to
estimate the theoretical values of stock index futures, practitioners should identify the degree of market
volatility for the markets they are participating in.
International Research Journal of Finance and Economics - Issue 10 (2007) 116
References
[1] Baig, T., and Goldfajn, I. (1999), Financial market contagion in the Asian crisis, IMF Staff
Papers, 46, 167-195.
[2] Bailey, W. (1989), The market for Japanese stock index futures: some preliminary evidence,
The Journal of Futures Markets, 9, 283-295.
[3] Cornell, B., and French, K. R. (1983a), The pricing of stock index futures, The Journal of
Futures Markets, 3(1), 1-14.
[4] Cornell, B., and French, K. R. (1983b), Taxes and the pricing of stock index futures, The
Journal of Finance, 38(3), 675-694.
[5] Cox, J. C., Ingersoll, J. E. Jr., and Ross, S. A. (1981), The relation between forward prices and
futures prices, Journal of Financial Economics, 9, 321-346.
[6] Cox, J. C., Ingersoll, J. E. Jr., and Ross, S. A. (1985a), An intertemporal general equilibrium
model of asset prices, Econometrica, 53, 363-384.
[7] Cox, J. C., Ingersoll, J. E. Jr., and Ross, S. A. (1985b), A theory of the term structure of interest
rates, Econometrica, 53, 385-407.
[8] Eytan, T., and Harpaz, G. (1986), The pricing of futures and option contracts on the value line
index, The Journal of Finance, 41(4), 843-857.
[9] Figlewski, S. (1984), Explaining the early discounts on stock index futures: the case for
disequilibrium, Financial Analysts Journal, 40, 43-47.
[10] Fung, J. K. W., and Draper, P. (1999), Mispricing of index futures contracts and short sales
constraints, The Journal of Futures Markets, 19(6), 695-715.
[11] Gay, G. D., and Jung, D. Y. (1999), A further look at transaction costs, short sale restrictions,
and futures market efficiency: the case of Korean stock index futures, The Journal of Futures
Markets, 19(2), 153-174.
[12] Hemler, M. L., and Longstaff, F. A. (1991), General equilibrium stock index futures prices:
theory and empirical evidence, Journal of Financial and Quantitative Analysis, 26(3), 287-308.
[13] Kamara, A. (1988), Market trading structures and asset pricing: evidence from the Treasury-bill
markets, Review of Financial Studies, 1, 357-376.
[14] Modest, D. M., and Sundaresan, M. (1983), The relationship between spot and futures prices in
stock index futures markets: some preliminary evidence, The Journal of Futures Markets, 3(1),
15-41.
[15] Nelson, D. (1991), Conditional heteroskedasticity in asset returns: a new approach,
Econometrica, 59, 347-370.
[16] Resnick, B., and Hennigar, E. (1983), The relationship between futures and cash prices for U.S.
Treasury bonds, Review of Research in Futures Markets, 2(3), 282-299.
[17] Tuluca, S. A., Zwick, B., and Seiler, M. J. (2003), International versus U.S. sector
diversification strategies in the wake of the Asian crisis, American Business Review, 21(1), 67-
74.