This paper explores the application of cointegration and error correction techniques to study the daily futures and cash prices on the S&P 500 index. Cointegration analysis is used to examine the temporal causal linkage between the S&P 500 stock index and futures daily closing prices for the year 1998. If the S&P 500 index and futures markets are efficient, the best forecast of next period's price is the price of this period. However, based on root mean square error comparisons, results from this paper indicate that the proposed error correction specification outperforms naive univariate forecasts. Thus, excluding transaction costs, it appears that this modeling strategy has the potential to profitably predict price changes.
Over the past 40 years, stock market prices have been analyzed in many different ways by
investors and researchers alike to determine whether price changes are forecastable or not.
Efforts to successfully predict stock price changes have not been very fruitful. In 1981,
Granger developed a promising technique called cointegration, where two variables may
move together although they are nonstationary. The idea behind the concept of
cointegration is that there exists a long-run equilibrium relationship between the two
variables. In the short-run they may deviate from each other but market forces,
government intervention, etc., will bring them back together in the long run. Engle and
Granger (87) extended this concept and showed that cointegrated series has an error
correction representation. With the error correction representation, a proportion of the
disequilibrium in one period is expected to be corrected in the next period.
Researchers have applied the cointegration and error correction models to various
markets only to bring some mixed results. Barnhart and Szakmary (91) find spot and
forward exchange rates are cointegrated and estimate the appropriate error correction
model. Bessler and Covey (91), using daily data from August 21, 1985 through
August 20, 1986, find evidence supporting weak cointegration between the live cattle futures market and the corresponding cash market. Chowdhury (91) examines the markets
for copper, lead, tin, and zinc on the London Metal Exchange and finds some evidence
of cointegration between the spot and futures price for copper. Wahab and Lashgari (93)
conclude that spot and futures prices for both the S&P 500 and FT-SE 100 indexes appear
to be simultaneously related on a daily basis over the period from 1988-1992. Gosh (93),
using intra-day data for 1988, tests for causality between the S&P 500 index spot and
futures prices and finds that both systems are in fact cointegrated. Booth, So, and Tse
(1999), using cointegration methods on German’s Stock index, index futures and index
options market find that markets are informationally linked. |
