EFFICIENCY TESTS OF AGRICULTURAL COMMODITY
FUTURES MARKETS IN CHINA
H. Holly Wang and Bingfan Ke1
June 4, 2002
Abstract
In this paper, we study the efficiency of the Chinese wheat and soybean futures markets
and assess the conditions in agricultural commodity futures and cash markets in China. Formal
statistical tests are conducted through Johansen’s cointegration approach using three different
cash prices along with different futures forecasting horizons ranging from one week to six
months. Results suggest a long-term equilibrium relationship between the futures price and cash
price for soybeans, and a weak short-term efficiency of the soybean futures market. The futures
market for wheat is inefficient, which may be caused by over-speculation and government
intervention.
1 The authors are, respectively, Assistant Professor and Graduate Research Assistant at the
Department of Agricultural and Resource Economics, Washington State University, USA.
Efficiency Tests of Agricultural Commodity Futures Markets In China
Introduction
China started a significant economic reform in 1978, and in the subsequent years, it had
accomplished the nearly miraculous feat of transforming its centrally planned economy to a
largely market-oriented one. In December 1990, the first agricultural wholesale market, the
Zhengzhou Grain Wholesale Market (ZGWM), was established with the assistance of the
Chicago Board Of Trade (CBOT). The ZGWM was initiated as an experiment to examine the
feasibility of a futures market for agricultural commodities. After three years of successful
operation of forward contracts, the China Zhengzhou Commodity Exchange (CZCE) was
established in 1993 basing its operation on the ZGWM and specializes in the trading of
agricultural commodity futures contracts
The first futures commodity traded in the CZCE was mung bean and its trading had
flourished in the earlier years. Because mung bean is not a major agricultural product, the active
trading was almost exclusively a result of speculation. Almost 300 million tons were traded
during a one-year period between August 1998 and August 1999, yet the total production was
only less that one million tons per year. In order to avoid a market disorder, the required trading
margin of mung bean was increased to 20% in February 2000, a prohibitively high level for most
speculators. As a result, mung bean futures have largely been phased out of the market. On the
other hand, wheat futures trading in CZCE has shown significant growth. Over 14 million
contracts were traded in 2001, compared to less than 200 thousand in 1996.
In the years following 1993, more than sixty other futures markets had emerged, many of
which were located in the same cities and trading the same commodities (Yao, 1998) and some
1
operated illegally. After a period of chaos in the market, the Chinese government started the
crackdown on these illegal futures operations in late 1994. Most of those existing futures
exchanges were closed or merged and only three remained by the end of 1998: the CZCE, the
Dalian Commodity Exchange (DCE), and the Shanghai Futures Exchange (SFE). The SFE
specializes in trading metals, while both the CZCE and the DCE remain as markets for
agricultural commodities: primarily wheat in CZCE and soybeans in DCE. Now the DCE is the
second largest soybean futures market in the world with trading volume 3.8 times higher than
that in the Tokyo Commodity Exchange, and ranks just behind the CBOT (Food China, 2001).
One of the objectives for the establishment of agricultural futures markets in China was
to develop a price indicator that is solely determined by market demand and supply. Such an
indicator is particularly important to China because it had adopted a “dual price system” since
1985, in which different prices for the same agricultural commodity existed. This policy was
aimed to encourage farmers to participate in the market during economy transition, while at the
same time the government still kept control of a proportion of the agricultural products for food
security consideration. However, this system has given rise to market disorders and inefficiency
(Cater and Estrin, 2001). The establishment of the agricultural futures markets was one of the
steps aimed at the elimination of the dual price system.
The goal of this study is to test the efficiency of the futures markets for agricultural
commodities in China. An efficient market is one in which prices always “fully reflect” available
information and where no traders in the market can make a profit with monopolistically
controlled information (Fama, 1970). In other words, an efficient commodity futures market can
provide effective signals for the spot market price and eliminates the possibility that profit can be
2
guaranteed as part of the trading process. This price reflects the equilibrium value for suppliers
and demanders in the market.
The study of market efficiency in agricultural commodity futures markets is important to
both the government and the producers/marketers in China. From the government policy point of
view, an efficient market means a better alternative to market interventions such as imposing
price stabilization policies. For processors/marketers, it provides a reliable forecast of spot prices
in the future to allow them effectively manage their risks in the production or marketing process.
It is also the interest of international market participants from countries like Canada, US,
Australia, and EU, who are the major grain exporters to China (USDA FAS, 2002). Thus this
study can provide international exporters/importers of agricultural commodities with some
knowledge of the conditions in Chinese agricultural commodity futures and cash markets. At
present, agricultural futures contracts are only traded domestically, however, as China joins the
WTO and the markets become mature, futures treading in China will be inevitably open to the
world in the future.
Although China’s successful economic reform has attracted international attention from
economists (Carter and Rozelle, 2001; Martin, 2001; Lin, 1999), studies on China futures market
are rare. Most of existing studies are focused on legislative regulation and market development
and improvement and very few have investigated the agricultural commodity futures prices
quantitatively. After nine years in operation and ongoing improvement, China futures markets
are on the way toward maturity. In addition, several agricultural wholesale markets have been
developed across the country in the last few years. Given these facts, it is feasible to examine the
efficiency of the futures market, especially, whether the futures price is a good predictor of a
future cash price.
3
In this study we use a quantitative approach to test the efficiency of agricultural futures
markets. Specifically, we study or examine (1) the long-run equilibrium relationship between the
futures price and the cash price; (2) the efficiency of futures market as a predictor of cash
market, and (3) the relative performance of futures price in forecasting different cash prices. We
focus on the two major agricultural commodities traded in China: wheat on the CZCE and
soybeans on the DCE. The rest of the paper is organized as follows: the next section is a
summary of theoretical and empirical studies related to market efficiency tests; the
methodological section contains details of the statistical tests for efficiency of futures markets;
empirical results are presented after a brief explanation of the data; the last section is summary
and conclusions.
Literature Review
There are few studies on futures market in China. Most of the publications about these
markets are descriptive analyses with emphases on legislative and/or other developing issues.
Such examples include papers written by Tao and Lei (1998), Fan, Ding, and Wang (1999), and
Zhu and Zhu (2000). The historical development of futures markets in China and be found in
Yao (1998). Yao also provided a detailed structural analysis of the commodity futures markets
and the government’s legislative and regulatory attempts along the way.
Williams, et al. (1998) described the development and the characteristics of mung bean
trading in the CZCE. By analyzing the price spreads between different contracts in the same crop
year, they found that the conditions for arbitrage existed on the CZCE. The existence of arbitrage
is a sign of inefficiency. Durham and Si (1999) examined the relationship between the DCE and
the CBOT soybean futures prices. Using the law of one-price models, they conclude that the
4
soybean futures price in DCE is influenced by CBOT price, but that the relationship between the
two cannot be well represented by one single model.
There are numerous studies, both theoretical and empirical, that analyze the efficiency of
futures markets in developed countries like the US. We refer to Fama (1970) for a thorough
summary of the early works on market efficiency testing. In general there are three forms of
testing market efficiency: strong-form tests in which the current information set includes
everything relevant; semi-strong-form tests in which the obviously publicly available
information are considered; and weak-form tests in which the current information set contains
the historical price series only. Most of the studies so far have been focused on the weak-form
tests since both the strong and semi-strong tests are difficult to conduct empirically. Fama
examined the expected return model, the sub-martingale model, and the random walk model, all
of which were used intensively in the early years for market efficiency tests.
A simple linear regression model was used by Bigman, Goldfarb, and Schechtman (1983)
to test the efficiency of wheat, corn and soybeans trading at the CBOT. Based on F tests they
conclude that futures prices generally provide inefficient estimates of the spot price at maturity.
Later, Maberly (1985), Elam and Dixon (1988), and Shen and Wang (1990) pointed out that the
result is invalid based on such conventional F tests when the price series are non-stationary.
The development of cointegration theory by Engle and Grange (1987) provided a new
technique for testing market efficiency. Aulton, Ennew, and Rayner (1997) re-investigated the
efficiency of UK agricultural commodity futures markets using the cointegration methodology.
They found that the market is efficient for wheat but not efficient for pigmeat and potatoes. The
cointegration method can effectively account for the nonstationarity in price series. But one
5
limitation of this approach is that no strong inferences can be drawn for the parameters, which
are the central point of the efficiency tests (Lai and Lai, 1991).
Following Engle and Grange, Johansen (1988, 1991) and Johansen and Juselius (1990)
derived statistical procedures for testing cointegration using the maximum likelihood method.
These procedures are based on a vector autoregressive (VAR) model that allows for possible
interactions in the determination of spot prices and futures prices. Lai and Lai (1991) suggested
use of Johansen's approach to test for market efficiency and illustrated the procedure with an
example of the forward currency market in the US.
Based on Johansen’s approach, Fortenbery and Zapata (1993) evaluated the relationship
of two North Carolina corn and soybean markets with respect to the CBOT. Cointegration
existed between any pair of these markets and no strong evidence was found to reject the
efficiency hypothesis. Mckenzie and Holt (1998) tested the efficiencies of the US futures
markets for cattle, hogs, corn, soybean meal and broilers. Their results indicated that futures
markets for all the commodities except the broiler are both efficient and unbiased in the long run.
Kellard, et al. (1999) examined the efficiency of several widely traded commodities in different
markets, including soybeans on the CBOT and live hogs and live cattle on the Chicago
Mercantile Exchange. The results showed that the long-run equilibrium condition holds, but
there was evidence of short-run inefficiency for most of the markets studied. The degree of the
inefficiency was measured based on the forecast error variances.
In this paper, the Johansen approach is used to test the efficiency of the two well-traded
agricultural commodities in China: wheat and soybean. The relative performances of futures
markets in forecasting different cash markets are also evaluated.
6
Methodology
A nonstationary time series is said to be integrated in order one, often denoted by I(1), if
the series is stationary after the first-order differencing. An (n x 1) vector time series Yt is said to
be cointegrated if each of the series taken individually is I (1) while some linear combination of
the series A′Yt is stationary for some nonzero vector A (Hamilton, 1994). The theory of
cointegration relates to the study of the efficiency of a futures market in the following way. Let
St be the cash price at time t and Ft-i be futures price taken at i periods before the contract matures
at time t, where i is the number of periods interested. If the futures price can provide a predictive
signal for the cash price i periods ahead, then some linear combination of St and Ft-i is expected to
be stationary--- that is there exist a and b such that zt is stationary with mean 0:
(1)
zt = St - a - bFt-i.
If both St and Ft-i are I (1), a condition that usually holds for prices, the vector process (St, Ft-i) is
cointegrated. This cointegration between St and Ft-i is a necessary condition for market efficiency
(Lai and Lai, 1991). Cointegration ensures that there exists a long-run equilibrium relationship
between the two series. If St and Ft-i are not cointegrated, they will drift apart without bound, so
that the futures price provides little information about the movement of the cash price.
In addition to cointegration, market efficiency also requires an unbiased forecast of
futures price on cash price---that is, a = 0 and b = 1 in equation (1). Therefore, the market
efficiency should be tested in two steps: first to examine the cointegration relationship between
the two price series St and Ft-i; if cointegration exists then the parameters restriction a = 0 and b
= 1 is tested. The second step may consist of multiple tests: a = 0 and b = 1 jointly or each
individually. The constraint b = 1 is a more important indicator for market efficiency, because a
is non-zero under the existence of risk premium and/or transportation costs even when the
7
market is efficient. That is why we also test them separately though a = 0 and b = 1 are often
tested jointly. The cointegration relationship and the parameter restrictions can be tested using
Johansen’s approach and the likelihood ratio test, respectively, as outlined below.
Cotintegration tests
Before testing for cointegration, each individual price series should be examined for I(1)
first. Augmented Dickey-Fuller (ADF) and the Phillips-Perron unit root tests are the common
methods (Chowdhury, 1991; Lai and Lai, 1991; Mckenzie and Holt, 1998).
If both the futures price and cash price are I(1), Johansen’s cointegration tests can then be
conducted. Consider a general kth order VAR model:
k-1
(2)
∆Yt = D + ΠYt-1 + ∑ Γ i ∆Yt-i + ε t
i=1
where Yt is an (n x 1) vector to be tested for cointegration, and ∆Yt = Yt − Yt −1 ; D is the
deterministic term which may take different forms such as a vector of zeros or non-zero
constants depending on properties of the data to be tested; П and Г are matrices of coefficients;
and k is chosen so that ε t is a multivariate normal white noise process with mean 0 and finite
covariance matrix.
The cointegration relationship can be detected by examining the rank of the coefficient
matrix Π , because the number of cointegration vectors equals the rank of Π . In particular, the 0
rank --- that is Π = 0 --- implies no cointegration. In a bi-variable case, i.e., n = 2, the two
variable are cointegrated only if the rank of Π equals 1 (Johansen and Juselius, 1990).
Johansen (1988) suggested two test statistics to test the null hypothesis that there are at
most r cointegration vectors. The null hypothesis can be equivalently stated as the rank of Π is
8
at most r, for r = 0, 1, …, n-1. The two test statistics are based on trace and maximum
eigenvalues, respectively,
(3)
(4)
λtrace = −T
n
∑ ln(1 − λˆ )
i
i = r +1
λmax = −T ln(1 − λˆr +1 ) ,
where λ̂1 … λ̂r are the r largest squared canonical correlations between the residuals obtained by
regressing ∆Yt and Yt-1 on ∆Yt-1, ∆Yt-2, …, ∆Yt-k-1 and 1 respectively. The critical values are
provided by Osterwald-Lenum (1992).
In our test for efficiency of futures market, Yt = ( St , Ft − i )′ , n = 2, and the null hypothesis
should be tested for r = 0 and r = 1. If r = 0 cannot be rejected, we will conclude that there is no
cointegration vector, and therefore, no cointegration. On the other hand, if r = 0 is rejected, and
r = 1 cannot be rejected, we will conclude that there is a cointegration relationship.
Tests of Restrictions on Cointegration Vectors
Inefficiency can be concluded if the futures price and the cash price are not cointegrated
since cointegration is a necessary condition for market efficiency. If the futures price and the
cash market price are cointegrated, we can then test restrictions on the parameters in equation
(1). Cointegration implies that there exists a cointegration vector β such that β ′Yt* is stationary,
where in the case of equation (1), β ′ = (1, −b, − a ) and Yt* = ( St , Ft − i ,1)′ , so that zt = β ′Yt* is
stationary. The market efficiency hypotheses can therefore be tested by imposing restrictions on
the cointegration vector β. For example, the hypothesis of a=0 and b=1 can be expressed
β ′ = (1, −b, −a ) . We can then apply the standard likelihood ratio test in this case. The test
statistics is twice the negative log ratio of the two maximum likelihood values under the
9
restricted model versus the full model. Specifically, the test statistics can be expressed by the
canonical correlations as (Johansen and Juselius, 1990):
r
(5)
Lr = T ∑ ln{(1 − λi* ) /(1 − λˆi )}
i =1
where λ1* … λ*r are the r largest squared canonical correlations under the null hypothesis, i.e., the
restricted model; and λ̂1 … λˆr are the r largest squared canonical correlations under the full or
unrestricted model. The test statistic follows an asymptotic χ2 distribution with degree of
freedom equaling the number of restrictions imposed.
Data
Two agricultural commodity futures markets are included in this study: the CZCE for
wheat and the DCE for soybean. Two cash markets, the Zhengzhou Grain Wholesale Market
(ZGWM) and the Tianjin Grain Wholesale Market (TGWM), are chosen to test the efficiency of
futures markets for both wheat and soybeans. The ZGWM is located in Central China – the
major wheat production area, and was the first agricultural wholesale market established in the
country. TGWM, established in 1994, is another major agricultural wholesale market in China
(Figures 1 and 2). Another cash price, the national average price, is also included in the tests.
The national average price is calculated based on price from major markets across the country.
Although it does not relate to a specific market, the national average price is often used as a cash
market price index, especially in some macroeconomic and international trade studies.
Weekly futures price data of wheat and soybeans during the period of January 1998 to
March 2002 are provided by the CZCE and the DCE respectively. Cash prices are obtained from
CNgrain online database (http://www.Cngrain.com). There are six contracts each year for both
10
wheat and soybeans futures: January, March, May, July, September and November. Cash prices
are taken at the third week of each maturity month of the six futures contracts. Futures market
efficiency is tested for six forecasting horizons, ranging from one week to six months.
Accordingly, futures prices are taken at one week, two weeks, one month, two months, four
months and six months prior to the maturity of each contract.
In summary, we have six cash price series, one for each crop in each market consisting of
prices taking at the third week of all maturity months. We also have twelve futures price series,
six for each crop. For either crop, each futures price series consists of prices taken at a particular
period prior to the cash price observation week, which is assumed to be the maturity week. The
label of each series is listed on the left column of Table 1. The number of observations of each
series is the total number of contracts, 26, in our data set.
Results
Each of the price series is first examined for I (1), which is carried out in two steps. We
conduct the unit root tests using both the ADF and the Phillips-Perron methods as implemented
in SAS (the %DFTEST macro and AUTOREG procedure). The results are reported in Table 1,
from which we see that all the p-values are more that 0.06 and the majority exceeds 0.10
suggesting the existence of unit root in each of the price series. Further tests indicate that all the
price series data are stationary after the first-order differencing (Table 2). The p-values are now
very small, basically less than 0.01. Therefore we conclude that each of the price series is I (1),
and proceed to the next step, Johansen’s cointegration tests.
Soybeans
11
Cointegration testing results for soybean prices are shown in Table 3. The null hypothesis
of r = 0 is rejected at a significant level of 0.05 by both test statistics for each of the nine soybean
series, while the corresponding hypothesis of r = 1 cannot be rejected. This suggests that the
futures price of soybeans taken at up to four months before its maturity is cointegrated with the
cash price in the TGWM, and the futures price of soybeans, taken at up to two months prior to its
maturity, is cointegrated with its national average cash price. These results suggest that a longrun equilibrium relationship has been established between the soybean futures price and cash
price of the TGWM, and between the futures price and its national average cash price. These
results also indicate that the futures price has a closer relationship with the cash price in a shorter
forecasting horizon than in a longer one. This is reasonable in that the futures price contains
better information about the supply and demand of the commodity when it gets closer to its
maturity. However, soybean futures price is not cointegrated with its cash price in the ZGWM,
which is different from what was expected. Because the ZGWM was established earlier, and it is
larger and more influential than the TGWM, a closer relationship between the ZGWM and the
futures market was expected. There are at least two factors that may explain the disparity. First,
Tianjin is located closer to the major soybean production area --- Northeastern China, and the
soybean futures market (Dalian) is right across the bay, as shown in Figure 2. This geographical
closeness may help to smooth the flow of information. Secondly, transportation has been a
critical factor in the circulation of agricultural products in China. Because Tianjin serves as large
port for both domestic and imported/exported grain transportation, convenient transportation
facilities there make the price more responsive to market demand and supply.
Cointegration is only a necessary condition for market efficiency. Efficiency also requires
the futures price to be an unbiased predictor of the cash price, i.e., a = 0 and b=1 in equation (1).
12
Restrictions on the cointegration vector are tested for the cointegrated soybean prices using the
likelihood ratio test outlined previously. We have tested three hypotheses: a=0 and b = 1 jointly,
and each individually. The results are shown in Table 4. The null hypothesis of a = 0 and b = 1
is rejected in every case at a significance level of 0.01 except for the 1 week and 4 month prices
of TGWM, which are also rejected at 0.05 level. This means the soybean futures price is not an
unbiased predictor for cash prices for either market at any time. However, the unbiasedness
assumption is too strong to imply market efficiency. As discussed earlier in the paper, the
unbiasedness hypothesis may be rejected with the existence of a risk premium and/or a
transportation cost even when the market is efficient. Therefore, more inferences can be drawn
from the separate tests of a=0 and b=1.
The null hypothesis of a = 0 is rejected for all cases at 0.05 or higher significant level
except for the one week forecasting horizon for Tianjin cash price, which is also rejected at 0.1
level. This indicates that there exists a non-zero risk premium from traders in the futures market
and/or a transportation cost between Dalian and Tianjin. The null hypothesis of b = 1 cannot be
rejected at 0.1 level for the one week forecasting of TGWM. This indicates the soybean futures
market is efficient in the short-term. Notice, although b = 1 is still rejected at 0.05 or even 0.01
for all other cases, the p-values are generally larger than those corresponding cases for a = 0.
This suggests that the market efficiency is less likely to be rejected than the zero risk premium
and transportation cost, and the latter is the main contributor to the rejection of joint test
hypothesis.
Wheat
Results of cointegration tests on wheat price data are also reported in Table 3. None of
the test statistics is large enough to reject the null hypothesis of r = 0 at 0.05 significant level for
13
any series. This shows that the wheat futures price is not cointegrated with cash prices. This
result holds for all forecasting horizons and all cash markets. There is no long-run equilibrium
relationship between the wheat futures market and any of its cash markets. Since cointegration is
a necessary condition for futures market efficiency, we can conclude that China’s wheat futures
market is inefficient. One major factor that may account for the market inefficiency is overspeculation or market manipulation. There have been a number of cases of market manipulation
since the establishment of the futures market. Although there might be also over-speculation
problems in the Dalian soybeans market, it is not as serious as in the CZCE wheat market
(Durham and Si, 1999). Different government policies on the two commodities might be another
factor that affects the relative performance of the wheat futures market and the soybeans futures
market. As the most important food grain, wheat is one of the commodities mostly related to
national food security—a high priority concern of the government in making policy. For this
reason, such commodities are still regulated by the government directly or indirectly. For
example, wheat import and export are tightly controlled by the government. On the other hand,
soybeans are used as feed and oilseeds and the market is less regulated. Soybean imports are no
longer controlled and imports of the product have increased significantly in recent years.
Participation in the world market helps to improve the price prediction role of the futures market.
Summary and Conclusions
After nine years in operation, the agricultural commodity futures markets in China are
among the most active ones in the world. However, few quantitative studies have been conducted
to evaluate the efficiency of these futures markets. In this paper, we perform formal statistical
14
tests on the efficiency of futures market for two major agricultural commodities, wheat and
soybeans.
Our approach calls for Johansen’s cointegration test and the likelihood ratio test. These
tests are conducted on three different cash prices with six different forecasting horizons ranging
from one week to six months. Results suggest that long-run equilibrium relationships between
the DCE soybean futures price and the TGWM cash price, and between the soybean futures price
and the national average cash price have been established. A weak short-term efficiency is also
implied by the data for the soybean futures market in terms of predicting the price on the Tianjin
cash market, despite of the existence of a risk premium from traders in the futures market and
some transportation cost between the futures and the cash market. The geographic proximities
between the cash market in Tianjin and both the soybean production area and the futures market
in Dalian can partially explain the better performance of DCE in predicting cash price of TGWM
than that of ZGWM. However, the wheat futures market in China is still inefficient. The wheat
futures market is not cointegrated with any wheat cash markets.
Over-speculation and
government policy may account for the inefficiency.
These results provide an outlook of the agricultural commodity futures markets in China
and provide helpful information to domestic producers, processors and policy makers. The
results are also interesting to the international community that watch closely the Chinese
economy and are involved in agricultural commodity trading with China.
15
Reference
Aulton, A. J., C. T. Ennew and A. J. Rayner, “Efficient Tests of Futures Markets for UK
Agricultural Commodities.” Journal of Agricultural Economics, 48(1997): 408424.
Bigman, D., D. Goldfarb and E. Schechtman, “Futures Market Efficiency and the Time
Content of the Information Sets.” The Journal of Futures Markets, 3(1983):
321-334.
Carter, C. A. and A. J. Estrin, “Market Reforms versus Structural Reforms in Rural China.”
Journal of Comparative Economics, 29(2001):527-541.
Carter, C. A. and S. Rozelle, “Will China Become a major Force in World Food Markets?”
Review of Agricultural Economics, 23(2001):319-331.
Chowdhury, A. R. “Futures Market Efficiency: Evidence from Cointegration Tests.” The
Journal of Futures Markets, 11(1991): 577-589.
Durham, C. and W. Si “The Dalian Commodity Exchange’s Soybean Futures Contract:
China’s Integration with World Commodity Markets.” Chinese Agriculture and
the WTO Proceedings, Western Coordinating Committee, December 1999.
Elam, E. and B. L. Dixon, “Examining the Validity of a Test of Futures Market
Efficiency.” The Journal of Futures Markets, 8(1988): 365-372.
Engle, R. F. and C. W. J. Granger, “Co-intergration and Error Correction: Representation,
Estimation, and Testing.” Econometrica, 55(1987): 251-276.
Fama, E. F., “Efficient Capital Markets: A Review of Theory and Empirical Work.”
The Journal of Finance, 25(1970): 383-417.
Fan, Y., X. Ding, and H. Wang, “Factors Affecting the Development of Agricultural
16
Commodity Futures Markets in China.” Farm Economic Management, 1(1999): 35-36.
Food China, www.foodchina.com Online News, June 28, 2001.
Fortendery, T. R. and H. O. Zapata, “An Examination of Cointegration Relations
Between Futures and Local Grain Markets.” The Journal of Futures
Markets, 13(1993): 921-932.
Hamilton, J. D., Time Series Analysis, Princeton University Press, Princeton, New
Jersey, 1994.
Johansen, S., “Statistical Analysis of Cointegration Vectors.” Journal of Economic
Dynamics and Control, 12(1988): 231-254.
Johansen, S., “Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian
Vector Autoregressive Models.” Econometrica, 59(1991): 1511-1580.
Johansen, S. and K. Juselius, “Maximum Likelihood Estimation and Inference on
Cointegration—with Applications to the Demand for Money.” Oxford Bulletin of
Economics and Statistics, 52(1990): 169-210.
Kellard, N., P. Newbold, T. Rayner and C. Ennew, “The Relative Efficiency of Commodity
Futures markets.” The Journal of Futures Markets, 19(1999): 413-432.
Lai, K. S. and M. Lai, “A Cointegration Test for Market Efficiency.” The Journal of
Futures Markets, 11(1991): 567-575.
Lin, J. Y., “Technological Change and Agricultural Household Income Distribution: Theory and
Evidence from China.” Australian Journal of Agricultural and Resouce Economics,
43(1999):179-194.
Maberly, E. D., “Testing Futures Market Efficiency—A Restatement.” The Journal of
Futures Markets, 5(1985): 425-432.
17
Martin, W. “Implications for Reform and WTO Accession for China’s Agricultural Policies.”
Economics of Transition, 9(2001):714-742.
Mckenzie, A. M. and M. T. Holt, “Market Efficiency in Agricultural Futures Markets.”
Selected papers presented at American Agricultural Economics Association Annual
Meeting, Salt Lake City, 1998.
Osterwald-Lenum, M., “A Note with Quantiles of the Asymptotic Distribution of the
Maximum Likelihood Cointegration Rank Test Statistics.” Oxford Bulletin of Economics
and Statistics, 54(1992): 461-72.
Shen, C. and L. Wang, “Examining the Validity of a Test of Futures Market Efficiency:
A Comment.” The Journal of Futures Markets, 10(1990): 195-196.
Tao, J. and H. Lei, “Futures Market and the Grain Circulation Reform.” Economic
Problems, 9(1998): 29-31.
USDA FAS, “China Grain and Feed Annual 2002: US Embassy.” GAIN Report #CH2010,
USDA Foreign Agricultural Service, March 4, 2002.
Williams, J., A. Peck, A. Park and S. Rozelle, “The Emergence of a Futures Market:
Mung Beans on the China Zhengzhou Commodity Exchange.” The Journal of Futures
Markets, 18(1998), 427-448.
Yao, C., Stock Market and Futures Market in the People’s Republic of China, Oxford
Express, Oxford, New York, 1998.
Zhu, L. and J. Zhu, “Futures Market and Contract Agriculture.” Journal of Zhengzhou
Grain College, 21(2000), 29-34.
18
-1.49
F-Soybeans 6 Months
-1.02
-0.25
-0.36
-0.38
-0.49
-0.59
-0.90
-1.44
-0.72
-1.11
C-Soybeans National
F-Wheat 1 Week
F-Wheat 2 Weeks
F-Wheat 1 Month
F-Wheat 2 Months
F-Wheat 4 Months
F-Wheat 6 Months
C-Wheat Tianjin
C-Wheat Zhengzhou
C-Wheat National
0.23
0.39
0.13
0.32
0.45
0.49
0.53
0.54
0.58
0.27
0.25
0.23
0.12
0.09
0.10
0.26
0.34
0.33
-0.29
-0.97
-1.01
-2.14
-2.02
-1.68
-1.37
-1.36
-1.46
-1.30
-2.04
-1.61
-2.00
-2.45
-2.37
-1.48
-1.59
-1.43
Statistic
0.91
0.75
0.73
0.23
0.28
0.43
0.58
0.58
0.53
0.61
0.27
0.46
0.29
0.14
0.16
0.53
0.47
0.55
P-Value
Case 2
ADF Tests
-3.09
-2.84
-2.11
-2.33
-2.37
-2.08
-1.71
-1.92
-2.18
-1.48
-2.17
-2.01
-1.77
-2.05
-1.99
-1.75
-1.99
-1.80
Statistic
0.13
0.20
0.51
0.40
0.38
0.53
0.71
0.61
0.48
0.81
0.48
0.57
0.68
0.55
0.57
0.70
0.57
0.67
P-Value
Case 3
-1.46
-1.11
-1.92
-0.69
-0.98
-0.84
-0.71
-0.52
-0.52
-1.46
-1.43
-1.27
-1.01
-1.58
-1.29
-0.97
-0.91
-0.94
Statistic
0.13
0.23
0.06
0.41
0.28
0.34
0.40
0.48
0.48
0.13
0.14
0.18
0.27
0.11
0.18
0.29
0.32
0.30
-0.93
-1.57
-1.10
-1.96
-2.13
-2.23
-2.47
-2.52
-2.88
-2.88
-2.43
-2.39
-1.58
-2.15
-2.52
-2.40
-2.51
-2.26
19
0.76
0.48
0.70
0.30
0.24
0.20
0.13
0.12
0.06
0.06
0.15
0.15
0.48
0.23
0.12
0.15
0.12
0.19
Statistic P-Value
Case 2
Phillips-Perron Tests
P-Value
Case 1
Notes: 1. F represents futures prices and 1 week means futures price as of one week before maturity, and so on for the rest.
2. C represents cash prices.
-1.05
C-Soybeans Zhengzhou
-1.11
-1.67
F-Soybeans 4 Months
C-Soybeans Tianjin
-1.60
F-Soybeans 2 Months
2
-1.05
F-Soybeans 1 Month
-0.87
P-Value
Case 1
Statistic
-0.85
1
F-Soybeans 2 Weeks
F-Soybeans 1 Week
Prices
Table 1 Unit Root Tests on Price Series
-2.52
-2.49
-1.86
-2.30
-2.25
-2.33
-2.56
-2.88
-3.27
-2.78
-2.47
-2.71
-2.05
-2.25
-2.66
-2.87
-3.01
-2.64
0.32
0.33
0.65
0.42
0.44
0.40
0.30
0.19
0.09
0.22
0.34
0.24
0.55
0.45
0.26
0.19
0.15
0.27
Statistic P-Value
Case 3
-4.10
F-Soybeans 6 Months
-5.55
-5.23
-4.92
-4.80
-3.87
-4.29
-3.82
-4.08
-3.81
-3.92
C-Soybeans National
F-Wheat 1 Week
F-Wheat 2 Weeks
F-Wheat 1 Month
F-Wheat 2 Months
F-Wheat 4 Months
F-Wheat 6 Months
C-Wheat Tianjin
C-Wheat Zhengzhou
C-Wheat National
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
-4.08
-3.80
-4.53
-3.84
-4.27
-3.79
-4.68
-4.81
-5.10
-5.55
-3.82
-4.55
-4.34
-4.32
-5.59
-5.16
-4.50
-4.65
Statistic
0.00
0.01
0.00
0.01
0.00
0.01
0.00
0.00
0.00
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
P-Value
Case 2
ADF Tests
-4.05
-3.67
-4.50
-3.73
-4.19
-3.69
-4.58
-4.75
-5.76
-5.21
-3.74
-4.43
-4.42
-4.51
-5.74
-5.05
-4.40
-4.53
Statistic
0.02
0.04
0.01
0.04
0.02
0.04
0.01
0.00
0.00
0.00
0.04
0.01
0.01
0.01
0.00
0.00
0.01
0.01
P-Value
Case 3
-4.98
-5.17
-4.07
-4.03
-4.28
-4.56
-5.76
-6.93
-7.44
-4.08
-4.96
-4.61
-5.14
-5.19
-5.19
-5.71
-6.26
-5.33
Statistic
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
-5.18
-5.08
-4.53
-4.02
-4.27
-4.47
-5.63
-6.79
-7.28
-4.19
-5.13
-4.65
-5.28
-5.41
-5.24
-5.71
-6.22
-5.29
20
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Statistic P-Value
Case 2
Phillips-Perron Tests
P-Value
Case 1
Notes: 1. F represents futures prices and 1 week means futures price as of one week before maturity, and so on for the rest.
2. C represents cash prices.
-3.76
C-Soybeans Zhengzhou
-4.46
-3.99
F-Soybeans 4 Months
C-Soybeans Tianjin
-5.31
F-Soybeans 2 Months
2
-5.13
F-Soybeans 1 Month
-4.66
P-Value
Case 1
Statistic
-4.53
1
F-Soybeans 2 Weeks
F-Soybeans 1 Week
Prices
Table 2 Unit Root Tests on First-order Differences of Price Series
-5.06
-5.15
-4.50
-3.93
-4.18
-4.37
-5.49
-6.65
-7.13
-4.21
-5.22
-4.58
-5.36
-5.24
-5.21
-5.64
-6.10
-5.18
0.00
0.00
0.01
0.03
0.02
0.01
0.00
0.00
0.00
0.02
0.00
0.01
0.00
0.00
0.00
0.00
0.00
0.00
Statistic P-Value
Case 3
Table 3 Statistics of Johansen's Cointegration Tests
Soybeans
Wheat
λtrace
λmax
λtrace
λmax
H0:r=0
H0:r ≤1
H0:r=0
H0:r ≤1
H0:r=0
H0:r ≤1
H0:r=0
H0:r ≤1
Tianjin, 1 week
21.63*
5.11
16.52*
5.11
12.64
3.73
8.91
3.73
Tianjin, 2 weeks
25.19*
5.11
20.08*
5.11
12.51
3.01
9.50
3.01
Tianjin, 1 month
35.18*
5.81
29.37*
5.81
11.68
3.30
8.39
3.30
Tianjin, 2 months
34.38*
6.28
28.11*
6.28
13.37
3.64
9.73
3.64
Tianjin, 4 months
27.63*
6.20
21.43*
6.20
11.44
4.03
7.41
4.03
Tianjin, 6 months
16.62
4.19
12.43
4.19
13.89
4.40
9.49
4.40
Zhengzhou, 1 week
19.23
5.00
14.23
5.00
10.63
2.27
8.36
2.27
Zhengzhou, 2 weeks
18.47
5.34
13.13
5.34
9.06
2.25
6.81
2.25
Zhengzhou, 1 month
17.02
5.05
11.97
5.05
8.37
2.42
5.96
2.42
Zhengzhou, 2 months
17.26
5.36
11.90
5.36
8.66
2.56
6.11
2.56
Zhengzhou, 4 months
13.53
4.76
8.77
4.76
9.80
2.78
7.02
2.78
Zhengzhou, 6 months
19.49
6.94
12.54
6.94
14.17
2.99
11.18
2.99
National, 1 week
31.63*
6.67
24.96*
6.67
8.94
1.99
6.95
1.99
National, 2 weeks
24.31*
5.58
18.74*
5.58
8.95
1.87
7.09
1.87
National, 1 month
23.74*
5.54
18.21*
5.54
8.18
2.03
6.15
2.03
National, 2 months
22.37*
5.63
16.75*
5.63
9.89
2.22
7.67
2.22
National, 4 months
12.55
2.96
9.59
2.96
9.33
2.23
7.11
2.23
National, 6 months
19.68
4.33
15.35
4.33
11.60
2.51
9.08
2.51
Critical Values
19.96
9.24
15.67
9.24
19.96
9.24
15.67
9.24
Note: An * indicates that the null hypothesis is rejected; critical values are taken at a significance level of
0.05.
21
Table 4 Tests of Restrictions on Cointegration Vectors for Soybeans
Estimates
a
b
H0: a=0 and b=1
Lr
P-Value
Lr
H0: a=0
P-Value
Lr
H0: b=1
P-Value
Tianjin, 1 week
-0.65
1.248
7.26**
0.026
3.24*
0.072
2.48
0.115
Tianjin, 2 weeks
-0.92
1.348
11.65***
0.003
6.26**
0.012
4.89**
0.027
Tianjin, 1 month
-0.63
1.224
16.52***
0.000
6.76***
0.009
4.67**
0.031
Tianjin, 2 months
-0.61
1.225
15.05***
0.000
6.35**
0.012
4.63**
0.030
Tianjin, 4 months
-0.63
1.249
8.95**
0.011
4.12**
0.042
4.23**
0.040
National, 1 week
-1.47
1.591
16.45***
0.000
10.74***
0.001
9.33***
0.002
National, 2 weeks
-2.32
1.953
14.05***
0.001
10.86***
0.001
9.92***
0.002
National, 1 month
-1.80
1.728
13.86***
0.001
10.38***
0.001
9.38***
0.002
National, 2 months
-1.75
1.714
13.33***
0.001
9.95***
0.001
9.07***
0.003
Notes: 1. An * indicates that the null hypothesis is rejected at a significance level of 0.10.
2. An ** indicates that the null hypothesis is rejected at a significance level of 0.05.
3. An ***indicates that the null hypothesis is rejected at a significance level of 0.01
22
Figure 1 China’s Wheat Production Map (2000) and Selected Markets
Data Source: China Statistical Yearbook 2001.
23
Figure 2 China’s Soybeans Production Map (2000) and Selected Markets
Data Source: China Statistical Yearbook 2001.
24