22
Price Discovery in the South African Yellow Maize Futures Market 1 PRICE DISCOVERY IN THE SOUTH AFRICAN YELLOW MAIZE FUTURES MARKET ABSTRACT This study examines the price discovery process in the South African futures and spot markets for yellow maize. Engle-Granger and Johansen tests of cointegration are performed, thereafter, Error Correction Model (ECM) and Vector Error Correction Model (VECM), representing the long-run equilibrium relationship between spot and futures for yellow maize. In order to reinforce on the findings of the former tests, the Granger-Causality test was also conducted and a two-way feedback between the two time series data sets was established. This meant that price discovery occurred in both the spot and futures markets. The paper concludes by discussing the policy implications of the findings. Key words: Price discovery, spot market, futures market, cointegration, error correction model INTRODUCTION Price discovery in the study of futures markets plays a vital role in determining market equilibrium. Many researchers have given different ways of defining price discovery. Schreiber and Schwartz (1994) defined price discovery as the process through which markets attempt to reach equilibrium prices. Other authors such as Yang et al (2001) commonly defined price discovery as the use of futures prices to determine expectations of (future) cash market prices. In the same regard Lapan (1991) also noted that through its price discovery function, futures markets help apportion resources and stabilize prices and incomes by establishing forward contracts. Furthermore, price discovery according to other writers can be classified in two different ways, the static and the dynamic sense (Yang and Leatham, 1999). In a static sense, price discovery simply means the existence of equilibrium prices. In the dynamic sense, the price discovery process describes how information is birthed and disseminated across the markets. Price discovery is a vital and major function for the commodity futures market. According to Chiang and Fong (2001) and Garcia et al (1997), the futures market functions in such a way that all the obtainable information at a precise point in time will be reflected in the current futures

Price Discovery in the South African Yellow Maize Futures Market PRICE DISCOVERY IN THE SOUTH AFRICAN YELLOW MAIZE FUTURES MARKET

Embed Size (px)

Citation preview

Price Discovery in the South African Yellow Maize Futures Market

1

PRICE DISCOVERY IN THE SOUTH AFRICAN YELLOW

MAIZE FUTURES MARKET

ABSTRACT

This study examines the price discovery process in the South African futures and spot markets

for yellow maize. Engle-Granger and Johansen tests of cointegration are performed, thereafter,

Error Correction Model (ECM) and Vector Error Correction Model (VECM), representing the

long-run equilibrium relationship between spot and futures for yellow maize. In order to

reinforce on the findings of the former tests, the Granger-Causality test was also conducted and

a two-way feedback between the two time series data sets was established. This meant that price

discovery occurred in both the spot and futures markets. The paper concludes by discussing the

policy implications of the findings.

Key words: Price discovery, spot market, futures market, cointegration, error correction model

INTRODUCTION

Price discovery in the study of futures markets plays a vital role in determining market

equilibrium. Many researchers have given different ways of defining price discovery. Schreiber

and Schwartz (1994) defined price discovery as the process through which markets attempt to

reach equilibrium prices. Other authors such as Yang et al (2001) commonly defined price

discovery as the use of futures prices to determine expectations of (future) cash market prices. In

the same regard Lapan (1991) also noted that through its price discovery function, futures

markets help apportion resources and stabilize prices and incomes by establishing forward

contracts. Furthermore, price discovery according to other writers can be classified in two

different ways, the static and the dynamic sense (Yang and Leatham, 1999). In a static sense,

price discovery simply means the existence of equilibrium prices. In the dynamic sense, the price

discovery process describes how information is birthed and disseminated across the markets.

Price discovery is a vital and major function for the commodity futures market. According to

Chiang and Fong (2001) and Garcia et al (1997), the futures market functions in such a way that

all the obtainable information at a precise point in time will be reflected in the current futures

Price Discovery in the South African Yellow Maize Futures Market

2

price. This is consistent with the Efficient Markets Hypothesis (EMH) developed by the

founding father, Eugen Fama’s paper of 1970. Information on price discovery is vital because

markets are largely used by firms which are involved in production, marketing and processing of

commodities. Decisions of production and consumption mainly depend on the proficient price

signals from the market.

Futures markets are argued to serve a social function by aiding participants make better estimates

of future prices in order to make their consumption and investment decisions efficiently.

Wiseman (1999) argues that futures price strive towards being the approximate of the future cash

price at the day of contract termination and, in addition to cash markets, adds positive

dimensions in competition by enlarging the domain of price – making forces. It is generally

stated also by many authors such as Leuthold et al (1989) that price discovery in the commodity

futures markets is more efficient than that in the spot markets. The literature in the next section

will mainly review on whether futures rather than spot (cash) markets are primarily a source for

price discovery. This mainly addresses the issue of as to whether futures prices lead the spot

prices or vice versa. The paradigms presented here are commonly concerned with a specific

commodity that is traded in a single centralized cash and/or futures markets.

Agricultural commodity derivatives were first introduced in South Africa 1996 after the

deregulation of the previous marketing system. Maize production in South Africa plays a major

role in both consumption and economic growth. Particularly yellow maize plays a major role in

providing feed for livestock. Livestock is both used for domestic consumption and export to

neighboring countries. This paper will therefore seek to test the hypothesis that price discovery

occurs in the futures market by examining the spot and futures price data for the yellow maize

contracts traded on the Johannesburg Stocks Exchange (JSE). The yellow maize will be used as

it represents a storage commodity and it is one of the largest and important commodities traded

on the JSE by volume (JSE, 2014). The study will mainly address the relationship between

yellow maize spot prices and futures prices to see whether the yellow maize traders can use

futures prices successfully in price discovery process. This study about yellow maize has not

been mainly addressed as a single commodity in recent South African studies.

The study employed cointegration analysis and an Error Correction Model (ECM) developed by

Engle and Granger, and Johansen, in order to explore the structure of the yellow maize market

Price Discovery in the South African Yellow Maize Futures Market

3

price discovery. Employing both the Error Correction Model (ECM) and the Vector Error

Correction Model (VECM), the study will test the reactions of both spot and futures prices to

short-run shocks and find evidence that in the market for yellow maize spot prices lead futures

prices and vice versa. This paper will seek to contribute to the knowledge of the South African

futures market regarding price discovery of the yellow maize futures market by using the ECM

and VECM mechanisms. The results found will have an important implication to market

participants who use futures prices as predictors of expected future spot prices.

LITERATURE REVIEW

Past studies on the price discovery function of futures market have concentrated on the price

correlation between futures and spot prices. The studies sought to address the issue as to whether

the current futures price is an unbiased predictor of future spot prices and if it can provide direct

evidence in favor of price discovery occurring primarily in the futures market. This type of test is

called the test of futures market efficiency and is closely associated with the price discovery role.

Fama’s (1970) definition of an efficient futures market came to be known as the Efficient Market

Hypothesis (EMH) (LeRoy, 1989; Zulauf and Irwin, 1998). According Fama (1970), he stated

that an efficient market is one that correctly incorporates all known information in determining

price (Zulauf and Irwin, 1997).

In a more descriptive definition Barnett and Serletis (2000) note that the hypothesis claims that

asset prices are rationally or reasonably related to economic realities and always incorporate all

information available to the market. Barnett and Serletis (2000) further warn that futures market

efficiency shouldn’t be confused with allocative efficiency, better known as “Pareto Optimality”.

Striving towards Pareto optimality would entail changes in the combination of goods produced

or in the combination of inputs used in such a way that the changes benefit or profit some people

without anyone else being worse off (Sloman, 1991). Therefore, allocative efficiency concerns

the relative quantities of different commodities to be produced (Lipsey et al, 1990), while futures

market efficiency involves the market’s ability or capacity to reflect all available information,

and this is known as the inter-temporal allocation of resources (Sabuhoro and Larue, 1997).

Price Discovery in the South African Yellow Maize Futures Market

4

Furthermore, a general description of price discovery is simply the impounding or restoration of

new information into a security’s price (Hasbrouck, 1995; Choy and Zhang, 2010). In this regard,

when two markets are inter-twined, as it is the case between spot and futures markets on the

same underlying commodity, then two prices subsist which are motivated or driven by the same

underlying primary information and the question then emerges as to whether price changes in

one market guide changes in the other. Yang et al (2001) and Kavussanos and Nomikos (2003)

as earlier stated in this study, for instance, define price discovery in futures markets as the use of

futures prices to project expectations of future spot market prices. The limitation of this approach

is it’s assumption that futures must lead spot prices when in reality the evidence regarding the

informational role between spot and futures prices is a mixed one (yang et al, 2001).

Pavabutr and Chaihetphon (2010) argued that in general, futures markets respond faster to new

information than spot markets. The most common explanations provided for futures prices to

lead spot prices is that transaction costs are lower in the futures market (Andersen et al, 2007)

and futures markets offer greater leverage making it easier for speculators to profitably utilize or

exploit new information in the futures market (Pavabutr and Chaihetphon, 2010). Kavussanos

and Nomikos (2003) made a point that limitations in short-selling typically exist in the spot

market which makes the futures market more attractive for traders seeking to utilize new

information. Thus, Wahab and Lashgari (1993) conclude that leverage, transaction costs and

possibility of short-selling make trading in stock index futures more appealing than trading in the

market for underlying stocks. Stoll and Whaley (1990) furthermore, argue that such trading

drives futures prices first after which index arbitrage “pulls’ the stock prices to react or respond

to deviations from the cost-of-carry relationship.

To further consolidate the assumption that futures prices lead spot prices, a significant number of

empirical studies provide support for the assumption. In their findings of comparing stock index

and stock index futures, Stoll and Whaley (1990) discovered that although the effect was not

completely unidirectional, futures prices led spot prices and that this incident had grown stronger

as futures markets had matured. Similarly, Pizzi et al (1998) found that both three and six

months stock index futures prices led spot market by at least 20 minutes but some causation from

the spot market to the futures market was also evident. Hasbrouck (2003) looked at U.S. equity

index markets including Exchange Traded Funds (ETFs) and found that electronically traded

Price Discovery in the South African Yellow Maize Futures Market

5

futures contracts subdued price discovery with the exceptions of the Standard and Poor’s (S&P)

400 index where ETF provided significant price discovery.

By employing GARCH approach on the Hang Seng index market, So and Tse (2004) found that

the volatilities of the index and futures market spilled over to each other but that effect was

stronger from the futures market to the index signifying that the futures market dominated in the

price discovery process. Choy and Zhang (2010) examining the Hong Kong market also found

that stock index futures play a dominant or leading role in price discovery albeit the relative

significance of mini futures had amplified over time. They further concluded that since both

regular and mini futures contracts traded on the same trading platform, the dominant role of

regular futures contracts in price discovery is due to their somewhat lower transaction costs. In

recent studies, Zelda and Efe-Omojevwe (2013) in their study of Efficiencies of Maize and

Wheat futures markets in India, found that futures prices for both maize and wheat had a positive

impact on the spot prices meaning futures prices led spot prices. Yan and Reed (2014) in

examining the Chinese corn futures with comparison to soybean futures concluded that market

participants can use corn futures prices to predict future spot prices and hedge their corn

positions to reduce their price risk. In this regard, a conclusion was reached that corn futures

prices led the corn future spot prices.

Whilst it is clear that many studies have found evidence to support the hypothesis that futures

prices serve as the primary source of price discovery, it is also evident that many of the studies

reported mixed results with some evidence that price discovery also flows from spot to futures

markets albeit to a lesser degree. Whilst less common, some studies, on the other hand have also

found results signifying that spot prices can lead futures prices. For instance, Wahab and

Lashgari (1993) found evidence of a feedback relationship between stock index and stock index

futures markets but concluded that their results confirmed that the lead from spot to futures was

more prominent. Similarly, Leng (2002) found for one of the sub-periods that spot price led the

futures price. Chan, Chan and Korolyi (1991) concluded that new market information circulates

in both markets and that the spot and futures markets execute or perform imperative price

discovery roles.

In the South African studies of price discovery, Ferret and Page (1998) found that stock index

futures price changes guided or led those of the underlying spot index by up to three days in

Price Discovery in the South African Yellow Maize Futures Market

6

reflecting new information. Several suggestions have been presented in literature to explain

findings that the spot price leads the futures price. To this regard Ferret and Page (1998)

recommended that changes in the spot market form part of the knowledge futures traders use to

make decisions thus changes in the spot price may influence futures traders and in turn affects

futures prices. Srinivasan and Bhat (2009) also speculated that it may indicate that speculative

traders, who are seeking profit making situations, would favor to use a commodities futures

market owing to suppleness in terms of investment strategies. Their movement away from the

spot market would then result in the spot market having less noise trading and decreased

informational asymmetries which in turn would enhance market depth, market efficiency and

liquidity resulting in the spot market being better positioned to react to new events first.

DATA AND METHODOLOGY

Data Description

Maize futures market has been traded in South Africa since 1996. Agricultural derivatives from

the JSE provide a platform for price discovery and efficient price risk management for the grains

market in South and Southern Africa. Trading on a formal exchange that connects buyers and

sellers provides transparent price discovery and all transactions are assured through the

derivatives clearing structure. Futures contracts have a futures expiry date and both parties have

to honor the position at the traded price on that date.

The data used in this study consists of price data on yellow maize futures contracts traded on the

SAFEX under the JSE. The settlement price (close) is the last price for a futures contract on any

trading day. These historical, spot and futures prices were obtained directly from the JSE website

(JSE, 2014). Both the white and yellow maize futures contract in South Africa have expiry dates

March, May, July, September and December which translates to either a two or three months

interval between contracts. Data is collected over the period of 2010 to 2014 and the yellow

maize futures contract for this study had two data points collected. The first was the maturity or

spot price and the other was the futures price quoted on each hedging month. The cash or spot

price is the proxy of the spot (near) month and the futures price being the furthest.

Price Discovery in the South African Yellow Maize Futures Market

7

From this it is important to note that there are no official spot prices for yellow maize in South

Africa hence theoretically the spot price and futures price are equal on the day of the contract

maturity of a specific futures contract. This is consistent with the methodology used from

previous studies by Aulton et al (1997) to test for cointegration. This price was used as the spot

price for the underlying commodity on the following based assumptions which include; there are

no price manipulations by players that could affect the futures price in such a way that it would

not reflect the true value of the commodity on the day of maturity, and the nearest month futures

price at maturity is the best available price from which the cash price can be derived on that

specific day.

Methodology

As the focus of this paper is to examine the price discovery process of the yellow maize futures

market of South Africa, the methodology of this paper focuses on describing the procedures

directly related to such. A mature futures market should perform the function of price discovery,

that is, in such a market the spot price, the futures price or a combination thereof will adjust to

new events in order to maintain equilibrium. Such a long run relationship is tested for by

examining the spot and futures price series for cointegration. Cointegration requires that both

series be non-stationary in level form and integrated of the same order and so both series must be

tested for the presence of a unit-root. This is basically done by means of an Augmented Dickey-

Fuller test or a Phillips-Perron test of which this study will only employ the Augmented Dickey-

Fuller test of unit root (Dickey and Fuller, 1981).

The relationship between spot and futures prices is described by a cointegrating relationship.

Once it has been established that the data series are non-stationary, the next step is to test

whether there is a cointegrating relationship between the two series. That is, it is only after the

stationarity of the data has been established and it is found that both series in question are non-

stationarity processes. In this regard, that’s when the next step can be taken which is to test for

cointegration. Therefore, the relationship between spot and futures price is described by a

cointegrating relationship. The relationship according to Zulauf and Irwin (1998) and Danthine

(1997) provides an equation for simple discussion of the major concepts underlying the EMH.

Price Discovery in the South African Yellow Maize Futures Market

8

Two approaches exist through which the presence of cointegration may be established; namely

the Engle-Granger (EG) and Johansen’s methods. In this study both approaches will be

employed in order to generate the output required to formulate an Error Correction Model

(ECM).

Cointegration and Augmented Dickey-Fuller (ADF) testing

The test for cointegration is based on the test for unit roots of the cointegration regression. In

simulation experiments, Engle and Granger (1987) examined a number of alternative methods of

testing the residuals for stationarity. It transpired that critical values for the test statistics

considered tend to depend on the model used to simulate the data. Nonetheless, the experiments

suggested that the tests for which critical values were least model-sensitive were Augmented

Dickey-Fuller (ADF) test. These were also found to have the greatest power of all the tests

considered.

Since most of the variables or series are non-stationary, unit root tests are useful to determine the

order of integration of the variable in study. In this regard, variables of spot and futures prices

are tested for unit root at level form and if found to have unit root then they are tested again at

first differencing I(1) in the ADF model in order to determine stationarity of the data. Thus, the

typical ADF test or unit root model is given as follows:

… (1)

Spurious Regression

Regression of a non-stationary time series on another non-stationary time series may cause a

spurious regression or nonsense regression. To this regard, a spurious or nonsense model is not

desirable. In this study, there are two variables Spot and Futures prices (Sp and Fp) as indicated

in equation 2 below with the precondition being that Sp and Fp have unit root at level form,

meaning non-stationary. The cointegration equation on the other hand is given as follows:

= + Fpt+ … (2)

where is the spot price at time t, Fpt is the futures price at time t, are parameters,

and is a random error term. The symptom of a spurious regression is R-square ( ) value

Price Discovery in the South African Yellow Maize Futures Market

9

would be greater than the Durbin Watson statistics.Furthermore, the residuals obtained from the

cointegrating equation will contain the short run disturbances, that is, the error term captures the

deviations from the long-run equilibrium. The ECM brings these extracted residuals into the

model as an explanatory variable. Wang and Ke (2002) show that a common method of detailing

this relationship is to show that there exists a linear combination of Spt and Fpt-1 that is stationary

with residues that have a mean of zero (0) and this combination can be seen by the following

equation below:

Therefore, the residuals of the model will be tested for stationarity using the ADF model in level

form or I(0) by using the absolute Engle-Granger critical values for unit root testing of 5% = 3.34

and 10% = 3.04 (Davidson and Mackinnon, 1993; Mackinnon, 1991). If found stationary then it

can be deduced that the model one in equation 2 is not spurious meaning the model can be

accepted. This means that variables of the Futures Prices and Spot prices are cointegrated or they

have a long-run relationship between them. This would mean that model one is a long-run model.

Testing of unbiasedness and Efficiency

Testing for unbiasedness is an integral part in testing for market efficiency. The conventional

efficiency tests using regression analysis requires that the intercept in equation 2 is not

significantly different from zero, also that the slope in equation 2 is not significantly different

from one and that the residuals are white noise (Kellard et al, 1999; Aulton et al, 1997). After

finding these results it can be inferred or argued that the futures price at contract purchase is an

unbiased predictor of the spot price at contract termination (Lai and Lai, 1991).

Error Correction Mechanism (ECM)

If the variables such as Sp and Fp are cointegrated, then an error correction model (ECM) is run.

The ECM determines the short run disequilibrium relationship between the futures and the spot

prices (Engle and Granger, 1987). This model can be represented in first difference form by the

following equation using the given variables of spot and futures prices:

… (4)

Price Discovery in the South African Yellow Maize Futures Market

10

where are the first differenced variables, is the intercept, is the short run

coefficient and is the white noise error term. is the one period lag residual of equation

above. It is also known as the error term of one period lag. It is the error correction term that

guides variables of the system to restore back equilibrium. In other words it corrects

disequilibrium in the model. The sign of the coefficient of the error correction term should be

negative after estimation. This coefficient tells the rate at which it corrects the previous period

disequilibrium of the system or model. Thus, when coefficient is significant and contains

negative sign, it validates that there exists a long run equilibrium relationship among variables

stated in the model of equation 1.

Johansen Cointegration Test

If the two variables spot price and futures price are integrated of the same order, the Johansen-

Juselius maximum likelihood of cointegration can be applied in order to obtain the number of

cointegrating vector(s). This method is preferred over the Engle-Granger method due to

statistical reasons. The Johansen-Juselius Multivariate Cointegration Model is given below:

∑ … (5)

where is a 2x1 vector (spot and futures) respectively, is a symbol of difference operator,

is a 2x1 vector of residuals. The VECM model has information about the short and long-run

adjustment to changes in , via the estimated parameters and , respectively. Here the

expression is the error correction term and can be factored into two separate matrices

and such as , where denotes the vector of cointegrating parameters while is the

vector of error correction coefficients measuring the speed of convergence to the long-run steady

state.

Granger Causality

If spot and futures prices are found to be cointegrated, an ECM can be specified and estimated

using standard methods and diagnostic tests. Cointegration however indicates that causality

exists between the two series but it fails to show us the direction of the causal relationship. Engle

and Granger (1987) suggests that if cointegration exists between two variables in the long-run,

then, there must be either unidirectional or bidirectional Granger-causality between these

Price Discovery in the South African Yellow Maize Futures Market

11

variables. In other words, according to Granger, if there is evidence of cointegration between two

or more variables, then a valid ECM should also exist between the two variables. Thus, as spot

and futures prices are cointegrated a VECM representation could have the following form as

shown by the two equations below:

∑ ∑

… (5)

∑ ∑

… (6)

where are the short-run coefficients, are Error Correction terms,

and and are residuals in equations 5 and 6 respectively. The is the lagged value of

the residuals derived from the cointegrating regression of spot price on futures price in equation

5 and is the lagged value of the residuals derived from the cointegrating regression of

futures price on spot price in equation 6. Unidirectional causality from spot price to futures price

(spot price Granger causes futures price) will occur in equation 5 if > and similarly,

unidirectional causality from futures price to spot (futures price Granger causes spot price) will

occur in equation 6 if >.

Therefore, if the set of estimated coefficients on the lagged spot price ( ) coefficients are non-

zero, it means it’s a short-run causality while the error correction coefficient of is

significant, it means long-run causality. If both variables Granger cause each other, then it is

said that there is a two-way feedback relationship between the two variables in this case spot and

futures prices. In both the Engle-Granger ECM and VECM, the lag length is an important

consideration and it is necessary to consider the choice of lag length carefully (Enders, 2004).

Following Mahalik et al (2009) and Srinivasan and Bhat (2009) the AIC and SIC criteria were

used to ascertain the appropriate lag length.

RESULTS AND DISCUSSIONS

Augmented Dickey-Fuller (ADF) test was conducted to test for stationarity. The data for both

spot and futures was found to be non-stationary at level. The data sets were further tested in first

Price Discovery in the South African Yellow Maize Futures Market

12

difference form and became stationary showing that they are both I(1). The coefficients on the

other hand had the negative signs as expected. The results are shown in table 1 below.

Table 1: Augmented Dickey-Fuller test for unit root

Dependent variable:

Parameter Yellow Maize

Coefficient ADF absolute value

-0.967 (2.863) 35.652

-1.017 (2.863) 37.468

Figures in the parentheses are Test critical values at 5% significance level

Spurious and Cointegration testing

Furthermore, the cointegration equation 2 was tested for the symptoms of a spurious model and

was found to be spurious because the condition of a spurious model was met, meaning the R-

squared was more than the Durbin Watson test with values 0.88 and 0.043 respectively. Thus,

the residuals were further tested for stationarity in level and found to be stationary at both the 5%

and 10% Engle-Granger critical values with ADF test value of 3.75. This meant that the time

series data of spot and futures prices were cointegrated under the Engle-Granger testing of

cointegration. Thus, this indicates that there is an existence of a long-run relationship between

spot and futures prices.

The Johansen test of cointegration was also conducted to further justify the sensibility of the

model. The results in table 2, below show that there is at least one cointegrating equation

between the two time series, spot and futures prices indicating the presence of a long-run

cointegration relationship. In this case both of the approaches used (that is, Engle-Granger and

Johansen’s cointegration tests) conclude that there exists a long-run equilibrium relationship

between the spot and futures prices.

Table 2: Johansen’s Cointegration test

Trace Test Maximum Eigen Value Test

No Cointegrating Relationship 16.869*

14.954*

At most 1 Cointegrating Relationship 1.915 1.915

* denotes rejection of the hypothesis at the 5% significance level

Price Discovery in the South African Yellow Maize Futures Market

13

Tests of Unbiasedness and Efficiency

A test of unbiasedness and efficiency on the results presented in Table 3 using equation (2)

showed that the null hypothesis is not rejected for the commodity yellow maize; suggesting that

the market is an unbiased and efficient predictor of the spot prices for the commodity. Also,

futures price is shown to have a significant positive effect on the spot price for the commodity.

This can be seen in the calculations below:

Test of Unbiasedness: ;

Using ⁄ ( ), Then Pr[0 2(25.499) 0+ 2(25.499)] = 1

= -50.998 50.998

Test of Efficiency: ;

Pr[1 2(0.013) 1+ 2(0.013)] = 1

= 0.974 1.026

Table 3: Statistics of cointegration regression

Dependent variable:

Parameter Yellow Maize

Coefficient t-Statistic

Intercept -569.339 (25.499) -22.327

1.339 (0.0134)*

100.089

*Significant at both 1% and 5% significance level and Standard Errors parentheses

EG Error Correction Model

An EG ECM model was estimated in order to describe the short-run deviations from the long-run

equilibrium that has been shown to exist due to the presence of cointegration. In the short run

using the ECM derived from equation (4), futures prices are found to have a positive impact on

the spot prices. The significant coefficient of the error correction terms in Table 4 below

indicated disequilibrium for the commodity. However, the coefficients of the futures price for

both commodity = 0.7026 was found to be significant with p-value = 0.0000 as shown in table

Price Discovery in the South African Yellow Maize Futures Market

14

4 below in the long-run. On the other hand, the coefficient = -0.0139 of the error correction

term that guides the variables (spot and futures) to equilibrium was negative after estimations as

anticipated. This means that, the previous level of disequilibrium for both commodities was

corrected at the rate 1.4%. This validates that there exists a long-run equilibrium relationship

between variables spot and futures prices as stated in equation (2).

Table 4: Estimates of Error correction model (ECM)

Dependent variable:

Parameters Yellow maize p-value

Intercept 0.1093 0.8931

0.7026 0.0000

-0.0139 0.0043

Johansen’s Vector Error Correction Model (VECM)

A number of different lag lengths were applied in order to evaluate the behavior of the AIC and

SIC figures so as to ascertain the most appropriate lag length (which will be seen where the SIC

and AIC values are lowest). Choosing the lowest AIC or SIC the better the model, thus according

to the results presented below it can be observed that lag one has the second lowest AIC and

lowest SIC criterion values.

LAGS AIC SIC

1 19.012 19.035

2 19.015 19.053

3 19.011 19.065

As it has been noted that lowest figures obtained are in lag one, therefore, in this model the

Johansen’s VECM formed was based on one lag on both the spot and futures prices as shown in

table 7 below. Therefore in line with the findings of the EGECM it is observed that there is one

positive and one negative error correction term labeled CointEg1. This suggests that there is a

cointegrating relationship between the futures and the spot price which contains a price

discovery process. As it has been noted in the EG ECM model, the error correction term

Price Discovery in the South African Yellow Maize Futures Market

15

(0.010783) associated with the futures price in the VECM, shown on the top right column is

statistically significant though slightly smaller than the value from the EG ECM. This means

there is approximately 1.1% adjustment to the futures price in response to a unit shock to the spot

price.

Table 5: Johansen’s Vector Error Correction Estimates

Standard errors in ( ) & t-statistics in [ ]

Cointegrating Eq: CointEq1

SPOT_PRICE(-1) 1.000000

FUTURES_PRICE(-1) -1.424103

(0.12863)

[-11.0712]

C 727.3181

Error Correction: D(CASH_PRICE) D(FUTURES_PRICE)

Price Discovery in the South African Yellow Maize Futures Market

16

CointEq1 -0.006515 0.010783

(0.00567) (0.00422)

[-1.14980] [ 2.55433]

D(SPOT_PRICE(-1)) 0.068827 0.043114

(0.03188) (0.02375)

[ 2.15869] [ 1.81501]

D(FUTURES_PRICE(-1)) -0.086787 -0.044011

(0.04265) (0.03178)

[-2.03485] [-1.38506]

C 0.365044 0.344273

(0.95313) (0.71011)

[ 0.38299] [ 0.48482]

Granger-causality test

Price Discovery in the South African Yellow Maize Futures Market

17

Tables 8 indicate the different Granger causality tests employed at different lag lengths of the

futures and spot prices for commodity in study. As it can be noticed from the tables, both the

spot and futures prices Granger cause each other with spot prices being the leading Granger

causality contracts. Though in an ideal market anticipation of Granger-causality, it would be

expected that the futures prices would Granger-cause the spot prices but in this case both the

futures and spot prices Granger-cause each other at all the lags employed in this model. It can be

therefore be concluded that both the spots and futures contracts are more active, hence serving as

a better pricing guide for each other, that is, both futures and spots contracts have a two-way

feedback relationship between them.

Table 6: Granger causality test of yellow maize

Direction of

Causality

P-value Lags Decision Outcome

5.22% 1 Don’t reject null hypothesis doesn’t Granger cause

3.72% 1 Reject null hypothesis does Granger cause

1.73% 2 Reject null hypothesis does Granger cause

4.60% 2 Reject null hypothesis does Granger cause

2.50% 3 Reject null hypothesis does Granger cause

1.61% 3 Reject null hypothesis does Granger cause

CONCLUSION

This research conducted was set out to test the price discovery process between the spot and the

futures markets for yellow maize in South Africa. Two prominent tests developed by Engle and

Granger, and Johansen for cointegrations were conducted between spot and futures. Though

cointegration is a necessary but not a sufficient condition for market efficiency was found

between the spot and futures prices of the individually non-stationary price data series. The

presence of cointegration was an indication that interdependence between spot and futures price

series. This supports the findings of unbiasedness of the futures price in predicting the future spot

price. In this case the futures market is an optimal forecaster of future spot prices hence able to

play the role of price discovery.

Price Discovery in the South African Yellow Maize Futures Market

18

In order to reinforce the findings of equilibrium convergence of the price series, an ECM,VECM

and Granger causality tests were formulated. The ECM results indicated that there existed a

long-run equilibrium relationship between the variables of spot and futures price.The VECM on

the other hand, also showed that a long-run equilibrium relationship existed between the two data

series by showing that spot prices lead futures prices. The Granger-causality results also a two-

way feedback relationship between spot and futures prices meaning that a shock in either of the

market will lead the other.

Based on these findings, it can be established that the South African futures market for yellow

maize is performing its roles or functions in the price discovery process. This means that market

participants can use yellow maize futures price to predict the future spot prices and vice versa.

This in turn will help the participants hedge their yellow maize positions to reduce their price

risk. Therefore, farmers, commodity trades, regulators and policymakers should therefore excise

caution using both futures and spot prices to predict the other.

REFERENCES

Andersen, T.G.; Bollerselv, T.; Diebold, F.X. and Vega, C.(2007).“Real-time price discovery in

global stock, bond and foreign exchange markets.”Journal of International Economics.73,

pp. 251-277.

Aulton, A.J., Ennew, C.T and Rayner, A.J.(1997). “Efficiency Tests of Futures Markets for UK

Agricultural Commodities.” Journal of Agricultural Economics. 48(3), pp. 408-424.

Barnett, W.A. and Serletis, A.(2000). Martingales, nonlinearity, and chaos. Journal of Economic

Dynamics and Control. 24: 703-724.

Brooks, C.(2008). Introductory Econometrics for Finance. 2nd Edition. Cambridge: Cambridge

University Press.

Chamberlain, G.(1982).The general equivalence of Granger and Sims causality. Econometrica,

50:549–582.

Chan, K.; Chan, K.C. and Korolyi, G. A. (1991). Intraday Volatility in the Stock Index and Stock

Index Futures Markets”. The Review of Financial Studies.4(4), pp. 657-684.

Price Discovery in the South African Yellow Maize Futures Market

19

Chiang, R. and Fong, W.(2001). Relative information efficiency of cash, futures, and options

market: the case of an emerging market. Journal of Banking and Finance.25: 355-375.

Chowdhury, A.R.(1991). Futures market efficiency: evidence from cointegration tests. Journal of

Futures Markets. 11,5: 577 – 589.

Choy, S. and Zhang, H.(2010). “Trading costs and price discovery”. Review of Quantitative

Finance and Accounting.34(1), pp. 37-57.

Danthine, J.P.(1977). Martingale, market efficiency and commodity prices. European Economic

Review. 10,1: 1 – 17.

Danthine, J.P.(1978).Information, futures prices, and stabilizing speculation. Journal of

Economic Theory. 17,1: 79 – 98

Davidson, R. and Mackinnon, J.G.(1993). Estimation and inference in Econometrics. New York,

Oxford University Press, Inc.

Dickey, D.A. and Fuller, W.A.(1979.) Distribution of the estimators for autoregressive time

series with a unit root. Journal of the American Statistical Association. 74: 427 – 431

Dickey, D. A. and Fuller, W. A.(1981). “Likelihood Ratio Statistics for Autoregressive Time

Series Model Specification.”Econometrica, 48: 1057-1072.

Engle, R.F. and Granger, C.W.J.(1987). Co-integration and error correction: representation,

estimation, and testing. Econometrica. 55,2: 251 – 276.

Enders, W. (2004). Applied Econometric Time Series.2nd Edition. New Jersey: John Wiley &

Sons, Inc.,

Fama, E. F.(1970). “Efficient Capital Markets: A Review of Theory and Empirical Work.”

Journal of Finance, 25(2): 383-417.

Ferret, A. and Page, M.J.(1998). “Cointegration of South African Index Futures Contracts and

the Underlying Spot Market Indices.”Journal for Studies in Economics and

Econometrics.22(1), pp. 69-90

Price Discovery in the South African Yellow Maize Futures Market

20

Garcia, P., Irwin, S.H., Leuthold, R.M. and Yang, L.(1997). The value of public information in

commodity futures markets. Journal of Economic Behavior and organization.32: 559-

570

Hasbrouck, J.(1995). “One Security, Many Markets: Determining the Contribution to Price

Discovery.” The Journal of Finance.50(4), pp. 1175-1199

Hasbrouck, J.(2003 ). “Intraday Price Formation in U.S. Equity Index Markets.” The Journal of

Finance. 58(6), pp. 2375-2399.

Johansen, S.(1988). Statistical analysis of cointegration vectors. Journal of Economic Dynamics

and Control. 12: 231 – 254.

Johansen, S. and Juselius, K. (1990).Maximum likelihood estimation and inference on

cointegration – with application to the demand for money. Oxford Bulletin of Economics

and Statistics. 52,2: 169 – 210.

JSE Overview. (2014, September 29). Retrieved from https://www.jse.co.za/about/history-

company-overview

Kavussanos, M.G. and Nomikos, N.K. (2003). “Price Discovery, Causality and Forecasting in

the Freight Futures Market”. Review of Derivatives Research.6(3), 203-230

Kellard, N., Newbold, P., Rayner, T. and Ennew, C. (1999).The relative efficiency of commodity

futures markets. Journal of Futures Markets. 19,4: 413– 432.

Leng, H.M.J. (2002). “The South African Share Index Futures and Share Markets: Efficiency

and Causality Revisited.” Journal for Studies in Economics and Econometrics. 26(3), pp.

1-18.

LeRoy, S.F. (1989). Efficient capital markets and martingales. Journal of Economic Literature.

27,4: 1583 – 1621.

Leuthold, R. M., J.C. Junkus, and J.E. Cordier. (1989). The Theory and Practice of Futures

Markets. Lexington Books.

Price Discovery in the South African Yellow Maize Futures Market

21

Lipsey, R.G., Steiner, P.O., Purvis, D.D. and Courant, P.N. (1990). Economics (9e). New York,

Harper and Row Publisher.

Mackinnon, J.G. (1991.). Critical values for co-integration tests. Chapter 13 in Long run

Economic Relations: Reading in co-integration, edited by R.F. Engle and C.W.J. Granger.

Oxford University Press

Mahalik, M.K., Acharya, D. and Babu, M.S.(2009). “Price Discovery and Volatility Spillovers in

Futures and Spot Commodity Markets: Some Empirical Evidence from India.” Paper

Presented to Fourth Annual International Conference on Public Policy and Management:

Quantitative Approaches to Public Policy. August 9-12. Available online at:

http://www.igidr.ac.in/pdf/publication/PP-062-10.pdf

Pavabutr, P. and Chaihetphon, P.(2010). “Price discovery in the Indian gold futures market”.

Journal of Economics and Finance.34(4), pp. 455-467.

Pizzi MA, Economopoulos AJ, O’Neill HM. (1998). An examination of the relationship between

stock index cash and futures markets: a cointegration approach. J. Futures Markets. 18(3):

297-305.

Sabuhoro, J.B. and Larue, B.(1997). The market efficiency hypothesis: the case of coffee and

cocoa futures. Agricultural Economics.16: 171-184.

Schreiber, P.S. and Schwartz, R.A.(1986). “Price Discovery in Securities Markets,” Journal of

Portfolio Management 12, 43-48.

Schroeder, T.C., &Goodwin, B.K.(1999). Price Discovery and Cointegration for live hogs. The

Journal of Futures Markets, 11:685–696.

Sloman, J. (1991). Economics (3e). New York, prentice Hall.

Srinivasan, P. and Bhat, K.S.(2009). “Spot and Futures Markets of Selected Commercial Banks

in India: What Causes What?” International Research Journal of Finance and

Economics.31, pp. 28-40.

Price Discovery in the South African Yellow Maize Futures Market

22

So, R.W. and Tse, Y.(2004). “Price Discovery in the Hang Seng Index Markets: Index, Futures,

and the Tracker Fund”. The Journal of Futures Markets. 24(9), pp. 887-907.

Stoll, H.R. and Whaley, R.E. (1990). “The Dynamics of Stock Index and Stock Index Futures

Returns”. Journal of Financial and Quantitative Analysis.25 (4), pp. 441-468.

Wahab, M. and Lashgari, M.(1993). “Price Dynamics and Error Correction in Stock Index and

Stock Index Futures Markets: A Cointegration Approach”. The Journal of

FuturesMarkets.13(7), pp. 711-742.

Wang, H.H. and Ke, B.(2002). “Efficiency Tests of Agricultural Commodity Futures Markets in

China.” Australian Journal of Agricultural and Resource Economics.49(2), pp. 125-141.

Wiseman J.A., Darroch M.A., and Ortmann G.F.(1999). Testing the efficiency of the South

African futures market for white maize. Agrekon, 38(3), 321-335.

Yan, Y. and Reed, M.(2014). Price Discovery in the Chinese Corn Futures Market with

Comparison to Soybean Futures. Agribusiness an International Journal.

Yang, J., Bessler, D.A. And Leatham, D.J.(2001). Asset storability and price discovery in

commodity futures markets: a new look. Journal of Futures Markets. 21,3: 279 – 300.

Zelda A. and Efe-Omojevwe. (2013). A Study of the Efficiencies of Maize and Wheat Futures

Markets in India. IOSR Journal of Agriculture and Veterinary Science (IOSR-JAVS) e-

ISSN: 2319-2380, p-ISSN: 2319-2372.Volume2, Issue 4, PP 09-14 www.iosrjournals.org

Zulauf, C.R. And Irwin, S.H. (1997). Market efficiency and marketing to enhance income of

crop producers. [online]. Available: http://www.ace.uiuc.edu/ofor. Urbana-Champaign,

College of Agricultural and Environmental Sciences, University of Illinois.[Accessed 6

June 2001].

Zulauf, C.R. And Irwin, S.H. (1998). Market efficiency and marketing to enhance income of

crop producers. Review of Agricultural Economics.20,2: 308 – 331.