CC BY
65
Section 3. Marketing
DOI: http://dx.doi.org/10.20534/EJEMS-17-2-14-23
Akinbode S. O.,
Department of Economics, Federal University of Agriculture,
Abeokuta, Ogun State, Nigeria Corresponding Author E-mail: deleakinbode@yahoo.com Adekunle C. P.,
Department of Agricultural Economics and Farm Management, Federal University of Agriculture, Abeokuta,
Ogun State, Nigeria E-mail: goodlord4mercy@yahoo.com
Price Transmission and Market Integration of Yellow Maize in Southwest, Nigeria: Co-integration and Vector Error Correction Model Approach
Abstract: Yellow maize constitutes the bulk of world production and international maize trade while Nigeria is the largest maize producer in Africa but the knowledge about the mechanism of its price transmission is limited thereby denying the economy the advantages derivable from such. This study examined price transmission and integration in yellow maize markets in Southwest Nigeria within the framework of cointegration and Vector Error Correction Model (VECM) using monthly rural and urban retail market price data between January 2004 and December 2015 obtained from Agricultural Development Programmes in Lagos, Oyo and Ogun States. The study found that the price series were generally integrated of order one i.e I (1) series, with one cointegrating equation existing among their linear combinations and results based on normalization of the restricted VAR system (VECM) in respect of urban market price of yellow maize in Lagos State and its determinants revealed that rural price of maize in Lagos State at p<0.01 and urban price of maize in Oyo State at p < 0.05 exerted significant and positive influence on Urban Price of Maize in Lagos State in both the long run and the short run. The equilibrium relationship was found to be stable, with exogenous shocks being corrected within 54 days. The speed of adjustment from the short run to the long run equilibrium was low. This suggested that there was weak integration and price transmission relating to the reference Lagos urban market and this could be attributed to bad communication among the various yellow maize markets in Southwest Nigeria. It was recommended that government should come up with efficient pricing policy and improvement of infrastructure which may aid price transmission and integration in maize markets, and help industry players understand price behaviour. This could enhance agricultural development and food security while ensuring good returns to actors along the market channel.
Keywords: Market, yellow maize, cointegration, speed of adjustment, Vector Error Correction model.
Introduction grown in the southern part of the country is rain fed.
Maize is one of the staple crops widely grown in Meanwhile, due to the scanty rainfall in the northern part
Nigeria. It is produced in abundance in the Northern of the country, maize growers support their production
part of the country compared with other regions. Maize with fertilizer and irrigation. Maize as a cereal crop is
high yielding, easy to process and readily digested. It is a versatile crop that grows across a range of agro ecological zones. That explains why almost all farmers in Nigeria grow maize. Maize is one of the most important cereals in the world alongside rice, wheat and millet but the most important in Africa [1].
Maize has grown to be a local "cash crop" especially in the southwest part of Nigeria where at least 30 percent of the cropland has been put to maize production under various cropping systems [2]. Growing maize in farms of 1-2 hectares can overcome hunger in the household and the aggregate effect could double food production in Nigeria [3].
Nigeria is the 10th largest producer of maize in the world, and the largest producer in Africa [3]. While maize (both yellow and white varieties) are grown across the country the North Central region is the main producing area. In Nigeria, Lagos and Kano represent the two main centers where goods are marketed due to their proximity to the two most active borders for informal trade between Nigeria and Benin and between Nigeria and Niger, as well as due to the proximity between Lagos and the ports of Lomé and Cotonou [4].
Maize is traded at both local and international levels with a considerable percentage filtering from Nigeria into Niger, Chad, Mali, Benin Republic and some other countries in the West African sub-region and sometimes vice versa. White and yellow maize are sold almost in all markets in Nigeria with the main markets for the commodity being Dawanau market in Kano, Dandume and Jibia markets in Katshina, Giwa market in Kaduna, Shinkafi and Talata Mafara in Zamfara, Bodija market in Ibadan, Osi market in Onitsha and Mile12 market in Lagos.
North Central Region produces one third of maize in the country. Although, most of processing facilities are in the South West (Lagos and Ibadan) and in the North (Kaduna and Kano) [5]. Furthermore [6], identified Lagos and Ibadan as the main wholesale and retail maize markets where imported and locally produced yellow maize compete, particularly for the feed (poultry) industry [7]. confirmed the North-Central Region (and the Central Belt in general) as the main surplus area in Nigeria, with flows directly from the North to the deficit areas, mainly towards the South of the country, as well as to neighboring countries.
A market system in which there is synchronous movement ofprices in different market locations over time is said to be integrated [1]. Market integration is a concept with application in spatial, temporal and product market inter-
relatedness. Without market integration, price signals will not be transmitted from food deficit (i. e. import) to food surplus (export) markets. Producers will fail to specialize according to comparative advantage and gains from trade will not be achieved [8]. This is due to the fact that, if traders do not have up to date information about prices in other markets, they will not respond quickly to profitable opportunities. This therefore impedes the process of spatial arbitrage that transmit price from one market to another.
Meanwhile, there is little knowledge about the pattern of price transmission and market integration for yellow maize in southwest region of Nigeria. This paper is therefore conceptualized to bridge this knowledge gap by determining the order of integration of the retail prices; determine the extent of cointegration between rural and urban prices, and, determining the speed of price transmission between the markets. Findings from this paper is expected to be useful for agricultural and food policy makers, farmers, agricultural product marketers and operators of agro-based industries as regards policy formulation and planning among others uses. The rest of the paper is organized as follows: methodology, results and discussion and conclusion. Methodology Study Data
Monthly data on rural and urban market prices of yellow maize in Naira per Kilogram (N/kg) from January 2004 to December 2015 were obtained from the Lagos State Agricultural Development Programme (LASADA), Oyo State Agricultural Development Programme (OYSADEP) and the Ogun State Agricultural Development Programme (OGADEP). However, Lagos State urban maize market was selected as the reference market because it is the most urban terminal market in Southwest Nigeria.
Model Specification
It is assumed for this study that urban prices of maize in Lagos State is determined principally by their rural prices and the urban and rural prices in the neighbouring states (Ogun and Oyo). The model for estimation in this study is specified in double logarithmic form as follows:
InPU = a + B.InPrt +XInPytu +
It ! 1 It It t \
+nInPyr + tylnPt" + InPg InPt = log of urban price of yellow maize in Lagos State (N/kg)
InPt = log of rural price of yellow maize in Lagos State (N/kg)
InPy = log of urban price of yellow maize in Oyo State (N/kg)
InPy = log of rural price of yellow maize in Oyo State (N/kg)
InPg = log of urban price of yellow maize in Ogun State (N/kg)
InPg = log of rural price of yellow maize in Ogun State (N/kg)
Analytical Techniques Stationarity Test
A stationary time series is a type of series whose statistical properties such as mean and variance are constant over time and non-stationary time series are those having time dependent statistical properties [9]. In effect, a stationary series has a finite variance, transitory innovations from the mean and a tendency for the series to return to its mean value. Stationarity test involves the use of procedures such as Dickey-Fuller (DF) test and the Augmented Dickey-Fuller (ADF) test [10]. If one identifies the series to be non-stationary, the first difference of the series is tested for stationarity to determine the order of integration. The number of times (d) a series is differenced to make it stationary is termed as the order of integration, I (d). In this study, the ADF test was used to determine the data properties due to the advantages it possesses and its common application in the time series literature. The ADF test as mentioned considers the null hypothesis that a given series is non stationary. The test is applied by running a regression of the following form: Ap =d +d2Pt-! + APt - + eit (2) Where A is the difference operator and Pi denote price series of the yellow maize markets and i = 1, 2 ..., 3 (1-Lagos; 2-Ogun; 3-Oyo) at different time t.
If the coefficient is not statistically different from zero, then the series has a unit root and therefore is non-stationary. Johansen's Cointegration Test If a linear combination of two non-stationary series is stationary, then the two series are considered to be coin-tegrated [9]. The ADF test which is a test for stationarity is supplemented by Johansen and Juselius maximum likelihood method. This method is preferred to the others because it addresses endogeneity and simultaneity problems associated with other bivariate models as well as its ability to test more than two variables at a time. Here, a hypothesis of the presence of cointegrating vector is imposed on a group of stationary series, as the hypothesis of reduced rank of the long run impact matrix. Maximum likelihood tests are applied to derive test statistic for the hypothesis of a given number of co integrating vectors and their weights. The specific linear combinations to be tested are the residuals from a static cointegrating regression as:
Apt =a + X rk AP-k + npt +e, (3) Vector Error correction Model An Error Correction Model (ECM) is an efficient way of combining the long run cointegrating relationship between the levels variables and the short run relationship between the first differences of the variables. It has the merit that all the variables in the estimated equation are stationary; thus there is no problem of spurious regression. The procedure of differencing results in the loss of valuable long run information in the data and so an error correction term is introduced in the theory of cointegration to link the short run dynamics of the series with its long run value. The residuals obtained from the equation are introduced as explanatory variables into the system of variables in the short-run model. The error correction term thus captures the adjustment towards long run equilibrium.
[11] demonstrated that once a number of variables are found to be cointegrated, then, there existed a corresponding error correction representation which implied that changes in the dependent variables are a function of the level of disequilibrium in the cointegrating relationship as well as changes in other variables. An error correction model is specified to relate the changes in the dependent variable to the independent variable as well as the error correction term where the error correction term measures the deviation from the long run equilibrium. VECM that captures the interactions between Lagos urban yellow maize prices and determinant variables takes the following form following [12]:
Mnp; =a+o (( -pinp;^)+
+SMnP;V) + pAInP(;.1) + et
Where:
InPU = log of urban price in Lagos State InPt"'r = log of rural prices in Lagos, urban and rural prices in Oyo, and, urban and rural prices in Ogun states (the hypothesized determinants) A = difference operator, so APt = Pt - Pt-1 a is the constant term, ft is the vector of coefficients of the long run model, 0 is the error correction term while p and S are estimated short-run parameters, and st = error term
Results and Discussions Summary Statistics and Trend of the price series The trend graphs (Figures 1-6) show the trend of both rural and urban price series in the three selected states of Lagos, Ogun and Oyo all located in the south-western part of Nigeria. A careful look at the graphs revealed that prices in each of the states (rural
and urban) seemed to have specific pattern peculiar to each of them. For instance, prices in Lagos state (both rural and urban prices) had sudden upward surge between 2007 and 2009. Ogun price series (rural and urban) had similar pattern of gentle upward trend with URBAN LAGOS
RURALOGUN
RURALOYO
Figure 1. Trends of price series of yellow
The descriptive statistics of the price series shown in table 1 revealed the mean, median, maximum, minimum, standard deviation, skewness and kurtosis of the series. The Jaque-Berra test revealed that out of the six series, Lagos urban, Lagos rural and Ogun urban price series were normally distributed while Ogun rural, Oyo rural
some fluctuations visible in years 2010 and 2013. Oyo state price series showed trending and upward fluctuations from 2008 to 2015. However, the obvious is that the series generally showed upward movement over time.
RURAL LAGOS
URBAN OGUN
URBAN OYO
maize in rural and urban south-west Nigeria
and Oyo urban were not normally distributed as indicated by the probability level. Stationarity of the price series The unit roots tests of price series were undertaken to ascertain the order of integration or test for the stationarity of the prices. The Augmented Dickey Fuller
(ADF) unit root test procedure was adopted in this case. plied that inclusion of first differences as variables in the The result (Table 2) indicated that the price series were model, instead of normal price series, will eliminate the stationary at first difference i.e 1 (1). This result im- stochastic trend to which the nominal series are exposed.
Table 1. - Descriptive statistics of the price series
Pu Pr Pgu Pgr Pyu Pyr
Mean 95.70512 85.75130 61.29417 57.44759 62.92435 57.53213
Median 91.35000 86.76000 61.68000 60.43500 66.75500 61.91000
Maximum 176.4706 158.0000 80.35000 75.21000 112.1600 98.46000
Minimum 42.25000 38.58000 43.95000 39.78000 26.10000 23.54000
Std. Dev. 35.55234 30.20679 9.955939 11.60053 21.48134 20.86865
Skewness 0.270795 0.306614 - 0.197003 - 0.189859 - 0.132232 - 0.093677
Kurtosis 2.242880 2.542823 1.983064 1.661747 1.687566 1.704551
Jarque-Bera 3.899479 2.632764 5.352297 8.707976 8.065910 7.709807
Probability 0.142311 0.268104 0.068828 0.012855 0.017722 0.021176
Sum 10336.15 9261.140 6619.770 6204.340 6795.830 6213.470
Sum Sq. Dev. 135244.6 97632.19 10605.92 14399.25 49374.95 46598.56
Observations 108 108 108 108 108 108
Source: Authors' computation, 2016
Table 2. - Augumented Dickey-Fuller Unit Root Tests
Level I (0) First Difference I (1)
No Intercept no trend With Intercept No Trend With intercept and Trend No Intercept no trend With Intercept No Trend With intercept and Trend I (d)
InP" 0.3127 - 1.8926 - 2.5415 - 11.9586*** - 11.9322*** - 11.8948*** I (1)
InPr 0.3092 - 1.9601 - 2.8833 - 12.4094*** - 12.3824*** - 12.3278*** I (1)
InPgu 1.0331 - 1.7842 - 2.1224 - 10.8897*** - 10.9713*** - 10.9341*** I (1)
InPgr 1.0160 - 1.4831 - 2.3240 - 8.7191*** - 8.8321*** - 8.7924*** I (1)
InPyu 0.5487 - 2.4026 - 2.6617 - 9.0403*** - 9.0279*** - 9.0217*** I (1)
InPyr 0.8365 - 1.7654 - 2.4839 - 8.3398*** - 8.3670*** - 8.3101*** I (1)
***, ** and * denote rejection of the null hypothesis at the 1%, 5% and 10% significance levels.
The respective critical values at the 1%, 5% and 10% significance levels are - 3.49, - 2.89 and - 2.57 for the ADF test.
Source: Data analysis 2016
Cointegration of the price series
The Johansen cointegration test indicated one coin-tegrating vector at 5% levels of significance (Table 3). This result implied that rural and urban market prices of yellow maize in Southwest Nigeria were integrated. Though, the price changes may vary in the short run between the different levels they were expected to move together as a system in the long run. This necessitated the estimation of the movement ofprices in the long and short run, using Vector Error Correction Mechanism (VECM).
The Johansen test indicated that there was significant long-run relationship. i. e. examination of the results in Table 3 shows that the null hypothesis of no cointegra-
tion in the model was rejected by both the trace and maximum eigen value tests. This shows that price of yellow maize in Lagos State Urban Markets (lnPu) and the hypotheized determinants rejected the null hypothesis of no cointegration at p < 0.05. This implies that although the urban market price of yellow maize in Lagos State and their hypothesized determinants are generally I (1) series, some stable long run equilibrium relationship existed among the series, which could be given some error correction representations [11]. It also showed that the finding of no causality in the relationship between them in the [13] sense was ruled out [14; 15]; just as the possibility of the estimated relationship being spurious was also ruled out as asserted by [16].
Table 3. - Results of Cointegration Tests
Hypothesis Eigenvalue Trace statistic 0.05 Critical Value Prob Hypothesis Max-Eigen Staitstic 0.05 critical value Prob
r = 0* 0.288927 95.25892 94.75366 0.0431* r = 0* 40.12099 35.07757 0.0491
r = 1 0.249160 59.13793 69.81889 0.2628 r = 1 29.51602 33.87687 0.1519
r = 2 0.137194 29.62191 47.85613 0.7380 r = 2 15.19926 27.58434 0.7323
r = 3 0.096680 14.42264 29.79707 0.8161 r = 3 10.47288 21.13162 0.6994
r = 4 0.032280 3.949758 15.49471 0.9075 r = 4 3.379661 14.26460 0.9183
r = 5 0.005520 0.570098 3.841466 0.4502 r = 5 0.570098 3.841466 0.4502
Source: Data analysis, 2016
Direction of Causality between Rural and Urban Market Prices in the Selected States
In time series analysis, the most frequently asked question is whether or not one economic variable can help in the prediction of another. When two series are stationary and cointegrated, one can also test for granger causality. This is due to the fact that at least one Granger causal relationship exist in a group of cointegrated series. Table 4 shows the pairwise Granger causality of yellow maize prices in the various markets considered. The result shows that out of the fifteen (15) maize market links investigated for evidence of granger causality, nine (9) market links rejected their respective null hypothesis of no Granger causality out ofwhich 2 were bi-directional while seven showed unidirectional causality. For bi-directional granger causality, it was found that Urban Price of Maize in Oyo State (InPyu) granger caused Rural Price of Maize in Lagos State (InPr) and vice versa. In the same vein, Urban Price of Maize in Oyo State (InPyu) granger caused Urban Price of Maize in Lagos State (InPu) and vice versa. This is an evidence of existence of transmission mechanism between Oyo State Urban yellow maize markets and Lagos state maize markets. It implied that there existed a strong and instantaneous feedback mechanism from Lagos to Oyo and vice versa.
However, InPyu and InPu i.e Oyo and Lagos urban had strong exogeneity over rural and urban markets of other States. Few of the market series were spatially linked by trade. Therefore, there was only moderate market integration between rural and urban yellow maize markets. The implication of this is that price changes in one market are not manifested to an identical price response in some of the other markets. There was also inadequate free flow of maize prices between markets and the markets were only moderately linked by efficient arbitrage. However, the results obtained here means that maize market participants in southwest Nigeria, namely producers, retailers
and consumers, have not effectively use information in the urban maize market prices in determining the rural prices. The result of the Granger causality test confirmed that InPu and InPyu occupied the leadership position in price formation and transmission. This is because prices formed therein were somehow transmitted to the other (follower) markets (Ogun State rural yellow maize market, Ogun State urban yellow maize market, Oyo State rural yellow maize market and Lagos State rural yellow maize market) with minor distortions during the transmission process thereby corroborating [5] and [6].
Long run price integration in the selected markets
The existence of cointegration between the dependent variables and their hypothesized determinants including the establishment of granger causalities among the various series necessitated the specification of Vector Error Correction Model for this study. The estimated long run relationship based on normalization in respect of the Urban prices of yellow maize in Lagos State (lnPu) and their hypothesized determinants is presented in Table 5.
The results showed that Rural Lagos State- InPr (at 1 percent), Oyo State urban prices- InP7" (at 5 percent) and Oyo State rural prices - InPyr (at 5 percent level) were the significant variables affecting the prices of yellow maize in Lagos State urban maize markets in the long run. This might be due to the traditional flow channel of the products from Oyo and other neighboring states to Lagos state in general. Similar relationship was suggested between Oyo prices and urban Lagos in the granger causality tests (Table 4). This was in line with the assertions of [5] and that of [6] that Lagos and Ibadan (Oyo state) were the main maize processing points in Nigeria where local and imported maize compete in the markets. Ogun state prices were not significant in the long run possibly because Ogun state is located between Oyo and Lagos state, therefore, it might be considered as only a transit point as the prices were not emanating from it. The results showed that
a 1 percent increase in Lagos State Rural prices increases prices in the Lagos state urban markets prices by 1.07 percent. A complete (107%) ofproportional change in Lagos Rural Market was transmitted to Lagos Urban Market in the long run. This can be classified as an efficient price
transmission system. However, Oyo urban and Oyo rural prices caused 0.33 percent and 0.34 percent increase in Lagos urban prices respectively as a result of 1 percent increases. This could be considered as a weak transmission link between Oyo State and Urban Lagos maize market.
Table 4. - Pairwise Granger Causality Result of Maize Markets
Null Hypothesis: Obs F-Statistic Prob.
InPr does not Granger Cause InPu 144 2.02542 0.1373
InPu does not Granger Cause InPr 3.28313* 0.0416
InPgu does not Granger Cause InPu 144 0.77827 0.4619
InPu does not Granger Cause InPgu 3.19902* 0.0450
InP®r does not Granger Cause InPu 144 2.25133 0.1105
InPu does not Granger Cause InPgr 3.01791 0.0533
InP7" does not Granger Cause InPu 144 3.49365* 0.0341
InPu does not Granger Cause InPyu 3.10016* 0.0494
InPyr does not Granger Cause InPu 144 4.30500* 0.0161
InPu does not Granger Cause InPyr 1.85571 0.1616
InP^ does not Granger Cause InPu 144 0.88014 0.4179
InPu does not Granger Cause InPgu 2.84965 0.0625
InP? does not Granger Cause InPr 144 2.83371 0.0635
InPr does not Granger Cause InP®r 3.07737 0.0504
InPyu does not Granger Cause InPr 144 3.72595* 0.0275
InPr does not Granger Cause InPyu 3.59741* 0.0310
InPyr does not Granger Cause InPr 144 4.75937 0.0106
InPr does not Granger Cause InPyr 2.99365 0.0546
InP? does not Granger Cause InP®u 144 4.58954* 0.0124
InP^ does not Granger Cause InP®r 0.66733 0.5153
InPyu does not Granger Cause InP®u 144 0.44246 0.6437
InP^ does not Granger Cause InPyu 2.87967 0.0608
InPyr does not Granger Cause InP®u 144 0.16557 0.8476
InP^ does not Granger Cause InPyr 2.79461 0.0659
nPyu does not Granger Cause InPgr 144 0.02307 0.9772
InP? does not Granger Cause InPyu 2.30593 0.1049
InPyr does not Granger Cause InP®r 144 0.10868 0.8971
InP? does not Granger Cause InPyr 3.07943* 0.0503
InPyr does not Granger Cause InPyu 144 6.60389** 0.0020
InPyu does not Granger Cause InPyr 0.81335 0.4463
*sig at 5 percent level, **significant at 1 percent level Source: Authors' Computation, 2016
Table 5. - Long run Relationship of the price series with reference to Lagos urban
Variables Coefficient t-value
LUPMLS (- 1) 1.000000
LRPMLS (- 1) - 1.079115*** - 32.3241
LUPMOGS (- 1) 0.129953 0.75311
LRPMOGS (- 1) - 0.075721 - 0.58602
LUPMOYS (- 1) 0.330362** 2.41035
LRPMOYS (- 1) 0.304137** 2.44699
Constant 0.154010
***, ** and * imply significant at 1%, 5% and 10% respectively Source: Authors' computation, 2016
The Short-run model (VECM estimation)
Examination of the F-statistics and the adjusted R2 in Table 6 suggested that some variables in the VECM significantly explained short run changes in Lagos Urban market (lnPu) at the acceptable significance level accounting for about 49.7 percent of the short run variation in the series.
Table 6. - The Short-Run Vector Error Correction Model Estimates Dependent Variable: Log of Lagos Urban Market Prices
According to [17], the Error Correction Model (ECM) helps to determine if the Law of One Price (LOP) of a particular good in markets in different locations holds in addition to revealing the speed with which prices adjust to changes in other locations.
Variables D (InPu) D (InPr) D (InPgu) D (InPgr) D (InPyu) D (InPyr)
ECM (- 1) - 0.5530** (- 2.5725) 0.0051 (0.0143) 0.0443 (0.3626) 0.0838 (0.5573) 0.1588 (0.4540) - 0.3105 (- 0.8996)
D (lnPu (- 1)) - 0.0694 (- 0.1920) - 0.2803 (- 0.7730) - 0.1184 (- 0.9424) 0.0624 (0.4036) - 0.2673 (- 0.7436) - 0.4705 (- 1.3257)
D (inPu (- 2)) 0.4901** (2.5404) 0.3427 (1.0742) - 0.0617 (- 0.5585) - 0.0290 (- 0.2131) - 0.5030 (- 1.5902) - 0.4412 (- 1.4129)
D (lnPr (- 1)) 0.1250** (2.3361) 0.0108 (0.0289) 0.1252 (0.9685) - 0.0920 (- 0.5783) 0.3629 (0.9811) 0.5227 (1.4315)
D (lnPr (- 2)) 0.4949** (2.5590) - 0.2489 (- 0.7819) 0.1655 (1.5007) 0.0973 (0.7170) 0.5034 (1.595) 0.4862 (1.5605)
D (inPgu (- 1)) - 0.0879 (- 0.2249) - 0.2667 (- 0.6805) - 0.357 (- 2.6296) 0.0290 (0.1737) 0.4679 (1.2041) 0.4758 (1.2404)
D (inPgu (- 2)) - 0.4337 (- 1.1097) - 0.4667 (- 1.1906) - 0.0825 (- 0.6078) - 0.2339 (- 1.3995) 0.2195 (0.5649) 0.1985 (0.5172)
D (inP^ (- 1)) 0.2873 (0.8704) 0.4909 (1.4833) 0.3706 (3.2321) 0.1482 (1.0499) - 0.1805 (- 0.5499) - 0.2491 (- 0.7690)
D (inPs- (- 2)) 0.2453 (0.7344) 0.2400 (0.7162) - 0.0902 (- 0.7773) - 0.1377 (- 0.9639) - 0.0505 (- 0.1521) 0.0223 (0.0680)
D (inP7" (- 1)) 0.2560** (2.2790) - 0.1773 (- 0.8831) 0.0759 (1.0912) - 0.0270 (- 0.3157) - 0.1474 (- 0.7406) 0.2031 (1.0336)
D (lnP7" (- 2)) 0.2413** (2.3006) - 0.2496 (- 1.3410) 0.1825 (2.8308) 0.0171 (0.2167) - 0.2891 (- 1.5669) - 0.0631 (- 0.3463)
D (inP71 (- 1)) - 0.0038 (- 0.0193) - 0.0400 (- 0.2008) - 0.0515 (- 0.7453) 0.0623 (0.7330) 0.3306 (1.6723) - 0.0796 (- 0.4079)
D (inP71 (- 2)) - 0.0252 (- 0.1329) - 0.0531 (- 0.2796) - 0.1783 (- 2.7124) - 0.0806 (- 0.9959) 0.2337 (1.2422) 0.0874 (0.4703)
Constant 0.0130 (1.0433) 0.0137 (1.0887) 0.0046 (1.0530) 0.0068 (1.2651) 0.0048 (0.3872) 0.0080 (0.6533)
R-squared 0.5206 0.2420 0.2483 0.1679 0.1645 0.1829
Adj. R-squared 0.4967 0.1337 0.1409 0.1406 0.1354 0.1166
F-statistic 18.5601 2.2348 7.8756 2.4129 2.3787 1.5664
Log LF 76.5539 46.2514 187.55 165.72 77.15 78.49
Akaike AiC - 14.9592 - 1.1857 - 3.3058 - 2.8901 - 1.2029 - 1.2286
Schwarz SC - 12.6843 - 2.9623 - 2.9519 - 2.5361 - 0.8490 - 0.8747
Source: Author's computation, 2016
The error correction coefficient in the Lagos urban market equation was significant at p < 0.01, less than one and it was associated with the desirable negative sign. This shows that prices in Lagos State urban yellow maize market adjusted significantly to shocks to its equilibrium
relationship with its hypothesized determinants which were caused by exogenous changes in in the explanatory variables included in the model. Past values of Lagos Urban prices (lnPu (-1)), Lagos Rural prices (lnPr) and prices in Oyo state urban markets (lnPuy) were found to
be significant within acceptable risk level (Table 6). The error correction coefficient value of - 0.553 implied that about 55.3 percent of the effects on Lagos Stae urban yellow maize price of shocks that destabilize the equilibrium relationship between it and the hypothesized determinants in the previous year were corrected in the current year. This suggest that since monthly data were utilized for the study it may take about 54 days (close to two months) for the system to restore back to the long run equilibrium after an exogenous shock. The speed of adjustment of 55.3 percent from the short run to the long run equilibrium is moderate compared with a perfect adjustment of 100% threshold. This indicated that there was moderate integration and price transmission relating to the reference urban market and this could be attributed to improved communication among the various maize markets in southwest Nigeria both in the short and long run. This was considered moderate given that monthly data were utilized for the study. Higher frequency (e.g weekly or daily) data might have being more informative in this regard. Focusing on the short term coefficients (elasticities) results on Table 6, Oyo Urban had elasticities ofabout 0.24 while two month lag of Lagos Urban, one month lag of Lagos Rural and two month lag ofLagos Rural had elasticities values of 0.49, 0.125 and 0.495 respectively.
Conclusion and Recommendation
Based on recent development in time series modeling that points to a need to review previous research efforts aimed at explaining price behaviour in southwest Nigeria, this study adopted Vector error correction modeling framework in analysing price transmission and market integration of yellow maize markets in southwest Nigeria. Using monthly retail rural and urban markets prices of maize in Naira per Kilogramme (N/kg) from January 2004 to December 2015 obtained from States Agricultural Development Programmes (ADPs) in Lagos, Ogun and Oyo.
It was found that the price series were stationary at first difference with one (1) cointegrating equation existing among their linear combinations i. e. one cointe-gration vector in all variants of the system specified. It was also found from the VECM estimations that positive relationship and complete price transmission existed between urban prices in the reference State and its rural price in the long run (i.e between Lagos state urban and Lagos state rural markets) possibly due to the metropolitan nature of Lagos state which may facilitate quick price transmission. On the other hand, there was weak but significant transmission between prices in Oyo state and the reference market (Lagos urban maize market). Meanwhile, Ogun state prices (both rural and urban) did not significantly affect prices in the reference market. There was a weak relationship between the pairs of market prices of yellow maize as this commodity is mainly supplied by the nearby local farmers. Pairwise granger causality results confirmed that Lagos urban market and Oyo urban market have strong exogeneity over other rural and urban market price series and could be classified as the lead markets.
It is therefore recommended that when it is desired that a national pricing policy for increased consumption or production ofyellow maize be implemented, the identified leader markets (Lagos State Urban Market and Oyo State Urban Market) should be targeted. This is because prices formed in these two markets are efficiently transmitted to the other markets with very minor distortions during the transmission process. A price-based incentive policy can only have a long term positive impact on local yellow maize production if it is embedded in a strategy that enhance the provision of infrastructures such as good road networks, market structures and efficient market information network systems.
References:
1. Jonah E., Oluchi R. A. and Jonshon T. E. Economic Analysis of Maize Production in South-East Nigeria. African Journal ofAgricultural Research, - 2014, - 10 (2):170-179.
2. Degrande A., Duguma B. Adoption Potential of Rotational Hedge Row Intercropping in Humid and Extension Network. - 2000.
3. United States Agency for International Development (USAID) and Maximizing Agricultural Revenues and Key Enterprises in Targeted sites (Markets).Package of Practices for Maize Production in Nigeria. - 2010.
4. Cadoni P., Angelucci F., Analysis of incentives and disincentives for Maize in Nigeria. Technical notes series, MAFAP, FAO, - Rome. - 2013.
5. Hartwich J., Devlin P., Kormawa I. D., Bisallah B. O., Odufote I. M., Polycarp I. Unleashing Agricultural Development in Nigeria through Value Chain Financing", UNIDO - 2010.
6. United States Department ofAgriculture (USDA) Foreign Agricultural Service. GAIN Report (Global Agriculture Information Network), Nigeria G^AIN, Grain and Feed Annual. FAO Special Report - Markets, Prices,
Food Situation and Prospects for Benin, Niger and Nigeria. Based on a CILSS/FAO/FEWSNET/SIMA/WFP Joint Market Assessment Mission to Benin, Niger and Nigeria. - 2009.
7. Famine Early Warning Systems Network-FEWSNET. Ethiopia Food Security Specialist short-term consultant Scope of Work. USAID - 2012.
8. Baulch B. Transfer costs, spatial arbitrage, and testing for food market integration, American Journal of Agricultural Economics, - 1997. - 79 (2): 477-482.
9. Gopal M. S., Shanmugam T. R., Raveendaran N., and Rajan G. K., "Co integration and Market Integration": An Application to Indian and International Black Tea Auction Markets. Journal of Asian Business Management, ma - 2009. - 1 (2):193-205.
10. Dickey D. and W. Fuller. "Distribution of the Estimators for Autoregressive Time Series with a Unit Root," Journal of the American Statistical Association, - 1979. - 74, 427-431.
11. Engle R., and Granger C. Cointegration and error-correction: Representation, estimation, and testing. Econo-metrica. - 1987. - 55 (2): 251-256.
12. Minot N. Transmission of World Food Price Changes to African Markets and its Effect on Household Welfare Paper to be presented at the Comesa policy seminar Food Price Variability: Causes, Consequences, and Policy options. - 25-26 January - 2010 in Maputo, Mozambique under the Comesa-MSU-IFPRI African Agricultural Markets Project (AAMP). - 2010. - 53.
13. Granger W.J. Investigating casual Relations by Econometric Models and Cross-Spectral Methods. Econometric - 1969. - 16: 121-130.
14. Granger C. W. J. Developments in the study of cointegrated economic variables. Oxford Bulletin of Economics and Statistics, - 1986. - 48, 213-228.
15. Granger C. W.J. and Lee T.-H. Investigationofproductionsales and inventory relationships, using multicointegra-tion and non-symmetric error correction models.Journal ofApplied Econometrics, - 4, Suppl. - 1989. - 145-159.
16. Masih A., and Masih R., Are Asian stock market fluctuations due mainly to intraregional contagion effects? Evidence based on Asian emerging stock markets, Pacific-Basin Finance Journal, - 1999. - 7, 252-82.
17. Jaramillo J. L., Naude A. Y and Cote V. S. Spatial Integration of Mexico and United States in Grain Market: The Case of Maize, Wheat, and Sorghum. International Association ofAgricultural Economist Triennial Conference. Brazil. - 2012.
