Abstract

Food commodities prices have increased over the recent decade attracting the attention of market participants and policy makers. The short period between the recent price surges has therefore drawn concerns and raised questions on the causes and future prospects of commodity markets. Instability and variation in food production over years contribute immensely to food insecurity. This study focuses on the objectives include: to explore the trends of prices of selected agricultural food commodities and to examine the short run and long run relationship between the prices of the selected agricultural food commodities. In order to examine the series, the unit root tests (ADF and Phillips-Perron tests), identification of the number of lags and co-integration analyses were conducted. Unit root tests indicate that all series are non-stationary at level and are stationary (first difference) at 5 percent significant level. The Johansen co-integration test suggests that there is at least one co-integration vector, which describes the long run relationship between prices of Rice, white Teff, Maize and Sorghum. The appropriate number of lag identified was two. The VECM analysis findings showed the existence of integration between the prices and revealed that increase in prices of maize and sorghum in the past will decrease price of Rice in the long run. Similarly, the result of VECM showed as prices of white Teff and Rice have positive relationship.

1. Introduction

In recent years, price fluctuations and price spikes on global commodity markets have increasingly come to the forefront of public attention. In 2008, the world experienced a dramatic rush in the prices of commodities. The prices of food commodities, in particular maize, rice and wheat increased dramatically from late 2006 through to mid-2008, reaching their highest levels in nearly thirty years. Prices stabilized in the summer of 2008 and then decreased sharply in the midst of the financial and economic crisis ¹. Food price volatility is the variation in agricultural prices over time. There are multiple causes of food price dynamics including the supply and demand of crops, drought and other adverse weather conditions, and financial speculation. However, ignoring the source of price peaks and recognizing just the fact of high food prices, developing countries suffer the most from rising food prices, since poor people need to spend more of their disposable income for food.

The rapid rises of food prices, now days, are at the top agenda of the international community. The people of developing countries, particularly, are highly exposed to food insecurity as they are financially incapable to afford basic food crops. ² indicated that the high food price inflation has been the most adverse economic shock that has continued to adversely affect the Ethiopian economy where significant proportion of households had to adjust food consumption in response.

In the last decade, particularly since 2007/2008, the world food prices have experienced rapid increase volatility ³. The food price variability experienced had caused risk related agricultural policies to be debatable. According to ⁴, the rise in volatility experienced in the international markets has successfully exposed the European Union’s domestic prices to international price signals. It was further explained that the high agricultural commodity prices have suppressed the market measures such as border protection, subsidies, production quotas and the likes, and failed to achieve their aims. According to ⁵, price inflation is characterized by unexpected price changes which involved risk to farmers that react to it by reducing their output supply and investments in productive inputs. Evidence has shown that the effect of price volatility in the global markets is not limited to farm gate, but it has extended its tentacle to the downstream sector, that is, the consumers ⁶.

In Ethiopia, inflation was not a concern until 2002/03 ⁷, but afterwards began to increase at an alarming rate. For instance, the annual average food inflation rate between July 2011 and June 2012 was 36.9% and in October 2012 compared to October 2011 was 30.4% ⁸.

Food price inflation has been soaring throughout the country and the regional distribution of food inflation took the same pattern as the general inflation ⁹. Eastern Ethiopia is one of the food insecure parts of the country sharing similar food inflation phenomena. As ¹⁰ indicates, Dire Dawa Administration experienced highest inflation rates and high price instability between 2017and 2018. The average annual food inflation rate, as reported by the NBE in 2017/18, was 49.3% for Dire-Dawa. This indicates that the area is vulnerable to high food price inflation. The impact of food price is very evident on the consumers and producers to the extent of reducing their purchasing powers. It leads people to poor people by limiting their food consumption causing ill‑health in the short and long run ¹¹.

Instability and variation in food production over years contribute immensely to food insecurity as a result of high volatility and co‑movement of food commodity prices. Price volatility (that is, price instability) also interacts with price levels of commodities to affect the welfare and food security. However, ¹² explained that the higher the price of commodities, the stronger the welfare consequences of volatility for consumers and vice‑versa for the producers.

Furthermore, up to this recent time, many researches have been conducted on volatility and co‑movement of food commodity prices in the world and even in Sub‑Saharan Africa. These research works have revealed diverse measurement of dynamics of food commodity prices, but little was done in Ethiopia. Therefore, this study wants to bridge the gap and contribute to this area in analyzing the Prices of the selected food commodity (namely, prices of Rice, Maize, white Teff and Sorghum) in Dire Dawa. This study focuses on the objectives include: to explore the trends of prices of selected agricultural food commodities and to examine the short run and long run relationship between the prices of the selected agricultural food commodities.

2. Methodology

The data used for the study were Monthly time series data for the producers’ price of food commodities (Rice, White Teff, Maize and Sorghum) spanning from December 2011 to July 2018 which was obtained from Ethiopian Central Statistical Agency (CSA). The prices were recorded in Ethiopian birr (ETB).

The test for stationarity of the time series was explored using the unit root test. A time series is said to be stationary when its mean and variance are the same over the given time and the covariance that exists between the two variables does not depend on the observed time, but rather on their lag length of time. Out of many tests available to determine the stationarity, the most common tests used were the Augmented Dickey‑Fuller and Philips Peron tests were employed for this study.

The ADF test constructs a parametric correction for higher order correction by assuming that the series follows and AR(p) process and adding p lagged difference terms of the dependent variable to the right hand side of the test regression. The ADF test here consists of estimating the following regression:

Where, and are the coefficients, t is the stochastic trend, is a white noise error term and etc.

In the fully specified regression model there is a presumption that the disturbances are a stationary, white noise series. But this presumption is unlikely to be true if and are integrated series. Generally, if two series are integrated to different orders, then linear combinations of them will be integrated to the higher of the two orders. Thus, if and are that is if both are trending variables, then we would normally expect to be regardless of the value of not (i.e., not stationary).

If and are each drifting upward with their own trend, then unless there is some relationship between those trends, the difference between them should also be growing, with yet another trend. There must be some kind of inconsistency in the model. On the other hand, if the two series are both then there may be a β such that is Intuitively, if the two series are both then this partial difference between them might be stable around a fixed mean. The implication would be that the series are drifting together at roughly the same rate. Two series that satisfy this requirement are said to be co-integrated, and the vector [1, −β] (or any multiple of it) is a co-integrating vector. In such a case, we can distinguish between a long-run relationship between and that is, the manner in which the two variables drift upward together, and the short-run dynamics, that is, the relationship between deviations of from its long-run trend and deviations of from its long-run trend. If this is the case, then differencing of the data would be counterproductive, since it would obscure the long-run relationship between and Studies of cointegration and a related technique, error correction, are concerned with methods of estimation that preserve the information about both forms of covariation.

The vector autoregressive (VAR) model is a general framework used to describe the dynamic interrelationship among stationary variables. So, the first step in time-series analysis should be to determine whether the levels of the data are stationary. If not, take the first differences of the series and try again. Usually, if the levels (or log-levels) of the time series are not stationary, the first differences will be.

If the time series are not stationary at level, then the VAR framework needs to be modified to allow consistent estimation of the relationships among the series. The Vector Error Correction model (VECM) is just a special case of the VAR for variables that are stationary in their differences (i.e., I(1)). If the series are co-integrated, the series move together in the long run. A VAR of the first differences does not capture the long run relationship. Therefore, it is appropriate to use VECM rather than VAR in order to reveal the short term and long term tendencies between the series. The VECM can also take into account any co-integrating relationships among the variables.

Consider a VAR with p lags

(1)

Where is a vector of variables, V is vector of parameters, are matrix of parameters and is vector of disturbances. Any VAR(p) can be rewritten as a VECM. The VEC model is specified as follows:

(2)

Where, and is an identity matrix.

Specifically, when we use our variables the VECM can be specified as follows: (Equations 3-6)

(3)

(4)

(5)

(6)

In the formulations above, is the difference of price of rice at time t, is the difference of price of white Teff at time t, is the difference of price of maize, is the difference of price of sorghum at time t, EC is the error correction term, which is the estimated residual from the co-integration regression and is the random disturbance term at time t. Here, Rice is the food commodity consumed by majority of the households of eastern part of Ethiopia, particularly in Dire Dawa administrative. Therefore, the study tried to find the short run dynamics and long run relationship between price of Rice and price of other commodities considered.

3. Results and Discussions

To achieve the research objectives, the monthly data span from December 2011 to July 2018 was analyzed. In the empirical analysis, four series namely, the prices of Rice, Maize, White Teff and Sorghum were used. Some descriptive statistics including the mean, the standard deviation, the coefficient of variation, minimum and maximum values of the series under study are presented in Table 1. The results show that the values of summary statistics are more or less similar except standard deviation which indicates relatively high dispersion for White Teff.

The plots of the selected prices of agricultural products were displayed as shown in Figure 1 and suggested that the series of the endogenous variables are non-stationary and shows increasing trend except maize price. Additionally, from the figure we can observe that from 2017 the prices are highly increasing with some spikes around the middle of 2018.

The time series under consideration should be checked for stationary before one can attempt to fit a suitable model. That is, variables have to be tested for the presence of unit root(s) thereby the order of integration of each series is determined. The stationarity of the series were tested by using an Augmented Dickey-Fuller test and a Phillips and Perron test. The hypothesis to be tested is: H₀: the series has unit root against H₁: the series has no unit root.

The results of ADF and PP tests, with constant and trend both at level and first difference for each series are presented in Table 2. Test results indicated that the null hypothesis that the series in levels contain unit root could not be rejected for all the four series. That is, the respective p-values are greater than conventional significance levels α =0.05 and 0.01. Since the null hypothesis cannot be rejected, in order to determine the order of integration of the non-stationary time series, the same tests were applied to their first differences. The order of integration is the number of unit roots that should be contained in the series so as to be stationary.

Since the variables are integrated of order one, we proceed to test for co-integration. Johansen (1995) co-integration test is applied at the predetermined lag 1. In these tests, Maximum Eigenvalue statistic or Trace statistic is compared to special critical values. The maximum eigenvalue and trace tests proceed sequentially from the first hypothesis no co-integration to an increasing number of co-integrating vectors. The results of co-integration tests for Rice, Maize, Sorghum and white Teff are reported in Table 3. The trace statistic indicates that at least one co-integrating vector (r ≥ 1) exists in the system at the 95 percent confidence level (estimated LR statistic, 29.68>29.01, 95 percent critical value).

From the Johansen co-integration test, it was determined that the rank of co-integration matrix to be equal to one. Consequently (Table 5), the co-integrating vector is given by β= (1, -1.064054, 0.0287828 and -1.809817). The values correspond to the co-integrating coefficients of Rice (normalized to one), Maize, white Teff and Sorghum, respectively. Thus, the vector above can be expressed as follows (Table 5):

(7)

Ho: there is no co-integration among variables

Ha: there is co-integration

Akaike Information Criterion (AIC), Hannan-Quinn Information Criterion (HQIC) and Schwarz Bayesian Information Criterion (SBIC) were used to determine the appropriate lag length (Table 4). Lag two are selected to be the optimum lag lengths, as these are selected by the majority of the criterion.

Co-integration and non-spurious regression are the fundamental requirements of VECM. Results of co-integration test (Table 3) provide enough evidence on the long run relationship between the variables under consideration as there is a co-integration relationship. Result of ADF and PP tests provides enough evidence of stationarity of the prices (Table 1) at first difference. Since the series are co-integrated and non-spurious it enables us to estimate VECM (Table 5). The results of VECM are given in Table 5.

In the Rice (ΔPR) equation, the lagged EC term ( = -0.1006366, P = 0.000) is negative and the coefficient is significant at 5% level meaning that system corrects its previous period disequilibrium at a speed of 10.06% monthly. It implies that the model identified the sizable speed of adjustment by 10.06% of disequilibrium correction monthly for reaching long run equilibrium steady state position. The coefficient of the lagged EC term for white Teff (= 0.0891136, P=0.002) and the coefficient of the lagged EC term for sorghum (=0.1617882, P=0.001) are positively significant at 5% level.

The short-run dynamics are captured through the individual coefficients of the difference terms. The examination of the short-term relationship between the variables demonstrated that price of Rice is negatively affected by only the first lag of price of maize (= -0.1006366, P=0.000) and its own past values (= -0.1006366, P=0.000). The other short-run coefficients in this equation are not significantly different from zero. This implies that the prices of white Teff and Sorghum have no temporal short term effects on the price of Rice. In the white Teff regression, the effects of the one lag of price of sorghum (= 0.1831521, P=0.012) is positively significant. For the Maize equation, the short-run coefficient of the first lag of price of white Teff (= -0.0155244, P=0.014) is negatively significant, but the other short-run coefficients in this equation are not significantly different from zero. In the Sorghum regression, the effects of the one lag of price of Rice (= 0.0228854, P=0.000) is positively significant.

4. Conclusions

The purposes of this study were to see the structure of the selected series and to determine whether integration exists between the selected agricultural food commodity prices. The analysis was based on the monthly data from December 2011 to July 2018. The series used in this study are monthly prices of Rice, white Teff, Maize and Sorghum. Over the time period considered, all the three series showed an increasing pattern, that is, there is the sign of non stationarity in each of the series. In order to examine the series, the unit root tests (ADF and Phillips-Perron tests), identification of the number of lags and cointegration analyses were conducted. Unit root tests indicate that all series are non stationary at level and are stationary at first difference at 5 percent significant level. The Johansen cointegration test suggests that there is at least one cointegration vector, which describes the long run relationship between prices of Rice, white Teff, Maize and Sorghum. The appropriate number of lag identified was two.

Previous studies on integration in various countries and commodity prices that are discussed in the theory section did not clearly indicate whether price integration occurs but suggested it depends on the location and commodities being analyzed. Thus, research specific to the study area is important. Analyzing the prices movements, the integration was observed between the prices of the selected food commodities. This observation was then empirically tested and verified using a VECM analysis. The VECM analysis findings showed the existence of integration between the prices and revealed that increase in prices of maize and sorghum in the past will decrease price of Rice in the long run. Similarly, the result of VECM showed as prices of white teff and Rice have positive relationship.

References