Figure 1: Annual Rice Production (2001-2012)

Source: FAOSTAT

From the above figure, it is deduced that rice production has been recording an upward trend, peaking in 2010, there after moving downwards persistently between 2010 and 2012. The downward trend that characterized the period between 2010 and 2012 can be explained by insufficient rainfall, low use of agricultural inputs as well as low mechanized agricultural area. However, generally production increased two folds from 867,692 MT in 2001 to 1.8 million MT in 2012.

1.2 Rice Consumption

Tanzania has one of the fastest growing urban populations in East Africa, rising at 4.7 percent per year; the growing middle class prefer rice over other staples. Rice is the second most preferred staple after maize. It is especially preferred in urban areas, in institutions such as hospitals and schools and in restaurants. Two reasons might explain the observed preference in rice consumption. Firstly, most of the people in non-rice growing urban areas of Tanzania are poor and cannot afford to buy rice on a regular basis and secondly, the preference for rice in restaurants and institutions is mainly due to its convenience in terms of catering. It is estimated that rice constitutes about 16.6 percent of cereal consumption in Tanzania. In terms of consumer preference, rice from Kyela –Mbeya is considered to be of the highest quality, followed by rice from Shinyanga, Tabora, Morogoro and Moshi in order of decreasing preference. These preferences are fully reflected in the prices of the different types of rice in various markets nationwide.

            Given the importance of rice as food and commercial crop in Tanzania, and the rapid growth of cities and population that we witness, increases in rice price and odd price fluctuations tend to affect consumers adversely. Consumers in rice deficit regions and markets tend to experience the hardest hit. The existence of market integration in the country can lessen the negative consequences brought about by increases in rice price and odd price fluctuations as Goletti et al., 1995 argues that market integration ensures that a regional balance occurs among food deficit, surplus and non-cash crop producing regions.

2.0 LITERATURE REVIEW

2.1 Previous Studies on Price Transmissions across countries

Spatial market integration signifies the time lapse for the exogenous shocks to transform and reach the various geographically separated markets. The shorter the time lapse, the better, since longer time lapse leads to the conveyance of an inaccurate price signal that might distort producers’ marketing decisions.

Price transmission and domestic market integration have implications for food security. The integration of markets can be considered either as: the existence of physical trade between markets, the difference in prices between two markets being equal to the cost of transporting goods between the markets, or the prices in the markets moving together over time (World Food Programme [WFP], 2007).

Jha et al. (2005) examined market integration in 55 wholesale rice markets in India using monthly data from the period January 1970 to December 1999 by employing techniques used by Gonzalez-Rivera and Helf and to identify common factors across various markets. It was discovered that market integration was far from complete in India and a major reason for this was the excessive interference in rice markets by government agencies. As a result, it was hard for scarcity conditions in isolated markets to be picked up by markets with abundance in supply. A number of policy implications were also considered.

Ifejirika et al (2013) in their study on Price Transmission and Integration of Rural and Urban Rice Markets in Nigeria observed that Rice markets in the study area were integrated, but the level of integration was low. The Vector Error Correction Model had a low coefficient. The Vector Error Correction Model had a coefficient of -0.0061872 which was significant at 1% level and was negative. The Market Integration Function had Coefficient of Determination (R2) of 0.78 showing that the independent variables explained about 78% of the variations in the prices of rice in the rural and urban rice markets. Transportation cost, toll fee, processing cost and storage cost significantly affected the level of market integration.

Suryaningrum et al (2013) investigated a spatial market integration of Thailand and Vietnam rice market in Indonesia. The results show that there is the existence of long-run relationship among Thailand, Vietnam, and Indonesian rice in the domestic market. While, VECM indicates that price change in Indonesia is mainly influenced by its seasonal dummy and import ban in 2004 – 2005. Thai and Vietnam rice prices are mainly influenced by Indonesian price at the previous period and Indonesian import ban policies. Adjustment speed of Vietnam toward long-run equilibrium is faster than Thailand; however speed adjustment of Indonesia is not significant.

Dawson and Dey (2002) studied the spatial market integration among major rice markets in Bangladesh. An integrated empirical framework tested long-run spatial market integration between price pairs using a dynamic vector autoregressive model and co-integration. Hypotheses tests of market integration, perfect market integration, and causality were conducted sequentially using monthly prices from rice markets in Bangladesh since trade liberalization in 1992. The results showed that rice markets in Bangladesh were perfectly integrated with each other. The results of the causality analysis showed that Dhaka market dominated nearby markets but was itself dominated by more distant markets.

Ghafoor and Aslam (2012) studied the Market Integration and Price Transmissions in Rice Markets of Pakistan and found out that five major rice markets of Pakistan were integrated with each other on overall basis. The pair-wise integration of these markets showed the presence of co-integrating vectors. As such it may be revealed that rice markets in the country are integrated with each other and price signals and related arbitrage are well practiced. The adjustment vector measured through Error Correction Mechanism showed that it takes almost 4-5 months for adjusting any short run shock in the rice markets of Pakistan. These markets adjust 8-19 percent of the disequilibrium per unit of time i.e. one month. In case of price transmission from international markets to domestic rice markets of Pakistan, it was found that co-integration was absent. It may be due to the reason that Pakistan itself is a big exporter of rice and depend less on international markets for price formation in its domestic rice markets.

            Ojo et al (2015) investigated the long and short-run price integration analysis of rice marketing in Kwara State, Nigeria. The results revealed that stationary status of the price series was eliminated after the first differencing and that there was a stable long-run equilibrium relationship among the markets.  The  vector  error  correction  estimates shows that most of the markets were not well integrated in the short –run, and finally, the causality test revealed that no single market dominated the price formation either in the rural or urban markets in the study area.

This paper aims at examining the level of domestic market integration and price transmission in the major wholesale rice markets of Tanzania targeting select surplus and deficit markets namely Mbeya, Morogoro, Dar-Es-Salaam, Mtwara, Dodoma and Shinyanga. The paper has resorted to examine rice market integration because rice is the second most important food and commercial crop in Tanzania after maize. Thus in a situation of food shortage in some parts of the country, rice consumption becomes an option and the integration of rice market can help address food shortage in other markets, since price changes in one market due to increased supply can be transmitted to another market. Further, there are limited studies in Tanzania that have focused on rice market integration specifically rice price transmission between surplus, deficit and major consumption markets; it is against this background that this study covers the existing gap.

3.0 DATA SOURCES AND METHODOLOGY

3.1 Data Sources

The study aimed at assessing price transmission and market integration among selected rice markets in Tanzania. The assessment was done for monthly prices of six domestic markets in Tanzania namely; Mbeya, Dar-Es-Salaam, Mtwara, Dodoma, Morogoro and Shinyanga for the period of January 2004 to December, 2012. The selection of these markets was based on being surplus rice producers (Morogoro, Shinyanga and Mbeya), deficit (Mtwara and Dodoma) and major rice consumption market (Dar-Es-Salaam). The data used in the analysis is rice wholesale prices (quoted in Tanzanian Shillings per bag of 100 kilograms) sourced from the Ministry of Industry and Trade in Tanzania.

3.2 Methodology

The price data series for all the markets were tested for the presence of unit roots employing Augmented Dickey-Fuller (ADF) test .The ADF test considered the null hypothesis that the series have a unit root, that is,. is non – stationary.

The typical ADF test is given below:

 Δy_t=y₀+y_1t+ (β-1) yt_-1+δ1ΔУ_t-1+ϵ_{t……………………………………………………………………………………}…...1.1

Johansen’s Co-integration Test

The next step is testing the Johansen’s Co-integration. The dependent variables y_t (i= 1, 2, 3...6) are prices of different rice markets (i=1, 2, 3…... 6) and the independent variables x_t (i=1, 2, 3. 6) are prices of other rice markets.  For instance, the co-integration equation for Dar-Es-Salaam (major consumption) market with the other markets is given as follows:

Dar= β₁+ β₂Dodoma₊ β₃Mbeya+ β₄Mtwara+ β₅Morogoro+ β₆Shinyanga+ ϵ_t………..1.2

The third step is about determining the Error Correction Model (ECM) to capture the long-run equilibrium and short-run disequilibrium situations as well as adjustments between prices. An ECM formulation, which describes both the short-run and the long-run behaviors of prices, is expressed as follows:

∆P_Bt   =γ₁+γ₂∆P_At -πῦ_Bt-1 + _Ѵit………………………………….……………..…………1.3

In this model,

γ2 = the impact multiplier (the short-run effect) that measures the immediate impact that a change in PAt will have on a change in PBt.

π = the feedback effect or the adjustment effect that shows how much of the dis-equilibrium is being corrected, that is the extent to which any disequilibrium in the previous period affects any adjustment in the PBt period.

Granger Causality

The fourth step is determining the direction of causality by Granger causality test. When two price series are co-integrated and stationary, one may proceed to carry out the Granger causality test. When Granger causality run one way (uni-directional) or it could also be bi-directional which indicates that both series influence each other (e.g. X causes Y and Y also causes X). The market which Granger-causes the other is tagged the exogenous market and Exogeneity can be weak or strong (Adeoyo et al 2011).

The Granger model is represented as follows;

……………………………….……1.4

Where m and n are the numbers of lags determined by a suitable information criterion. Rejection of the null hypothesis indicates that prices in market j Granger-cause prices in market

H_o: price of rice in one market does not determine (granger cause) the price in the other market

H₁: price of rice in one market does determine the price in the other market (not granger cause)

4.0 RESULTS AND DISCUSSION

Table 1 shows the results of ADF Unit Root test of major wholesale rice price series to ascertain the time series properties of the data. The results confirm that all the price series were stationary and integrated of the same order after first difference.

Co-integration test was carried out on all the variables to determine the existence of long-run relationship between the price variables. Table 2 presents the result of the co-integration test involving the use of Johansen Maximum Likelihood test to determine the number of co-integrating relations. Both the trace statistics and Maximum Eigen statistics show that out of 12 market pairs investigated all market pairs are co-integrated at 5% level of significance. Based on this finding for both Trace and Maximal Eigen value tests, it is evident that a substantial number of rice markets in Tanzania are well linked with each other and price signals are fairly transmitted (Table 2). 

The results of Johansen co-integration test indicate that we can continue the analysis to vector error correction model (VECM) which means that there is a long-run equilibrium relationship among Dar-Es-Salaam, Dodoma, Mbeya, Morogoro, Mtwara and Shinyanga rice prices. To estimate co-integrating coefficients (β), we used the Johansen co-integration test. After normalization, the first co-integrating vector on Dar normalized co-integrating coefficients were estimated as follows:

Dar = -2.392 Dodoma +1.121 Mbeya +1.894  Morogoro -2.963 Mtwara+1.741 Shinyanga…………...1.5

               (-4.95)**              (2.85) **           (3.19) **                 (-6.42) **           (5.43) **

Note: All figures in Parentheses indicate t values

** Significant at 5%

The equation indicates that a 1% increase in prices in Dodoma results in a 2.39% decrease in prices in Dar, whereas a 1% increase in prices in Mbeya increases prices in Dar by 1.12%. This implies that Dodoma being a deficit market, its price increase does not translate directly into price increases in Dar because Dar is a rice recipient market (consumption market) from various sources including Mbeya, Morogoro and Shinyanga (surplus markets) as a result, the respective markets have equally recorded positive relationships with Dar-es-salaam in the above co-integrating equation.

The causal relationship of the price series was examined using Granger causality test at 1 lags level. This analysis was necessitated by the statement of Gujarati (2004) that although regression deals with the dependence of one variable on other variables, it does not necessarily imply causation. The results are presented in the Table 3.

NOTE: Values with asterisk shows granger causality. Prob>F is higher at 1 (***) and 5 (**) percent statistical significance level.

There are four bidirectional causal relationships and eleven unidirectional relationships among markets as indicated in table 3. Dar-Es-Salaam is a major consumption market and has bidirectional granger causality with Mtwara which is a deficit market and has recorded a unidirectional relationship with the rest markets.  Other pairs of markets with bidirectional relationships among them are Mbeya-Mtwara; Morogoro-Dodoma and Dodoma- Shinyanga. Mbeya shows a unidirectional causality with Morogoro, Dodoma and Shinyanga whereas Mtwara market shows a unidirectional causality with Morogoro, Dodoma and Shinyanga. This implies that Mbeya granger causes price formation in Morogoro, Dodoma and Shinyanga but not vice versa. In the same manner, Mtwara granger causes price formation in Morogoro, Dodoma and Shinyanga, but no reverse mechanism of price formation was found among them. On the basis of the granger causality results, we conclude that all these markets granger causes price formation in each other so linked in efficient manner to transmit signals and related arbitrage.

On the basis of Vector Error Correction Model as indicated in Table 4 above, the coefficient of Vector Error Correction Model (VECM) is -1090.383 which is significant at 1% level of significance and it is negative. Thus, it will rightly act to correct any deviations from long run equilibrium. The significant Error Correction Mechanism showed that the speed of adjustment of rice prices to long run equilibrium is 60%. This means that the speed of adjustment of rice prices to the equilibrium is relatively higher.

4.1 Impulse Response Function Analysis for Rice markets

In further examining the short-run dynamic properties of rice prices, impulse response function (IRF) and forecast error variance decomposition (FEVD) were analyzed among rice markets. The impulse response function graphs are as shown in annex 1(from figure 2 to figure 10). The plots (Figures 2 to 10) show that the markets reacted positively and negatively when one standard error shock was given to deficit, surplus or major consumption market. In figure 2, a one standard error shock to Dodoma has a positive effect on Dar, whereas its effects dissipate after 6 months. In figure 3, a one standard error shock to Mbeya has a positive effect on Dar and its effect disappears after two months. In figure 4, a one standard error shock to Dar has a positive effect on Morogoro and its effects dissipate after 4 months. In figure 7, a one standard error shock to Dar has a negative effect on Shinyanga, its impact dies after 4 months. In figure 8, one standard error shock on Mtwara has initially a positive impact on Dodoma then after two months, it becomes negative throughout the period and its impacts dies after 6 months.

4.2 Variance Decomposition Analysis for Rice Markets

Based on variance decomposition analysis, the results reveal t the predominant sources of price fluctuations across markets are due largely to own shocks and Dar market, to a very small degree, shocks are coming from the rest as it is indicated in Annex 2 (Table 5). This implies that Dar-es-salaam as a major consumption market has a significant position in influencing the rice price behavior in Tanzania and can be used to predict future prices.

5.0 CONCLUSION AND POLICY IMPLICATION

The study found out that six markets that include one major consumption market, three surplus and two deficits markets are co-integrated. Further, based on the results of the vector error correction model whereby its error correction term is negative and significant, the markets have the long run relationship among them. The results of VECM further show that the speed of adjustment of rice prices to long run equilibrium is 60%. This means that the speed of adjustment of rice prices to the equilibrium is relatively higher.

Based on variance decomposition analysis, the results reveal t the predominant sources of price fluctuations across markets are due largely to own shocks and Dares-salaam market. This means that Dar can be used as well to predict future prices. Based on the study findings, it implies that the government can use market-based policies for food security which are appropriate if markets are integrated, so that the effect of policy intervention in one market would be transmitted to other markets.

REFERENCES

Adeoye I.,  Dontsop  M., Badmus M. and Amao  O. (2011). Price Transmission and Market Integration of Banana and Plantain in Oyo State, Nigeria

Dreze, J. and A. Sen (1995). The Political Economy of Hunger, Oxford: Clarendon Press

Dawson, P. J. and P. K. Dey. (2002). Testing for the Law of One Price: Rice Market Integration in Bangladesh. Journal of International Development, Volume 14, issue 4, 473-484.

Ghafoor and Aslam (2012). Market Integration and Price Transmissions in Rice Markets of Pakistan, SANEI Working Paper Series No. 12 – 08.

Goletti, F., Ahmed, R. and Fraid, N. (1995). Structural determinants of market integration: The case of rice markets in Bangladesh, The Developing Economics, 33(2):185-202.

Gujarati, D.N. (2004). Basic Econometrics, Tata McGraw- Hill Publishing Co. Ltd, New Delhi.

Gonzalez-Rivera G and Helfand  S (2001). Economic Development and the Determinants of Spatial Integration in Agricultural Markets

Henry De-Graft Acquah, John Andoh Micah, Rebecca Owusu (2012). Price Transmission in Domestic Agricultural Market: Evidence from Selected Cassava Markets in Ghana, Journal of Sustainable Development in Africa (Volume 14, No.5, 2012).

Ifejirika C.A., Arene C.J., Mkpado M. (2013). Price Transmission and Integration of Rural and Urban Rice Markets in Nigeria, Journal of Agriculture and Sustainability, Volume 2 (2013), Issue No 1, pages 66-85.

Jha, R., K. Bhanu and M. A. Sharma. (2005). Market Integration in Wholesale Rice Markets in India. ASARC Working Paper 2005/03. The Australian University of Delhi, India.

Ojo A. O., M. A.Ojo, A. Ogaji, A. Oseghale and R. Kutigi (2015). Long and short-run price integration analysis of rice marketing in Kwara State, Nigeria, Agricultural Science Research Journal 5(6); pp. 98- 104, June 2015

Suryaningrum D.A.,  Chang W.I., Anindita R. (2013). Analysis on Spatial Integration of Thailand and Vietnam Rice Market in Indonesia, Greener Journal of business and management studies, volume  3 (7), pp. 333-342, Issue N0. 2276-7827.