4.1. Trends In Meat Export
Ethiopia has great potential for meat and meat related products exports. In 2017/8, the volume of global meat produced was 308.5 million tons and Ethiopia share of export was 659305 tones. In Ethiopia cattle, goats, sheep, camel and poultry, are used as resource for meat production; however, the first three species are the most common. Despite the fact that Ethiopia is the tenth largest livestock population in the world, the production of meat is still low and contributed for about 0.2 percent of the world total meat production, of which most is sheep and goat meat. The reason behind for this include but not limited to very low off take rates; large number of animals that by-pass abattoirs and are exported live both illegally and legally, producers who are not commercially oriented and sell only in need of cash or when draught affects the environment or when their animals getting too old and other limiting factors such as failure in meeting the international standards by meat processors (Alem et al., 2015; Kefyalew and Tarkegn, 2013).
Recently, the production of meat and meat related products has been growing in the last decades emanated from both demand side and supply side conditions. The demand side conditions was because of increase in the level of income both in developing and developed countries and the supply side conditions was facilitate by the lowing the cost of meat and meat related products because of economies of scale. The main market for Ethiopia meat export countries were in the Middle East and North African countries, mainly Saudi Arabia, United Arab Emirates, Bahrain, Egypt, Yemen and Congo.
The country meat export also consider as the buyers driven commodity, it is expected that supply to the external market does not have a unique relationship with the behavior of prices. So far, no empirical evidences have been produced to justify the presence of such responsiveness in the Ethiopian livestock markets. However, it could be hypothesized that increased export activity brought through the rise in export demand and this associated with higher prices usually prompt a supply response compared to the level we would see if no exports were involved. The changes that are occurring today in the types of meat and meat products produced and exported from Ethiopia are a function of end use markets and the development of different strategies to supply to those end use markets.
Sources: own manipulation
The share of meat sources livestock was indicated was beef followed by sheep, goat and chicken meats. Beef contributed for more than half of the total meat produced in 2017. Ethiopia cannot influence the prices of meat in the world market. Hence, foreign demand for Ethiopian meat can be considered as perfectly elastic at the given world price. This implies, noting the small country assumption, export prices are determined exogenously.
As it indicate in the figure...that meat export volume and value are expected to grow. It is worthwhile to mention here possible expected pushing factors for meat export growth during the strategic plan period. The driver for this includes increasing human population and urbanization, rising per capita income, increasing demand for meat, milk and milk products, increasing in domestic price of live animal and meat products, improving infrastructure, favorable enabling environment and globalization and market opportunities. However, for meat product export promotion, there should be upgrading the country’s meat and its product export to the importing countries standard market requirement with quality improvement, expanding commercialization of production and marketing of livestock, diversifying into other products such as the processing of sausages and other similar types of meat and boosting domestic consumption.
4.2. Trends In Live Animal Export
The total live animal exports are estimated as high as 1.6 million live heads exported from the country annually, although the vast majority of these (approximately 1.2 million or one-fourth) pass through informal channels (Belachew, 2009). In addition to these internal challenges, Ethiopian cattle exporters face stiff competition from Brazil, India, Pakistan, Australia and New Zealand. Brazilian beef is price competitive as a result of low production costs, while India has a ready-made market, catering primarily to the large non-resident Indian population in the Gulf for its beef (buffalo meat) exports. Faced with such competitors, Ethiopian live cattle exporters find it difficult to compete on price and quality requirements for the short-term (Belachew, 2010) 2010). All those involved in live cattle exports in Ethiopia must adopt improved practices in production, transportation, processing and packaging of products to maintain and increase market share.
Exporting live animals takes both formal and informal export routes. As some estimates indicate, between 80 and 90 percent of live animal exports are informal exports (USAID, 2013). Live animal exports through formal channels to various importing countries from 2005–2012. Live bull exports gradually grew from 38 000 in 2005 to 205 000 heads in 2012, increasing by more than fourfold. Of all live animal exports, live bull exports are the first largest foreign exchange earner. The majority (more than 70 percent) of live bull exports are from pastoral regions, particularly from Oromia region in the Borena zone. Other live animal exports include sheep, camels and goats, for which export volumes and related foreign currency earnings have increased substantially during the period 2005–2012. Ethiopia's live animal exports are concentrated in a few destinations. Sudan accounted for the largest share (35 percent) of live bull exports, followed by Somaliland (30 percent). The two biggest importers accounted for 65 percent of the total live bull exports, mainly re-exporting the imported live cattle, especially Somaliland, which mainly supplies the countries of the Arab peninsula. The United Arab Emirates, Djibouti, Egypt and Yemen accounted for 10, 8, 7 and 6 percent, respectively, and together these six countries constituted a 96 percent share, with the remaining four destined for a few other countries.
In the last nine months, the country has exported over 232,228 live animals which included the export of 71,105 cattle, 11,527 camels, and 149,595 sheep and goats. In this export performance, the country has totally earned 58.89 million USD; of which, about 42.72 million USD was from cattle, 6.57 million USD from camels and 9.6 million USD from sheep and goats. Compared with the previous year performance, the number of exported animals and the volume of foreign currency have been reduced by half. In 2015/16, the country has secured over147 million USD from the export of 667,005 live animals. Of these, the sheep market took the leading volume of 398,333 while the cattle market followed with 153,051. The remaining 28,271 is accounted for camels, and 82,724 are for goats.
4.3. Trends In Leather And Leather Products Export
Ethiopia holds a huge livestock resource endowment, this is good opportunity exists for the development and competitiveness of the leather and leather related products. Ethiopia’s modern tanning industry was started in the mid-1920s by Armenian immigrants, and since then the country has increased the number of tanning companies to 29 (ELIA, 2014). During 1970 to 1980, the export of leather from hides and skins and finished leather products were some of the top exports of the country. Hides and skins export ranked second to coffee in the 1970s and early 1980s. However, the export ban imposed on hides and skins in 1986 resulted in a decline in the export volume (Abebe & Schaefer, 2013). The finding of this study indicate that the Ethiopian export of leather and leather products shows an increasing trend (Fig. 4). Ethiopia leather (specially raw hides and skins and semi-processed leather products) export were fluctuated more during the year 1971 to 2003 and significantly decrease starting from 2007 to 20019. However, after 2011 the export leather and leather product increase significantly. The highest export of leather and leather product in Ethiopia was recorded in 2014/15.
4.4. Unit Root Testing For Stationary
It is well know that most of the macro economy data were have the non-stationary properties and they tend to have a deterministic and/or stochastic trend. In time series analysis, it is important to consider a number of issues that can potentially influence the estimation. In order to give good reason for the economic theory, stationary properties of the variables behind the time series models shall be conducted in order to avoid spurious regression, meaning, there will not be have meaningful relationship between those variables. Thus, it is a good move to test the stationary properties of the variables for the next step.
This study have tested the stationary properties of the variables using the commonly used method of Augmented dickey fuller (ADF) test. The result from the ADF test below revealed that almost all variable have non-stationary property at level and they became stationary after first differencing, they have fulfill and leads to apply the co-integration or long run equation test (Table 3).
Table 1
Augmented Fuller Dickey Test
|
at level(0)
|
at level(1)
|
Variable
|
P-value for Z(t)
|
Test statistics
|
P-value for Z(t)
|
Test statistics
|
Meat Export
|
0.6522
|
-1.905
|
0.0037
|
-4.254
|
Animal Export
|
0.6031
|
-1.997
|
0.0001
|
-5.672
|
Leather Export
|
0.3283
|
-2.499
|
0.0004
|
-4.846
|
Exchange Rate
|
0.0416
|
-3.48
|
0.0001
|
-6.903
|
Real GDP
|
0.9512
|
-0.944
|
0.0001
|
-5.364
|
Consumer Price Index
|
0.7393
|
-1.726
|
0.0113
|
-3.921
|
Total Labor
|
0.1108
|
-3.081
|
0.0001
|
-9.872
|
Gross Fixed capital Formation
|
0.0498
|
-3.412
|
0.0001
|
-7.005
|
Values of Mackinnon test for ADF
1% critical value: -4.187
5% critical value: -3.516
10% critical value:-3.19
|
Values of Mackinnon test for ADF
1% critical value: 4.196
5% critical value: -3.52
10% critical value: -3.192
|
4.5. Lag length Determinations
The lag length determination essential to avoid the previous year observations dependent variable or independent variable to be considered as the explanatory variable. This implies that,
Since the dependent variable is time series type the previous year observation either the dependent or the independent variable may be considered an explanatory variable. To determine the lag length, there are well known methods and criteria which lead to different outcomes and needs subjective judgments. In this study, the maximum four lag of dependent as well as the independent variable considered based on AIC and HQIC criteria (Table 2).
Table 2
Lag length Determinations
Lag
|
LL
|
LR
|
FPE
|
AIC
|
HQIC
|
SBIC
|
0
|
122.963
|
|
7.40E-13
|
-5.22561
|
-5.10531
|
-4.90121
|
1
|
401.478
|
557.03
|
4.50E-17
|
-14.9763
|
-13.8936
|
-12.0567*
|
2
|
453.746
|
104.53
|
1.00E-16
|
-14.443
|
-12.3978
|
-8.92821
|
3
|
554.012
|
200.53
|
4.80E-17
|
-16.0915
|
-13.0839
|
-7.98152
|
4
|
705.924
|
303.82*
|
9.2e-18*
|
-20.0875*
|
-16.1175*
|
-9.38231
|
4.6. Co-Integration Rank Test
The ADF stationary test results presented previously indicate that all the variables are not level stationary and they became stationary after first differencing. This would imply that there is a meaningful long run relationship among the variables. Thus, the presence and the number of such co-integrating relationships are checked using the trace and the maximum eigen value methods from Johansen co-integration test. The Johansen method of co-integration rank test result is very much dependent on the deterministic trend assumption in the underlying VAR structure, in addition to the number of lags of the endogenous variables. Hence, since the results may differ with the alternatives, a decision must be made as to which one to choose for the purpose of further analysis. As shown in the below table the both trace test and maximum Eigen value test suggests that there are about six co-integrating equations at 5% level of significance (Table 4). This is a proof for the long-run relationship among variable. This co-integration rank test also led for the application of the johansen method instead of the single equation based equation-based Engle-Granger two-step procedure.
Table 3
Johansen Tests For Cointegration Trend: constant Number of obs = 44Sample: 1975–2018 Lags = 4
Max. rank
|
Parms
|
LL
|
Eigenvalue
|
Trace Statistic
|
5% Critical Value
|
0
|
200
|
-2539.371
|
.
|
537.3036
|
156
|
1
|
215
|
-2438.7495
|
0.98968
|
336.0605
|
124.24
|
2
|
228
|
-2384.85
|
0.91370
|
228.2616
|
94.15
|
3
|
239
|
-2345.3029
|
0.83430
|
149.1673
|
68.52
|
4
|
248
|
-2317.2878
|
0.72012
|
93.1372
|
47.21
|
5
|
255
|
-2296.4709
|
0.61180
|
51.5033
|
29.68
|
6
|
260
|
-2279.9758
|
0.52753
|
18.5132
|
15.41
|
7
|
263
|
-2272.3117
|
0.29416
|
3.1851*
|
3.76
|
8
|
264
|
-2270.7192
|
0.06983
|
Max. rank
|
Parms
|
LL
|
Eigenvalue
|
Trace Statistic
|
5% Critical Value
|
0
|
200
|
-2539.371
|
.
|
201.243
|
51.42
|
1
|
215
|
-2438.7495
|
0.98968
|
107.7989
|
45.28
|
2
|
228
|
-2384.85
|
0.91370
|
79.0943
|
39.37
|
3
|
239
|
-2345.3029
|
0.83430
|
56.0301
|
33.46
|
4
|
248
|
-2317.2878
|
0.72012
|
41.6339
|
27.07
|
5
|
255
|
-2296.4709
|
0.61180
|
32.9901
|
20.97
|
6
|
260
|
-2279.9758
|
0.52753
|
15.3282
|
14.07
|
7
|
263
|
-2272.3117
|
0.29416
|
3.1851
|
3.76
|
8
|
264
|
-2270.7192
|
0.06983
|
source: own analysis (2018) |
4.7. Vector Error Correction Model Estimation
The co-integration rank test revealed that the data has six co-integration relationship based on the johansen co-integration test. Thus, VECM consists of two parts: the matrix of long-run co-integrating coefficients that is used to derive the long-run co-integrating relationship, and the short-run coefficients which is for the short-run analysis.
4.7.1. Long run co-integration
The long run relationship test tries to show the co-integration of the dependent variable and the independent ones, which emphasizes the existence of long run co-movement In the two types of variables. These imply that the study is trying to see long run first order of integration. Based on the VECM matrix of long-run co-integrating coefficients result, out of the seven variable, five variable are found to be statistically significant and the sign of two variables (consumer price index and live animal export) were affecting real GDP negatively while meat export, exchange rate and gross fixed capital formation affects real GDP positivity in the long run.
Table 4: Johansen normalization restriction imposed
|
Dependent variable: DReal GDP
|
Variables
|
Coefficient
|
St. Error
|
z
|
P>|z|
|
[95% Conf. Interval]
|
DMeat Export
|
1.444609
|
0.203105
|
7.11
|
0.00
|
1.04653
|
1.842688
|
DAnimal Export
|
-0.26376
|
0.06159
|
-4.28
|
0.00
|
-0.38448
|
-0.14305
|
DLeather Export
|
65.95929
|
137.6363
|
0.48
|
0.632
|
-203.803
|
335.7214
|
DExchange Rate
|
51933.53
|
10950.81
|
4.74
|
0.00
|
30470.35
|
73396.72
|
DCPI
|
-12980
|
2255.338
|
-5.76
|
0.00
|
-17400.3
|
-8559.58
|
DTotal Labor
|
-0.01524
|
0.009146
|
-1.67
|
0.096
|
-0.03316
|
0.002689
|
DGFCF
|
17733.83
|
6247.774
|
2.84
|
0.005
|
5488.413
|
29979.24
|
Constant
|
274188.8
|
.
|
.
|
.
|
.
|
.
|
chi2: 195.2181
P>chi2: 0.0000
|
|
|
|
|
|
|
source: own analysis (2018)
In order to clearly understand the above and interpret the result, we can estimate the vector error correction equilibrium relationship normalized on real GDP. The equation of this model incorporates a corrective mechanism by which previous disequilibrium in the relationship between real GDP and one or more determinants are permitted to affect current change and this model is direct extension of the above long-run result after considering the significant variables in the co-integration equation. thus, the vector error correction model if formulated in the following form;
Real_GDP=274188.8+1.444609Meat_Export-0.26376Animal_Export+51933.53Exchange Rate -12980 CPI - 0.01524total labor + 17733.83 GFCF...............................................................5
The above equation showed the in the long run real GDP can be explained by the Meat export, animal export, Exchange Rate, CPI, total labor and GFCF.
4.7.2. Short run relationship
The result from the short run matrix showed that the coefficient of error correction term for the equation is significant and negative, which indicates there is a reasonable adjustment towards the long rung steady state. This guarantees that the real GDP may temporally deviate from its long equilibrium value and converge to its equilibrium. The value of error correction term (-0.31146) indicated that about 31% of the deviation of real GDP from its equilibrium value is eliminated every year, as a result the full adjustment would require a period of more than three years. Live animal export is significant and its coefficient is negatively related to real GDP, indicating that it has unfavorable effect on real GDP in short run. Export of meat, Exchange rate, total labour, and GFCF were found is significant and its coefficient is positively related to real GDP in the short run.
Table 5
short run with the dependent variable D.Real_GDP
Error corrections
|
Coefficient
|
Std. Error
|
z
|
P > z
|
[95% Conf.
|
Interval]
|
ECM1
|
-0.31146
|
0.050918
|
-6.12
|
0.00
|
-0.41126
|
-0.21166
|
D.Meat_Export
|
1.995448
|
0.259076
|
7.7
|
0.00
|
1.487668
|
2.503228
|
D.Animal_Export
|
-0.90058
|
0.153939
|
-5.85
|
0.00
|
-1.2023
|
-0.59887
|
D.Leather_Export
|
58.36762
|
43.17532
|
1.35
|
0.176
|
-26.2545
|
142.9897
|
D.Exchge_Rate
|
13468.19
|
5062.878
|
2.66
|
0.008
|
3545.127
|
23391.25
|
D.CPI
|
-673.639
|
1288.899
|
-0.52
|
0.601
|
-3199.84
|
1852.558
|
D.Total_Labour
|
0.023192
|
0.008128
|
2.85
|
0.004
|
0.007262
|
0.039122
|
D.GFCF
|
6027.584
|
1686.198
|
3.57
|
0.00
|
2722.696
|
9332.473
|
Constant
|
-29640.4
|
12904.98
|
-2.3
|
0.022
|
-54933.7
|
-4347.15
|
4.8. Wald Causality Test
The co-integration test can determine the existence of causality between variables it cannot determine the direction of this relation. If two variables are co-integrated then a relationship will exist that can be measured by Granger causality test. The result from the causality test revealed that there is a short run causality between meat export, live animal and real GDP. In addition, there is bi-directional causality between meat export and real GDP and unidirectional causality between live animal, leather export and real GDP.
Table 6
causality test between Meat, Live animal and leather export and real GDP
N0.
|
Nuall hypothesis
|
chi2( 1)
|
Prob > chi2
|
Decision
|
1
|
Real GDP does not cause Meat Export
|
5.18
|
0.0228**
|
Reject
|
1
|
Meat Export does not cause Real GDP
|
7.22
|
0.0072*
|
reject
|
2
|
Real GDP does not cause Live animal Export
|
0.03
|
0.8599
|
Accept
|
2
|
Live animal Export does not cause Real GDP
|
15.59
|
0.0001*
|
reject
|
3
|
Real GDP does not cause leather Export
|
4.67
|
0.0307**
|
reject
|
3
|
Leather Export does not cause Real GDP
|
0.02
|
0.8812
|
accept
|
*significant at 1% and **5 significant at 5% |