EXPORT PERFORMANCE AND ECONOMIC GROWTH IN ETHIOPIA

1. EXPORT PERFORMANCE AND ECONOMIC GROWTH IN ETHIOPIA By Kagnew Wolde March 2007

2. Introduction • Like other SSA economies, the Ethiopian economy is essentially agricultural based and highly dependant on earnings of fragmented household agricultural activities. • The performance of the economy is guided by the performance of the agricultural sector. • Agricultural commodities dominate the country’s export baskets. • Coffee is the principal export product. • The share of non-coffee exports has been rising remarkably in recent years attributed to the dropping in the coffee export earnings. • The share of non-agricultural exports is very narrow.

3. Introduction con’d • The major manufacturing export commodities are leather and leather products, frozen meat, sugar and textiles. • Ethiopia did not succeed in increasing manufactured exports. • The Ethiopian economy had recorded a promising growth performance during the imperial regime, which was halted after the mid 1970s. • In the Derge regime the overall economic performance was gloomy and real aggregate variables decelerated. • Since 1992, Ethiopia has embarked on reform package with the aim of reversing the deteriorating economic conditions and put the economy in a sustainable growth momentum. However, the economy remains weak and sensitive to shocks.

4. Table 1: Growth rate of Real GDP and Real Export of Ethiopia

5. Introduction Cont’d • The over all performance of the export sector has been gloomy in the last four decades due to structural problems and policy constraints. • The average growth in real value of goods and services exported: - 1960/61-2003/04 = 13.04%. - 1960/61-1973/74 = 8.2%. - 1974/75-1990/91 = 4.7% - Has showed improvement in the period 1991/92-2003/04. • The contribution of exports in financing imports has been continuously contracting to reach about 45.1% in 2001/02. • The decline in export revenue relative to the steady rise in import bill has: - widened the balance of trade deficit, - restrained the import of essential intermediate and capital goods, - exacerbated the rise of external financing in the form of aid and credit. • Not only does the export pattern dominated by primary commodities but there is also market concentration of the country’s exports.

6. Introduction Cont’d • The country needs to find out ways that makes it possible to diversify and radically augment its exports on sustainable basis. • This study examined a data set for a period of four decades and could assess the impact of the government strategies on export performance and output growth. Statement of the Problem • The Ethiopian economy is agrarian and agricultural commodities dominate the export basket, basically Coffee. • Although the focus of the economic reform program has been to make export as an engine of growth, it does not seem that the government’s attempt has brought the required results and thus whether exports determine GDP growth needs to be empirically probed. • This study is, therefore, undertaken to fill the gap; may be an up to date test of the export growth linkage in Ethiopia.

7. Objective of the Study • To examine the relationship between export performance and economic growth in Ethiopia using co-integration and vector error correction techniques. • To highlight possible intervention areas for export growth and diversification. Significance of the Study • Policy makers can utilize the information generated to design appropriate policies. • Can serve as a reference to subsequent research works in the area of export-led growth in the context of Ethiopia.

8. A Review of the Export policy in EthiopiaThe pre 1974/75 Periods • The private sector, mainly foreign capital, had occupied the sheer weight of both exports and imports activities. • The development plan had three phases (Imperial Government of Ethiopia, 1955, 1962, 1968). The first five-year development plan (1957/58-1961/62): • Import substitution industrial promotion and infrastructure facilities like road development were the focus. • Diversification of the export structure by exploiting the large livestock population, and the products of agro-processing industries to secure average annual export growth of 9% and 11% share of exports in national income.

9. The second five-year development plan (1962/63-1966/67): • Structural change and export diversification received priority. • New export products of industrial origins and mining products were supposed to play key role. • The average annual export growth rate was expected to reach 11%. - the share of agricultural exports to trim down to 72.3% in 1966/67 from 93.6% in 1962/63 - manufactured products to wind up to 24.2% from 5.2%. • The policies: the formation of government foreign trade corporations, revisions of existing customs tariff to protect domestic products and stimulate exports, directing credit and subsidy policies towards the production and promotion of exports, conclusion of a series of bi-lateral and multilateral agreements and better participation at international trade fairs.

10. The third five-year development plan (1968/69-1973/74): • Geographic diversification of traditional export produces (coffee, livestock products and oilseeds) and the development of non-agricultural exports were the focus. • Envisioned to reduce the share of traditional exports from that of 86% in 1967/68 to 75% in 1973/74. • The share of coffee was planned to plummet from 55% to 40% in the same period partly due to the addition of new products in the export basket. • The measures adopted include overvaluation of the exchange rate, high tariff rates, wide-ranging foreign exchange control and non-tariff barriers like restrictions on some items and heavy tax on export.

11. The Derge Regime (1974/75- 1990/91) • The overall policies favored the expansion of collective and public enterprises while private enterprises were kept at cove for long. • An inward looking strategy behind a highly protective tariffs and quantitative restrictions was the development strategy. • The government undertook a ten-year perspective plan (1985/86- 1994/95): - aimed at orienting export structures of the country towards high value added products and increasing the amount and composition of manufactured exports, expanding the foreign exchange earnings, and increasing the socialization of the export sector. - However, particular attention was given to state owned export companies heedless of their inefficiency. - geographic diversification of exports towards the markets of socialist countries and neighboring African countries.

12. - average annual export growth rate was targeted to stand at 15.4%. State owned export companies were expected to take up 90% of the export business. - to reduce the share of all traditional exports (coffee, hides and skins, pulses and oilseeds) to 53.2% from 73.5%. - the share of other export products (live animals, meat products, fruits and vegetables, spices, sugar and molasses, natural gum, chat and others) was targeted to rise from 26.5% in1985/86 to 46.8% in 1994/95. • The tools employed: provision of favorable tax, tariffs and foreign exchange rate measures, improving exports in terms of quality, quantity and variety and providing current information on world market prices and other factors in international market to exporters and producers. • The other efforts: the introduction of the export subsidy scheme in 1983/84 and the directive issued to ban the export of raw hides and skins in 1989/90.

13. The Post 1990/91 Period • Since 1992, Ethiopia under the support and guidance of the IMF and the World Bank has undergone liberalization and enhanced Structural Adjustment Programs (SAPs) to restrain internal and external imbalances of the economy. • One of the basic tasks of the new policy regime is to increasingly open the economy to foreign competition with a view of benefiting the economy from expanded markets. • The tools implemented include: - devaluation of the Birr and step-by-step liberalization of the foreign exchange market, - streamlining import and export licensing system, - tariff reduction, and provision of incentives to exporters, - abolishing taxes on exports and subsidies to parastatal exporting enterprises, - encouraging export-oriented investment,

14. The Post 1990/91 Periods Cont’d - introduction of duty draw back and foreign exchange retention scheme, - minimizing administrative and bureaucratic procedures. • Promulgating an export development strategy, • Established export support institutions, • Instituted export specific incentive schemes, • Foreign exchange retention scheme, • Export duty incentive schemes such as duty draw back scheme, voucher scheme and bonded manufacturing warehouse scheme, • Export credit guarantee scheme, • External loan and suppliers’ or Foreign partners’ credit

15. Literature Review • Much has been said in the literature regarding the role of the export sector to the overall economic performance. • Export-led growth theory suggests that export-oriented polices enhance economic growth. Proponents of this theory argues that export has strong correlation with economic growth and can play key roles to enhance overall economic performance of a country. • Export expansion brings about technological progress resulting from foreign competition that is crucial for improvement of factor productivity and better use of resources (Kavoussi, 1984, Moschos, 1987). • Export may benefit economic growth through generating positive externalities on non-exports (Feder, 1982), increased scale economies, improved allocative efficiency and better ability to generate dynamic comparative advantage (Sharma and Panagiotidis, 2004). • Exports ease foreign exchange constraints and can thereby provide greater access to international market. The foreign exchange earnings from exports allow the import of high quality intermediate inputs, mainly capital goods, for domestic production and exports, thus expanding the economy’s production possibilities (McKinnon, 1964).

16. Literature Review Cont’d • According to Esfahani (1991), export enables developing countries to relieve the import shortage they may face up to. Speaking differently, revenue from exports can fill “the foreign exchange gap” which is perceived as barrier to growth. • Michaely (1977) used cross-section data set for sample of 41 less developed countries to the export-economic growth relationship by applying spearman rank correlation coefficient. The estimated coefficient suggested a positive and significant relationship between exports and economic growth among the more developed economies but not among the least developed ones. He concluded that export performance affects output growth once countries attain some minimum level of development. • Tyler (1981) took a sample of 55 middle-income developing economies to investigate the impact of exports on growth and found a positive and significant relation between export growth and income growth.

17. Literature Review Cont’d • Kavoussi (1984) considered 73 middle and low-income developing countries and found a strong relation of higher rate of economic growth with higher rates of export growth. He showed that the positive correlation between exports and growth holds for both middle- and low-income countries but the effects tend to diminish according to the level of development. • Esfahani (1985) used sample of 31 semi-industrialized countries and employed a simultaneous equation model to deal with simultaneity problem between GDP and export growth. He found that the economic growth of most of the sample countries is due to the import supply effects of exports. • Lussier (1993) employed cross-section and panel data analysis to establish the direction of causality between the growth of export and real output, by taking sample of 24 and 19 African countries. The result supports the hypothesis for panel data but fails to find any positive linkage when using export growth as a share of GDP.

18. Thornton (1996) used Engle-Granger co-integration and Granger causality tests within a two variable framework and found a positive and significant causal relationship running from exports to economic growth in Mexico. • Amoating and Amako (1996) run causality test for 35 African countries by introducing foreign debt service as a third variable within a tri-variate causality analysis of exports and economic growth. The results showed a joint feedback effect between export revenue, external debt service and economic growth. • The study of Doraisami (1996) strongly supports for bi-directional causality between exports and growth in Malaysia, and that a positive long-run relationship existed between these results. • Al-Yousif (1997) employed time series analysis for four Arab Gulf countries and the results support the hypothesis in the short-run but fail to find long-run relationship, i. e. does not find co-integration. • Shan and Sun (1998) established a Vector Autoregressive model in the production function context in case of China. They found a bi-directional relationship and hence their results rejected the export-led growth hypothesis of uni-directional linkage. Nonetheless, both exports and industrial output contribute positively to each other in the course of the Chinese economic development.

19. Chang etal (2000) used multivariate causality analysis incorporating imports as a factor in the relationship between exports and output in the case of Taiwan and found no support for the export-led growth hypothesis during the period of rapid growth in Taiwan (1971-1995). • Keong, Yusop and Liew (2003) run vector autoregressive model and multivariate co-integration for Malaysia and found a long-run association between the variables considered. The results of the error correction model revealed that all variables except exchange rate Granger cause economic growth in the short run. This led them to confirm the validity of the export-led growth hypothesis in the case of Malaysia both in the short - and long - term. The result further suggests that the growth rate of capital formation and imports have positive impacts on economic growth, while labour has a negative impact in the short run. • Sharma and Panagiotidis (2004) test the export-led growth hypothesis for the case of India using different approach and the results strengthen the arguments against the export-led growth hypothesis for the case of India.

20. Methodology and Sources of DataSources and Type of Data • The main sources of data for this study were the National Bank of Ethiopia (Annual and Quarterly bulletins), the Ministry of Finance and Economic Development, Export Promotion Agency, and the Ethiopian Economic Association statistical data base. • Annual data was due to inadequacy of reliable quarterly data for most of the variables. • Long years of data are not available for most of the variables considered in the case of Ethiopia; the ideal national account data available to undertake time series based studies is from 1960 onwards. • Even the available data lack accuracy. • Total population is taken as a proxy for labor force though the use of population in an empirical study could result in overestimating the contribution of labor as a factor of production to the rate of economic growth. • Gross fixed capital formation is used as a proxy for capital.

21. Framework of Analysis • As indicated in recent empirical literatures, excluding relevant variables may understate or overstate the causality between variables (See Al-Yousif (1999), E. J. Medina-Smith (2001), Keong, Yuop and Liew (2003). • The econometric technique employed in this study is a multivariate co-integration and error correction procedure with the hypothesis that GDP is a function of aggregate exports, imports, capital, labor force and exchange rate, which can be stated as follows: • The signs above the variables suggest the anticipated relationship between each explanatory variable with the dependent variable (real GDP).

22. Introducing logarithm to the variables in equation (2) yields: Where LRGDPt, LLt, LRGCFt, LRERt, LRMt, LRXt are the logs of output, labour, gross capital formation, exchange rate and export variables respectively. The coefficients β1….β5 are parameters and Ut is the random disturbance term. Equation (3) will form the basis of estimations in this study.

23. Tests for Order of Integration • The primary step in testing whether variables share a common trend in such a way that they can be considered a long run equilibrium relationship is to find out whether each series contains a stochastic trend. • Estimating regressions using non-stationary variables based on ordinary least square lead to spurious and inconsistent results (Gujerati, 1995, Enders, 1995). • It is also difficult to conduct hypothesis testing in non-stationary variables as the classical assumptions on the property of the disturbance term is violated (Rao, 1994). • One could achieve stationarity by applying appropriate differencing called order of integration.

24. Augmented Dickey-Fuller (ADF) Test • The standard Dickey-Fuller test is conducted by estimating the following regression equation: Where is the differencing operator, Yt represents the variables to be estimated (i.e. LRGDPt, LLt, LRGCFt, LRERt, LRMt, LRXt), is constant, is a the trend coefficient, U is the white noise residual of zero mean and constant variance and t is the time or trend variable. The null and alternative hypotheses may be written as follows:

25. Accepting the null implies there is a unit root (the series is non-stationary) where as rejecting the null implies Yt is a stationary time series. • According to Rao (1994) the problem associated with the simple Dicky-Fuller unit root could be avoided by running the ADF test, which is derived from the regression equation: Where ΔYt-1 is equal to (Yt-1- Yt-2), ΔYt-2 is equal to (Yt-2- Yt-3), etc. and m is the maximum lag length on the dependent variable to ensure that Ut is the stationary random error. • The null hypothesis of a unit root is rejected if the t-statistic associated with the estimated coefficients exceeds the critical values of the test. • The ADF specification accounts for possible autocorrelation in the error process Ut through the lagged dependent variable on the right hand side. • The practical rule for establishing the value of m (i.e. the number of lags) is that it should be relatively small in order to save degrees of freedom, but sufficient to remove serial correlation in the residuals. The weakness in this test is that the power of the test may be adversely affected by mis-specifying the lag length (Rao, 1994).

26. Tests for Cointegration • The next step is to find out whether the variables share a common stochastic trend, i.e. to test whether two or more variables are cointegrated. • Cointegration can be regarded as the empirical counterpart of the theoretical notion of a long-run relationship among the variables. • In other words, a cointegration of two or more variables suggests that there is a long run, or equilibrium relationship between the variables (Rao, 1994). • Cointegration technique provides a means of identifying and hence avoiding spurious regressions generated by non-stationary series. • When variables are cointegrated, the OLS estimates from the cointegrating regression will be super-consistent.

27. The Johansen Approach • To determine the long run relationship between export growth and economic growth in this study, the Johansen (1991) multivariate co-integration test will be employed, which involves three steps. • For an intuitive insight into the Johansen method, consider a two-variable case. Suppose that two I(1) variables, x and y, are determined by the following Auto-Regressive Distributed Lag (ADL) equations with a maximum of two periods: • Equations (5) and (6) can be re-written as follows:

28. Equations (7) and (8) can be expressed in matrix form as follows: Where Zt = ……………………………9a β1= ………………………9b β2 = …….………9c t = …………………..……9d

29. In a multivariate case of m variables that are all I(1), there will be m equations similar to (5) and (6). These equations will have a matrix formulation similar to (9) except that β1 and β2 are now both m  m matrices and t now contains m disturbances. • If the m variables in (9) are cointegrated, it means that the m equations are free from spurious regression problems. However, standard estimation methods are not applicable. The Johansen procedure not only determines the number of cointegrating vectors but also provides estimates of these vectors. • For the purpose of testing the number of cointegrating vectors, Johansen and Juselius (1988, 1990) propose the use of two likelihood ratio test statistics namely, the trace test and the maximum eigenvalues test.

30. Trace test • The trace statistic for the null hypothesis of r cointegrating relations is computed as follows: Maximum Eigenvalue Test • The maximum eigenvalue static tests the null hypothesis of r cointegrating relations against r +1 cointegrating relations and is computed as follows.

31. Vector Error Correction Model (VECM) • If variables are cointegrated, an error correction model exists. Error Correction Model combines both the short run dynamics and long run properties and at the same time eludes the ‘spurious regression’ problem. • The VECM can be expressed as follows. Where, ZYt-1 and Zxt-1 represent the error terms lagged by one period for the real output and real export equations, respectively. The coefficient measures the long run equilibrium relationship while 1, …, 6 and , … measure the short run causal relation.

32. Empirical ResultsTests for Stationarity • The ADF test was employed to the data in levels and in first differences (with series lagged once and twice and with the option of intercept and trend) to examine the stationarity of the variables in the model. • All variables under investigation are found to be non-stationary at levels suggesting that the estimated values of all variables in levels are not statistically different from zero (table 2A). • However, after first differencing the null hypothesis of the variables are stationary at the 1%, 5% and 10% levels of significance and hence characterized by one order of integration I(1) processes.

35. Co-integration and Error Correction Model • The fact that all the variables appear to be integrated in an order of one (i.e. I(1)), and their first differences appear to be stationary, they are all candidates to be included in a long run relationship. • The next procedure is to test for co-integration. The Johansen procedure was used for detecting the number of cointegrating vectors. • The results show that the variables are co-integrated with a maximum of two cointegrating equations. • The null hypothesis of no co-integration among the variables is rejected at the1% and 5% levels of significance. • The results thus demonstrate that the considered variables are cointegrated in that there is a long run equilibrium relationship among them and the existence of causality for instance, between export and GDP in at least one direction.

36. Table 3: The Trace and Eigenvalues Tests r

37. Long run Relationship • The residual generated from estimation of the long-run output growth model is stationary suggesting that the variables are co-integrated. • The results of the long-run model reveals that all the variables have the anticipated signs and are of reasonable magnitude: export, exchange rate, labor, and capital positively affect output growth while import is negatively related with output. • The estimated parameter of real import is highly significant but of a negative sign: - indicates that the much dependence on import has negatively affected the growth of the economy, - tells us that the rise in imports slows down economic growth through affecting the country’s international reserves. - the country had used to expend a great deal of foreign exchange on the purchase of military armaments during the regime of the Military junta. - may also suggest that the imports of consumer durables and non-durables outweigh the imports of capital goods. • Real exchange rate is positively and significantly related to growth suggests the need to shift in the structure of both production and trade towards products with demand elastic and high value added produces.

38. Table 4: The Estimates of the Long run Parameters (Johansen Method)

39. Short Run Dynamics • After estimating the coefficients of the long-run model, the next step is estimating the coefficients of the short-run dynamics. • A VECM was estimated that incorporates the short-run interactions and the speed of adjustment towards long run equilibrium. • In estimating the model dummy for drought, war and government change was introduced to capture their effect from the regression analysis. • The estimated results inclusive of the dummy are reported below (t ratios in parenthesis). D(LGDP) = 0.424 *D(LLt) + 0.443 *D(LRXt) + 0.127 *D( LRERt) (2.957) (2.15) (0.594) - 0.073 * D(LRMt) + 0.027*D(LRGCFt) – 0.017 *Dummy - 0.525*ECMt (-0.184) (0.325) (-1.943) (-3.356)

40. Coefficients of the short run dynamics reveals that both export and labor are statistically significant showing their contribution to the growth of the economy in the short run. The regression result confirms the hypothesis that exports positively affect the growth of the Ethiopian economy. • Though statistically insignificant other variables have the expected signs. • The literature relates dynamic technological spill over with manufactured exportable produces, i.e. scholars argued that income elastic goods (manufactures) have a larger effect on economic growth than traditional agricultural goods. • However, the results obtained in this study evidenced that the relationship between export and economic growth holds despite the Ethiopian export basket is dominated by traditional primary produces and in the face of an inward oriented trade strategy. • The results thus led to skepticism on the argument that a positive and significant linkage in export and output growth is achieved only when the country’s export basket is dominated by higher value added manufactured exports.

41. The magnitude of the error correction coefficient is - 0.525 implying that with in one year it adjusts about 52.5% of the disequilibria. In other words deviation from the long run equilibrium is adjusted fairly quickly 52.5% of the disequilibrium is removed each period. • Where as the negative coefficient of the dummy variable suggests that the prolonged civil war and the border conflict, the cyclical drought and famine that hits the country and the change of government with a new ideology has negatively affected the growth of the economy. • The results of the Granger causality test at different lag lengths reveal that the causation is weak implying that export growth affects output growth through another channel, which could be ascertained in further research works. • Estimating the model with other specification may bring about a better result.

42. 6.0. Conclusion and Policy Recommendations • The paper empirically examined the export growth and income (output) nexus in the Ethiopian context using multivariate time series approach. • Consistent with expectation, export growth and output growth were found to be positively related supporting the export-led growth hypothesis. Thus export expansion brings economic growth in various ways. • Despite the focus on export diversification in the development plans of the country, the export pattern is still dominated by traditional produces and more so on coffee whose world price is fluctuating; - Symptomatic of the lack of economic transformation and absence of structural change in the mode of production. - Still primitive mode of production predominates in the major sector. • The country has failed to fully utilize the opportunities offered by the African Growth Opportunities Act (AGOA) of the United States of America.

43. Despite the country has implemented an ADLI strategy, there remains a more long distance to go to use this sector as the basis for industrialization and sustainable progress. • It is impossible to see a vibrant Ethiopian economy without structural transformation. • I.e. Ethiopia should in the long time horizon diversify its economy towards higher value added and income elastic products. In this regard: - the ardent need for a concerted effort to improve internal conditions and bring about diversification of the productive structure and curtailment of supply side constraints. - Rehabilitating the domestic manufacturing industries and encouraging both private domestic and foreign investments.

44. The study suggests that the country needs to strengthen export capacity and to promote diversification in the sector to fully exploit the benefits of the sector and achieve sustainable growth. - Maintaining export friendly effective exchange rate, lowering tax burden on exports and imports of inputs, providing exporters with easy access to foreign exchange for the purchase of intermediate items and preferential and/or lower interest rates on bank loans. - Designing export promotion strategies, policies and support services conducive towards stimulating competitiveness remains crucial. Targeted and concrete export support services, product specific export market research, active participation in international trade fairs and instituting officially sponsored trade missions. - Addressing sector specific bottlenecks. - Exerting intensified efforts to identify key commodities to which Ethiopia has a comparative advantage and could diversify (e.g. expanding the production of horticultural and flower products, fruits and vegetables, livestock products, and exploiting apicultural possibilities).

45. Identification of key supply constraints that handicap the performance of the export sector and take appropriate measures to improve domestic conditions for business development. • Improving problems in infrastructure such as roads, air transport, railways and energy and other facilities, which would greatly contribute for facilitating export diversification and effectively competing in the international market. • Success in export diversification and promotion is an amicable means to extricate the country from excessive dependence on foreign loans and aid and eradicate abysmal poverty. - The government needs to work together with the business community inculcating an atmosphere of mutual trust and confidence building to persuade them engage in the export diversification policy making. • Due attention should also be given to exploit abundant mineral resources, the export of which would be a good source of foreign exchange. There is therefore, a need to develop new policies and strategies to encourage investments in these untapped resources.

46. The success to bring about economic transformation, to increase production, and to diversify exports and ultimately, to break the vicious and visible circles of underdevelopment relies on addressing the problems of governance, avoiding bureaucratic chains, creating an enabling supervisory, regulatory and legal environments and increasing private investment particularly in the productive sectors. • The results obtained from the regression indicate that labor growth positively and significantly affects economic growth both in the short - and long - term horizon. In this regard, Ethiopia needs to educate its active labor force with the required skills. • Ethiopia has decided to become a member of the WTO. The country should thus improve its position in the international economy so as to benefit from its accession to the WTO.