Birr devaluation and its effect on trade balance of Ethiopia: An empirical analysis

This study examined the effect of Birr devaluation on trade balance of Ethiopia for the period 1970-2014 using the Vector Error Correction Model. The key results of the present study revealed that Birr devaluation deteriorates the trade balance of Ethiopia in the short run and improves it in the long run. Moreover, the result from the long and short run models showed that real effective exchange rate, money supply, domestic real income and term of trade are the major determinants of the trade balance of Ethiopia both in the short and long run. Besides the short and the long run model, the impulse response function and the descriptive analysis revealed that a J-curve phenomenon exists for the trade balance of Ethiopia. That means, the finding of this study showed that the Marshall-Lerner Condition holds only in the long run. So, the present study revealed that the elasticity, monetary and absorption theories are significant in explaining the trade balance of Ethiopia. The policy implication is that policies that encourage productivity improvements, diversification of the export sectors and expansion of import computing industries are alternative policies for devaluation. Moreover, government may need to be conservative in using devaluation (exchange rate policy) to improve trade balance as it may worsen the situation in the short run. Thus, the country on the process of industrialization, first needs to promote import computing industries and then, once the production gets its way, devaluation would be clear.


INTRODUCTION
Exchange rate is one of the key barometers of economic performance, indicating output growth, demand conditions, levels and trends in monetary and fiscal policy stance.Exchange rate policy emerged as one of the controversial policy instruments in developing countries in the 1980s, with strong opposition to devaluation for fear of its inflationary impact, among other effects.Due to their vulnerability to external shocks, less developed countries have considered the exchange rate as the central policy issue over the years.The external economic shocks, as argued, contributes to the deterioration of the trade balance, domestic inflationary pressure and the overall competitive position of less developed countries (UNCTAD, 2005).
The legal currency of Ethiopia was issued on 23 July, 1945 by defining the monetary unit as the Ethiopian E-mail: bekatfech@gmail.com.dollar.The currency Proclamation of 1945, defined the monetary unit of the country as the Ethiopian dollar with a value of 0.355745 g of fine gold.The linkage with fine gold was in accord with the monetary system established by the Bretton Woods Agreement of 1944 and automatically established the exchange rate between the national currency and other currencies with the same arrangement.
Accordingly, the official exchange rate of Ethiopian currency with US dollar was created, with the official exchange rate of 2.48 birr per United State dollar, on July 23, 1945.After almost two decades, that is, 1 January1964, the Ethiopian birr was slightly devalued to 2.50 birr per US dollar.During the collapse of the Bretton Woods System in 1973, Ethiopian birr was revalued to 2.07 birr per US dollar.Since then, the Ethiopian currency was pegged to the US dollar at the rate of 2.07 birr per dollar until massive devaluation of October 1992 (Degefa, 2001).
In May 1991, the Ethiopian landscape was markedly overwhelmed by major economic and political changes.The military government that terrorized the country for 17 years collapsed and the transitional government of Ethiopia assumed the political power.Under the new reform program, foreign trade and exchange rate regimes were liberalized; prices of inputs and outputs were decoupled from arbitrary government regulation and interference; the financial market was reformed to allow private sector participation in commercial banking, insurance and micro credit services; export tariffs were abolished; export subsidies to domestic, export-oriented firms were eliminated and were replaced by incentives that provided the duty-free importation of raw materials.
Moreover, in October 1992, Ethiopia's national currency was devalued by 142% from its pegged rate of 2.07 per US dollar to 5 per US dollar, showing the first major change on the value of birr.Currency devaluation is designed to improve the external imbalance, boost output growth in general and export growth in particular.Though still fragile and vulnerable to the vagaries of nature and aid money, the export sector in Ethiopia has shown tangible improvements since the country abandoned the fixed exchange rate regime in 1991.For instance, export receipts have increased by 35 fold between 1990/1991 and 2009/2010.The export industry has also shown significant diversification away from its dependence on coffee.In 1991/1992, when the reform package was launched, coffee brought 53% of the country's total export revenue but in 2008/2009, its share declined to 26% while the shares of other goods such as oilseeds, chat, gold, flower, leather and leather products increased substantially.Though much of this diversification is within the same industry, the overall result shows a significant departure from the traditional, mono-crop dominated export sector.However, at worst, the problem is that import is near unresponsive to currency devaluation and its value increases by more than export value and this lead to further deterioration of the trade balance of Ethiopia.For instance, according to the data from Ethiopian Custom Authority, import spending has increased by 58 fold between 1990/1991 and 2009/2010.The total export values of Ethiopia during 2012Ethiopia during /2003Ethiopia during , 2013Ethiopia during /2014Ethiopia during and 2014Ethiopia during /2015 were 3115.8, 3300.1 and 3019. were 3115.8, 3300.1 and 3019.3 million US dollar, while the total import values during the same periods were 11,460.6, 13,712.3 and 16,458.6 million US dollar, respectively.So, the export earnings of Ethiopia in 2014/2015 reduced by 8.5% as compared to export earnings in 2013/4 and this is mainly due to slowdown in export earnings from oilseeds, gold, chat, pulses, live-animals and electricity.But, total imports in 2014/2015 increased by 20% over last year and this is as a result of a rise in imports of capital goods, consumer goods, semi-finished goods and raw materials.As a result, the trade deficit in merchandise trade during 2014/2015 stood at USD 13.4 billion, widened by 29.1% relative to the preceding fiscal year mainly due to the significant growth in total import bills coupled with low performance in the growth of total export proceeds (NBE, 2015).
Having faced with widening gap between export and import values, lack of diversification, and shortage of foreign currency, Ethiopian government has undertaken a continuous devaluation of its currency, birr.But, the gap between export receipts and import payments has been rising following the massive devaluation of Ethiopian birr by 142% in 1992.It seems, therefore, that birr devaluation has failed to narrow the gap between export receipt and import payment and even leads to further deterioration of the trade balance of Ethiopia.
Study by Agbola (2004) on Ghana Economy, Shahbaz and Abdul (2010) on Pakistan Economy, Aziz (2008) on Bangladesh Economy, El-Ramly and M. Abdel-Haleim (2006) on Egyptian Economy and Ratha (2010) on China economy, found that currency devaluation deteriorates the trade balance of a country.Frank Agbola also found that there is no J-curve phenomenon for the trade balance of Ghana.However, movement of the trade balance in response to change in exchange rate appears to be characterized by an M-curve phenomenon.Similarly, Shahbaz and Abdul (2010) also tested the Jphenomenon for the trade balance of Pakistan and they found that there was no evidence of J-curve phenomenon.
Marial and Ahmad (2005) examined the effect of devaluation of currency on trade balance for three North Africa countries: namely Morocco, Egypt and Tunisia.They found that devaluation improves only the trade balance of Morocco in the long run.Moreover, Oylinlola (2011) examined the effect of devaluation on trade balance of four countries of West African Monetary Zone, namely Gambia, Nigeria, Sierra Leone and Ghana, and he found that devaluation improves only the trade balance of Nigeria in the long run.The study also found the J-curve phenomenon only for Nigeria.
But, the study by Edward and Wilcox (2003) on South African economy, Kamoto (2006) on Malawi economy, Hailemariam (2011) on Ethiopian economy and Lawrence (2001) on South African economy, found that currency devaluation improves the trade balance of a country.Loto (2011) examined the Marshall Lerner condition for the trade balance of Nigeria.The study adapted the elasticity approach of the Marshall-Lerner condition to the balance of payment adjustment mechanism.The empirical results showed that devaluation does not improve the trade balance of Nigeria since the sum of demand elasticities for imports and exports is less than unity (0.78).On the contrary, Bahmani-Oskooee (1998) applied the Johansen and Juselius cointegration technique to estimate the trade elasticities for Greece, Korea, Pakistan, Philippines, Singapore and South Africa and found that the Marshall-Lerner condition holds for these countries in the long run.
Thus, despite the colossal effort that Ethiopian government has made to buoy export performance, and curtail imports via its expenditure switching policy (devaluation), there has been a persistent rise in trade deficit of Ethiopia over the past 25 years.In spite of the wide interest in the impact of devaluation on the trade balance of a country, most empirical research on this topic that focused on African countries in general and Ethiopia in particular, has been extremely limited.Taye (1999) applied a macro-simulation approach to a macroeconomic model for Ethiopia.His results indicated that devaluation has a positive impact on the trade balance because of the reduction in imports and stagflationary via its impact on output and employment.
This study therefore, tried to assess devaluation works via supply channel or demand channel by examining the effects of birr devaluation on the trade balance of Ethiopia using econometric approach as opposed to macroeconomic-simulation method that was used by Taye (1999).Taye did not examine the J-curve phenomenon for the trade balance of Ethiopia but this study does.
The overall objective of this study is to test the effect of currency devaluation on the trade balance of Ethiopia for the period 1970-2014 using Vector Error Correction Model (VECM) which was developed by Johansen (1987) and Johansen and Juselius (1990).In line with this general objective, the study also tried to achieve the following specific objectives with the methodology at hand.
1.This study aimed to investigate the effects of Birr devaluation on trade balance of Ethiopia.2. This study also planned to check whether the J-curve phenomenon exists for the trade balance of Ethiopia 3. In addition, the present study tried to check the relative relevance of the absorption theory, monetary theory and elasticity theory of trade balance in explaining the trade Eshetu 105 balance of Ethiopia.

SURVEY OF THE RELEVANT LITERATURE
Starting from the advent of floating exchange rate in 1973, there has been a heated debate on the effect of devaluation on the trade balance.Several theories have been used to analyze the effects of exchange rate movements on the trade balance in developing countries and these include the structuralists' (elasticity theory), the Keynesian and the monetary theories.The elasticity theory argues that the price effect outweighs the volume effect in the short run while the volume effect dominates the price effect in the long run (Krueger, 1983).
The monetary theory on the other hand, insists that devaluation increases the domestic price and this leads to a fall in real money supply.As a result of the fall in real money supply, imports decreases and trade balance improves.On the other hand, monetary theory also argues that higher money supply leads to trade deficit, whereas lower money supply causes surplus in trade balance.The last but not the least one is the Keynesian theory or absorption approach which assumes that devaluation improves a trade balance of a nation if the substitution towards domestic goods in response to the change in relative price boosts output more than spending (Pilbeam, 1998).
Furthermore, empirics on the impact of currency devaluation on trade balance of devaluating nation in general has shown a contradicting results presented in this chapter.Thus, this section presented the various theories and empirical results on the effect of currency devaluation on trade balance of a nation.

The elasticity theory trade balance
Whether devaluation improves the trade balance of the devaluating nation depends on the elasticity of foreign demand for the nation's export and the elasticity of domestic demand for imported goods in that nation, according to elasticity approach.Put differently, if the foreign demand for a nation's export is inelastic, then a devaluating nation will not be able to increase its foreign exchange earnings.This is because, even if devaluation has made the nation's export cheaper for the foreigners, the foreign countries' demand for a nation's export has failed to increase.This will not improve the trade balance of devaluating countries.
In the same way, if the devaluating nation's demand for imported foreign goods is inelastic, the level of a nation's import will remain the same as before.However, devaluation has made imports costly to a devaluating nation and this implies that a devaluating nation will now spend more dollars on fixed amount of imports thereby leading to the worsening of the trade balance of the nation.But, if on the other hand, both the demand for import of a nation and the demand for export of a nation are elastic, devaluation improves the trade balance of a nation.Thus, the success and the failure of devaluation in correcting the disequilibrium in foreign sector depend mainly on the elasticity of demand for export of a nation and elasticity of demand for imported foreign goods.
Moreover, if the export demand is unitary elastic, currency devaluation has no impact on the trade balance situation of a nation.Simply put, if the export and import elasticities are more elastic, devaluation will help reduce the trade balance disequilibrium of a nation but, when these elasticities are lower, devaluation is helpless in correcting the disequilibrium in the foreign sector of a nation.Even worst, it will increase the size of deficit and worsen the trade balance situation of a nation.However, import demands for developing countries like Ethiopia, are inelastic as their imports are primarily composed of capital goods, semi-finished goods, fuels and the like of which, a nation cannot cut their imports.Similarly, small developing countries have in elastic export supply curve as their export is mainly composed of primary agricultural commodities with a longer gestation period (Mannur, 1995).

The absorption theory of trade balance
Under elasticity approach, devaluation is assumed to work on trade balance via its price effects, so that devaluation directly increase the prices of imports and reduce the price of exports, and leads to improvement in the trade balance of a nation if import and export demand elasticities are large enough.So, under elasticity approach, the effects of currency devaluation are on external variables viz.exports(X) and imports (M) (Keynes, 1936).
However, the Keynesian theory considers macro approach towards the impact of devaluations which considers the income effect.According to this theory, devaluation affects domestic variables viz.consumption, investment and national income and this need to be taken into account in analyzing the effects of devaluation on the balance of payment situation of a nation.The national income equation of an open economy is given as: (1) Where, total national incomes (GDP), imports, domestic consumption, investment spending, government purchases and exports, respectively.Furthermore, the left sides of the above equation represents the total supply in an economy which is the sum of the total domestic supply (Y) and total imports (M), whereas the right hand side of the equation shows the way these total supply in an economy are spent or the total demand viz.consumption, investment, government purchase and exports.The above national income equation can be written as: (3) Where, TB is a trade balance or net export, is a domestic expenditure or absorption, is domestic output level (GDP).This implies that trade balance can be improved if domestic output level exceeds domestic spending.Devaluation, according to this theory, reduces trade deficit if the substitution towards the domestic goods as a result of devaluation increases output more than domestic expenditures.Put differently, devaluation improves a trade balance of a nation only if it raises a nation's income (Y) more than a domestic expenditure or if there is expenditure switching effect in the correct direction and amount.Equation 3 also helps to determine a number of key variables that have effect on imports and exports and hence trade balance of Ethiopia.

The monetary theory of trade balance
Mundell and Johnson (1960) developed the monetary approach towards the balance of trade and stated that the balance of trade is essentially a monetary phenomenon.According to this theory, balance of trade deficit and surplus is caused by easy and tight monetary policy, respectively.Currency devaluation has an impact on the balance of trade via its impact on the real money supply.This means that, when there is currency devaluation, domestic prices rise and real money supply decreases which will lead to lower imports.However, if devaluation of currency is followed by further rises in the nominal money supply, the initial disequilibrium will reestablish and the positive impact of devaluation will be negated.Therefore, according to the monetary theory of trade balance, devaluation leads to improvement in trade balance of a nation via its demand channel by decreasing imports as long as the money supply remains unchanged (Blanchard, 1993).
The essence of this theory is that the trade deficit and surplus in the balance of payments is denoted as disequilibrium in the money market.So, the trade deficit or surplus or disequilibrium in the money market is a transitory phenomenon which lasts only until government responds by changing money supply.According the monetary theory, the country's supply of money is given by: (4) Where, H, D and F are the nation's total money supply, domestic components of the nation's monetary base and the foreign components of the nation's monetary base, respectively.This money supply equation can be written as: (5)

Where
, and denote the change in net foreign assets, total money supply of a nation, and the central bank's extension of domestic credit.The above equation states that the change in the central bank's holding of foreign assets is equal to the change in the stock of high powered money minus the change in the domestic credit.The important point about the above equation is that is the balance of payment.The domestic component of the nation's monetary base (D) is the domestic credit created by the nation's monetary authority.Thus, the external balance is given by: (6) But, ∆R is a change in international reserve and the same as ∆NFA.Thus, Equation 6shows that import minus exports is equal to foreign saving (Fs) less than the change in international reserve (∆R).Therefore, the relationship between trade balance (external balance) and the monetary account can be written as follows from Equations 5 and 6: Equation 7 shows how the external account and the monetary accounts are related and the monetary theory of trade balance states that change in domestic money supply is negatively related to the trade balance of a country.This means an excess stock of money supply will lead to an out flow of reserves (import increases) or a balance of payment deficit.It is clear from Equation 7 that devaluation improves the trade balance of a nation in line with the monetary theory.If there is disequilibrium in the money market, this will lead to greater import or export which in turn brings equilibrium in the money markets via its impact on the nation's monetary base.This implies that when the money market is in equilibrium, so does the balance of payments.Therefore, the nation's balance of payments surplus/deficit is a temporary and selfcorrecting phenomenon (Dornbusch, 1990(Dornbusch, , 1998;;Dornbusch and Fischer, 1996).
Eshetu 107 The J-curve phenomenon The J-curve phenomenon predicts that the long run and the short run effects of devaluation on the trade balance are not the same.The time path which the balance of trade follows generates a J-curve phenomenon.The nation's trade balance may worsen sooner after devaluation or depreciation, before improving later.This is mainly due to the tendency of the domestic currency price of imports to rise faster than export price, soon after devaluation, with quantities not changing very much.But, in the long run, there are two volume effects which move in opposite direction.The volume of export should increase and the volume of import should decrease because imports are more expensive.Therefore, in the very short run, the price effect will dominate the volume effect while the volume effect will dominate the price effect in the long run.This is mainly due to the difference in elasticity of demand and supply in the short and long run, with the long-run elasticities generally exceeding the short-run elasticities.This difference between the long and short-run elasticities laid the foundation for the Jcurve effect of devaluation (Lindert, 1993).

Model specification
In this study, three different theories were used to specify an empirical model which helps to achieve the stated objectives of the study.The absorption, the elasticity and the monetary theories of trade balance were used for the analysis of trade balance of Ethiopia.The trade balance is conventionally measured as the total value of exports less imports.This study, however, uses the ratio of exports to imports values.The merit of this choice is that the unit of measurement is unimportant and the constructed variable can be conveniently interpreted as nominal or real trade balance.As mentioned above, the absorption theory, the elasticity theory and the monetary theory of trade balance were used to analyze the trade balance of Ethiopia.The relative relevance of the above three theories of trade balance in explaining the trade balance of Ethiopia were also tested in the present study together with the response of the trade balance of Ethiopia to birr devaluation.The absorption theory (Keynesian theory) was used to relate exchange rate with trade balance and develop a theoretical frame work for our trade balance model which incorporates the various theories of trade balance.Thus, the theoretical frame work for the trade balance of Ethiopia is given as follows: Where TB is a trade balance, X is export revenue, M is import expenditure, Px is export price in home currency, Qx is quantity of export, E is the nominal exchange rate, Y* is foreign income, P*m is the price of import in foreign currency, Qm is the quantity of import and Y is the domestic national income.Therefore, the main determinants of a trade balance of a nation are exchange rates, domestic real income, interest rate, money supply, government expenditure and relative price or term of trade.Thus, the Keynesian/absorption theory asserts that if devaluation is to succeed in eliminating the disequilibrium in the foreign sector or trade deficit, then a nation as a whole must be producing more outputs of goods and services than it is consuming or absorbing by way of higher consumption and investment expenditures (Pilbeam, 1998).Therefore, using Equation 1, a comprehensive model that incorporates the synthesis of the various theories of trade balance (the absorption, elasticity and monetary theories) was specified for Ethiopia as follows: Where, is the log of trade balance which is defined as the ratio of exports value to imports, such that an increase implies an improvement of the trade balance and , , , and refers to the log of domestic real income, real effective exchange rate, government expenditure, domestic money supply, term of trade and interest rate, respectively.The coefficient of , β1, is of key interest here.The regression of the trade balance as it appears in Equation 2on real effective exchange rate 1 , domestic real income, money supply, term of trade or relative price and interest rate can be driven from the perspective of three approaches to the balance of trade analysis.The elasticity approach focuses on exchange rate as the determinants of the trade balance and thus the relevant coefficient is β1.The monetary approach assumes money supply as the determinants of the trade balance and hence the relevant coefficient to analyze the trade balance of a nation is β4.Furthermore, the absorption approach considers domestic real income level as key determinants of the trade balance of a nation (Salvatore, 2001).
In Ethiopian case, an increase in real domestic income ( ) increase the volumes of imports from the trading partners and hence the estimate of β2 is expected to be negative.The coefficient of domestic money supply ( ), interest rate ( ) and term of trade ( ) are expected to be negative in line with economic theories.The coefficient of real effective exchange rate ( ) in the above model is indeterminate.

Types and sources of data
The study exclusively used secondary data for the period 1970-2014.This means, 44 years data were used in this study to estimate our model and answer the basic objectives of the study.The study used annual data because there are no quarterly data for most macroeconomic variables in Ethiopia.The main sources of these data include the National Bank of Ethiopia, World Bank, Ethiopian Economic Association Data Base, Ministry of Finance and Economic Development and UNICTAD.

Methods of data analysis
Before carrying out the estimation of the above models, the time series characteristics of all data were examined and vector auto regressive (VAR) model was used.This means, VAR model was used because it implies that all the variables in the model are treated as endogenous.In many time series analysis, one of the most important preliminary steps in regression analysis is to test the 1 Nominal exchange rate is the amount of US dollars that will be obtained for one birr in foreign exchange market or it is merely the price of one currency in terms of other currency.But, the real exchange rate is the nominal exchange rate adjusted for relative prices between the nations under consideration.Similarly, NEER is a trade weighted average of the bilateral nominal exchange rate of all the major trading partners of Ethiopia, whereas the REER is a trade weighted bilateral real exchange rate of all the major trading partners of Ethiopia.Therefore, in this study, a fall in REER and NEER refers to devaluation.
stationarity of the data used in the analysis.The main goal of under taking stationarity test is to get a variable which has a constant mean, variance and covariance.The regression results from the above models may be spurious if trade balance, real effective exchange rate, money supply, real GDP, term of trade and interest rate are all non-stationary.So, test of co-integration to check the existence of long run relationship between all variables in the model has a paramount importance in time series data analysis.
This study used the classical unit root test, namely, the Augmented Dickey-Fuller (ADF) test for the stationarity of the series (Dickey andFuller, 1979 and1981).Augmented Dickey-Fuller test is based on the null hypothesis that a unit root exists in the time series (Guajarati, 1995).Using this ADF test, all the variables included in the models were tested for unit roots both at level and at first difference.
As shown in Table 1, all the variables are not stationary at level with lag of zero.Similarly, at lag one and two, all the variables are non-stationary at level.One way of converting non-stationary series into stationary series is to differentiate the data and in this study, the first differences of all variables in the model were tested for unit root and the result showed that all variables are integrated of order one.This means, they are stationary at first difference and thus, the test result from Table 2, revealed that all variables in the model are integrated of order one.
One of the most important steps after stationary test is the determination of the lag length which should be incorporated in the model.Once the optimal lag length is determined for the model, the next step in Johansen's co-integration analysis is to determine the number of co-integrating vector for the model.In this study, a common lag length of two is selected for our trade balance equation using the various information criteria.
Having established that all variables are integrated of the same order, cointegration test is the appropriate method for detecting the existence of long-run relationship.Johansen (1990) and Johansen and Juselius (1990) methodologies was used to test for the existence of long run relationship between various variables in the model (Table 3).
As seen from the result of the test, the maximum Eigenvalue test indicated that the null hypothesis of no cointegration vector is rejected since the test statistic (66.39) is greater than the 95% critical value (46.23).But the null hypothesis of one co-integration vector is not rejected and it can be concluded that there is one cointegrating vector.

Descriptive data analysis
Here deals with the descriptive analysis of the present study.It presents the trend analysis of exchange rate, trade balance, export values, import values and the likes.Ethiopia has been experiencing three distinctive foreign trade policies such as the relatively liberal trade regime, the highly controlled trade regime and the market oriented trade regime corresponding to the Imperial Era, the Derg Era and the current Ethiopian government, respectively.In the imperial era, except for some essential consumer items, imports were free from licensing or other quantitative restrictions and exporters were required to give their foreign exchange to commercial banks at the prevailing official exchange rate and this led to the existence of macroeconomic stability and a relatively liberal trade regime (Befikadu and Kebre, Ethiopia as follows: Where,  is the log of trade balance which is defined as the ratio of exports value to imports,
Realizing the negative effect of an over-valuation of the birr on the country's trade balance and hence, on output level, the Transitional Government of Ethiopia (TGE) devalued the birr gradually many times with an official exchange rate reaching 22 birr per US dollar in 2016.
Needless to say, the Ethiopian economy showed a strong economic performance, especially during 2003 to 2015, due to improvement in the service, construction and agricultural sectors.Since then, the country has been on a sustained, strong path of economic performance with mean real GDP growth of eleven percent for the period of 2003 to 2015.On the other hand, Ethiopia has experienced a long period of unfavorable trade balance deficits that means the external sector is characterized by structural trade deficits.This may be partly due to the composition of imports and exports and the import intensive nature of Ethiopian economy.Put differently, the composition of the imports of Ethiopia mainly includes capital goods, fuels, raw materials and consumer durables for which the country cannot afford to cut their imports despite the higher price.On the export side, the country highly depends on agricultural exports such as coffee, oilseeds, hide and skins, chat and the like which has unstable demand in the international market.Despite the colossal effort that Ethiopia government has made to improve the external imbalance and output growth via its expenditure switching policy (devaluation), the country has experienced a persistent trade balance deficit.

Total shares of the imports of Ethiopia
Ethiopia's aggregate imports have shown substantially soar since the demise of the Socialist government in the country in 1991.The nominal value of merchandise imports rose from 429.1 million birr in 1970 to 251.05 billion birr in 2014.Thus, the growth rate of imports, which had averaged 3% annually in the 1980s, during the socialist government, rose to an annual average of 27% between 2000 and 2014.
The data in Table 4 shows the mean share of the various imports of Ethiopia over the indicated ranges of periods.So, Table 4 revealed that the import of capital goods dominated aggregate imports from 1970 to 2014, with 44 years average share of 32.92% followed by consumer goods, semi-finished goods and fuels imports with 44 years average share of 30.16, 16.07 and 15.78%, respectively.
All the same, the share of all imports of Ethiopia did not change significantly over the four decades despite the continuous devaluation of our currency over the last two decades.Thus, it seems that our imports are price inelastic and devaluation failed to cut imports significantly.In sum, over the last four decades, not only the relative shares of Ethiopian imports have remained unchanged significantly, the distributional pattern of imports did not change with the import of capital goods leading and followed by consumer goods.The data also shows that the imports of semi-finished goods and fuels in Ethiopia stood third and fourth, respectively in aggregate imports.The proximate determinant of this outcome is the Agricultural Development Led Industrialization strategies which heavily relied mainly on imported inputs, particularly capital goods.Of course, this meant dependency on imported machinery and equipment that are basic to production in the economy.

Total shares of the exports of Ethiopia
Exports are important for the process of economic growth.They generate scarce foreign exchange reserves that are necessary to finance imports of goods such as fuels and investment goods which are crucial for the formation of capital and economic growth.Moreover, exports can also reduce the impact of external shocks on the domestic economy.Ethiopia's domestic exports, which are largely comprised of agricultural commodities, are subject to sharp fluctuations from internal and external disturbances.As in most developing economies, the commodity concentration of Ethiopia's exports has long been regarded as a major factor contributing to the short-term instability of exports.
As can be seen in Table 5, prior to the 1991 financial and economic reforms in Ethiopia, the mean share of coffee in the total export of Ethiopia was 63.61%.However, since then the share of coffee in the total exports of Ethiopia has been falling and this is partly attributed to export diversification via birr devaluation.Needless to say, the share of flowers, oilseeds, pulses and chat were insignificant, less than one percent, before the 1991 economic and financial reforms in Ethiopia.Indeed, their shares in the total imports have shown a significant rise through time since the reform with the continuous devaluation of Ethiopian birr.For instance, as seen in Table 5, the mean share of oilseeds, pulses and flowers was less than one percent before the liberalization of the foreign trade while it was 15.34, 6.34 and 6.20%, respectively between 2011-2014.Yet, the share of coffee and hides and skins in total exports have been falling over time despite the colossal efforts that the Ethiopian government has made to boost exports via birr devaluation and other incentives.Overall, the fall in the share of these two export items over time is attributed to two factors: birr devaluation which has boosted the exports of other items (diversification) and the variability of their prices in the international markets.The export volume of hides and skins has been falling over time and this indicates that the supply side factors (low production) also affect our export performance.
As shown in Table 5, the share of coffee and hides and skins has been falling over the three decades.The fall in the share of coffee exports is mainly attributed to the result of diversification that new items enter the export sectors (flower) and to some extent, to the variability of its price in international markets.However, the fall in the share of hides and skin exports is mainly attributed to supply side factor that is low domestic production.

Evolution of the trade balance of Ethiopia through time
Accurate understanding of trade balance response to changes in exchange rate is a crucial factor in the coordination and implementation of trade and exchange rate policies in Ethiopia.The traditional view is that a nominal devaluation of exchange rate improves the trade balance in the long run while making it to worsen in the short run.This is because the change in the exchange rate has two effects on the trade balance, the price effect and volume effect.The price effect works to make imports more expensive and export cheaper for foreign buyers.But, the volume of exports and imports do not adjust instantaneously in the short run, trade balance may initially experience some deterioration in the short run following the currency devaluation.Put differently, the price effect will tend to dominate the volume effect in the short run whereas, the volume effect will dominate in the long run.
As shown in Figure 1, Ethiopia has experienced a long period of unfavorable trade balance deficits which means the external sector is characterized by structural trade deficits.In order to overcome this adverse situation, the government initiated measures to bring about pragmatic changes in the trade balance via its expenditure switching policy such as currency devaluation.It is clear from Figure 1 that Ethiopian had experienced a trade balance surplus only in the two periods, 1972/73 and 1973/4 where the trade ratio( export/import) were 1 and 1.1, respectively.As shown in Figure 1, the trade ratio (export/import) of Ethiopia was rising during the imperial government which persuaded a somewhat liberal foreign trade policy.However, during the Derg regime where a highly controlled foreign trade policy was in place, the trade ratio of Ethiopia was falling during the whole periods which were partly due to the fall in exports as shown in Figure 1.But, following the demise of the socialist government in 1991, the trade ratio2 started to rise due to the then liberalization policy in Ethiopia.However, Figure 1 shows that the trade balance of Ethiopia was rising (improving) during the Imperial regime before 1975 due to the rise in exports despite our overvalued currency.On the other hand, the trade balance of Ethiopia was falling for the whole period during the socialist (1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991) and this continuous fall in the trade balance of Ethiopia during the Derg Era was mainly due to the then highly controlled foreign trade policy which affected the export sector of Ethiopia adversely.As shown in Figure 1, following the liberalization of Ethiopian economy in 1992, there was an increase in the trade balance till 1997.This is mainly due to the rapid rise in the export value in this period.
But, since 1998, the trade balance of Ethiopia has been falling (deterioration) despite the continuous devaluation of the currency and this is attributed to the rise in the import values as shown in Figure 2. But, since 2003, while the REER of Ethiopia has been rising, the imports and the trade deficit of Ethiopia have been rising at higher rate.On the other hand, Ethiopian economy has been growing on average at a rate of more than 11% since 2003 and it seems that some Ethiopia's frequent trade balance deficit have been the results of good opportunities for new capital formation in Ethiopia such as building new roads, expansion of textile industries, power development, irrigation development and expansion of cement factories which all depend on the imports of raw materials from abroad.
Figure 1 shows that Ethiopia has experienced a trade balance surplus only in two periods in 1972 and 1973 and since then the gap between export and import values (trade deficit) has been widening and reached its maximum value during the end of the civil war in 1991.At worse, the import values of Ethiopia has been rising at higher rate despite the continuous devaluation of the currency which gave birth to the continuous deterioration of trade balance.The variation in the trade balance of Ethiopia is mainly explained by the variation in exports as compared to that of imports.Thus, it can be said that the trend of the trade balance of Ethiopia mimics the trend of the export of Ethiopia.
Therefore, according to this visual representation, though devaluation promotes export, it failed to cut imports and even the growth in values of import outweighs that of exports and led to continuous deterioration of the trade balance.Moreover, Figure 2 shows that there has been a continuous variation in exports of Ethiopia during the last two regimes.The implication of this visual representation is that Ethiopia cannot afford to cut its imports as most of its imports are composed of capital goods and fuels which are very important for the development of an economy.Despite the continuous devaluation of birr, there has been a continuous rise in the values of imports and the rise in the values of exports failed to out weight that of imports and at the end of the day, led to the deterioration of the trade  balance.

Determinants of the trade balance of Ethiopia
This study used the vector autoregressive model to assess the relationships between dependent and various independent variables in the model.This is due to the fact that economic theory rarely tells us anything about short-run dynamics and it is rather mainly based on long run equilibrium relations.However, this model allows for a representation which differentiates between long-run equilibrium relations and short-run adjustment processes.Therefore, the major findings of this study which includes the long and short run dynamics along with the impulse response functions (IRFs) of the trade balance of Ethiopia are presented.This deals with the trade balance regression which helps to examine the effects of devaluation on trade balance of Ethiopia.Impulse response functions (IRFs) and the coefficient of short and long run dynamics were used to check whether the Jcurve phenomenon holds or not for trade balance.showed a single co-integrating vector, there is only one equation with one dependent variable.The dependent variable in the above long run model is the log of the trade balance of Ethiopia and since all the variables entered in log form, their long run coefficients are elasticities.

As shown in
Table 6 summarizes the estimation results of the long run trade balance model.Fortunately, all the variables in the model: money supply, real GDP, real effective exchange rate, interest rate, government expenditure and term of trade are statistically significant in affecting the trade balance of Ethiopia.The log of REER with a coefficient of -0.469 shows a negative long run relationship between the REER and trade balance.That means, 10% increase or appreciation of Ethiopian birr will lead to 4.69% deterioration in the trade balance or 10% depreciation or decrease of Ethiopian birr will lead to a 4.69% improvement in the trade balance of Ethiopia in the long run.
This result is also consistent with economic theory, more specifically, the elasticities approach to a trade balance adjustment, which states that, following currency depreciation, import prices would rise, leading to decreased demand and a decline in the volume of imports, ultimately leading to an improvement in the trade balance on the import front due to the price effect of imports.On the other hand, if Ethiopian birr appreciates, foreign goods (imports) become cheaper; hence the volume of imports is likely to rise, thus worsening the trade balance in the long run.
The real national income (LY) has also negative impact on the trade balance in line with economic theory and was found to be significant in determining the trade balance.Thus, the main determinants of the trade balance of Ethiopia in the long run are real effective exchange rate, domestic real income (economic growth), money supply, interest rate and the domestic and foreign general price level (TOT) according to this long run model.
The negative sign of the coefficient of real domestic income supports the Keynesian view or absorption approach that increases in domestic real income will encourage citizens to buy more imported goods and thus worsen the trade balance both in the short and long run.This means, higher domestic income will lead to higher demand for foreign goods and thereby deteriorate the trade balance of a country.
The coefficient of domestic money supply is also negative which confirms the monetary approaches to the trade balance of a country.The monetary approach states that a fall in domestic money supply will improve the trade balance, whereas the rise in domestic money supply will worsen the trade balance of a nation.Thus, the coefficient of money supply in our trade balance equation is in agreement with the monetary theory of the trade balance of a nation.
Moreover, the coefficient of real effective exchange rate is negative and significant implying that the sum of the price elasticity of export and import demand is greater than one or the Marshall-Lerner condition holds in the long run for the trade balance of Ethiopia.Moreover, in line with the elasticity approach, there is strong and significant relationship between exchange rate movement and the trade balance of Ethiopia.Thus, the relative relevance of the various theories of the trade balance in Ethiopian context is also tested and the result revealed that the elasticity approach, monetary approach and the absorption approach are significant in explaining the trade balance of Ethiopia.
In this long run model, all the variables are in log form and this implies that the coefficient of each variable in the model gives its elasticity with respect to the trade balance.Thus, the long run elasticity showed that devaluation improves the trade balance of Ethiopia in the long run.In totality, in the long run, the trade balance of Ethiopia is determined by domestic economic growth, exchange rate, money supply, interest rate and the relative price level.

Regression results of the short run model
After determining the long run relationship among the variables in the model and their long run coefficients, the next step in Johansen's co-integration analysis is to determine the coefficients of the short run dynamics.The coefficient of short-run dynamics can be obtained by regressing the first difference of the dependent variable on its lag, the lagged error correction term which is derived from the long run static model, the first difference of all exogenous variables and their lags.However, the determination of the coefficient of short run model uses the general to specific model selection technique to obtain vector error correction model (ECM).
So as to obtain the following parsimonious short run model, a testing down procedure was used in which nonsignificant and wrong signed lags are excluded from our model.The various diagnostic tests also showed that the model is correctly specified.Moreover, the coefficient of vector error correction term (ECM -1 ), which shows the rate at which the trade balance equation adjust to shocks in the system, has a negative sign as expected and statistically significant with 1% level of significance and thus, we can conclude that the model is correct.
Furthermore, the various results of the diagnostic test showed that our model is correctly specified.The coefficient of multiple determinations (R 2 ) shows that 67.7% of the variation in the trade balance is due to the explanatory variables included in the model (Table 8).The test summary result as shown above indicates that there is no problem of serial correlation as evidenced from the Durbin-Watson's statistics of 2.19.The test for the existence of heteroscedasticty in the short run model also indicated that there is no problem of heteroscedasticity since the Breusch Pagan test is statistically insignificant.This implies that, the variance of the residual from the short run model is independent of the explanatory variables in the model.However, the result of the test indicated that there is problem of residual non-normality.But, in Johansen's co-integration analysis, normality is not a problem.In addition, the coefficient of error correction term (ECM_ 1 ) has a negative sign as expected and statistically significant at 1% level of significance.So, the coefficient of adjustment is 0.5 which implies that if there is disequilibrium, the system adjusts towards the long run equilibrium by 50%  each year.That means it takes the system two years to reach its long run equilibrium.

Impulse response function
The impulse response functions (IRFs) are used to uncover the dynamic relationship between the dependent variable (TB) and the other independent variables thereby helps to trace the response of the dependent variable to a unit and one time shocks to other variable in the system.Thus, it traces the response of the endogenous variable to a change in any of the innovations in the model and it is a test for identification of the sign and duration of the effect of the shock on the endogenous variable.The impulse response functions plotted below shows a 10-years response of trade balance to an initial and one-time shock in each variable.
Figure 3 shows the response of the trade balance of Ethiopia to 1 standard deviation shock to real effective exchange rate.This implies that devaluation (an increase in REER 3 ) deteriorates the trade balance immediately following the devaluation of birr.But, just after two years, it starts to improve the trade balance of Ethiopia.This may be due to the fact that devaluation may not necessarily cut imports in developing countries as they are import dependent.That means, devaluation initially deteriorates the trade balance of Ethiopia and starts to improve the trade balance of a nation in the long run.
The impulse response function is, therefore used to check whether the J curve phenomenon holds or not in Ethiopia using time series data for the periods 1970-2014.Accordingly, Figure 3 revealed that the responses of the trade balance of Ethiopia follows the J-curve phenomenon and Birr devaluation deteriorates the trade balance of Ethiopia in the short run and improves the trade balance in the long run.
In similar vein, domestic money supply and government expenditure affect negatively and statistically significantly, the trade balance of Ethiopia in the short run.It could be seen that any shocks of the explanatory variable makes the impulse response dies out to zero.This implies that the error correction model (ECM) is stable.Moreover, the direction of the trade balance's response to innovation in money supply and domestic real income is theoretically acceptable in all the cases.Trade balance responded negatively to domestic income and money supply.
Figure 4 shows that, in line with economic theory, an increase in domestic money supply led to a deterioration of the trade balance of Ethiopia in the short run.The result is also consistent with economic theory, more specifically, the monetary approach to the balance of payments.
With regard to domestic real income, as economic theory predicts, the impulse response function also revealed that the domestic real income has negative effect on the trade balance of Ethiopia.The impulse response function showed a negative relationship between domestic real income (Y) and TB in the short run (Figure 5).This means that when the domestic real income increases, domestic consumers will increase their consumption of imported goods.Therefore, the descriptive data analysis, econometric analysis and impulse response function (IRF) revealed that Birr devaluation deteriorates the trade balance of Ethiopia in the short run.Traditional theory predicts that devaluation boosts export performance and curtails import and thereby lead to an improvement in the trade balance and output growth for developing countries.But, others argue that devaluation improves the trade balance and buoys output growth only if the country is initially export dependent.If the country is initially import dependent, devaluation worsens the trade balance and leads to lower output growth (Cooper, 1971).Despite the colossal effort that Ethiopian government has made to improve the external balance (trade balance) and output growth via devaluation, it failed to cut imports as can be seen in trend analysis.This is mainly due to the fact that most of the imports (72% including fuel in 2014) are related to domestic production and therefore, devaluation increases the cost of production.
Therefore, Ethiopia can influence or change its trade balance mainly by affecting its export performance since the imports are less responsive to birr devaluation.

CONCLUSION AND POLICY IMPLICATIONS
This study aimed at examining the effect of Birr devaluation on trade balance of Ethiopia for the period of 1970 to 2014 using the Johansen co-integration analysis.Regarding the time series characteristics of the data, all the variables were tested for unit roots using Augmented Ducky Fuller test and the test result revealed that all the variables are non-stationary at level and become stationary only after first difference.Moreover, since all the variables are non-stationary at level, the long run relationship (co-integration) among the variables was tested using the Johansen approach and the result revealed that there is a long run relationship (cointegration) among the variables.This means that, the analysis of long and short run dynamics is possible for the trade balance model.The trend analysis, coefficient of long and short run dynamics and the impulse response function (IRF) revealed that birr devaluation deteriorates the trade balance of Ethiopia in the short run and improves it in the long run.This means, devaluation of currency shows a Jcurve phenomenon for the trade balance of Ethiopia.In other words, devaluation of Ethiopian birr deteriorates the trade balance of Ethiopia in the short run and improves the trade balance in the long run which means devaluation shows a J-curve phenomenon for the trade balance of Ethiopia.This result is in agreement with the study conducted by Michael (1996) for Ethiopian economy, Frank (2004) for Ghana, Shahbaz and Abdul (2010) for Pakistan economy.
Moreover, the relative relevance of the various theories of trade balance in Ethiopian context was also tested and the result revealed that the elasticity approach, monetary approach and the absorption approach are significant in explaining the trade balance of Ethiopia.According to the study, money supply, real effective exchange rate, domestic real income and term of trade are the major determinants of the trade balance of Ethiopia both in the short run and long run.
Thus, devaluation may not be the optimal approach to improve export performance and curtail imports.This is because devaluation feeds into domestic inflation, as well as increase the cost of imported capital, intermediate and final goods.Policy should rather focus on factors that constrain the export growth.Thus, infrastructure availability and the relative productivity of export supply are the main determinants of exports.Over all, there are two alternative policies to devaluation, to boost export performance of Ethiopia and thereby lead to improvement in trade balance.

Export productivity and diversification
The first alternative policy for devaluation, to boost Ethiopian export performance is through the rise in export productivity and diversification.Ethiopian government may stimulate export diversification in the area of agriculture, agro-investment and agro-allied industries and mining industries which will improve the country's trade balance and foreign exchange earnings.Some of the potential areas for Ethiopian exports include coffee, hides and skins, oilseeds and pulses, horticulture and floriculture, textiles, sugar, fruits, skilled labor, and electricity.

Import computing industries
The second alternative policy to devaluation is to give due attention to the import computing industries like chemical industries (fertilizers), cement factories, steel factory and textile factories.This helps to improve Ethiopian trade balance by decreasing dependence on imported goods.
To sum up, policies that encourage productivity improvements and diversification of the export sectors and expansion of import computing industries like, metal industry, cement industry and chemical industry (fertilizers industry) are an alternative policies to devaluation which enhance export growth and curtails import growth and thereby leads to improvements in the country's trade balance, output growth and a downward pressure on domestic price level (inflation).Therefore, Ethiopian government may need to be conservative in using exchange rate policy to improve external balance.
Thus, it seems that devaluation gives with one hand (export) and takes with the other hand (imports), and the effect of devaluation on trade balance, therefore, depends on the net effect.This means that devaluation improves the trade balance and leads to output growth only if the country is initially export dependent.But, if the country is initially import dependent like Ethiopia, devaluation may worsen the situation.Devaluation would have positive impact on the Ethiopian Economy (via the export sector), once the country establishes export oriented firms and international markets.Thus, the country on the process of industrialization first needs to import capital goods which cannot be favored by devaluation.Then, once the production gets its way, devaluation would make a sense.

)
This equation states that domestic supply (Y) in an open economy is given by domestic absorption plus net exports.That means in an open economy, domestic output level equals domestic spending plus trade balance.By substituting A for domestic spending (C, I and G) and (X-M) by TB, we can drive a trade balance equation for an open economy as follow:

Figure 1 .
Figure 1.Trend of the trade balance of Ethiopia for the period 1970 to 2014.Source: Computed from Ethiopian Custom Authority Data (2016).

Figure 2 .
Figure 2. Trends of the log of export and import volumes of Ethiopia in millions of birr.Source: Own Computation (2016).

Figure 3 .
Figure 3.The response of DL (TB) to Cholesky one standard error DL (REER) innovation.

Figure 4 .
Figure 4. Response of DL (TB) to Cholesky one standard error DL (MS) innovation.

Figure 5 .
Figure 5. Response of DL (TB) to Cholesky one standard error DL(Y) innovation.

Table 1 .
The result of unit root tests of variables at level using ADF test.

Table 2 .
The result of unit root tests of variables at first difference using ADF test.

Table 3 .
Johansen's cointegration test for the trade balance equation.

Table 4 .
Share of the values of imports of Ethiopia for the periods 1970-2014.
Source: Own computation from Ethiopian Custom Authority Data (2016).

Table 5 .
Share of the volume of the major exports of Ethiopia for the period 1985-2014.

Table 6 .
Estimation results of the static long run model of the trade balance equation.

Table 6
, the maximum Eigenvalue test indicates that there is one co-integrating equation in the model and so, there is one dependent variable (TB) and money supply, real effective exchange rate, real GDP,

Table 7 .
Regression results of the short run model or error correction model.

Table 8 .
Description of variables in the model.