The Effect of Changes in the Real Exchange
Rate on Income and Bilateral Trade Balance
The Case of Ethiopia and the United States
University of Amsterdam, Faculty of Economics and Business Bachelor Thesis
12 Credits June 2015
Supervised by Maximilian Hoyer
Written by Christina Gies
1
Statement of Originality
This document is written by Student Christina Gies who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2
Abstract
Economic theory often infers that a currency devaluation improves a country’s trade balance through a gain in international competitiveness of export goods. This paper aims to investigate whether a devaluation of the Ethiopian Birr leads to an improvement of the bilateral trade balance of Ethiopia with the United States and whether the abandonment of the Ethiopian Birr peg to the U.S. Dollar in 1992 led to an increase in Ethiopia’s income. In order to identify these relationships two ARDL error correction models are specified. The results indicate that the abandonment of the peg led to an increase in national income and that a devaluation of the Birr is associated with an improvement in income and the bilateral trade balance in the short run from 1981 to 2008 and from 1993 to 2010 respectively. No long run relationship between the real effective exchange rate and the total trade balance or income could be identified, which is why this paper concludes that a devaluation to improve the total trade balance is not a sound strategy for Ethiopia. However, more research needs to be conducted on this issue.
Table of Contents
1. Introduction ... 3
2. Background on Ethiopia... 5
2.1 Economic Background and Bilateral Trade ... 5
2.2 Exchange Rate System and Development ... 6
3. Theoretical Framework ... 8
3.1 The Exchange Rate and the Trade Balance ... 8
3.1.1 Elasticity Approach ... 8
3.1.2 Absorption Approach ... 9
3.2 Discussion and Empirical Evidence ... 11
4. Methodology ... 13
5. Data ... 15
6. Analysis and Results ... 17
7. Conclusion ... 22
Bibliography ... 23
3 0 1 2 3 4 5 6 7 8 9 10 1 9 9 3 q 3 1 9 9 4 q 2 1 9 9 5 q 1 1 9 9 5 q 4 1 9 9 6 q 3 1 9 9 7 q 2 1 9 9 8 q 1 1 9 9 8 q 4 1 9 9 9 q 3 2 0 0 0 q 2 2 0 0 1 q 1 2 0 0 1 q 4 2 0 0 2 q 3 2 0 0 3 q 2 2 0 0 4 q 1 2 0 0 4 q 4 2 0 0 5 q 3 2 0 0 6 q 2 2 0 0 7 q 1 2 0 0 7 q 4 2 0 0 8 q 3 2 0 0 9 q 2 2 0 1 0 q 1 2 0 1 0 q 4 2 0 1 1 q 3 2 0 1 2 q 2 2 0 1 3 q 1 2 0 1 3 q 4 E T B /U SD Date RER 1. Introduction
As one of the most populous countries on the African continent, Ethiopia has exhibited an above average economic growth of about 10 percent among the Sub-Saharan African countries within the last decade and is thus a major potential market in Africa. However, despite these developments Ethiopia is still one of the poorest countries worldwide, with a per capita gross domestic product of about 500 U.S. Dollars in 2013 (The Worldbank). Ethiopia is largely dependent on foreign aid and imports of food and manufactured goods, which has been leading to large and persistent trade deficits. Compared to other countries Ethiopia exhibits very low total foreign exchange, which is a possible explanation for the country’s economic struggles (Harisson, 1996).
In July 2014, The World Bank released its third economic report on Ethiopia and stated that the Ethiopian currency, the Ethiopian Birr, is overvalued and that the country could boost its export earnings by devaluing it (World Bank Group, 2014). Since 2010, when the Birr was last devalued, it gained in value by more than 50 percent in real terms against the U.S. Dollar, leading to an overvaluation of 17 percent, according to Lars Moller, the World Bank group’s lead economist and program leader for Ethiopia. Moller also states that a devaluation of 10 percent in real terms may increase export earnings by up to 5 percent per year and income growth by up to 2 percent (Davidson, 2014).
Figure 1 - Real Exchange Rate (ETB/USD)
Source: Own calculation Note: q1 refers to the first three months of the year based on the Gregorian calendar
Since its official adoption in 1945, the Ethiopian Birr was pegged to the U.S. Dollar at a rate of 2.50 Birr per Dollar. However, due to the macroeconomic policies of the Ethiopian Derg regime, which
4
reigned from 1974 until 1991, and the unwillingness of the regime to devalue the currency, the Birr was largely overvalued in this period (Kibret, 1994). In September 1992 the Ethiopian People’s Revolutionary Democratic Front (EPRDF), which constituted the new regime beginning from 1991, decided to adopt a managed floating exchange rate policy and devalued the Birr to 5 Birr per Dollar in September, with the aim to reduce the currency’s overvaluation and to improve Ethiopia’s balance of payments, specifically its trade balance (Taye, 1999), which is defined as the value of net exports throughout this analysis. Since then, the Birr depreciated against the U.S. Dollar in real terms until the early 2000s when it started to appreciate again. In 2010 the Birr was devalued once more, but then again appreciated in real terms afterwards. This development is visualized in Figure 1.
The debate about the relationship between exchange rates and the trade balance is very controversial and still ongoing. The same holds for the debate about the importance of a country’s exchange rate system for accomplishing economic prosperity. According to Ghura and Grennes, the exchange rate policy is one of the most important factors in determining economic growth and performance in a country (1993). The aim of this paper therefore is to determine whether there exists a long run relationship between changes in the real exchange rate and the bilateral trade balance between Ethiopia and the U.S., as well as whether there exists a long run relationship between the real exchange rate and the Ethiopian national income. The analysis further aims provide insights to whether the government’s goal to improve the balance of payments was actually achieved through a devaluation of the Ethiopian Birr in 1992.
This paper is structured as follows. In section two some insights into the history and important factors about the Ethiopian economy will be explained, in order to get a better understanding of the underlying backgrounds. In section three a theoretical framework is presented, where the relevant underlying economic theory is outlined and empirical evidence as well as the current discussion on the issue is presented. In order to derive a conclusion with respect to the central research question, the following methodology will be used in section four: A bounds testing approach to cointegration will be conducted, using an autoregressive distributed lag error correction model to investigate whether a long run relationship exists between changes in the real exchange rate and the bilateral trade balance of Ethiopia with the U.S. from 1993 to 2011. Furthermore the relationship between the real exchange rate and national income, taking into consideration the country’s monetary and fiscal policy, from 1981 to 2008 is analyzed also using an error correction model. An Engle Granger cointegration test will be conducted in order to test for the existence of a long run relationship between the real effective exchange rate and the total trade balance of Ethiopia, with the aim to provide some insights into the multilateral situation and thus to facilitate the provision of some policy advice. Finally, this paper will end with a conclusion, where the main findings are summarized, the limitations of this paper are mentioned and ideas for further research are addressed.
5
2. Background on Ethiopia
2.1 Economic Background and Bilateral Trade
An important aspect to consider when analyzing the Ethiopian economy is that Ethiopia is one of the main recipients of foreign aid worldwide. According to the theory of the so called Dutch Disease, large increases in income can cause the country’s currency to appreciate and thus lead to a loss of competitiveness of the economic sectors. In his research, Martins (2010) analyzed the effect of capital inflows on the competiveness of Ethiopia’s goods. He found that foreign aid has no statistically significant effect. He concludes, that Ethiopia has been able to effectively manage the large capital inflows through a prudent fiscal policy and is not suffering from the symptoms of the Dutch Disease, and hence successfully avoids periods of macroeconomic instability following the receipt of aid flows. Therefore the inflow of foreign aid is assumed to have no influence on the competitiveness of Ethiopian goods throughout this analysis.
Furthermore the Ethiopian economy is mainly based on agricultural products, amounting to about 50 percent of GDP in 2007. Exports amounted to about 2.4 percent of total GDP in 2007 (IMF, 2008) and constituted predominantly primary goods. The main export goods of Ethiopia are coffee, accounting for about 28 percent of total foreign exchange revenue in 2011, closely followed by sesame. Primary goods are usually subject to large international price fluctuations, which means that the revenues from exports in Ethiopia are very volatile. Food imports constituted 11 percent of total imports in 2010 (Trading Economics, 2015), which implies that a persistent currency devaluation may constitute a problem regarding Ethiopia’s large dependency on food imports. About 5 percent of total imports and exports are to and from the U.S., and include mainly agricultural products (African Growth and Opportunity Act, 2015). A graph of the bilateral trade balance is presented in Figure 2 in the Appendix.
As mentioned earlier, under the Derg regime which reigned in Ethiopia from 1975 to 1987, the country suffered from political instabilities which also affected the economy. Ethiopia experienced very high inflation during the end of the regime and additionally suffered from a severe famine in 1984. In 1991 however, the ERDF, which later became the Transitional Government of Ethiopia, took over the power, which provided an opportunity for the economy to recover. During the 1990s several reforms to liberalize trade flows, agricultural markets and foreign exchange markets were undertaken, which took the form of a structural adjustment program. A tight fiscal and monetary policy was adopted and ceilings on the interest rate were abandoned (Degefe, 1994). Inflation averaged 3.8 percent in the years following the transition to the new regime (Dercon, 2000), however the above mentioned reforms led to downward pressures on the real exchange rate, which was finally devalued in the end of 1992 (Martins, 2010). After the devaluation total imports increased from 15 to 30 percent of GDP in the period from 1993 to 2011, whereas exports increased from around 6 percent in
6
1993 to 16 percent in 2011. However, according to Dercon (2000), much of the economic growth during the 1990s has to be viewed as a recovery from the crisis following the poor policy environment of the 1980s.
From 1998 to 2003 Ethiopia was in a war with its neighbor country Eritrea, which led to very high government spending on defense expenditures. The war led to the abandonment of most aid flows into Ethiopia (Dercon, 2000), which further worsened the fiscal deficit. During the war period the Ethiopian economy experienced a negative growth rate, however after 2003 the growth rate has been above 10 percent per year.
In 2010 a five year growth and transformation plan has been implemented which had the aim to improve the economy in Ethiopia and lead to an annual economic growth of 11 to 15 percent. This was supposed to be achieved by enhancing the productivity of agriculture, by strengthening marketing systems and by improving the engagement of the private sector (IFPRI, 2010). Foreign investments should be encouraged, the country’s infrastructure should be improved and Ethiopia should aim to become a member of the World Trade Organization.
2.2 Exchange Rate System and Development
Since the proclamation of the Ethiopian Birr in 1945, it has been valued at 2.48 Birr per U.S. Dollar. It remained relatively stable until its first depreciation in January 1964, when it was slightly devalued to 2.50 Birr per Dollar. This exchange rate persisted until the collapse of the Bretton Woods system, which led to the following revaluations of 2.30 Birr per Dollar in 1971 and to 2.07 Birr per Dollar in 1973 (Martins, 2010).
From 1974 until 1987 Ethiopia was ruled by the Marxist Derg Regime. Due to their expansionary macroeconomic policies and as can be seen from the significant black market premium for foreign exchange as well as a higher real than nominal exchange rate during this time, the Birr has been largely overvalued, which depressed economic growth and led to misallocation of resources and the loss of international competitiveness (Taye, 1999; Kibret, 1994). In 1974 the price level in Ethiopia increased, which led to the emergence of an alternative unofficial foreign currency market. According to Kibret (1994) the emergence of the parallel market for exchange rates is an important indicator for the misalignment of the exchange rate. He argues that the value of the exchange rate which is determined on the parallel market is a good indicator for the actual value of the exchange rate, allowing for risk premium. Since 1976 foreign exchange is controlled and rationed in Ethiopia, which led to the parallel market exchange rate assigning only half the value to the currency compared to the official rate (Kibret, 1994). Between 1973 and 1992, when the currency was devalued, the official nominal rate was 2.07 Birr per Dollar, whereas the unofficial exchange rate had an average value of 6 Birr per Dollar, which clearly demonstrates the overvaluation of the Birr.
7
At the end of the Derg regime there was high inflation, which led to a deterioration of export competitiveness. Moreover, there were low foreign exchange reserves and the country was running fiscal and trade deficits, which put pressure on the currency to devalue. Because of these circumstances the monetary authorities in Ethiopia decided to abandon the peg to the Dollar at the end of 1992 and to devalue its currency in order for the country’s goods to gain more international competitiveness, to improve the balance of payments and to boost the economy (Martins, 2010). The Birr was devalued to 5 Birr per Dollar. In 1993 a foreign exchange “Dutch Auction” system was introduced, which has been mainly financed through export earnings and foreign loans and grants (Geda, 2006). This auction based exchange rate system initially worked next to the official exchange rate, however in 1995 the official exchange rate and the auction based exchange rate were unified and in 2001 a foreign exchange interbank market has been established (Geda, 2006). Martins (2010) finds in his study that the parallel premium of the exchange rate is only significant until 1997, after which both rates are almost the same.
Table 1 – Chronology of Main Events and Exchange Rates (ETB/USD)
Period Event Official Parallel Premium (%)
1945:07 Currency Proclamation 2.48 n/a n/a
1964:01 Devaluation 2.50 n/a n/a
1971:12 Revaluation (Collapse of Gold Standard) 2.30 2.44 6.1
1973:02 Revaluation 2.07 2.16 4.3
1992:10 Devaluation 5.00 12.75 155.0
1993:05 Introduction of an Auction System (fortnight) 5.00 13.30 166.0
1995:07 Unification of the Official and Auction Rates 6.25 10.55 68.8
2001:10 Introduction of an Interbank Market (daily) 8.53 8.70 2.0
Source: P.M. Martins, 2010, Do capital inflows hinder competitiveness? The real exchange rate in Ethiopia (No. 10/07). CREDIT Research
Paper.
These developments are summarized in Table 1 above. Currently the exchange rate system of Ethiopia is best described as a managed float to the U.S. Dollar (Martins, 2010). Since the abandonment of the peg, the Birr has remained relatively stable in real terms compared to other Sub-Saharan African countries’ currencies, due to conservative monetary policy and large amounts of foreign exchange reserves.
8
3. Theoretical Framework
3.1 The Exchange Rate and the Trade Balance 3.1.1 Elasticity Approach
There are several factors and theories to take into account when examining the effects of exchange rate changes on the trade balance. The most commonly used theory in trying to explain how changes in the exchange rate can influence the trade balance is the Elasticity Approach, which states that import and export demand and supply of a country are closely related to the valuation of the country’s currency (Pilbeam, 2013). If the currency is relatively cheap, imports become relatively more expensive and exports become more competitive compared to other country’s goods. Based on this, the Elasticity Approach identifies to two main effects of a change in the exchange rate, namely a price and a volume effect. The price effect refers to the change in relative prices for foreign and domestic goods. The elasticity approach assumes constant prices in producer’s currency, which means that the supply curves of domestic goods are horizontal and a change in demand has no effect on the price. Thus the price effect of exports is equal to zero. The price effect of imports equals one, which indicates that the price of imports changes proportionally to changes in the exchange rate since
(1) PM=ER*PX*
where PM denotes the price of imports, ER denotes the nominal exchange rate and PX denotes the price of exports. The size of the volume effects depend on the import and export demand elasticity which are denoted by ηx and ηm and are specific to each country. The total effect of a currency depreciation
according to the elasticity approach is therefore equal to
(2) ηx + ηm – 1
which can be transformed in to Marshall-Lerner condition which states that if a country’s trade is initially in balance, then a depreciation of the currency will lead to an improvement in the trade balance if
(3) ηx + ηm >1
The elasticity of import and export demand are commonly estimated to be rather small in the short run due to fixed contracts and slow habit reformation, which is why in the short run the Marshall-Lerner condition is usually not fulfilled (Pilbeam, 2013). However in the long-run the elasticity is estimated to be larger, which is why it is often assumed that in the long run a depreciation of currency in fact
9
leads to an improvement in the trade balance of a country. This relation is commonly referred to as the J-Curve effect (Pilbeam, 2013).
This model is however highly simplistic and makes assumptions that are not certain to be met in reality, especially the ceteris paribus assumption is often criticized (Pilbeam, 2013). The following approach aims to incorporate changes in the trade balance and income into a more comprehensive framework.
3.1.2 Absorption Approach
The idea of the Absorption Approach is that changes in the trade balance affect income, which in turn affects the trade balance and thus this approach tries to incorporate the effects of changes in the exchange rate as well as changes in income in trying to explain the total effects of a devaluation of the currency on the trade balance. The basis identity for the Absorption Approach, put forward by Alexander (1952), can easily be derived from the national income identity:
(4) Y = C + I + G + X + M
where Y denotes national income, C denotes consumption, I denotes investment, G denotes government spending, X denotes exports and M denotes imports. The following basic identity states that the current account imbalance equals the difference between domestic output and domestic spending, with the latter representing absorption, A, which is composed of consumption, government spending and investment (Pilbeam, 2013).
(5) CA = X – M = Y – (C + G + I) = Y - A
CA denotes the current account and is equal to exports minus imports. In this paper measures of the current account and the trade balance are used interchangeably, not accounting for net factor income. Transforming this equation into difference form leads to the following equation:
(6) dCA = dY – dA
which implies that the effect of a devaluation on the current account depends on how income is affected relative to absorption (Pilbeam, 2013). The effects of a devaluation on absorption can be divided into two parts: a direct and an indirect effect. The indirect effect of a devaluation on absorption is through a rise in income. Higher income leads to more absorption, how much more is determined by the marginal propensity to absorb. The direct effect incorporates all the other effects of a devaluation on absorption. The total effect on absorption is thus equal to:
10 (7) dA = adY + dAd
where a denotes the marginal propensity to absorb. From the last two equations, it can be derived that for a devaluation to improve the trade balance it is necessary that the following inequality holds:
(8) (1-a)dY > dAd
which states that any change in income, which is not spend on absorption must exceed any change in direct absorption. To investigate whether the above mentioned condition is likely to be satisfied it is necessary to understand what effect a devaluation has on income and direct absorption.
There are two main effects of a devaluation on national income: The employment effect and the terms of trade effect. The employment effect can be explained as follows: If there is unemployment in a country, this country’s income has the potential to rise. If the Marshall-Lerner condition is fulfilled, a devaluation will lead to an improvement of the trade balance and thus to an increase in national income through the foreign trade multiplier (Pilbeam, 2013). However, in case the Marshall Lerner condition is not satisfied, a devaluation would worsen the trade balance, which has adverse effects on national income. Subsequently we can conclude that it is not clear whether the employment effect has positive or negative effects on national income.
A devaluation of currency however lowers the terms of trade of a country, which is defined as the ratio of the price of exports to the price of imports, and thus will lower national income. This is called the terms of trade effect. The total effect of a devaluation on national income is thus ambiguous in case the Marshall Lerner condition is fulfilled, since both effects work in different directions. The exact effect of a rise in income on the trade balance depends on the marginal propensity to absorb. In case the marginal propensity to absorb is less than one, a rise in income will improve the current account due to a rise in the income to absorption ratio. If the marginal propensity to absorb is larger than one, an increase in income would worsen the trade balance.
In case of full employment, for a devaluation to improve the current account, direct absorption needs to be reduced. The effects of a devaluation on direct absorption include the real balance effect, the income redistribution effect, the money illusion effect, the expectational effect and the Laursen-Metzler effect, which are explained in detail in Pilbeam (2013). Pilbeam concludes that the effects of a devaluation on direct absorption are ambiguous.
The most important insight provided by both the elasticity and the absorption approach is that a devaluation is more likely to succeed in case the Marshall-Lerner condition is fulfilled and a devaluation is accompanied by monetary and fiscal restraints in order to increase income relative to domestic absorption.
11
3.2 Discussion and Empirical Evidence
As outlined above, traditional theories suggest that the trade balance improves due to higher export demand and lower import demand following a currency devaluation, in case the Marshall Lerner condition is fulfilled. The existing discussion and empirical evidence on the relationship between exchange rates and the trade balance and income is however inconclusive and there have been different findings on the issue.
According to Bahmani-Oskoee and Gelan (2013), a devaluation improves the country’s goods competitiveness, but at the same time it also creates contractionary problems in the economy, leading to reduced output, which might even outweigh the benefits of an improved competitiveness. As costs that are associated with a currency devaluation, they for example mention a decrease in income in case of foreign currency denominated debt burdens. In this case a devaluation carries the risk of increased unemployment and bank bankruptcies, when facing the increased debt burden. Another aspect is that since a devaluation will make export goods more competitive on the international market, a country’s production will concentrate on the traded goods sectors, compared to the non-traded goods sectors, which may constitute a problem in case that imports constitute a large fraction of a country’s factors of production, because then production will become more expensive as well. Higher cost of production may in turn lead to decreased aggregate demand or decreased real wages, which will depress the economy (Bahmani-Oskoee & Gelan, 2013).
Supporting this view, Taye (1999) stresses the importance of the use of an appropriate macroeconomic policy in line with a depreciation, in order for the depreciation to actually lead to an increase in national income instead of the increase being cancelled out by inflation. Specifically, he argues that restrictive monetary and fiscal policies are needed in order to neutralize some negative effects such as reduced output and employment.
There have been many researches in this field and different findings will now be discussed. Several researches find a positive relationship between a devaluation of the currency and the trade balance. For example Bahmani-Oskooe (1998) estimates the long-run trade elasticity in several LDCs from 1973 to 1980, using a cointegration technique, and finds that the Marshall Lerner condition is satisfied in the long run for the majority of the countries. He concludes that a devaluation of the currency for these countries will lead to an improvement of their trade balance. This finding is supported by a study conducted by Bahmani-Oskooee (1991), who uses an Engel Granger cointegration method to show that a long-run relationship between the trade balance and the real effective exchange rate of eight LDCs exists, which shows that a currency devaluation leads to an improvement of the trade balance in the long run in these countries.
Gebeyehu (2014) analyzed the impact of a change in the exchange rate on the trade balance in Ethiopia from 1974 to 2010. He finds that the Marshall-Lerner condition is satisfied in the long run and there in fact exists a J-curve relationship. He concludes that a real devaluation of the Birr is an
12
effective tool to correct trade balance problems the country is facing (Gebeyehu, 2014). This finding is supported by Kibret (1994).
More studies that find an improvement in the trade balance after a currency devaluation in different countries are Rawlins and Praveen (1993), Shirvani and Wilbratte (1997) and Arize (1994).
Other studies suggest however, that a devaluation of the currency worsens the trade balance. For example Shahbaz et al. (2012), who analyze the effects of changes in the real exchange rate on the trade balance in Pakistan from 1980 to 2006, find that a currency devaluation leads to a deterioration of the trade balance. The same is found by Ogundipe et al. (2013) who analyze this relationship for Nigeria. Rose (1990) could not find a significant positive or negative impact of a currency devaluation on the trade balance in several developing countries.
There has also been some literature that focused on the relationship between changes in the exchange rate and national income. A study is conducted about the impacts of a devaluation on macroeconomic performance in the case of Ethiopia by Taye (1999). Using a simulation approach he estimates that a depreciation of the Birr would lead to a substitution away from foreign to domestic goods and thus to a decrease in import demand, due to the increased competitiveness of Ethiopian products, which means that the current account balance is indeed improved. However, he finds that output growth and employment are decreased in the new equilibrium, which could be an indication for the increase in income being due to expenditure reducing effects (decrease in import demand) rather than expenditure switching (from foreign to domestic) effects. As Taye (1999) argues, this may be of great concern to the country in the long run since its large dependency on food imports. The obtained results from the analysis are supported by the findings of Kibret (1994).
Gylfason and Schmid (1983) conducted a study on the effects of devaluation on national income for several developed and developing countries. They find no clear evidence to support or abandon their central research question all together, however they argue that for several countries some positive effects of a currency devaluation on income have been found. It is concluded that a currency devaluation to improve the country’s balance of payments may be a sound strategy to increase national income (1983).
To support this finding, a real depreciation of the currency is found to have expansionary effects on the Ethiopian economy by Bahmani-Oskooee and Gelan (2013), who analyzed the effect of a currency devaluation for 22 African countries from the period 1971 to 2009. To analyze this effect, an ARDL error correction model is used. They find that the increase in demand due to increased competitiveness of exports exceeds the decline in supply and thus conclude that a currency devaluation of the Birr has positive effects on the Ethiopian economy.
Based on this discussion and the different empirical findings it becomes clear that no consensus on the effect of a currency depreciation on national income and the trade balance can be found, however the existing empirical evidence of this effect for Ethiopia seems to suggest that a currency devaluation indeed leads to an improvement of the current account and that it should be
13
accompanied by tight fiscal and monetary policy in order to sustain the positive effects on national income. In the next section the methodology used to analyze the relationship between the real exchange rate and the trade balance and national income respectively is presented.
4. Methodology
In order to test whether the adoption of a managed floating exchange rate regime led to an improvement of national income as was hoped for by the Ethiopian government, and whether there exists a long run relationship between the real exchange rate and the trade balance, two Autoregressive Distributed Lag (ARDL) bounds testing models will be used. The model to identify the relationship between the real exchange rate and the trade balance is based on the methodology of Bahmani-Oskooee et al. (2013) and Yazici and Islam (2014), who outline that this method has become the standard for similar time series studies, since it provides short-term as well as long-term estimates of the coefficients, the results are robust to small samples and stationary as well as non-stationary variables can be incorporated (Bahmani-Oskooee et al., 2013).
Quarterly data from 1993 to 2010 will be used to estimate the Ethiopian trade balance as a function of Ethiopia’s and the U.S.’s income and the real exchange rate. The log-linear model is specified as follows:
(9) 𝑙𝑛 (𝑀𝑋𝑡
𝑡) = 𝛼 + 𝛽𝑙𝑛𝑌𝐸𝑇𝐻,𝑡+ 𝛾𝑙𝑛𝑌𝑈𝑆,𝑡+ 𝛿𝑙𝑛𝑅𝐸𝑅𝑡+ 𝜀𝑡
Where the trade balance is specified as the ratio of Ethiopian exports to the U.S. to Ethiopian imports from the U.S., because this allows the model to be in a log-linear form and it will lead to insensitivity of units of measurement. YETH,t represents the national income of Ethiopia at time t, whereas YUS,t
measures the income of the United States at time t. RER specifies the real exchange rate.
An increase in Ethiopia’s income is expected to decrease the trade balance due to increased import demand and thus the coefficient of the Ethiopian income is expected to be negative. Increases in the real exchange rate, which implies a currency depreciation of the Birr, as well as an increase in the U.S. income are expected to lead to an improvement of the trade balance through increased exports.
The above described model however specifies only the long-run relationship between the variables. Since the mentioned J-Curve effect is a phenomenon that includes the short-run effects, these need to be incorporated in the model. Therefore an error-correction specification is estimated:
14 (10) ∆ ln (𝑀𝑋𝑡 𝑡) = α + ∑ 𝛽𝑗 𝑛1 𝑗=1 ∆ ln ( 𝑋𝑡−𝑗 𝑀𝑡−𝑗) + ∑ 𝛾𝑗∆ln𝑌𝐸𝑇𝐻,𝑡−𝑗+ 𝑛2 𝑗=0 ∑ 𝛿𝑗∆𝑙𝑛𝑌𝑈𝑆,𝑡−𝑗+ ∑ 𝜆𝑗∆𝑙𝑛𝑅𝐸𝑅𝑡−𝑗+ 𝜃1ln (𝑀𝑋𝑡−4 𝑡−4) + 𝜃2𝑙𝑛𝑌𝐸𝑇𝐻,𝑡−4+ 𝑛4 𝑗=0 𝑛3 𝑗=0 𝜃3𝑙𝑛𝑌𝑈𝑆,𝑡−4+ 𝜃4𝑙𝑛𝑅𝐸𝑅𝑡−4+ 𝜀𝑡
Equation (10) will be estimated using Ordinary Least Squares. The short-run effects will be obtained by the estimates of the differenced variables, whereas the long-run estimates are inferred by the estimates of θ1-θ4 (Bahmani-Oskooee et al., 2013).
In order to test whether the devaluation had an effect on total output, a second model is developed, which incorporates the level of government spending (G) and the country’s money supply (M) as proxies for monetary and fiscal policy, because of their crucial role in sustaining the real devaluation (Ayen, 2014). For this approach yearly data is used from 1981 to 2008. A dummy variable will be included in the model in order to account for the period in which the exchange rate was pegged to the U.S. Dollar (Yazici & Islam, 2014; Bahmani-Oskooee et al., 2013). The dummy will be 1 in the periods before 1992 and 0 afterwards. Following Ayen (2014) and Bahmani-Oskooee and Gelan (2013) the error-correction model is specified as follows:
(11) ∆𝑙𝑛𝑌𝐸𝑇𝐻,𝑡 = 𝛼0+ 𝛼𝑃𝐸𝐺𝑡 + ∑𝑛1𝑗=1𝛽𝑗∆𝑙𝑛𝑌𝐸𝑇𝐻,𝑡−𝑗+ ∑𝑛2𝑗=0𝛾𝑗∆𝑙𝑛𝑀𝑡−𝑗+ ∑𝑛3 𝛿𝑗∆𝑙𝑛𝐺𝑡−𝑗
𝑗=0 + ∑𝑛4𝑗=0𝜆𝑗∆𝑙𝑛𝑅𝐸𝑅𝑡−𝑗+ 𝜃1𝑙𝑛𝑌𝐸𝑇𝐻,𝑡−1+ 𝜃2𝑙𝑛𝑀𝑡−1+ 𝜃3𝑙𝑛𝐺𝑡−1+ 𝜃4𝑙𝑛𝑅𝐸𝑅𝑡−1+ 𝜀𝑡
Higher money supply is expected to have a positive effect on income in the short run, however according to the quantity theory of money, money has no real effects on income in the long run and thus the coefficient measuring the long run relationship is expected to be zero whereas the coefficient of the differenced variable is expected to be positive. Government spending will increase income and so the coefficient of the variable denoting government spending is expected to be positive. As discussed above, the effect of changes of the exchange rate on total income are expected to be ambiguous.
The model selection criterion used in this paper is chosen following the approach in Yazici and Islam (2014), who state that using the Akaike Information Criterion (AIC), which is employed by Bahmani-Oskooee et al. (2013), may lead to unreliable results, since cointegration and diagnostic tests to the selected model are applied only after the optimum model is selected, and thus the model is used without regard to whether or not the necessary conditions are satisfied (Yazici & Islam, 2014). Using their approach, first cointegration and diagnostic tests will be applied to the available models with different lag combinations, given a maximum lag length. A model selection criterion will then be applied to the subset of models which satisfy the condition of cointegration and the diagnostics.
15
Using the ARDL bounds testing approach, cointegration will be tested with the use of an F-test statistic. There exists no cointegration in the case that θ1- θ4 are jointly equal to zero, whereas if
one of those values is significantly different from zero, cointegration can be established. Thus the following hypothesis will be tested against each other
(12) H0 = θ1 = θ2 = θ3 = θ4 = 0
(13) H1= θ1 θ2 θ3 θ4 0 (at least one is not equal to zero)
As Yazici and Islam (2014) outline, the distribution of the F-statistic is non-standard under the assumption of the null hypothesis, which is why the critical values for small samples, provided by Narayan (2005) are used. In case that the estimated F-statistic exceeds the critical values, the null hypothesis can be rejected and cointegration can be concluded.
With all the lag combinations that satisfy the OLS assumptions and the condition of cointegration, the optimal lag length for the parameters n1 to n4 is determined, using the Akaike
Information Criterion (AIC). Finally, the error-correction models are estimated using OLS.
In addition to this models, following Bahmani-Oskooee (1991) an Engle Granger cointegration will be conducted to see whether there exists a long run relationship between the real effective exchange rate and the total trade balance of Ethiopia. This will add to this analysis by allowing for a comparison of the specific case studied in this paper with the multilateral situation. The first step is to test whether the variables are integrated of the same order. To test this, the Augmented Dicky Fuller test will be applied. In the next step an OLS Regression of both variables is conducted and the residuals are estimated. If the obtained residuals have a unit root, the two variables are said to be cointegrated and thus exhibit a long run relationship.
Based on previous empirical findings and the Ethiopian background, it is expected to find an improvement of the trade balance following a currency devaluation. Furthermore the effect of a currency devaluation on income is expected to be positive due to the tight fiscal and monetary policy which was implemented in Ethiopia after the abandonment of the peg.
5. Data
In this paper quarterly data from the period 1993-3 to 2011-4, as well as annual data from 1981 to 2008 are used. In 1992 Ethiopia adopted a managed floating exchange rate system and since the objective of this paper is to investigate whether this adoption led to an improvement of the trade balance and how changes in the exchange rate affect the trade balance and income, these periods seem appropriate for the analysis. Unfortunately no data could be found on the bilateral trade between Ethiopia and the United States in the period before 1993, which is the main reason for including the second model which is described by (11) and which is supposed to provide insights into the effect of
16
the abandonment of the peg in 1992. Due to the unavailability of quarterly data during this period, unfortunately the analysis on the relationship between the abandonment of the currency peg and national income is subject to inaccuracy due to the fact that the peg was abandoned in September instead of at the beginning of the year. To improve the analysis, both cases (using 1992 and 1993 as the year of the abandonment of the peg) were analyzed.
All data is seasonally adjusted and indexed using the year 2000 as the base year. The following annual data has been used for the analysis: The GDP, the nominal exchange rate and government spending of Ethiopia, which are provided by the The Global Economy database (The Global Economy, 2015a; 2015b; 2015c) on annual basis. Government spending is reported in billion Dollars in nominal terms. This is transformed into real balances using the CPI of Ethiopia, which is retrieved from the Economic Research database (Economic Research, 2013). Data for the Ethiopian money supply, defined as money and quasi money, is retrieved from the IMF (IMF, 2015), where it is reported in nominal terms and in current Birr. This is transformed in real balances denominated in U.S. Dollars.
The following quarterly data have been used: For the U.S., data for the nominal GDP is retrieved from the Bureau of Economic Analysis (Bureau of Economic Analysis, 2015). Since for the GDP of Ethiopia only annual data is provided by the The Global Economy database, a linear match interpolation method is used to estimate quarterly data. The quarterly nominal exchange rate (E) is retrieved from the Oanda database (Oanda, 2015). The real exchange rate (RER) is then calculated as follows:
(14) RER=E*(P*/P)
where P* and P denote the countries price levels respectively, specifically P* denotes the price level for the U.S.. Price levels for both countries are based on the Consumer Price Index (CPI) and are retrieved from the Economic Research database on quarterly basis for the U.S. (Economic Research, 2015) and on an annual basis for Ethiopia (Economic Research, 2013). Using a linear interpolation method, quarterly data for the Ethiopian CPI is estimated. Data on bilateral imports and exports are provided by the U.S. Census Bureau on a monthly basis. The data is provided at a nominal and non-seasonally adjusted basis and has been converted into non-seasonally adjusted real quarterly data. This was done using the sum of the monthly observations to obtain quarterly estimates. The Ethiopian CPI was used to obtain real rather than nominal observations and a seasonal factor was calculated in order to adjust the data for seasonal fluctuations.
The data of the real effective exchange rate is obtained from a database constructed by Darvas and is defined as the exchange rate of Ethiopia weighted against a basket of currencies of Ethiopia’s trading partners (Darvas, 2012). Measures of the total trade balance of Ethiopia are obtained from The Gobal Economy database (The Global Economy, 2015d; 2015e).
17
The Ethiopian fiscal year starts in July, however all data is based on the Gregorian calendar and thus the first quarter refers to the months January to March throughout the entire analysis.
6. Analysis and Results
As mentioned above, the bounds testing approach to cointegration is used to analyze the relationship between the real exchange rate and trade balance between Ethiopia and the United States. This approach does not require that the involved variables exhibit the same order of integration. However, an Augmented Dicky Fuller (ADF) test will be applied, to establish the integrating properties of the variables, since the approach is based on the assumption that the order of integration of the variables should not exceed one. In order to fulfill this requirement, the variables should be stationary or become stationary after being differenced once.
The results of the ADF test indicate that this requirement is fulfilled. All quarterly and yearly variables become stationary after being differenced once and are thus integrated of order one. Only the variable denoting the trade balance is integrated of order zero, which means it is stationary. For the level estimates the ADF test was conducted with and without a trend whereas for the differenced variables no trend was included. The results of the ADF test are sensitive to the lag length that is used, which is why the lag length has been chosen carefully for each variable, by testing the significance of each lag coefficient. The results of the unit root test for the quarterly as well as the yearly variables are summarized in Table 2 and Table 3 in the Appendix respectively.
The maximum lag length for the ARDL model is chosen to be one for the annual dataset and four for the quarterly dataset, since it is assumed that after one year the explanatory variables have no effect on the endogenous variable anymore. To proceed, it will be established whether the models satisfy the assumptions of Ordinary Least Squares (OLS): normality, no serial correlation and no heteroscedasticity of the residuals at a reasonable significance level, which is here set at 10 percent. All lag combinations of θ1 to θ4, with a maximum value of one and four respectively, that satisfy the
above mentioned assumptions are identified, the other lag combinations are discarded. From the remaining lag combinations of both models, each lag combination will be tested for cointegration as explained above. The combinations for which the model does not infer cointegration are discarded. With the remaining combinations and the use of the Akaike Information Criterion (AIC), the optimal lag length for the parameters n1 to n4 of the error-correction models are estimated. The estimation of
the respective error correction model is shown in Table 4 and Table 5 below. To avoid biased estimates due to the conducted interpolation methods, Newey-West standard errors are used in the OLS estimations, since they are robust to autocorrelation.
18
Table 4 – Error Correction Model Estimation: Trade Balance, Quarterly
Variables Coefficients Newey-West Std. Error t P>t 95% Confidence Interval ∆TB (Dependent Variable) ∆TBt-1 -0.86138(***) 0.1946256 -4.43 0 -1.25141 -0.47134 ∆TBt-2 -0.79927(***) 0.1865478 -4.28 0 -1.17312 -0.42542 ∆TBt-3 -0.78175(***) 0.2356981 -3.32 0.002 -1.2541 -0.3094 ∆YETH -22.2158 36.21171 -0.61 0.542 -94.7857 50.35408 ∆YETH;t-1 17.66315 35.12402 0.5 0.617 -52.727 88.05326 ∆YETH;t-2 7.116119 15.39757 0.46 0.646 -23.7413 37.97355 ∆YETH;t-3 12.51918 18.57637 0.67 0.503 -24.7087 49.74705 ∆YETH;t-4 -0.2 16.29001 -0.01 0.99 -32.8459 32.4459 ∆YUS -8.05588 8.543735 -0.94 0.35 -25.1779 9.066144 ∆YUS;t-1 9.740102 9.479619 1.03 0.309 -9.25748 28.73768 ∆YUS;t-2 -0.66632 11.44406 -0.06 0.954 -23.6007 22.26809 ∆YUS;t-3 6.946324 12.93146 0.54 0.593 -18.9689 32.86155 ∆YUS;t-4 3.925224 10.45135 0.38 0.709 -17.0198 24.8702 ∆RER -1.00845 2.546642 -0.4 0.694 -6.11203 4.095137 ∆RERt-1 0.162022 2.417096 0.07 0.947 -4.68195 5.005991 ∆RERt-2 0.566818 3.387292 0.17 0.868 -6.22147 7.355102 ∆RERt-3 1.748359 3.203847 0.55 0.587 -4.67229 8.169012 TBt-4 -0.66952(**) 0.2979356 -2.25 0.029 -1.2666 -0.07244 YETH;t-4 -0.78101 0.9763451 -0.8 0.427 -2.73765 1.175631 YUS;t-4 1.974628 3.579023 0.55 0.583 -5.1979 9.147152 RERt-4 -0.55788 1.213947 -0.46 0.648 -2.99069 1.874921 constant -16.4781 29.08714 -0.57 0.573 -74.77 41.81386
Source: Own calculation; Note: t denotes the time, 1 time period corresponds to 3 months; all variables are in natural logarithms; TB
denotes the Trade Balance, RER denotes the Real Exchange Rate,YETH and YUS denote the level of national income of Ethiopia and the U.S. respectively, ∆GDPETH denotes the growth rate of national income of Ethiopia, ∆GDPUS denotes the growth rate of national income of the U.S.,∆RER denotes the change in the Real Exchange Rate, ***significant at the 1% level, **significant at the 5% level, *significant at the 10% level
From Table 4 it can be seen that the only coefficients that are significantly different from zero are those belonging to the trade balance and its first differenced lagged variables. Specifically, a one percent increase in the trade balance level in the long run decreases the growth of the current trade balance by 0.67 percent. Higher growth rates of the trade balance in previous quarters are associated with a lower current growth rate of the trade balance. These findings are consistent with
19
macroeconomic theory. Even though no other coefficient is significantly different from zero, throughout all models with different lag combinations, that satisfied the regression diagnostics and exhibited cointegration of the variables, the following trends could be identified:
The income growth of Ethiopia seems to have a large negative effect on the change in the trade balance in the very short run, which complies with the expectations and can be explained by the increase in import demand due to higher income, which leads to a deterioration of the trade balance. In the medium run however, income growth seems to have positive effects on the trade balance. An explanation for this outcome could be that there exists a reverse causality in the estimation, meaning that high income growth might be due to high net exports and not the other way around.
In the short and long run a high income growth and income level in the U.S. seems to have positive effects on the Ethiopian trade balance, which confirms with the expectation, since this will increase the import demand in the United States, which in turn increases net exports from Ethiopia.
The short run estimates for the effect of changes in the exchange rate on the trade balance suggest that there is a negative effect in the very short run which then turns into a positive effect in the medium run. This is consistent with the theory of the J-Curve and implies that a devaluation of the currency in fact improves the trade balance, as was found by Gebeyehu (2014). The long run coefficient however indicates, that a high exchange rate level has a small and insignificant negative effect on the trade balance. This can be explained by the above mentioned problems regarding a cheap currency, such as a deterioration of the terms of trade, which implies an increase in expenditures on imports (e.g. food imports), on which Ethiopia is highly dependent. However, since the coefficient is not significantly different from zero, the model estimation infers that there is no long run relationship between the real exchange rate and the trade balance. Still, this estimation confirms the expectation that a devaluation of the Birr in fact leads to an improvement in the trade balance in the short run.
From Table 5 it can be seen that it is estimated that there does not exist a long run relationship between the level of the real exchange rate and Ethiopian income growth, since the coefficient of the variable RERt-1 is not significantly different from zero. The short-run impact of changes in the real
exchange rate on the trade balance can be found from the coefficient of the first differenced real exchange rate parameter. It can be seen that there is a significant short run effect of changes in the exchange rate on income growth. Specifically, a 1 percent increase of the real exchange rate, increases national income growth by 0.49 percent in the very short run. After one year this increase in the exchange rate is found to still have a positive impact on income growth, specifically income growth increases by 0.25 percent. This result implies that a devaluation of the Birr indeed has a positive effect on income growth. This finding is consistent with the finding of Bahmani-Oskooee and Gelan (2013), who found that a devaluation of the Birr leads to an increase in national income growth.
Surprisingly, the coefficient for the dummy variable, which denotes the peg to the U.S. Dollar, is positive, which means that under the peg, income growth is estimated to be 0.22 percent higher than it was after the abandonment of the peg. However, the data suggests that income growth
20
was on average 0.076 percent higher after the abandonment of the peg than it was before. A possible explanation for this contradictory estimation could be the Ethiopian-Eritrean war from 1998 to 2003, which resulted in negative growth rates during this period. In addition it is found that this result is sensitive to the selection of the year in which the peg was abandoned, which was chosen to be 1992 since the peg was abandoned in September 1992. If instead the year in which the peg was abandoned is chosen to be 1993, the coefficient of the peg is estimated to be negative. This result is shown in Table 6 in the Appendix. This estimation indicates that income growth is 0.15 percent lower under the peg than it is after the abandonment of the peg, which suggests that the devaluation of the Birr in 1992 indeed led to an increase in income. The other coefficients from this adjusted model coincide with the coefficients of the original model, except for the coefficient of the level of the real exchange rate in the long run (RERt-1). In the adjusted model, a higher real exchange rate in the long run is estimated to
negatively affect income growth. This finding is consistent with the finding from the quarterly model which analyzed the effects of changes of the real exchange rate on the trade balance, and which found that a higher level of the real exchange rate in the long run negatively affects the trade balance, though the coefficient is not significant.
Table 5 – Error Correction Model Estimation: Income, Annual
Variables Coefficients
Newey-West
Std. Error t P>t
95%
Confidence Interval
∆YETH(Dependent Variable)
Peg 0.199921(*) 0.0987758 2.02 0.062 -0.01193 0.411775 ∆YETH;t-1 -0.16441 0.1749753 -0.94 0.363 -0.5397 0.210874 ∆M 0.18786 0.1490651 1.26 0.228 -0.13185 0.507573 ∆G -0.07943 0.112618 -0.71 0.492 -0.32097 0.162116 ∆Gt-1 0.213354(*) 0.1176122 1.81 0.091 -0.0389 0.465607 ∆RER 0.488928(***) 0.1411742 3.46 0.004 0.18614 0.791717 ∆RERt-1 0.24581(*) 0.1226615 2 0.065 -0.01727 0.508893 YETH;t-1 0.330908(**) 0.1339903 2.47 0.027 0.043528 0.618289 Mt-1 0.214996 0.1419988 1.51 0.152 -0.08956 0.519553 Gt-1 -0.21323(*) 0.1124065 -1.9 0.079 -0.45432 0.02786 RERt-1 0.008212 0.0777928 0.11 0.917 -0.15864 0.175061 constant -0.87894(***) 0.2607187 -3.37 0.005 -1.43813 -0.31976
Source: Own calculation; Note: t denotes the time, 1 time period corresponds to 1 year; all variables are in natural logarithms; peg2 denotes
the dummy variable describing the currency peg to the U.S. Dollar, which has the value 1 from 1981-1992 and the value 0 afterwards, RER denotes the Real Exchange Rate, M denotes the Money Supply, G denotes Government Spending, ∆RER denotes the change in the Real Exchange Rate, ∆M denotes the growth rate of Money Supply, ∆G denotes the growth rate of Government Spending, ***significant at the 1% level, **significant at the 5% level, *significant at the 10% level
21
During the analyzed period, money supply does not seem to have a relevant effect in the long or short run since both coefficients (∆M and Mt-1) are not significantly different from zero at the 10 percent
level. The estimates however still suggest that there is a weak but positive effect of an increase in the money supply on income.
Government spending from the year before is estimated to have a positive effect on income, specifically a 1 percent increase in government spending in the previous year is estimated to increase income growth in the current year by 0.21 percent.
To test whether there exists a long run relationship between the real effective exchange rate and the total trade balance of Ethiopia, first an Augmented Dicky Fuller test will be applied. The results of the test can be found in the Appendix in Table 7 and Table 8. The test indicates, that both variables are indeed integrated of the same order. The results of the Engle Granger cointegration test further indicate that both variables are not cointegrated and thus there does not exists a long run relationship between the real effective exchange rate and Ethiopia’s trade balance. Both variables are illustrated in Figure 2 and Figure 3 in the Appendix respectively. This finding is not contradicting the results obtained considering only the case with Ethiopia and the United States, where only a weak and not significant relationship between a currency devaluation and the bilateral trade balance could be found. Furthermore no long run relationship could be established between the real effective exchange rate and the Ethiopian income. An explanation for this finding could be that Ethiopia is largely dependent on food imports as mentioned above, which means that the price elasticity of the import demand is relatively inelastic and thus positive effects of a currency devaluation (gain in international competitiveness and thus higher export earnings) might be cancelled out by negative effects (terms of trade deterioration and thus higher import expenditures). Another possible explanation might be that there are other exogenous factors influencing the trade balance of Ethiopia, which cause large price fluctuations of primary goods and thus a high volatility in export earnings in Ethiopia, and which are not included in this analysis.
Still, for the bilateral trade with the United States some positive effects on income growth and the trade balance in the short run could be identified, which is consistent with the findings of Moller (Davidson, 2014). In light of the findings of the multilateral analysis, this indicates that the positive effects of a devaluation of the currency seem to outweigh the negative effects in the bilateral case. In the long run however, a currency devaluation is estimated to have weak negative effects, which means that the negative effects outweigh the positive effects.
Robustness and diagnostics test for the error correction models are presented in Table 9 in the Appendix and infer that the model analyzing the relationship between the real exchange rate and income is appropriate to explain this relationship. The statistical model analyzing the relationship between the real exchange rate and the bilateral trade balance is estimated not fit the data well and the coefficient of determination is rather small.
22
7. Conclusion
In this paper the relationship between changes in the real exchange rate and the trade balance as well as national income was investigated. The findings of the analysis can be summarized as follows: Weak evidence on the existence of a J-Curve effect following a devaluation of the currency in the bilateral case was found. A devaluation was estimated to improve the trade balance in the short run, however, in the long run so significant relationship could be established. These findings suggests that the aim of improving the country’s balance of payments by devaluing the Birr indeed is a sound strategy in the case of the foreign exchange with the United States in the short run.
Increasing the exchange rate is found to have significant positive effects on national income growth in the short run, however as in the previous model, the level of the real exchange rate has no significant effect on income in the long run. The effect of the abandonment of the peg on the growth rate of the economy is sensitive to the selection of the year in which the peg was abandoned. If 1993 is chosen as the first year without the currency peg, then income growth is estimated to be slightly higher after the abandonment of the peg than before.
Furthermore, an Engle Granger cointegration analysis was conducted to analyze the relationship between the real effective exchange rate and the total trade balance of Ethiopia. The result suggests that there is no long run relationship between the variables, which suggests that contrary to the findings of previous studies, a devaluation is not a sound strategy to improve the countries total trade balance.
A limitation to this analysis is the use of seasonally adjusted data, since this introduces a moving average component into the time series data, which in turn might lead to the unit root tests being biased towards non-rejection of the null hypothesis. Furthermore, the interpolation methods used for the Ethiopian GDP and CPI introduced a source of serial correlation into the regression. Newey-West standard errors were used, however, the use of the interpolation may have decreased the significance of the analysis. The unavailability of data for the bilateral trade between Ethiopia and the United States before 1993 furthermore diminishes the explanatory power of the model to analyze the effects of the devaluation of the Birr following the abandonment of the currency peg.
All in all, it can be concluded that a devaluation of the Ethiopian Birr does improve national income growth in Ethiopia and the bilateral trade balance with the United States in the short run. However, no long run relationship between the real effective exchange rate and the total trade balance could be established. Thus it can be concluded that a devaluation of the currency does not seem to be an appropriate strategy to improve the trade balance for Ethiopia, however in order to give a meaningful policy recommendation, further research on the relationship between the exchange rate and the trade balance need to be conducted, including different trading partners. Therefore for further research it would be interesting to investigate these relationships.
23
Bibliography
African Growth and Opportunity Act. (2015). Bilateral Trade by Sector: Unites States – Ethiopia. Retrieved from: http://agoa.info/profiles/ethiopia.html
Alexander, S. S. (1952). Effects of a Devaluation on a Trade Balance. Staff Papers-International Monetary Fund, 2(2), 263-278.
Arize, A. C. (1994). Cointegration Test of a long-run Relation between the Real Effective Exchange Rate and the Trade Balance. International Economic Journal,8(3), 1-9.
Ayen, Y. W. (2014). The Effect of Currency Devaluation on Output: The Case of Ethiopian Economy. Journal of Economics and International Finance, 6(5), 103-111.
Bahmani-Oskooee, M. (1991). Is there a long-run Relation between the Trade Balance and the Real Effective Exchange Rate of LDCs?. Economics Letters, 36(4), 403-407.
Bahmani-Oskooee, M. (1998). Cointegration Approach to Estimate the long-run Trade Elasticities in LDCs. International Economic Journal, 12(3), 89-96.
Bahmani-Oskooee, M., & Gelan, A. (2013). Are Devaluations Contractionary in Africa?. Global Economic Review, 42(1), 1-14.
Bahmani-Oskooee, M., Harvey, H., & Hegerty, S. (2013). Currency Depreciations and the US–Italian Trade Balance: Industry-level Estimates. Research in Economics, 67(3), 215-225.
Bahmani-Oskooee, M., & Hegerty, S. W. (2011). The J-curve and NAFTA: Evidence from Commodity Trade between the US and Mexico. Applied Economics, 43(13), 1579-1593. Bureau of Economic Analysis. (2015). National Income and Product Accounts. Retrieved from:
https://www.bea.gov/newsreleases/national/gdp/gdpnewsrelease.htm
Darvas, Z. (2012). Real Effective Exchange Rates for 178 Countries: A New Database. Retrieved from: http://www.bruegel.org/publications/publication-detail/publication/716-real-effective-exchange-rates-for-178-countries-a-new-database/
Davidson, W. (2014). World Bank Urges Ethiopia to Devalue Birr to Boost Exports, Bloomberg Business, Retrieved from:
http://www.bloomberg.com/news/articles/2014-07-22/world-bank-urges-ethiopia-to-devalue-birr-to-lift-export-revenue
Degefe, B., & Moges, K. (1994). Post-Devaluation Ethiopian Economy: From Stagflation to
Stagflation. In The Ethiopian Economy: Problems of Adjustment: Proceedings of the Second
Annual Conference on the Ethiopian Economy, p. 61. Addis Ababa University Print. Press.
Dercon, S. (2000). Growth and Poverty in Ethiopia in the 1990s: an Economic Perspective. University
of Oxford: Centre for the Study of African Economies.
Economic Research. (2013). Consumer Price Index for Ethiopia. Retrieved from: https://research.stlouisfed.org/fred2/series/DDOE01ETA086NWDB
24
from: https://research.stlouisfed.org/fred2/series/CPALTT01USQ661S
Gebeyehu, A. B. (2014). Exchange Rate and Trade Balance; J Curve Effect in Ethiopia. Journal of Economics and Sustainable Development, 5(24), 185-191.
Geda, A. (2006). The Structure and Performance of Ethiopia's Financial Sector in the pre-and post- Reform Period with a Special Focus on Banking (No. 2006/112). Research Paper, UNU-WIDER, United Nations University (UNU).
Ghura, D., & Grennes, T. J. (1993). The Real Exchange Rate and Macroeconomic Performance in Sub-Saharan Africa. Journal of Development Economics, 42(1), 155-174.
Gylfason, T., & Schmid, M. (1983). Does Devaluation Cause Stagflation?. Canadian Journal of Economics, 641-654.
Harrison, A. (1996). Openness And Growth: A Time-Series, Cross-Country Analysis for Developing Countries. Journal of Development Economics, 48(2), 419-447.
IFPRI. (2010). Ethiopia’s Growth and Transformation Plan. Retrieved from: http://www.ifpri.org/sites/default/files/publications/esspwp30.pdf
IMF. (2008). The Federal Democratic Republic of Ethiopia: Selected Issues. Retrieved from: http://www.imf.org/external/pubs/cat/longres.cfm?sk=22219.0
IMF. (2015). Key Indicators for Ethiopia.
Retrieved from: http://elibrary-data.imf.org/FindDataReports.aspx?d=33060&e=161891 Kibret, H. (1994). Is the Ethiopian Birr Overvalued? A Preliminary Assessment and Policy
Implication. In The Ethiopian Economy: Problems of Adjustment. Addis Ababa
Martins, P. M. (2010). Do Capital Inflows Hinder Competitiveness? The Real Exchange Rate in Ethiopia (No. 10/07). CREDIT Research Paper.
Narayan, P. K. (2005). The Saving and Investment Nexus for China: Evidence from Cointegration Tests. Applied economics, 37(17), 1979-1990.
Oanda. (2015). Historical Exchange Rates. Retrieved from: http://www.oanda.com/currency/historical-rates/
Ogundipe, A. A., Ojeaga, P., & Ogundipe, O. M. (2013). Estimating the Long Run Effects of Exchange Rate Devaluation on the Trade Balance of Nigeria. European Scientific Journal, 9(25).
Pilbeam, K. (2013), International Finance, 4th edition, Basingstoke (UK): Palgrave MacMillan.
Chapter 3.
Rawlins, G., & Praveen, J. (1993). Devaluation and the Trade Balance: The Recent Experience of Selected African Countries. Center for Economic Research on Africa, Department of Economics and Finance, School of Business Administration, Montclair State.
Rose, A. K. (1990). Exchange Rates and the Trade Balance: Some Evidence from Developing Countries. Economics Letters, 34(3), 271-275.
25
Evidence from Pakistan. The International Trade Journal, 26(2), 139-153.
Shirvani, H., & Wilbratte, B. (1997). The Relationship between the Real Exchange Rate and the Trade Balance: an Empirical Reassessment. International Economic Journal, 11(1), 39-50.
Taye, H. K. (1999). The Impact of Devaluation on Macroeconomic Performance: The Case of Ethiopia. Journal of Policy Modeling, 21(4), 481-496.
The Global Economy. (2015a). Ethiopia Economic Indicators: Ethiopia GDP, Constant Dollars. Retrieved from: http://www.theglobaleconomy.com/Ethiopia/GDP_constant_dollars/ The Global Economy. (2015b). Ethiopia Economic Indicators: Ethiopia Dollar Exchange Rate.
Retrieved from: http://www.theglobaleconomy.com/Ethiopia/Dollar_exchange_rate/ The Global Economy. (2015c). Ethiopia Economic Indicators: Government Spending, in Dollars.
Retrieved from: http://www.theglobaleconomy.com/Ethiopia/government_spending_dollars/ The Global Economy. (2015d). Ethiopia Economic Indicators: Ethiopia Exports, in Dollars.
Retrieved from: http://www.theglobaleconomy.com/Ethiopia/exports_dollars/ The Global Economy. (2015e). Ethiopia Economic Indicators: Ethiopia Imports, in Dollars.
Retrieved from: http://www.theglobaleconomy.com/Ethiopia/imports_dollars/ The Worldbank. (2015). World Development Indicators. Retrieved from:
http://databank.worldbank.org/data/views/reports/tableview.aspx Trading Economics. (2015). Food Imports in Ethiopia. Retrieved from:
http://www.tradingeconomics.com/ethiopia/food-imports-percent-of-merchandise-imports-wb-data.html
World Bank Group. (2014). 3rd Ethiopia Economic Update – Strengthening Export Performance
Through Improved Competitiveness, Retrieved from:
http://www.wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2014/11/12/0 00470435_20141112125152/Rendered/PDF/895480REVISED00c0Update0v0final0web.pdf Yazici, M., & Islam, M. Q. (2014). Exchange Rate and Bilateral Trade Balance of Turkey with EU
