Evaluation of the Effect of Exchange Rate Variability on the Export of Ethiopia’s Agricultural Product: A Case of Coffee

Abstract

Research was conducted to find out the effects of exchange rate variability, terms of trade, competitiveness and gross domestic product on the dominant export crop of Ethiopia, coffee. This study employs annual time series data (1992–2010) and uses the autoregressive distributive lag (ARDL) model augmented by the Wald test. The results reveal that exchange rate variability has a negative effect on the export of coffee in the short run, but is insignificant in the long run. This implies that, over time, exchange rate changes in the country have been favouring the export performance of coffee. Regardless of exchange rate variability, the competitiveness of the country, explained by real effective exchange rates, improved, but the price of coffee did not increase relative to the price of imports, which has resulted in deteriorating terms of trade. To improve the worsening terms of trade and benefit from policy changes, export diversification and value addition are possible solutions the country should focus on.

JEL Classification: F130

Keywords

ARDL Exchange Rate Variability Export Coffee Devaluation Wald Test

1. INTRODUCTION

Over the past three decades, many least developed countries (LDCs) have implemented economic stabilisation and structural adjustment programmes (ESSAPs) under initiative and supervision of the International Monetary Fund (IMF) and World Bank. These programmes include currency devaluation, reduction of government expenditure and increase in taxes. Of these, exchange rate policy has been a major item on the economic and political agenda of Ethiopia for almost two decades. From the late 1940s through to the early 1990s, the Ethiopian currency, the birr, remained rigidly pegged to the US dollar (USD). During 1945–71, the birr/USD rate remained unchanged at 2.5. It was revalued to 2.3 in December 1971, and then to 2.07 in February 1973, and remained at that level until October 1992. The natural outcome of this passive exchange rate policy was the development of an illicit parallel market for foreign exchange, where at times the spread between the two rates reached as high as 230. The overvalued official exchange rate coupled with stringent foreign exchange rationing provided fertile ground for illicit cross-border trade, particularly in coffee and live animals (ECA, 2007; Kibrom, 2009). However, since 1992 onwards, there has been continuous devaluation of the birr; for instance, within three years (1992–95), the birr was devaluated 242 per cent (Kidane, 1999).

Although the focus of the economic reform programme has been to make exports an engine of growth, the government’s attempts do not seem to have achieved the desired results. This is because, despite the government continuously devaluing the currency over the last two decades, the evils of market imperfections such as negative terms of trade, as well as deteriorating terms of trade, and serious foreign currency shortages still persist.

Exchange rate variability, which was a result of the continuous devaluation, has been a source of risk for stakeholders. At times, large commercial firms tend to hoard USD assuming that the government will devalue the birr. This situation leads to foreign currency shortage in the market. Above all, firms involved in the import and export sector tend to speculate in the market and hold onto their goods, which also exacerbates the aforementioned problem (National Bank of Ethiopia [NBE], 2008). For instance, followed by a 20 per cent devaluation in 2009–10, total exports that year were about USD 1.45 billion, which was 1.2 per cent lower than the preceding year (NBE annual report, 2009). During the same year, the poor export performance was attributed to lower export earnings from coffee, pulses, leather products, fruits and vegetables, among others. All in all, the ratio of exports to gross domestic product (GDP) dropped from 5.5 per cent to 4.6 per cent reflecting structural rigidities in the export sector and a lack of competiveness (NBE annual report, 2009). Yet, trade in services, particularly travel and transportation, surged as a result of increased conference tourism and air transport services (NBE, 2008). Clearly, the exchange rate policy followed by the country has not resulted in the predetermined policy objectives.

The agriculture-based Ethiopian economy is highly dependent on Coffea Arabica, as it contributes more than 60 per cent of the country’s foreign exchange earnings. The labour-intensive tree crop also provides much employment in the rural areas and is the means of livelihood for over 15 million people in Ethiopia. Thus, as well as being an important export, coffee plays a vital role in both the cultural and socio-economic life of the country. Though coffee has been the dominant agricultural export, its share in the total export basket of the country has been decreasing (NBE, 2010).

So far, there has been little or no scientific statistical investigation into how exchange rate variability affects the export of the dominant exports in Ethiopia. Therefore, without concrete evidence that the policy is on the appropriate path, following the prevailing exchange rate policy could be problematic. Several empirical studies have investigated the effects of exchange rate variability on exports in different parts of the world. Some have found a positive significant relationship (Bernardina, 2004; Sorsa, 1999; Todani and Munyama, 2005), while others have reported the reverse (Abolagba et al., 2010; Aziakpono et al., 2005; Bahmani-Oskooee and Kara, 2003; Ogun, 1988; Oyejide, 1986; Sabuhi-Sabouni and Piri, 2008). Another group of studies have indicated that the effects of exchange rate variability are ambiguous (Du and Zhu, 2001; Klaassen, 1999).

2. METHODOLOGY

2.1 Data Source and Collection Method

Time series data over an 18-year period (1992–93 to 2009–10) were collected from the NBE, Central Statistical Authority of Ethiopia (CSA), Ethiopian Custom Authority (ECA) and Food and Agriculture Organization (FAO) relating to major private traders involved in import and export trade. The primary data helped augment the econometric estimation with information on the real situation at hand.

2.2 Theoretical Framework

Domestic producers and consumers react to prices in domestic currency: for a given world price, devaluation increases the internal price of the exportable, so that for a given world price, the price received per unit of exports or per unit of import substitute (tradable) in national currency increases relative to the price of non-tradables (Abbot et al., 2001; Romer, 2006).

Based on this theoretical background, it is possible to formulate the following export supply model which can capture factors related to export supply of a country.

The model can be presented as follows:

X_{t} = f (R E E R, G D P, T O T, E V) + ε

(1)

where:

X_t= real value of coffee exported;

REER = real effective exchange rate;

GDP = gross domestic product of the exporting country;

TOT = terms of trade;

EV = exchange rate variability; and

ε = the error term.

The following formulae were applied to find the real effective exchange rate (REER), terms of trade (TOT) and exchange rate variability (EV):

The real effective exchange rate (REER) measures the competitiveness of Ethiopia’s export sector vis-à-vis the rest of the world. It is important to construct the REER index, which is a measure of the price of the country’s goods relative to the price of its trading partner’s, both expressed in domestic currency.

R E E R = \frac{N E R \times W P I_{W}}{C P I_{D}}

(2)

where:

NER = trade-weighted nominal exchange rate of trading partner countries;

WPI_W = trade-weighted wholesale (producer) price index of trading partners; and

CPI_D = Ethiopian domestic consumer price index.

Terms of trade (TOT): This research uses the general terms of trade which is calculated as the value of exports divided by the value of all imports.

T O T = \frac{\sum P_{i} \times E x_{i}}{\sum P_{j} \times E M_{j}}

(3)

where:

Ex_i = volume of export of commodity i;

EM_j = volume of import of commodity j;

P_i = price of commodity i;

P_j = price of commodity j;

EV = exchange rate variability, which is an index used to measure the movement of exchange rate based on the competitive potential of the country.

Exchange rate variability can be calculated by (as cited by David, 2004; from Kenen and Rodrick, 1986):

E V = {[\frac{1}{3} \sum_{t = - 1}^{+ 1} {(R E E R_{t} - R E E R_{t - 1})}^{2}]}^{\frac{1}{2}}

(4)

2.3 The Autoregressive Distributed Lag (ARDL) Model

The ARDL approach is found to be an appropriate method to estimate the influence of continuous devaluation on export supply. Bahmani-Oskooee and Hegerty (2007) recommend that future studies of the effects of exchange rate variability on trade flows should rely on this method. According to them, all variables in a given trade flow model are non-stationary, while most measures of exchange rate variability are stationary (see also Abbott et al., 2001; Bahmani-Oskooee and Wang, 2007). The only cointegration and error correction method that allows some independent variables to be non-stationary (I(1)) and some to be stationary (I(0)) is the ARDL bounds testing approach advocated by Pesaran et al. (2001). It has the advantage of avoiding the classification of variables (the independent variables) of interest into I(0) or I(1), and unlike conventional cointegration tests, there is no need for unit root or stationary pre-testing (Pesaran et al., 2001).

The generic ARDL model, as adopted from Green (2002), looks like:

Y_{t} = μ + \sum_{i = 1}^{p} γ_{i} Y_{t - i} + \sum_{i = 1}^{r} β_{i} X_{t - i} + δ W_{t - 1} + ε_{t}

(5)

where:

Y_t = the value of the dependent variable in time t;

Y_t–i = the lagged values of the dependant variable Y;

X_t–i = the lagged values of the independent variable X;

W_t–1 = the first lag of the independent variables associated with long-run elasticities; and

ε_t = assumed to be the serially uncorrelated and homoscedastic error term.

2.4 Lag length Determination

The Akaike information criterion (AIC), Schwarz information criterion (SIC) and Hannan–Quinn criterion (HQC) are the most common techniques to determine the lag length of time series data. In this study, we have employed the parsimonious model (SIC) to select the smallest possible lag length. Since the ARDL model estimates (p + 1)k numbers of regression in order to find the optimal lag length, where p is the maximum number of lags to be used and k is the number of variables in the equation, this criterion will help us arrive at a given fit with the smallest number of parameters per observation (Green, 2002).

2.5 Specification of the Model

Engle and Granger (1987) and Johansen and Juselius (1990) suggest that the error correlation model can be applied to estimate the long-term relationship between the dependant variable and other explanatory variables. But due to constraints in this model, more appropriate techniques have been suggested, such as the ARDL which has been outlined and elaborated by Pesaran and Shin (2001). This approach involves two stages: first, it examines if there is a long-run relationship between the variables under investigation; and then, it estimates the coefficient of the long-run relations and the associated error-correlated models.

To set the dependant and independent variables at the same level and make the econometric estimates robust, differencing of all the variables involved is found to be crucial. The tabulated F-values (Pesaran et al., 2001) have upper and lower bounds where the upper bound assumes all the variables are I(1) stationary and the lower bound assumes all the variables are I(0) stationary. Therefore, examining whether a variable is I(1) or I(0) stationary, which requires differencing, helps arrive at a conclusion where the calculated F-statistic lies within the upper and lower bound.

Based on our earlier equations, the ARDL model for the export price of coffee can be specifically modelled as follows:

\begin{array}{l} Δ ln X_{t} = α_{0} + \sum_{j = 1}^{k} β_{0} Δ ln X_{t - j} + \sum_{j = 1}^{m} β_{1} Δ ln R E E R_{t - j} + \sum_{j = 1}^{n} β_{2} Δ ln G D P_{t - j} \\ + \sum_{j = 1}^{s} β_{3} Δ ln T O T_{t - j} + \sum_{i = 1}^{l} β_{4} Δ ln E V_{t - j} + μ_{0} ln X_{t - 1} \\ + μ_{1} ln R E E R_{t - 1} + μ_{2} ln G D P_{t - 1} + μ_{3} ln T O T_{t - 1} + ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ μ_{4} ln E V_{t - 1} + u_{i} \end{array}

(6)

where:

k, m, n, s and l indicate the optimum lag length of the variable under investigation;

ΔlnX_t–j = the differenced and lagged logarithmic value of the export of coffee measured in USD;

ΔlnREER_t–j = the differenced and lagged logarithmic index of the REER of the country using base year 1995 = 100;

ΔlnGDP_t–j = the differenced and lagged logarithmic value of the GDP of the country measured in USD;

ΔlnTOT_t–j = the differenced and lagged logarithmic value of the TOT of the country measured in percentages (calculated using equation 3);

ΔlnEV_t–j = the differenced and lagged logarithmic value of the exchange rate variability (calculated using equation 4);

ΔlnX_t–1, ΔlnREER_t–1, ΔlnGDP_t–1, ΔlnGDP_t–1, ΔlnTOT_t–1and ΔlnEV_t–1are logarithmic first-lag values of the same variables explained above;

β₀, β₁, β₂, β₃ and β₄ are short-run coefficients to be estimated; and

µ₀, µ₁, µ₂, µ₃and µ₄ are long-run coefficients to be estimated.

In order to determine whether there is a long-run relationship among the variables in cointegrating equation (1), the null hypothesis of no long-run relationship (that is, H₀: μ₁ = μ₂ = μ₃ = μ₄ = 0) against the alternative hypothesis of a long-run relationship (that is, H_A: μ₁ μ₂ μ₃ μ₄ 0) using the F-statistic (Wald test) was tested. The decision rule is that, if the computed F-statistic is greater than the upper critical bound as tabulated in Pesaran et al. (1999), then the null hypothesis can be rejected, suggesting cointegration. On the other hand, if the computed F-statistic is less than the lower critical bound, the test fails to reject the null hypothesis and it can be concluded that there is no cointegration. Given the case where the test statistic lies within the lower and upper critical bounds, conclusive inference can only be made once the order of the integration of the underlying regressors is known (Pesaran et al., 1999).

3. EMPIRICAL RESULTS

3.1 ARDL (0,1,1,1,1) Results

The ARDL (0,1,1,1,1) was found to be the appropriate model for this particular crop after employing the lag length-determining criterion (SIC) and all the necessary tests for time series analysis. These tests include: Breusch–Pagan/Cook–Weisberg test for heteroscedasticity; Ramsey RESET test using powers of the fitted values of the regressor (helps to test whether the model has an omitted variable or not); the Lagrange Multiplier (LM) test for autoregressive conditional heteroscedasticity (ARCH); and the Breusch–Godfrey LM test for autocorrelation augmented by the Durbin–Watson d-statistic.¹

All the tests in Table 1 indicated that the model was the best fit to explain the problem at hand. The regression results of short-run and long-run effects are given next. An important point to ponder about the forthcoming discussions is that variables are interpreted even if they are insignificant. This is because, although the significance implies the degree of relationship, since this is a time series analysis, the signs do have a meaningful implication.

Table 1
Diagnostic Test Results

Test Null Hypothesis (H₀) Test Statistics Prob > chi²

Breusch–Pagan/Cook–Weisberg test for heteroscedasticity Constant variance Chi²(1) = 0.04 0.8422

Ramsey RESET test Model has no omitted variables F(3, 3) = 0.34 0.80123

ARCH No ARCH effects Chi²(1) = 0.003 0.9559

Breusch–Godfrey LM test for autocorrelation No serial correlation Chi²(1) = 0.0251 0.8733

Test	Null Hypothesis (H₀)	Test Statistics	Prob > chi²
Breusch–Pagan/Cook–Weisberg test for heteroscedasticity	Constant variance	Chi²(1) = 0.04	0.8422
Ramsey RESET test	Model has no omitted variables	F(3, 3) = 0.34	0.80123
ARCH	No ARCH effects	Chi²(1) = 0.003	0.9559
Breusch–Godfrey LM test for autocorrelation	No serial correlation	Chi²(1) = 0.0251	0.8733

Source: Authors’ own computations.

Table 2

ARDL (0,1,1,1,1) Short-run Coefficient Estimates²

DlnVEC	Coefficient	Standard Error	p-value
L1DlnVEC	0.277809	0.2859011	0.369
DlnGDP	–0.229706*²	0.0964255	0.055
L1DlnGDP	–0.2711396**²	0.0935859	0.027
DlnREER	1.345314*	0.6153217	0.071
L1DlnREER	2.032021*	0.8969321	0.064
DlnEV	–0.1047839*	0.0488081	0.075
L1DlnEV	–0.0465739	0.0475047	0.365
DlnTOT	–0.0885879	0.4719257	0.857
L1DlnTOT	–0.794256	0.4819659	0.150
_cons	–0.0781353	0.0891529	0.415
Adj. R² = 0.57 Durbin–Watson d-statistic (10, 16) = 1.902658 p = 0.08

Source: Authors’ own computations.

Notes: (i) ** and * denote significance of the variables at the 5 per cent and 10 per cent significance levels, respectively.

(ii) All the variables are as specified in Section 2.5.

(iii) The L1D in front of some variables implies it is lagged by one year and differenced.

Consistent with the findings of Mohammad and Zulkornain (2010), the variable of interest (exchange rate variability) is found to be significant at the 10 per cent significance level and to affect exports of the country negatively in the short run (See Table 2). This is actually in line with the theory of risk aversion (Salvator, 2007), where producers speculate against the positive effect of continuous devaluation, which is the basic source of exchange variability in this particular case. According to the qualitative data collected, this finding is also found to be in consistence with the existing situation at hand. Another interesting result is that the terms of trade (TOT) of the country are found to be a negative function of exports. This is consistent with the theory of deteriorating terms of trade (Prebisch, 1950; Singer, 1950)³,which is typical of many LDCs. Its effects are not significant, because the data indicates that the competitiveness index (the REER) has been appreciating since 2005, which leads to an increase in the export of the country. Moreover, diversification in exports, particularly the increase in the exports of oil seeds and flowers, could be a possible reason for the insignificance of the TOT.

In line with Abolagba et al. (2010), the long-run competitiveness index of the country (REER) is significant at the 5 per cent significance level (Table 3). This is a good reflection of the J-curve effect of devaluation. It means that even if devaluation makes exports cheaper for the rest of the world, it is difficult to extract gains from trade in the short run because, for a country like Ethiopia, whose exports are dominated by agricultural products, it is nearly impossible to boost the volume of export immediately. For a crop like coffee, it is impossible to increase production due to the nature of the crop itself (which needs at least three to four years to bear the first berry). As a result, we may not see significant change in the exports of coffee in the short run, while in the long run, it is possible to boost output and increase foreign currency earnings.

Table 3

Long-run Coefficient Estimates

DlnVEC	Coefficient	Standard Error	p-value
L1lnVEC	–0.6655351**	0.2565162	0.025
L1lnGDP	–0.12360373	0.1302725	0.354
L1lnREER	1.295555**	0.5772853	0.046
L1lnEV	0.0305225	0.0486336	0.543
L1lnTOT	–0.2324551	0.3257396	0.490
_cons	5.260311	3.048434	0.1123

Source: Authors’ own computations.

Notes: (i) ‘L1’ in front of each variable implies the lag length of the variable under investigation.

(ii) ** denotes significance of the variables at the 5 per cent significance level.

The negative and insignificant relationship between GDP and the export of coffee is a good indicator of the tendency of export diversification in the country (Arize, 1995; Bahmani-Oskooee and Kara, 2003; Kidane, 1999). This has resulted in the decline in the share of coffee to the total export basket of the country. Empirically, coffee exports contributed more than 60 per cent to the foreign exchange earning of the country in 1992–2001, after which its share started to decline to below 50 per cent at present (data from ECA, 2010).

The TOT are found to be insignificant but with a negative impact in the long run. This implies an improvement in the TOT in the long run, because Ethiopia is trying value addition in the production of coffee. Thus, in addition to exporting raw and washed coffee, Ethiopia is now exporting roasted and decaffeinated coffee. Moreover, value addition in certain other exports, such as hides and skins and clothing (cotton), could be possible reason for the improvement in the TOT.

The previous discussion emphasises that there is a joint relationship between the variables in question. The test result for the existence of the long-run relationship is given in Table 4.

Since all the variables are stationary at their first difference, the upper critical values are used to make a decision on the level of the long-run relationship. Based on Table 4, the computed test statistic is greater that the upper bond critical value at the 5 per cent significance level. As a result, the null hypothesis of no long-run relationship is rejected at the 95 per cent confidence interval.

Table 4

Test Statistics (for the existence of a long-run relationship)

Ho (null hypothesis)	Computed Test Statistics (Wald Test)	1 per cent Critical Values		5 per cent Critical Value		10 per cent Critical Value
Ho (null hypothesis)	Computed Test Statistics (Wald Test)	I(0)	I(1)	I(0)	I(1)	I(0)	I(1)
No long-run relationship	3.98**	3.57	4.84	2.78	3.94	2.43	3.50

Source: Authors’ own computations.

Notes: (i) The critical values are based on a restricted intercept and no trend, as tabulated by Pesaran and Shin (1999).

(ii) ** denotes significance of the variables at the 5 per cent significance level.

3.2 Sensitivity Analysis

As shown earlier, diagnostic tests for serial correlation, autoregressive conditional heteroscedasticity and functional form were conducted. Finally, the stability of the long-run parameters, together with the short-run movements, was examined. A sensitivity analysis was performed using the cumulative sum (CUSUM) and cumulative sum squares (CUSUMSQ) tests proposed by Borensztein et al. (1998). The same procedure was utilised by Mohsen et al. (2002), Pesaran and Pesaran (1997) and Duasa (2007) to test the stability of long-run coefficients. The critical bounds were graphed and the plot of CUSUM was found to be within the critical 5 per cent bound; the CUSUMSQ statistics did not exceed the critical boundaries and that confirms the long-run relationships between the financial variables and indicates the stability of the determinants (Figure 1).

Figure 1

Source: Authors’ own computations.

4. CONCLUDING REMARKS

Both in the short run and the long run, the REER was found to be significant, implying that the competitiveness of the country has improved. On the other hand, the exchange rate variability is significant in the short run but not in the long run, implying that exports are not sensitive to exchange rate variability in the long run. Generally, we can conclude that the exchange rate policy of the country has had a positive impact on the exports of the dominant export crop, coffee.

In the world of uncertainty, there is no next-best as long as the outcome of any policy measure is unknown. However, in the field of economics, various literatures on different policy measures have shown their consequences for over half a century. As a result, this study recommends reliable alternative policy based on the outcomes observed. It appears that the exchange rate variability, which has emanated from continuous devaluation of the birr, has accomplished its target; though it is depleting the natural and capital resources of the country. Therefore, instead of focusing on the volume of exports, the focus should be on real benefits for the country from exports. As a result, this study recommends value addition to the product before export, so that it increases the final price of the product, thereby rewarding factors of production reasonably.

Footnotes

Notes

References

Abbott

Darnell

A.C.

Evans

(2001). The influence of exchange rate variability on U.K. exports. Applied Economic Letters, 8(1), 47–49.

Abolagba

E.O.

Onyekwere

N.C

Agbonkpolor

B.N.

Umar

H.Y.

(2010). Determinants of agricultural exports. Australian Journal of Agricultural Economics, 29(3), 181–84.

Arize

A.C.

(1995). Trade flows and real exchange-rate volatility: An application of cointegration and error-correction modelling. North American Journal of Economics and Finance, 6(1), 37–51.

Aziakpono

Tsheole

Takaendasa

(2005). Real exchange rate and its effect on trade flows: New evidence from South Africa. (Online). Retrieved on 15 December 2010, from http://www.essa.org.za/download/2005Conference/Takaendesa.pdf

Bahmani-Oskooee

Hegerty

S.W.

(2007). Exchange rate volatility and trade ﬂows: a review article. Journal of Economic Studies, 34(3), 211–55.

Bahmani-Oskooee

Kara

(2003). Relative responsiveness of trade flows to a change in prices and exchange rate. International Review of Applied Economics, 17, 293–308.

Bahmani-Oskooee

Wang

(2007). United States-China trade at the commodity level and the yuan-dollar exchange rates. Contemporary Economic Policy, 25(3), 341–61.

Bernardina

(2004). Price and income elasticities of Russian exports. The European Journal of Comparative Economics, 1(2), 175–93.

Borensztein

De Gregorio

Lee

J.W.

(1998). How does foreign investment affect growth? Journal of International Economics, 45, 115–35.

10.

Zhu

(2001). The effect of exchange-rate risk on exports: Some additional empirical evidence. Journal of Economic Studies, 28(2), 106–21.

11.

Duasa

(2007). Determinants of Malaysian trade balance: an ARDL bound testing approach. Journal of Economic Cooperation, 28(3), 21–40.

12.

ECA (Economic Commission for Africa). (2007). Economic report on Africa: Accelerating Africa’s development through diversification. Addis Ababa, Ethiopia.

13.

ECA (Economic Commission for Africa). (2010). Annual report, Addis Ababa, Ethiopia.

14.

Engle

R.F.

Granger

C.W.J.

(1987). Cointegration and error correlation representation estimation and testing. Econometrica, 55, 251–76.

15.

Green

W.H.

(2002). Econometrics analysis (5th edition). Upper Saddle River, New Jersey: Prentice Hall.

16.

Johansen

Juselius

(1990). Maximum likelihood estimation and inference on cointegration–with application to the demand for money. Oxford Bulletin of Economics and Statistics, 52, 169–210.

17.

Kenen

P.T.

Rodrick

D.R.

(1986). Measuring and analyzing the effects of short-term volatility in real exchange rates. The Review of Economics and Statistics, 68, 311–15.

18.

Kibrom

(2009). The galloping inflation in Ethiopia: A cautionary tale for aspiring ‘developmental states’ in Africa. MSc thesis, Addis Ababa University, Ethiopia.

19.

Kidane

(1999). Real exchange rate price and agricultural supply response in Ethiopia: The case of perennial crop. African Economic Research Consortium Research Paper 87, Nairobi, Kenya.

20.

Klaassen

(1999). Why is it so difficult to find an effect of exchange rate risk on trade? Econometric Society World Congress 2000, Paper No. 0133, University of Amsterdam.

21.

Mohammad

Zulkornain

(2010). Impacts of trade liberalisation on aggregate import in Bangladesh: An ARDL bounds test approach. Journal of Asian Economics, 21(1), 37–52.

22.

Mohsen

Bahmani-Oskooee

R.W.

(2002). Long-run demand for money in Hong Kong: An application of the ARDL model. International Journal of Business and Economics, 1(2), 147–55.

23.

National Bank of Ethiopia. (2008). Quarterly reports, 2007–2009. National Bank of Ethiopia, Addis Ababa, Ethiopia.

24.

National Bank of Ethiopia. (2010). Annual reports, 2008–2010. National Bank of Ethiopia, Addis Ababa, Ethiopia.

25.

Ogun

(1998). Real exchange rate movements and export growth: Nigeria, 1960–1990. African Economic Research Consortium. Research Paper No. 82.

26.

Oyejide

T.A.

(1986). The effects of trade and exchange rate policies on agriculture in Nigeria. IFPRI Research Report No. 55.

27.

Pesaran

M.H.

Pesaran

(1997). Working with Microsoft 4.0: Interactive econometrics analysis. Oxford: Oxford University press.

28.

Pesaran

M.H.

Shin

(1999). An autoregressive distributed lag modeling approach to cointegration, Chapter 11 in Econometrics and Economic Theory in the 20th Century. The Ragnar Frisch Centennial Symposium, Cambridge University Press.

29.

Pesaran

M.H.

Shin

Smith

R.J.

(1999). Pooled mean group estimation of dynamic heterogeneous panels. Journal of the American Statistical Association, 94(446), 621–34.

30.

Pesaran

M.H.

Shin

Smith

R.J.

(2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics, 16(3), 289–326.

31.

Prebisch

(1950). The economic development of Latin America and its principal problem. New York: United Nations, ECLA.

32.

Romer

(2006). Advanced macroeconomics (3rd edition). New York: McGraw Hill.

33.

Sabuhi-Sabouni

Piri

(2008). Consideration the effect of exchange rate volatility on agricultural products export price, the case study of Iran’s saffron. American-Eurasian Journal of Agricultural and Environmental Sciences, 2(1), 97–100.

34.

Salvator

(2007). International economics. 9^th edition, USA: John Wiley.

35.

Singer

(1950). The distribution of gains between investing and borrowing countries. American Economic Review, 40(2), 473–85.

36.

Sorsa

(1999). Algeria—The real exchange rate, export diversification, and trade, protection. Washington, DC: International Monetary Fund.

37.

Todani

K.R.

Munyama

T.V.

(2005). Exchange rate volatility and exports in South Africa. Pretoria, South Africa: South African Reserve Bank Research Department.