Niger
Selected Issues and Statistical Appendix
Author: International Monetary Fund
Publication Date:
24 Feb 2009
eISBN:
9781451828832
Language:
English
Keywords:
ISCR; CR; physical capital; impact aid; critical value; equilibrium exchange rate index; underlying current account; Institut National de la Statistique; current account balance; Current account balance
Download PDF (3.5 MB)

This Selected Issues Paper and Statistical Appendix for Niger are discussed. The Nigerien economic outlook has been improving since 1999. After a long period of decline in per capita income, growth accelerated through 2007, attaining an annual average of 4 percent, or about 1 percent in per capita terms. Economic reforms and political stability have attracted external aid and higher domestic and external private investment. The Equilibrium Real Exchange Rate Approach indicates a moderate undervaluation of 0.7 percent to 3.7 percent.

Abstract

This Selected Issues Paper and Statistical Appendix for Niger are discussed. The Nigerien economic outlook has been improving since 1999. After a long period of decline in per capita income, growth accelerated through 2007, attaining an annual average of 4 percent, or about 1 percent in per capita terms. Economic reforms and political stability have attracted external aid and higher domestic and external private investment. The Equilibrium Real Exchange Rate Approach indicates a moderate undervaluation of 0.7 percent to 3.7 percent.

Chapter i Assessment of the Real Exchange Rate and External Competitiveness1

I. Economic Background and Export Performance

1. The Nigerien economic outlook has been improving since 1999. After a long period of decline in per capita income, growth accelerated through 2007, attaining an annual average of 4 percent, or about 1 percent in per capita terms. Economic reforms and political stability have attracted external aid and higher domestic and external private investment. The total investment-to-GDP ratio has increased from an average of 12 percent in 1997–99 to 22 percent in 2005–07 (Table 1).

Table 1.

Niger and Sub-Saharan Africa Comparative Performance, 2005-07

article image
Source: REO, Sub-Saharan Africa, 2008.

Excluding Nigeria and South Africa.

2. Export performance has been similarly positive in recent years (Table 2). After relatively sluggish growth in 1995–99, merchandise exports doubled their nominal growth rate in 2003–07. The decline in uranium exports during the 1990s was reversed after 2002 when world prices rose dramatically due to years of underinvestment in production and revived world demand. However, the acceleration of Niger’s exports is not limited to uranium. Because modern production and marketing techniques have been introduced, agricultural exports like onions and cowpeas are growing annually at double-digit levels. On the other hand, livestock exports were stagnant in 2003–07 because of a drought in 2004 that led to a decline in the stock. In total, Niger exports of goods increased significantly, from US$280 million in 2000 to US$730 millions in 2007 and are projected to increase to about US$980 million in 2008, mainly because uranium prices are still high. As a result Niger’s share of total world exports rose by about 70 percent between 2000 and 2008 (see Figure 1). Although this is mainly driven by mineral exports, nonmineral exports have also accelerated because since 2004 Niger has specialized in onions and cowpeas in response to heavy demand in neighboring countries. Livestock exports have also grown solidly since 2004.

Table 2.

Niger: Average Annual Growth Rate of Exports In U.S. Dollar by Product, 1996-2007

(In Percent)

article image
Source: Central Bank of Niger, 2008.

Exports of gold started in 2004.

Figure 1.
Figure 1.

Niger: Share in World Exports, 2000-08

(In percent)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF WEO.

3. Despite the recent favorable export performance, there are still concerns about a possible deterioration of Niger’s competitiveness. The appreciation of the euro, to which the CFA franc is pegged, and the increase in imported commodity prices have partly offset the positive impact of the rise of uranium prices on the current account balance. Wage pressures have mounted in the public sector and the construction, energy and telecommunications sectors.

II. Research Strategy and Summary of the Results

4. Against the backdrop of recent economic developments, this chapter assesses the real exchange rate and competitiveness of Niger’s economy. The chapter uses four methods to evaluate the exchange rate: (i) purchasing power parity; (ii) the fundamental equilibrium exchange rate; (iii) the macroeconomic balance approach; and (iv) external sustainability based on an analysis of Niger’s sustainable net foreign assets (NFA) position. Competitiveness is assessed by various surveys. The results from applying these methods should not, however, be taken as a complete assessment of Niger’s external stability because Nigeria a member of the West African Economic and Monetary Union (WAEMU), so its real exchange rate is determined by economic, financial, and monetary developments throughout the WAEMU zone. For these reasons the 2007 Decision on Bilateral Surveillance over Members' Policies recommends that external stability in a currency union be assessed at the union level. The decision also states that this does not preclude assessment of the real exchange rate at the member level.

Exchange rate assessment

5. The purchasing power parity approach indicates that in 2006 Niger’s real exchange rate was moderately undervalued by about 12 percent, though it had been generally overvalued in the late 1970s and early 1990s.2

6. The Equilibrium Real Exchange Rate Approach indicates a moderate undervaluation of 0.7 to 3.7 percent. This approach shows a slight overvaluation in 1979 and a major overvaluation in the late 1980s and the early 1990s, in phase with the Purchasing Power Parity (PPP) findings.

7. The macroeconomic balance approach suggests an undervaluation of about 8 percent.

8. The external sustainability approach indicates that the exchange rate is consistent with a sustainable NFA position. Based on current substantial inflows of foreign direct investment to expand uranium production and start petroleum production, and a foreign debt level well below the policy-determined threshold, it is estimated that Niger could sustain an NFA position of about -86 percent of GDP. This level corresponds to a current account deficit of 6.1 percent of GDP, which is very modestly below the projected 2013–28 average deficit of 6.4 percent after a spike in 2009–12 due to heavy investment in mining and petroleum. Although applying the calculated elasticity of the current account to the real effective exchange rate implies a small overvaluation of 1.9 percent, given the margin of error of these estimates, we cannot conclude that there is an overvaluation.

9. On the basis of this analysis it appears that the real effective exchange rate of Niger is in line with the fundamentals. This conclusion is somewhat different (see below) from the recent assessment for the WAEMU region as a whole. The reasons are (i) Niger’s better performance on fundamentals that affect the equilibrium level of the real exchange rate, mainly the terms of trade; and (ii) the lower appreciation in Niger of the real effective exchange rate compared to other WAEMU countries in recent years, due to the higher weight of Nigeria’s naira in the basket of Niger’s trading partners.

Competitiveness

10. A variety of competitiveness indicators show that the business environment has improved significantly. Niger’s Index of Economic Freedom (IEF) with respect to comparator groups stands out because of its high IEF scores on monetary freedom. Diagnostic studies like the World Bank’s Investment Climate Assessment (ICA) and the Diagnostic Integration Study (DTIS) also recognize progress in stabilizing the economy and its positive effect on competitiveness. However, the World Bank’s report on Doing Business (DB) indicators suggests that Niger is among countries where business activities are most costly. This points to the need for improvement in the labor market, procedures to establish a business, access to credit, and cost of capital.

11. Global and sectoral productivity indicators show improvement. Data on annual wages, employment, and value added of the modern sector show that since 2000 productivity has improved in all sectors except services. Value added per worker has increased especially in manufacturing, energy, and transportation. However, the labor market is characterized by shortages of skills, rigidities, and wages cost that are higher than in comparable countries in sub-Saharan Africa.

III. Purchasing Power Parity Assessment

12. One indication of the potential over- or undervaluation of a country’s currency is whether, under the currently existing exchange rate, prices in its economy are higher or lower than in other countries, that is, whether there is absolute PPP against the rest of the world. There are several reasons why absolute PPP would not hold,3 but one that is particularly significant over the long run is the Harrod-Balassa-Samuelson effect (Harrod, 1933; Balassa, 1964; Samuelson, 1964), which states that countries with higher income per capita are expected to have more appreciated real exchange rates.4 Therefore, several studies have sought to assess a potential exchange rate overvaluation by assuming PPP holds after controlling for income per capita level (e.g., Rogoff, 1996; Frankel, 2004; Johnson, Ostry, and Subramanian, 2007; Rodrik, 2007).

13. This section conducts such an exercise for Niger. Following Rodrik (2007), the real exchange rate is estimated as the ratio of the nominal exchange rate to the PPP conversion factor.5 The logarithm of this variable is then regressed on the logarithm of the chained real GDP and on a fixed effect for a given period. The fitted values of the regression are the real exchange rate predicted by PPP taking the Harrod-Balassa-Samuelson effect into account. The deviation of the actual real exchange rate from the predicted values is the measure of overvaluation. Figure 2 shows the estimated overvaluation between 1960 and 20066.

Figure 2.
Figure 2.

Niger: PPP-Implied Estimate of REER Overvaluation, 1960-2005

(In Percent)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Calculation, 2008.

14. The PPP assessment indicates that while the real exchange rate was largely overvalued in the late 1970s and early 1990s, today it is moderately undervalued. The first episode of significant overvaluation occurred in the last half of the 1970s when the real exchange rate was 30 percent above the value indicated by PPP and Niger’s income per capita. This apparent was the result of an increase in the price of uranium that boosted the terms of trade and thus prompted appreciation of the real exchange rate.

15. The second large overvaluation occurred in the early 1990s and ended abruptly in 1994, when the WAEMU devalued the nominal exchange rate. This move brought the real exchange from an overvaluation of 30 percent in 1993 to an undervaluation of 30 percent in 1994. Although the undervaluation has significantly diminished in the last few years as the euro appreciated, the PPP assessment still points to a moderate undervaluation of about 12 percent in 2006.

IV. Equilibrium Real Exchange Rate

16. This section follows the equilibrium real exchange rate (ERER) approach pioneered by Edwards (1989) in assessing whether the real effective exchange rate (REER) is in line with the fundamentals in Niger. Theoretically, the ERER is defined as the rate that would yield equilibrium in the balance of payments—with three important qualifications: (i) no undue restrictions on trade flows, (ii) no special incentives for inflows or outflows of capital; and (iii) no excessive unemployment (Nurske, 1945). In other words, here the ERER is compatible with both internal and external equilibrium. Operationally, however, it is not easy to identify what constitutes internal and external equilibrium. Edwards (1989) considers internal equilibrium to be achieved when the market for nontradable goods clears in the present and the future. Others view it as being realized when there is no gap between domestic and foreign output.

17. ERER models relate the real exchange rate to economic fundamentals, such as net foreign assets, openness, and productivity growth. These models have been used to assess the magnitude by which the exchange rate needs to be adjusted to correct excesses of current account balances. Using the autoregressive distributed lag (ARDL) approach to estimate a panel of 39 African countries and 70 possible single-country models, Chudik and Mongardini (2007) found that the terms of trade, openness, government consumption, and productivity had significant impact on the real exchange rate. Roudet, Saxegaard, and Tsangarides (2007) employ robustness techniques, such as the ARDL and Johansen methods, to conclude that terms of trade, investment, government spending, and openness determine exchange rate equilibrium in WAEMU countries. There are also several country-specific studies using many of the same techniques. For instance, in the 2008 selected issues paper for the WAEMU regional consultation, the ARDL approach was used to compute the equilibrium exchange rate of all WAEMU members. The results did not suggest any significant overvaluation of the real exchange rate in recent years. For the WAEMU as a whole, recent estimates indicate that at the end of 2007, the REER exceeded by a small amount the 95 percent confidence band around the estimated equilibrium real exchange rate.

A. Estimation Strategy

18. The shortness of the time series (1970–2007) and the quality of the data available (Di Bella, Lewis & Marin, 2007) cause many difficulties in estimating the equilibrium exchange rate of Niger. For this reason Chudik and Mongardini (2007) found no significant long-run coefficients when trying to estimate the ERER for Niger, among other countries, even though they used seven determinants of the real exchange rate and estimated ten model specifications.

19. To deal with these challenges, the following strategy has been implemented: (a) taking into account the limited availability and low quality of the time series data, variables included in the regressions are those deemed particularly relevant for Niger that are available for a long period; and (b) the short length of the time series is accounted for by using the ARDL-bound test approach (Peseran, Shin and Smith, 2001) to test for cointegrating relationships for each model specified.

The model

20. The equilibrium REER data-generating process (DGP) of the Nigerien economy for 1970–2007 period can be approximated using the following model:

REERt=α+β1*LTOTt+β2*⁢    LGCGDPt+β3*LPRODt+β4*⁢LOPENt+β5*FLOWt+μ(1)

LTOT, LGCGDP, LPROD, and LOPEN are the natural logarithm of the terms of trade, the ratio of government consumption to GDP, productivity, and openness; FLOW is the flow variable; and μ is the disturbance term. Box 1 defines all these variables and the signs that could be expected from the estimation.

Unit root tests

21. The Dickey and Pantula strategy (1987) is used to carry out the unit root tests. The first step uses the augmented Dickey-Fuller (ADF) model to test on the first difference the existence of more than one unit root. If this null hypothesis is rejected, the second step is implemented; if accepted, the series is an I(2) series or higher. The second step is performed at the variable level to test for the presence of one unit root against the alternative hypothesis of no unit root.

22. The Dickey and Pantula tests, however, may not be robust due to autocorrelation and that heteroskedasticity that may be present given the limited time span and the possible presence of breaks. Therefore, this strategy has been supplemented with the Perron (1992) (PP) and Zivot and Andrews (1992) (ZA) tests. The PP test proposes a nonparametric correction of the error autocorrelation and heteroskedasticity; the ZA test makes it possible to pin down potential structural breaks in the series and then to again carry out the ADF taking into account these breaks. These tests reveal that all series plotted in Appendix 1 are I(1) except for the resource balance variable. Details of the tests are provided in Appendix 2.

Description of Variables

The real effective exchange rate (REER) is defined as

REER+E*⁢(P/P*)

where E is the nominal effective exchange rate, and P and P* are the CPI inflation rates of Niger and its trading partners. Nigeria is included as a trading partner.

The terms of trade is the ratio of the export and import price indices. The macroeconomic impact of the terms of trade on the REER is generally positive because the income effect tends to dominate the substitution effect, but that will depend on the share of international trade in economic activity.

Government consumption as a share of GDP: Increase in government consumption can affect long-run equilibrium in different ways. If the spending is biased towards nontradable goods, the increase will result in an appreciation of the real exchange rate; if the increase is directed to tradable goods, it may cause a depreciation. Therefore the sign could be either positive or negative.

Productivity relative to trading partners captures the Balassa-Samuelson effect. It has been proxied by relative real GDP per capita. Countries with higher productivity growth in the tradable sector experience a rise in relative wages. This increases the price of nontradables relative to tradables and thus results in an appreciation of the real exchange rate. The expected sign is positive.

Openness is usually proxied by the ratio of the value of imports and exports (at current prices) to GDP. However, in a resource-dependent country like Niger, this ratio seems to reflect more the vagaries of the uranium international market than changes in Niger’s trade policies. Therefore, other two proxies have been used: (i) value of imports to GDP and (ii) the value of total trade to GDP, controlling for the value of uranium exports. The latter is equal to trade minus the predicted values from a regression of trade on uranium exports.

The flow variable has been approximated in three different ways: (i) the resource balance (the ratio of real GDP to the values of exports at constant prices minus the value of imports at constant prices; (ii) aid net of reserves; and iii) net foreign income. The resource balance is further adjusted by multiplying the value of exports by the terms of trade.

Econometric Estimation
ARDL-OLS Estimation

23. The ARDL is well-suited to the estimation of time series with different levels of cointegration and performs well when the time series are short (Baffes et al., 1997). In the ARDL approach, model (1) above can be reformulated in the following form:

yt=b+∑i=1pai⁢yt−1+∑i=0nct′⁢xt−i+ωt(2)

This ARDL(p, n) is of order p (number of lags) and n (number of variables), yt being the vector of the dependent variables, xt the vectors of the independent variables, ai and ci the parameters to be estimated, b the constant term, and ωt the disturbance term.

24. This formulation, however, does not take into account short-run dynamics, in which case estimation of the long-run parameters can be biased (Banerjee et al, 1986). Therefore, following the Stock (1987) and Pesaran and Shin (1998) reparametrization techniques, model (2) has been respecified as a vector error correction model (VECM):

Δ⁢yt=c+δ⁢yt−1+θ′⁢xt−1+∑i=1p−1αi⁢Δ⁢yt−i+∑i=0n−1ϕ′⁢Δ⁢xt−i+ωt(3)

25. The estimation of the parameters using model (3) minimizes the collinearity among regressors and therefore the standard errors. It also facilitates identification of possible simplifications of the relationship (Johnston and DiNardo, 1997). Using model (3), the first step of the ARDL tests for the existence of a long run-run relationship using a bound testing approach. Pesaran, Shin, and Smith (2001) propose testing (1) H0 : δ = 0 and θ = 0, which means that we cannot reject the absence of cointegration vector, against the alternative (2) Ha : either δ or θ is significantly different from zero, which implies that the hypothesis of the existence of such a relationship cannot be rejected. In this test, the ARDL uses an F-test with lower critical value, corresponding to the case where all variables are I(0) and upper critical value corresponding to the case where all variables are I(1). If the test statistic is higher than the upper bound critical value, the null of no cointegration is rejected; it is lower than the lower bound critical value, it is not rejected. In the second step, once existence of a cointegrating relationship is established, the ARDL(p, q) long-run model can be estimated and the appropriate lag length selected. Appendix 3 reports the F-bound tests of the existence of the long-run relationship and shows that a long-run relationship cannot be rejected. It also shows that the best model, using both the bound test and the t-test as selection criteria, is the model in the first column. Appendix 4 uses the model of the first column of Appendix 3 to test for the appropriate lag length, which is two, because all coefficients using two lags are significant and also the error term is significant and less than one.

Robustness checking: The Johansen and Engle-Granger Estimations

26. Cointegration has been tested for robustness using the Johansen and Engle-Granger techniques. The Johansen method performs a cointegration test by comparing the estimated likelihood ratios to the asymptotic critical value (Johansen, 1992), which has led to rejection of the null hypothesis of no cointegration in favor of the existence of one cointegrating vector; this implies that the existence of a long-run relationship cannot be rejected. The results of this test are reported in Appendix 5. The Log-likelihood value (159.2) and the trace statistic (24.73) are above their critical values, which implies that the existence of “at most one” cointegrating relationship cannot be rejected. Likelihood ratio tests, however, are known to be sensitive to small size sample bias; therefore, the critical value has been adjusted using the Cheung and Lai (1993) method.7

27. The Johansen method, however, could present pitfalls where the number of observations is limited.8 Therefore, further tests have been carried out using the two-step method (Engle-Granger, 1987). The first step of the Engle-Granger OLS technique is applied to the long-run equation of the REER as a regressand and its main fundamentals as regressors. The results of this first step is reported in the first column of Appendix 4 and in Table 5. These results show that all the parameters are significantly different from 0. The estimates, however, may result from a spurious regression. Therefore, the second step is carried out in Appendix 6 to test for the normality of the residuals from the long-run equation. The DF-value is -3.209, which is less than the critical value at the 5 percent confidence level. The p-value is 0.019. This implies that the existence of a cointegrating vector cannot be rejected. Tests for robustness of the long-run parameters using both the Johansen and Engle-Granger methods are carried out in the following section.

Table 3.

Niger: External Sustainability and the Real Exchange Rate

article image
Source: IMF’s Staff Calculations, 2008.
Table 4.

Niger: Competitiveness Assessment

article image
Source : World Bank, 2007-2008.
Table 5.

Niger: Dependent Variable

(In REER)

article image
In parentheses are the t statistics *** p <0.01, ** p<0.05.Source: IMF Staff calcuation, 2008.

B. Econometric Results and Interpretations

28. Appendices 3 and 4 show all the results of the econometric estimation of equation (3), including different variables and number of lags,9 The following model is selected because all of its variables are significant and economically congruent, and according to its efficient error correction term, F-bound, test and information criteria:

R⁢E⁢E⁢Rt=α+β1⁢T⁢O⁢Tt+β2⁢G⁢C⁢G⁢D⁢P+β3⁢P⁢R⁢O⁢D+εt(4)

Thus, variables such as openness, aid net of reserves, net foreign income, and real resource balance are dropped from this preferred model.

29. When the robustness of model (4) to alternative estimation techniques is tested using OLS and Johansen methods, the estimates remain consistent and efficient, with the same expected signs as in the ARDL. Results are presented in Table 5.

30. The long-run parameters in Table 5 show that terms of trade elasticity is estimated to be in the range of 0.36 (OLS Engle-Granger) to 0.49 (Johansen), which means that with a 10 percent increase in the terms of trade, the REER would appreciate by 3.8 to 4.9 percent This estimate confirms results observed in other country studies that the income effect of the terms of trade tends to dominate the substitution effect in Niger. This result is in line with those of Roudet et al. (2007) for the WAEMU countries (2.5–3.1 percent), and the medium value of those of Elbadawi (1994) for Chile, Ghana, and India. It is also within the 0.23–0.68 range of the single-equation estimates of Chudik and Mongardini (2007).

31. A 10 percent increase in government spending is associated with an appreciation of the REER of 3.5 percent (ARDL) to 5.09 percent (Johansen). These results confirm that an increase in public spending tends to weigh more on nontradable than on tradable goods. The estimates are similar to those found in the literature: Chudick and Mongardini (2007) found parameters within a range of 0.27–0.67 for the single-equation estimations. Their coefficients for the panel estimation (0.556) are not far from the ones found through the ARDL (0.35) and Johansen (0.51) estimation. Moreover, a positive change in productivity is associated with an appreciation of the REER. A 10 percent increase in productivity would require a 3.8 percent (OLS Engle-Granger) to 4.5 percent (ARDL) appreciation.

32. The adjustment parameters presented in Table 5 strongly support cointegration; all the error correction adjustment elasticities are highly significant and less than one. The corresponding estimate for the ARDL, Johansen, and Engle-Granger estimators are -0.77, -0.75, and -0.51. These elasticities are in line with estimations such as those of Elbadawi (1994) for Chile, Ghana, and India. The error correction coefficients have then been used to derive the adjustment speed in terms of number of years required to adjust for a given exogenous shock. To eliminate 99.9 percent of an external shock, it would take 3.5 years for the ARDL model, 3.7 years for the Johansen, and 6.3 years for the Engle-Granger.

C. Calculating the Equilibrium Exchange Rate Indexes

33. The cointegrating relationships obtained by estimating the REER with their fundamentals (terms of trade, productivity, government spending) permit the computation of the ERER indices. ERERs are calculated using the long-run parameters in Table 5 and the permanent component of the fundamentals estimated using the Hoddrick-Prescott (HP) filter.

34. The computed indexes are summarized in Table 6 and reported in Appendix 7. The over- and undervaluations calculated from the different models are shown in Figures 3, 4, and 5. The three procedures (ARDL, OLS, Johansen) give similar results for the two main episodes of overvaluation: the late 1970s and the early 1990s.

Table 6.

Niger: Percentage of Overvaluation (+) or Undervaluation (-) under Three Different Estimation Methods, 1973-2007

article image
Source: IMF Staff calculation, 2008.
Figure 3.
Figure 3.

Niger: REER and Equilibrium REER (Using ARDL)

(REER in 2000=100)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Calculation, 2008
Figure 4.
Figure 4.

Niger: REER and Equilibrium REER by Econometric Method

(REER in 2000=100)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Calculation, 2008
Figure 5.
Figure 5.

Niger: REER and Equilibrium REER

(REER in 2000=100)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Calculation, 2008

35. The REER appears undervalued in 2007 in a range of 0.7–3.7 percent. The difference in the results of the WAEMU-wide analysis (2008 WAEMU regional consultation selected Issues Paper) is caused by a number of factors. On one hand, the REER has not appreciated in Niger as much as in the WAEMU because of the weight of the Nigerian naira in the effective exchange rate, and the bilateral real exchange rate with Nigeria has been stable. In addition, the equilibrium rate in Niger has appreciated recently because of improvements in the terms of trade and increases in government consumption.

V. The Macroeconomic Balance Approach

36. The MB approach to assessing exchange rates focuses primarily on whether the underlying current account, at prevailing real exchange rates, is consistent with normal or equilibrium saving-investment balances. The 2007 Decision on Bilateral Surveillance defined the underlying current account as “the current account stripped of temporary factors, such as cyclical fluctuations, temporary shocks, and adjustment lags.”

37. The MB assessment proceeds in three steps. In the first a trade-equation model is applied to calculate the underlying current account positions that would emerge, at prevailing real exchange rates, if all countries were operating at full potential output after taking out the transitory impact of lagged exchange rate changes. The second step uses a separate model to estimate an equilibrium or normal position for saving and investment balances based on the medium-term determinants of savings and investment. The third calculates how much the exchange rate has to be adjusted, other policies being unchanged, to bring the underlying current account in line with the current account norm.

38. The underlying current account can be approximated using the n-year moving average, the Hodrick Prescott-Filter, the World Economic Outlook projection method, and the IMF Consultative Group on Exchange Rate Issues (CGER) model.

39. The 5-year10 moving average applied for 1970–2007 leads to an estimation of the underlying current account balance of -8.3 percent of GDP. The 5-year lag corresponds to the time needed for the real exchange rate to get back to equilibrium as estimated in the ERER model (see above). The advantage of the n-year moving average is that it is both simple and commonly used in empirical works; its drawback is that it gives equal weight to all data points, although an author may choose to reduce the weights of outliers, as the HP filter does.

40. The HP filter applied for 1970–2007 gives an estimate of the underlying current account balance of about -8.5 percent of GDP for 2007. The filter differences the data to make them stationary and then smoothes the differenced data with an asymmetric moving average.

41. The 2008 Article IV exercise projects the average 2013–28 current account deficit at -6.4 percent, after petroleum and uranium investment spikes. This could be taken as the underlying current account because it is expected that temporary factors will die out during those years. This is realistic for Niger, where it is expected that foreign direct investment (FDI) and the impact on the current account generated by uranium and oil related investment will be completed by 2013.

42. The CGER approach estimates the underlying current account balance to be -8.1 percent of GDP. The CGER method approximates the underlying current account as it would prevail if domestic and foreign output were at potential. To calculate the current account, the method uses a model that has a standard structure: export volume depends on relative price and the level of foreign activity, and import volume is a function of relative price and the level of domestic activity.11,12

Figure 6.
Figure 6.

Niger: Underlying Current Account

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Estimates, 2008.

43. The second step is computation of the current account norm. This could be done by regressing the Niger current account balance on medium-term values of the fundamentals. However, given the limited Nigerien data, two other methods have been used: (i) the estimation obtained by Chinn and Presad (2003) for a panel of 71 countries (developed and developing) over a long period (1970–95); and (ii) the estimation of Lee et al. (CGER, 2008) for a panel of 54 developed and developing countries for 1973–2004, with a hybrid (mean group) pooled estimation. Table 8 presents the estimated coefficients from those panel regressions and the calculated current account balances for Niger.

Table 7.

Niger: Alternative Estimations of the Underlying Current Account

article image
Source: IMF Staff Estimates, 2008.
Table 8.

Niger: 2007 Current Account Norm Estimation

article image
Source: IMF Staff Estimates, 2008.

44. The current account norm based on Chinn-Presad indicates a somewhat larger deficit than the one implied by the hybrid pooled method. In the case of Niger, the Chinn-Presad method gives a norm of -9.56 percent of GDP and the hybrid pooled method a norm of -8.61 percent. The difference is due to the bigger coefficient on the government deficit and incorporation of foreign aid in the Chinn-Presad method. That method seems better suited to developing countries because the level of aid is an important determinant of the current account norm; a country that receives significant aid flows will spend it in large part on imports and therefore will have a higher current account deficit.

45. The last step in the MB approach is to use trade elasticities to compute the magnitude of the adjustment needed to eliminate the gap between the current account norm and the underlying current account and align the real exchange rate with fundamentals. Estimation of trade elasticities is based on earlier estimates on a panel of 77 countries for 1960–93 by Senhadji and Montenegro (1998), which were also used in the 2008 WAEMU Selected Issues Paper. Their study estimates that on average export elasticities with respect to price are -1.0 and to income 1.5, and import elasticities with respect to price are -1.1 and to income 1.5.

46. To summarize, the underlying current account estimated at -8.1 percent of GDP and the sustainable current account at -9.6 percent suggest a moderate undervaluation. The REER of Niger would need to appreciate by 8.1 percent to align the current account with the S-I balance that corresponds to its fundamentals, assuming other policies are unchanged. This assessment is in line with our earlier estimates using the ERER framework.

Table 9.

Niger: Impact of a Devaluation on the Current Account

article image
Source: IMF staff estimates, 2008.

Has been computed using the following formula: (EXPEL*EXP/GDP)-(IMPEL-1)*(IMP/GDP).

Figure 7.
Figure 7.

Niger: Current account norm is below the underlying current account trends which suggests a moderate undervaluation…

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Estimates, 2008.
Table 10.

Niger: Implied Results of the Macroeconomic Balance Approach

(In percent of GDP)

article image

VI. The External Sustainability Approach

47. The ES approach complements the ERER and the MB methods by focusing on the sustainability of the relationship between a country’s current account and net foreign asset (NFA) position and its current account. Considering an intertemporal budget constraint for the whole economy, the ES approach involves computing the current account balance-to-GDP ratio that would stabilize the NFA position at some assumed benchmark value. Based on the exchange rate elasticity of the current account and assuming all other policies are unchanged, the ES approach then calculates the size of the adjustment of the real exchange rate needed to close the gap between the NFA stabilizing current account and the underlying current account.

48. To compute the NFA stabilizing current account, the ES approach uses straightforward assumptions about GDP growth and inflation rates, applying the following formula:

c⁢as=g+π*⁢(1+g)(1+g)*⁢(1+π)*bs(5)

where cas is the current account norm as a share of GDP; g is the expected growth of the overall Nigerien economy, assumed to be 5.2 percent a year for 2007–16, driven mainly by investment and production in the mining sector and a sustained increase in agricultural value added (this level of growth is similar to average growth for 2000–07); π is the inflation rate expected to prevail in the medium term, proxied as the GDP deflator and is set at 2 percent in accordance with Niger’s DSA; and bs is the target value of the net foreign liabilities as a share of GDP projected in the following table.

49. The model shows that, under these assumptions, to stabilize the NFA position at the 2007 Net Foreign Asset Position (-18.9 percent of GDP) the current account would need to remain at -1.33 percent of GDP. This normative current account deficit, however, appears low if Niger were to maintain a high growth, to reduce poverty and to build basic capital infrastructure to accelerate its development process to meet the Millennium Development Goals.

50. It is preferable to set the NFA benchmark based on the expected levels of its components in the long run: FDI, portfolio investment, external official reserves, and external debt. In this case, special consideration is given to the inflows of FDI that Niger expects to receive; these are high in view of the investment expected to expand uranium production and start petroleum production. A realistic FDI projection assumes that Niger would be able to attract over the next five years annual FDI of about 10 percent of GDP, so that FDI would reach about 40 percent of GDP in about five years. Portfolio investment is expected to rise noticeably. Also, external reserves are expected to increase modestly, attaining 4 months of imports, equivalent to 10 percent of GDP. The target for external debt comes from the IMF and the World Bank debt sustainability framework, which suggests that for a country rates as a medium policy performer like Niger the NPV of debt should not exceed 40 percent of GDP. Considering that for Niger the ratio of nominal debt to the NPV of debt is 1.6, a target of 50 percent of nominal debt-to-GDP (equivalent to 30 percent of NPV of debt-to-GDP) seems reasonable, and is a significant distance from the threshold.

Table 11.

Niger: Net Foreign Asset Target

(In percent of GDP)

article image
Source: Staff Estimates, 2008.

51. Table 10 indicates that on the basis of these assumptions, a reasonable benchmark for Niger’s NFA position as a percent of GDP is -86 percent. Using equation 1, this benchmark implies a current account deficit of 6.1 percent of GDP. This should be compared with the current account deficit that will prevail after the spike in 2009–12 due to investment in mining and petroleum. The path of the current account deficit in the next 20 years is presented in Figure 8. After the spike in 2013–28 the current account deficit averages 6.4 percent—very slightly above the level consistent with stabilizing the NFA position at its benchmark level.

Figure 8.
Figure 8.

Niger: Average Current Account Deficit by Sub-period

(% of GDP)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: IMF Staff Estimates

52. The small discrepancy between the underlying current account and that which stabilizes the NFA position at the benchmark level would suggest a small overvaluation of 1.9 percent of the REER. However, given the margin of error of these estimates, we conclude that the exchange rate is consistent with a sustainable NFAP position.

VII. Non-Exchange-Rate Determinants of Competitiveness

53. Up until the last decade, Niger’s competitiveness was severely eroded by an unstable political environment, economic instability, and too large a role of the state in economic activity. The business environment improved after political stabilization and macroeconomic reforms in the late 1990s, which included trade liberalization, elimination of price controls, and a general disengagement of the state from productive sectors. The progress over the last decade is reflected in the significant improvement of Niger’s Index of Economic Freedom (IEF) (Figure 9), especially in the fiscal and trade areas.13 With respect to comparator groups, Niger currently stands out for its high IEF scores on monetary freedom, which recognize the prudent monetary policy of the regional central bank and lack of price controls (see Table 12). Diagnostic studies like the World Bank’s Investment Climate Assessment (ICA) and the Diagnostic Trade Integration Study (World Bank, forthcoming) also recognize the progress Niger has made to stabilize its economy and positive effect of that on competitiveness. Indeed, largely as a result of stability and liberalization reforms, the investment rate in Niger jumped from an average of 12.5 percent of GDP in 1995–2004 to 22 percent in 2005–07.

Figure 9.
Figure 9.

Niger: Index of Economic Freedom for Niger: 1998/2008*

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

* 0-100 Index, 100=most freedomSource: Holmes (2007)
Table 12.

Niger: Sensitivity Analysis

article image
Source: IMF Staff Estimates, 2008.

54. However, there is plenty of room to improve the business environment and competitiveness in Niger. Regulation in general is particularly cumbersome and imposes an unnecessary burden on the private sector. Considering mostly regulatory issues, the World Bank’s report on Doing Business (DB) lists Niger as one of the countries where business activities are most costly (Table 13). Although the country has raised its ranking in the last couple of years, it is still 169th of the 178 countries ranked. In addition, company managers surveyed in the ICA argued that regulation is not consistent, and stated that they spend nearly 15 percent of their time resolving administrative problems.

Table 13.

Niger: Index of Economic Freedom Score Averages for 2007 1 Niger and Comparator Groups

(Excluding Niger)

article image
Notes:

Scores for 165 countries. Scores range from 0 to 100, with 100 implying greatest economic freedom.

Defined as Sub-Saharan countries with at least 6% real GDP growth for 2006 - 2007 and are non-oil, non-island and not post conflict. For 2007, countries included are: Burkina Faso, Ethiopia, Ghana, Kenya, Mozambique, Tanzania, Uganda and Zambia.

Includes African and non-African countries.

Includes Burkina Faso, Chad and Mali.

55. Basic bureaucratic procedures have traditionally been lengthy and costly, thus limiting the flexibility of economic agents and imposing a barrier to the much-needed formalization of the economy. Not surprisingly, the DB ranks Niger far below comparators in all three areas of starting a business, dealing with licenses, and closing a business. Recently, though, the government has been taking action to speed up procedures for establishing a business.14

56. Diagnostic studies further identify rigid labor market legislation in Niger as an area that keeps the country at a competitive disadvantage. The DB classifies Niger as one of the countries in which it is hardest to employ workers, much harder than in comparators in sub-Saharan Africa. Based on the DB ranking, the IEF gives a very low score to Niger for labor freedom.15 Furthermore, the ICA notes, there is an absence of an educated work force; poor on-the-job professional training programs; and significant losses because of the ill health of members of the labor force (Table 14).

Table 14.

Niger: Doing Business Ranking Averages for 2007 1 Niger and Comparator Groups

(Excluding Niger)

article image
Notes:

Rankings among 178 countries.

Defined as Sub-Saharan countries with at least 6% real GDP growth for 2006 - 2007 and are non-oil, non-island and not post conflict. For 2007, countries included are: Burkina Faso, Ethiopia, Ghana, Kenya, Mozambique, Tanzania, Uganda and Zambia.

Includes African and non-African countries.

Includes Burkina Faso, Chad and Mali.

57. Firms also have major difficulties in accessing capital. The ICA finds that most firms cannot meet the basic requirements for a loan or an overdraft,16 And the DB rankings imply that Nigerien firms face more difficulties in getting credit than firms in all comparators,, essentially because of poor legal rights and scarce credit information. However, it is promising that credit to the economy has increased by an annual average of 24 percent since 2004, if from a very low base.

58. Particularly disappointing for a landlocked country like Niger is the fact that firms face obstacles in trading across borders even beyond those naturally imposed by geography. Mainly for reasons related to its landlocked location (i.e., time and costs to deliver goods to major international markets), the DB lists Niger among the countries with highest obstacles to engaging in world trade. Yet, it also ranks Niger low in this area for the policy-related reason that a large number of documents are needed for both exporting and importing. Moreover, the ICA highlights the costs imposed by the inadequate transport infrastructure in Niger: 41 of the firms surveyed considered transportation-related problems to be major issues hindering productivity. The government thus urgently needs to simplify international trade procedures and build up the infrastructure network to at least partly compensate for the natural impediments to international trade.

59. Competitiveness is also hindered by the cost of “informal payments” to ease business transactions. The IEF ranks Niger near the bottom in terms of freedom from corruption, with an index below the average in low-income countries and in comparators in sub-Saharan Africa. The ICA found the cost of informal payments to be high, ranging from 4.1 to 13.2 percent of a firm’s yearly turn-over depending on the sector. About 58 percent of entrepreneurs responding to ICA surveys cited corruption as a serious concern.

60. It thus appears that making the Nigerien economy more competitive requires reforms to ease regulatory procedures, make labor legislation more flexible, improve technical training, facilitate access to finance, build up infrastructure, and combat corruption. Accomplishing reform in turn requires sustained commitment from the government, recognizing that results may only be perceived in the long run. Nevertheless, progress in some areas can be dramatic, as illustrated by the improvement of Niger’s DB ranking on registration of property, where it climbed 40 positions between 2007 and 2008 by significantly reducing the duration and cost of registration. Besides the obvious benefits of such measures on competitiveness and the business environment, improvements in Niger’s performance on cross-country rankings is important per se because the lists affect the views of international investors considering doing business in Niger.

Table 15.

Niger: Major or Very Severe Constraints to Firm’s Growth According to the Investment Climate Assessment

article image
NB: Exclusively for firms in Manufacturing.Source: Niger’s Investment Climate Assessment (2008).

VIII. Productivity and Wages

61. Productivity is the main determinant of a nation’s competitiveness and the main source for improvement of its population living standards. Therefore, an analysis of the main driving forces of productivity is important for formulating and implementing welfare-enhancing policies. This section analyzes the evolution of productivity and unit labor costs in the main sectors of the Nigerien economy, using Senegalese productivity as a comparison.

62. Data on annual wages, employment, and the value of the modern sector have been computed using the Institut National de la Statistique du Niger database from 1998 to 2004 for mining, manufacturing, energy, construction, commerce and hospitality, transportation and services. Productivity of sector i (PRODi) is approximated as value added at current prices (VAi) per worker in that sector (WORKi), and the unit labor cost at current prices in sector i (ULCi) as the ratio of salaries at current prices (SAi) to productivity per worker at current prices in each sector (PRODi):

PRODi=VAiWORKiULCi=SALiPRODi

63. These indices, presented in Figure 10, indicate that from 1998 to 2000, productivity decreased in all sectors. Since 2000, productivity has improved in all sectors except services. Value added per worker rose considerably in manufacturing, energy, transportation, and commerce, declined slightly in mining, and declined more in services; unit labor cost also declined in mining as salary per worker fell.

Figure 10.
Figure 10.

Niger: Evolution of Unit Labor Costs and Productivity Indexes, 1998-2004

(1998 = 100)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: Nigerien authorities; and IMF staff calculations.

64. A comparison with Senegalese data from the World Bank 2006 report indicates that in most sectors, except services, construction, and commerce, productivity has increased faster in Niger than in Senegal with an accompanying improvement in unit labor cost. Even in the three exceptions, unit labor costs increased less in Niger than in Senegal.

65. Despite improvements in productivity and unit labor cost since 2000, wage costs are high in Niger. World Bank data based on enterprises surveys compare monthly wages (in US dollars) for unskilled workers in the 2000s across a number of African and non-African countries; they show that wages for unskilled worker in Niger, at US$88 a month in 2005 are 10 percent below those of Kenya, which has a much higher per capita income, but higher than in Nigeria, Benin, Mali, Uganda, and Tanzania (Figure 12). This confirms that further efforts are needed to improve productivity, in order to enhance overall competitiveness.

Figure 11.
Figure 11.

Niger and Senegal: Evolution of Unit Labor Costs and Productivity Indexes 1998-2004

(1998=100)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: Nigerien and Senegalese authorities and IMF staff calculations.
Figure 12.
Figure 12.

Niger: Monthly Wages for Unskilled Workers in the Manufacturing Sector

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: World Bank, Enterprise Surveys 2003-06.
uA01fig01

Evolution of Unit Labor Costs and Productivity Indexes 1998-2004

(1998 = 100)

Citation: IMF Staff Country Reports 2009, 070; 10.5089/9781451828832.002.A001

Source: Nigerien authorities and IMF staff calculations.

Appendix Tables

Table 1.

Niger: Unit Root Test for REER Fundamentals: Augmented Dickey Fuller

article image
* Critical values: 1%: -5.43, 5%: -4.80.
Table 2.

Niger: ARDL on Base and Alternative Specifications

(Using Two Lags)

article image
Z statistics in parentheses.*** p<0.01, ** p<0.05, *p<0.1.

Upper level bound for F-statistic is 6.36 (Pesaran and others, 2001).

Table 3.

Niger: ARDL on Base Specification with Different Number of Lags

article image
Z statistics in parentheses.*** p<0.01, ** p<0.05, * p<0.1.

Upper level bounds are 6.36 for two lags, 7.84 for one lag, 5.61 for three lags, and 5.06 for four lags.

Table 4.

Niger: Johansen Cointegration

article image

The Trace Statistic Test implies that the hypothesis of “at most one” cointegration vector is not rejected.

Table 5.

Niger: Engle-Granger Estimation: Test for Normality of the Error Correction Term

article image
Table 6.

Niger: Estimates of Equilibrium REER and Overvaluation

article image
Source: IMF staff calculation, 2008.
Table 7.

Niger. Underlying Current Account Balance, 2007

article image
Source: Staff calculations, 2008.

References

  • Adam, C., 1992, “Recent Developments n Econometric Methods: An Application to the Demand for Money in Kenya,” Special Paper Fifteen (Nairobi: African Economic Research Consortium).

    • Search Google Scholar
    • Export Citation

  • Azam, Jean P., 1999, “The Uranium Boom in Niger, 1975–1982,” in Trade Shock in Developing Countries, Vol. 1: Africa, ed. by Paul Collier., Jean Willem Gunning., and Associates (New York: Oxford University Press).

    • Search Google Scholar
    • Export Citation

  • Baffes, J., I. Elbadawi, and S. O'Connell, in Part III of Hinkle and Montiel, eds., 1997, “Single Equation of the Equilibrium Real Exchange Rate,” Paper No. 2.

    • Search Google Scholar
    • Export Citation

  • Balassa, Bella, 1964, “The Purchasing Power Parity Doctrine: A Reappraisal,” Journal of Political Economy, Vol. 72, pp. 584–96.

  • Banerjee, A., J. Donaldo, J.W. Galbraith, and D. Hendry, 1993, Co-integration, Error-Correction, and the Econometric Analysis of Non-Stationary Data (New York: Oxford University Press).

    • Search Google Scholar
    • Export Citation

  • Banerjee, A., J.J. Dolado, D.F. Hendry, and G. W. Smith, 1986, “Exploring Equilibrium Relationships in Econometrics Through Static Models: Some Monte Carlo Evidence,” Oxford Bulletin of Economics and Statistics, Vol. 48, pp. 253–77.

    • Search Google Scholar
    • Export Citation

  • Cassel, G., 2008, Foreign Investments, Harris Foundation Lectures (Chicago: University of Chicago Press).

  • Lee, J., G.M. Milesi-Ferretti, J. Ostry, A. Prati, and L.A. Ricci, Exchange Rate Assessment : CGER Methodologies, Occasional Paper No. 261 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Cheung, Yin W., D.C. Menzie, and F. Eiji, 2007, “The Overvaluation of Renminbi Undervaluation,” NBER Working Paper No. 12850 (Cambridge, MA: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

  • Cheung, Yin W., and S. Lai Kon, 1993, “Finite Sample Sizes of Johansen’s Likelihood Ratio Tests for Cointegration,” Oxford Bulletin of Economics and Statistics, Vol. 55, pp. 313–28.

    • Search Google Scholar
    • Export Citation

  • Chinn M. and E. Prasad, 2003, “Medium-Term Determinants of Current Accounts in Industrial and Developing Countries: An Empirical Exploration,” Journal of International Economics, Vol. 9, pp. 47–76.

    • Search Google Scholar
    • Export Citation

  • Chudick, A., and J. Mongardini, 2007, “In Search of Equilibrium: Estimating Equilibrium Real Exchange Rates in Sub-Saharan African Countries, IMF Working Paper WP/07/90 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Cline, R., and J. Williamson, 2007, “Estimates of the Equilibrium Exchange Rate of the Renminbi: Is There a Consensus? If Not, Why Not? Peterson Institute for International Economics Paper (Washington: Peterson Institute).

    • Search Google Scholar
    • Export Citation

  • Cline, William R, 2005, The United States as a Debtor Nation (Washington: Institute for International Economics and Center for Global Development).

    • Search Google Scholar
    • Export Citation

  • Cline, William R, 2007, “Estimating Reference Exchange Rates,” Paper presented to a Workshop at the Peterson Institute sponsored by Brueghel, KIEP, and the Peterson Institute (Washington: Peterson Institute).

    • Search Google Scholar
    • Export Citation

  • Coudert, V., and C. Cécile, 2005, “Real Equilibrium Exchange Rates in China,” CEPII Working Paper No. 9971. Paris.

  • Di Bella, G., M. Lewis, and A. Marin, 2007, “Assessing Competitiveness and Real Exchange Rate Misalignment in Low-Income Countries,” IMF Working Paper WP/07/201 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

  • Dickey, D. and S. and Pantula, 1987, “Determining the Order of Differencing in Autoregressive Processes,” Journal of Business and Economic Statistics, 15, 445–461.

    • Search Google Scholar
    • Export Citation

  • Edwards, S., 1989, Real Exchange Rates, Devaluation, and Adjustments: Exchange Rate Policy in Developing Countries (Cambridge, MA: MIT Press).

    • Search Google Scholar
    • Export Citation

  • Elbadawi, I.A., 1994, “Estimating Long-run Equilibrium Real Exchange Rates.” In Estimating Equilibrium Exchange Rates, ed. by John Williamson (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

  • Elbadawi, I.A., L. Kaltani, and K. Schmidt-Hebbel: “Foreign Aid, the Real Exchange Rate, and Economic Growth in the Aftermath of Civil Wars”, The World Bank Economic Review, 22 (1): 113–140. 2008.

    • Search Google Scholar
    • Export Citation

  • Engle, R.F., and C.W.J. Granger, 1987, “Cointegration and Error Correction: Representation, Estimation, and Testing,” Econometrica, Vol. 55, pp. 251–76.

    • Search Google Scholar
    • Export Citation

  • Frankel, Jeffrey, 2004, “On the Renminbi: The Choice Between Adjustment Under a Fixed Exchange Rate and Adjustment Under a Flexible Rate,” written for a High-Level Seminar on Foreign Exchange System, Dalian, China, May.

    • Search Google Scholar
    • Export Citation

  • Goldstein, M., and Nicolas Lardy, 2007, “China’s Exchange Rate Policy: An Overview of Some Key Issues,” paper prepared for the conference on China’s Exchange Rate Policy, Peterson Institute for International Economics, Washington, October 19.

    • Search Google Scholar
    • Export Citation

  • Harrod, Roy, 1933, International Economics (Cambridge, UK: Cambridge University Press).

  • Heston, Alan, Robert Summers, and Bettina Aten, 2006, Penn World Table Version 6.2, Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania, September 2006 (http://pwt.econ.upenn.edu/php_site/pwt_index.php).

    • Search Google Scholar
    • Export Citation

  • International Monetary Fund, 2006, “Methodology for CGER Exchange Rate Assessments” (Washington: IMF).