DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ 4. Balassa-samuelson effect: exploring the case of a sample of African countries Elhadj EZZAHID, Enseignant chercheur en économie, FSJES/LEA, Université Mohammed V de Rabat Abstract The main goal of this paper is to test the Balassa-Samuelson effect in the case of Kenya, Morocco, Nigeria, Senegal, South Africa, and Tanzania. That is we explore the contribution of intercountry differentials in tradable to nontradables sectors productivity (dual productivity gap) to the changes in the real exchange rate. In the way, we used the data of the International Comparison Program of prices conducted by the World Bank to test the Penn effect in the world and in Africa for the 2011 year. The results provide convincing support for the idea that prices are positively linked to the levels of per capita GDP in the World and Africa. The internal version of the BSE is valid in Kenya, Nigeria, and in South Africa. It is rejected in Senegal and in Tanzania. In Morocco, the internal BSE is insignificantly rejected. The use of per capita GDP as measures of productivity show that the BSE is valid in the cases of Kenya, Senegal, and Tanzania while in Morocco, Nigeria, and South Africa, this effect is not observed. The use of a disaggregated measure of productivity shows that only in Morocco a BSE is probably operating. Globally, the empirical evidence about the existence of BSE in our sample of countries provide conflicting results. Key words: Balassa-Samuelson hypothesis, Real Exchange rate, Tradable and nontradables sectors, dual productivity gap JEL Classification: F3, F11, F37 Effect Balassa-Samuelson : Etude de cas de six pays africains Résumé L’objectif de ce papier est de voir s’il existe ou non un effet de Balassa-Samuelson dans un échantillon de pays Africains. Cet échantillon comprend Kenya, le Maroc, le Nigeria, le Sénégal, l’Afrique du Sud et la Tanzanie. Nous explorons le lien entre la différence entre le ratio de la productivité des secteurs produisant les biens échangeables à la productivité des EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 367 PARTIE III secteurs produisant les biens non échangeables et le même ratio aux Etats-Unis d’Amérique dans cet échantillon de pays avec les taux de change réel. Trois résultats sont à signaler. Premièrement, les prix dans le monde et en Afrique sont corrélés positivement avec les niveaux de richesse. Deuxièmement, quand la productivité des échangeables augmente plus vite que la productivité des non-échangeables, les prix de ces derniers biens augmentent plus vite que les prix des échangeables dans les pays suivants : Kenya, Nigeria et l’Afrique du Sud. Les résultats concernant l’existence de la version externe de l’effet BS sont non systématiques. Mots-clés : Modèle de Balassa-Samuelson, Taux de change réel, Secteurs des biens échangeables et des biens non échangeables, Différentiel dual de la productivité 368 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ Introduction Many emerging economies base their development strategy on exports’ promotion policy. Therefore, their trade competitiveness relies partially on a not appreciated currency. Indeed, literature documents widely the fact that a devalued currency is linked with more exports and, henceforth, with more growth (Connolly, 1983; Rodrik, 2008). Misalignment of exchange rate is also potentially instrumental in exports’ diversification (Sekkat, 2015). Furthermore, exchange rate dynamics are also at the heart of financial instability as illustrated by the 1998 Asian crisis. Numerous factors may be at the origin of a currency appreciation such as higher capital inflows or a spur in current revenues (Jonglez, 2008). One of the most debated sources of currency appreciation is the famous Balassa-Samuelson effect due to the fact that productivity differential between tradable and non-tradable sectors is higher in the domestic developing country compared to its developed partner. This effect is a supply-side phenomenon and is frequently labeled productivity bias model. Empirical literature documented a significant contribution of the Balassa-Samuelson effect in the appreciation of the currencies of many developing countries. Consequently, policymakers have to assess the extent of this source of currency appreciation and develop adequate remedies in order to enhance the competitiveness of their economies. African economies are engaged in a catch up process. Their integration to World markets and the improvement of their population’s standard of living depend on the development of competitive exporting sectors. Managing their currencies far from becoming overvalued help these countries to preserve their competitiveness. The sources of the dynamics of African countries currencies is an insufficiently explored topic in economic literature. The objective of this paper is to explore the extent of Balassa-Samuelson effect in a sample of African economies namely Kenya, Morocco, Nigeria, Senegal, South Africa, and Tanzania. The paper proceeds as follows. Section 2 presents the Balassa-Samuelson effect. Section 3 summarizes some empirical papers about this topic. Section 4 contains a test of the Balassa-Samuelson effect in a sample of African countries. The last section is for the concluding remarks. EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 369 PARTIE III I. Balassa-Samuelson effect : a simple formalization 1. An enduring stylized fact The Balassa-Samuelson hypothesis constitutes an explanation of a will documented stylized fact in international economics. Contrary to what predict the PPP theory, prices translated to a unique currency are higher in developed countries compared to prices in less rich countries. This is clear from plotting (Figure 1) the log of price level196 on the log of per capita GDP in the World or in African countries using the ICP 2011 data (World Bank, 2015). The data are for the year 2011 and the price level is normalized by the price in the World. Figure 1. Price level vs per capita expenditure (World left and Africa right ) Source: Author’s elaboration, data from ICP 2011 (World Bank) Consequently, we observe that developing countries’ currencies tend to appreciate when, during the catch up process, they grow faster compared to the frontier economy (generally the U.S.A). Japan was a good example due to the appreciation of the Japanese Yen in the post war period (Ito, 1996). This stylized fact appears also when we compare market exchange rate to PPP exchange rates in developing and developed countries. In the later countries, market exchange rates are commonly overvalued, and undervalued in the former. This stylized fact is observed 196  The price level in country i compared to price level in country j is nothing but the ratio of purchasing- power parity of country i currency to its market exchange rate (Heston et al., 1994). Let the Dinar (D) and the Rupee (R) be the currencies of countries i and j. are prices in the two countries in their respective currencies. Suppose that the Dinar is indirectly quoted so 1 R=E*D. The price in country I in terms of Rupees is 370 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ also if we scrutinize the Big Mac index. The currencies of developing countries (Russia, India, Indonesia…) are usually undervalued and those of developed countries (Norway, Switzerland…) are overvalued (The Economist, 26 July 2014, p. 48). Figure 2. ln(EPPP/E) vs ln(per capita expenditure at market exchange rate) Source: Author’s elaboration, data from ICP 2011 (World Bank) In the following, we present the Balassa-Samuelson model which is an established explanation of why prices in developing countries tends to increase and, henceforth, their currencies tend to appreciate. 2. The model Our setup contains two countries: the home country and the foreign one. Let E be the number of domestic (home) currency units necessary to obtain one unit of the foreign currency. E is the nominal or market exchange rate. P and P* are prices’ indices in the domestic and in the foreign countries respectively. The subscripts T and N are for tradeable and non-tradable goods. PT and PN (respectively and ) are the prices of tradeable goods and non-tradable goods in the domestic country (respectively foreign country). The real exchange rate is the price of 197 a representative basket of goods in the foreign country in terms of its price in the domestic currency. Formally, the real exchange rate Q is equal to: (1) When Q increases (decreases) the domestic currency depreciates (appreciates) in real terms. Real depreciation of the domestic currency improves the competitiveness of the domestic country. Henceforth, consumers, whenever they are, will prefer to buy domestic products. The 197  Variables with asterisk are for the foreign economy EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 371 PARTIE III real appreciation of domestic currency may be due to a decrease of E (nominal appreciation) or to an increase of P* that is below the increase of P; that is dlnP*< dlnP. Thus, when inflationary pressures in the domestic country are stronger than what is going on in the foreign one, then the domestic currency appreciates. B. Balassa (1964) and P.A. Samuelson (1964) formulated a model linking real exchange rate dynamics with productivity differentials between the foreign economy and the domestic one. The Balassa-Samuelson model relies on four main assumptions. First, the domestic economy is small and open and consists of two sectors. One sector produces tradeable goods and the other sector produces non-tradable goods. The smallness of the domestic economy implies that the price of tradeable goods is exogenous. That means that the price of tradables is determined on World markets. The second hypothesis is about mobility of productive factors. Labor and capital are perfectly mobile within each economy. Labor is immobile between the two economies, whereas capital is perfectly mobile between economies. Therefore, the real rate of return to capital r is the same in the two countries. The third hypothesis states that prices follow wages. Wages in their turn follow productivity. Firms increase prices when wages increase in order to preserve their margins. The fourth hypothesis is the validity of PPP theory in the tradeable goods. That is, . This results from the fact that the law of one price holds in the tradeable sector. In other words, the price of tradeable goods is determined in international markets; whilst the price of non-tradable goods is determined within each country. In each country, firms use a Coob-Douglass technology to produce tradable (T) and nontradable (N) goods as follows: Producers in each sector maximize their profits. So, they use inputs (L and K) until their respective marginal revenues equal nominal wages wT and wN and interest rate r. The interest rate is determined in international markets given the perfect mobility of capital between countries. Remark that perfect inter-sectoral mobility of labor in each economy equalizes wages in the two sectors, i.e. wT=wN=w. Focusing on the domestic economy, the optimality conditions in using labor and capital implies by firms that: 372 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ Introducing logarithms and first differences in equations 6 through 9 and rearranging them allows to formulate proportional growth rate of the relative price of nontraded goods compared to traded goods as function of the differential between proportionate growth rate of total factors productivities in traded and non traded sectors (respectively aT and aN). We denote the proportional growth rate of a variable X by the lower case letter x198. So we will have: This equation implies that the ratio of non-tradable goods price to tradable goods price increases in the case of a faster productivity growth in the traded goods sector compared to productivity growth in nontraded goods sector. Naturally, if the production functions in traded and nontraded sectors in the foreign country have the same forms as their counterparts in the home country, we will have an analogous relationship in the foreign country, i.e.: The differential of the growth rates of nontradables and tradables’ prices in the domestic country compared to what is in the foreign country is: (10) We call the LHS of the last equation the dual inflation gap and the RHS the dual productivity gap. All else equals, the relative price of nontradables increases more rapidly in the home country if tradables’ productivity in the home country increases faster than tradables’ productivity in the foreign country. We turn now back to the real exchange rate to track the possible implications of equation 10 on its dynamics. The real exchange rate is Q=EP*/P. Let us rewrite it in proportional rates of growth terms to get q=e + p* - p. The price index P in each country is a geometric mean of traded goods and nontraded goods price indices. Thus, we have in the home country: . The growth rates equivalent version of this formula is p=δpT+(1- δ)pN. For the foreign country we have a symmetric formula and thus the growth rates version of the foreign price index is: . We can now rewrite the proportional growth rate of Q by introducing the last two equations to get after some rearrangements this decomposition: (11) We replace now the relative prices by their expressions found earlier that link in each country inter-sector differentials of prices and productivity pN-pT=β/αaT-aN and . 198  Let X be a variable in level and denote by x its rate of growth; we use this approximation: xt=lnXt-lnXt-1. EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 373 PARTIE III (12) It is assumed that purchasing power parity is valid in tradable goods, thus EP*T=PT implying that e + p*T - pT = 0 and consequently we have: (12bis) To get this equation that links intercountries inflation differential to the difference between the differential of tradable and nontradables sectors productivities in domestic country and the same differential in the foreign country. (13) This equation states that domestic to foreign country inflation differential depends on the rate of depreciation of nominal exchange rate and on inter-countries differential of tradables to non tradables sectors productivities. If in the domestic country, tradable sector productivity increases faster than non-tradables sector productivity compared to what is going on in the foreign country then prices at the domestic country increases faster than prices at the foreign country. This is synonymous of/leads to an appreciation of the domestic currency. II. Review of empirical literature In this section, we will review a limited sample of empirical literature about the Balassa- Samuelson effect in some countries. The findings are not systematic due to diversity of studied cases, of used datasets and variables, and of methods mobilized. Indeed, the BSE test produces mixed findings. J. R. Faria and M. Leon-Ledesma (2000) used the approach of bounds testing, developed by Pesaran, Shin, and Smith in their paper of 1999, to investigate the presence of the BSE. The dataset contains quarterly data and covers the period 1960: 1-1996: 4 and concerns Germany, Japan, United Kingdom, and the U.S. The two variables used are relative prices and productivity. The first variable is measured by the GDP deflator and the second variable is measured by the real per capita GDP (the domestic currency is used here). The main finding of the authors is the absence of evidence about the presence of the BSE. Paradoxically, the authors warn that “although the null of the Balassa-Samuelson effect is rejected, this does not mean that we can accept the possible alternative of PPP. In fact, the real exchange rate seems to have a long run impact on relative growth rate” (Faria and Leon-Ledesma, 2000, p. 10). This empirical evidence is not compatible with the fact that when exchange rate is determined as postulated by 374 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ the PPP doctrine then it does not affect the real economy. The potential presence of BSE in the case of central European countries is extensively explored. One of the studies in this vein is of Mihaljek and Klau (2003). The authors investigated the case of Croatia, Czech Republic, Hungary, Poland, Slovakia, and Slovenia. Data are for periods beginning in different dates in the first half of the 1990s and ending in 2001. “The paper finds clear evidence of the Balassa-Samuelson effect – in both its international and domestic versions - in all six countries. However the size of the effect is found to be relatively small” (Mihaljek and Klau, 2003, p. 1). E. U. Choudhri and M. S. Khan (2005) explored the case of a set of 14 developing countries and 2 new industrialized countries using a time-series data spanning the period 1976-1994 and panel econometrics. The reference country is the United States. The authors focus on long-run effects in a multicountry framework and found “strong evidence that the Balassa-Samuelson mechanism operates in developing countries” (Choudhri and Khan, 2005, p.  391). Indeed, the differential in productivity between traded and nontraded sectors exerts a significant positive effect on the relative price of nontradables. This induce an appreciation of these developing countries’ currencies. The paper of J. Lee and M.-K. Tang (2003) is devoted to the investigation of the links between productivity and real exchange rate dynamics using data about 12 OECD countries. The used econometric technique is panel cointegration technique. The authors found many results that worth to be highlighted. They found in each country a positive relationship between the relative prices of non-tradables to tradables and relative productivity. This result is confirmed using either labor productivity or total factor productivity. Concerning the links between inter- country tradables to nontradables productivity differentials and exchange rate, the authors find that “when productivity is measured by labor productivity, the correlation between the exchange rate and the productivity differential remains positive, largely working through wage differentials. When total factors productivity (TFP) is used to measure productivity, however, the correlation between the exchange rate and the productivity differential is not statistically significant, and their signs are, at times, negative” (Lee and Tang, p. 4). T. A. Pletonen and M. Sager (2009) explored some issues related to the links between productivity and the dynamics of real exchange rate in a panel of 68 countries. The data cover the period 1990-2004. The authors findings “provide only moderate support for Balassa – Samuelson” hypothesis (2009, p. 9). Furthermore, the results offer evidence in favor of the existence of important differences “in the relationship between real exchange rates and productivity differentials between currencies under fixed and floating exchange rate regimes” (2009, p. 6). The nominal convergence of the Central and Eastern European Countries (CEEC) to the conditions prevailing in member states of the European Union was the goal of CEEC negotiating or planning to do so in order to join the Union. The existence of a Balassa-Samuelson effect is EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 375 PARTIE III more likely given that the CEEC countries are expected to experiment a catch up to the standards levels of EU countries especially via the conduit of faster productivity increases. V.  Coudert (2004) surveyed a sample of empirical literature related to this question. A main conclusion from this sample of literature is the sensitivity of the findings to the specification used and to the assumptions about the parameters such as the part of nontraded goods in the overall index of prices. K. Lopcu and al. (2013) challenged the analysis of the Turkish Central Bank (CBRT) about the substantial role of Balassa-Samuelson effect in explaining the real effective appreciation of the Turkish Lira. The authors’ data cover the period 1990: Q1-2011: Q2. The BSE is tested using cointegration’ technique and allowing for multiple breaks in the dynamics of variables. Operationally, the explicit objective of the authors is to measure the extent to which the appreciation of the Turkish lira in the 2000s, is due to the relative productivity differentials recorded in favor of Turkey. One innovation of this paper is that “seven models related to the B-S Hypothesis are constituted by adding other variables such as net foreign assets and real interest rate differentials” (Lopcu et al., 2013, p. 617). The authors’ results provide little support to the analysis of the CBRT concerning a substantial contribution of relative productivity differential in explaining the appreciation of the Turkish currency. M.D. Chinn (2000) examined the links between real exchanges rates and productivity in 9 Asian countries namely China, Indonesia, Japan, Korea, Malaysia, Philippines, Singapore, Taiwan, and Thailand over the period 1970-1992. When the author used time-series regressions, he found that productivity ratios positively and significantly affects exchange rate in Japan, Malaysia, and the Philippines. This result corroborates the Balassa-Samuelson hypothesis. It is worth to note that government spending and terms of trade are not important factors in the dynamics of the exchange rate. Contrary to what is predicted by the so called Linder-demand hypothesis, M.D. Chinn (2000) found that the growing “preference toward services is not at the heart of the secular appreciation in most East Asian real exchange rates” (p. 40) M. Zakaria and E. Ahmad (2009) explored the ties between productivity shocks and movements of the Pakistanis Rupee exchange rate against a bundle of 16 currencies. These currencies are those of the major Pakistanis trading partners. The data are quarterly time series and the authors focus on the period 1983-Q1/2006-Q4. The authors regress, for each commercial partner of Pakistan, the nominal exchange rate growth rate on the relative price growth rates and on domestic and foreign productivity growth differential. “The results predict a close relationship between nominal exchange rates, relative price differentials and, relative domestic and foreign productivity differentials” (p. 186). They conclude that productivity differential translates to the real exchange rate through two channels namely: nominal exchange rate and inflation rates (p. 186). M. Bahmani-Oskooee (1995) focused on the dynamics of the Iranian Rial using data for the period 1960-1990. He explored the links between real exchange rate and productivity ratio. The 376 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ study is devised to test the hypothesis of productivity bias in the case of Iran with its seven major commercial partners in this period. The author used cointegration and ECM methodology to capture simultaneously short-run and long-run links between real exchange rate and productivity ratio. The author documented a robust and positive influence of relative productivity on the real value of the Iranian Rial. D. Steenkamp (2013) tested the existence of the Balassa-Samuelson effect in the case of New-Zealand with its major economic partners namely Japan, Australia, USA, and United Kingdom using annual data for the period 1977-2006. The available data shows that New Zealand’s productivity declined whereas its real effective exchange rate was trendless. The internal version of the BSE asserts that the tradable to nontradables productivity differential is positively linked to the differential of nontradables to tradables price. The author was not able to document the existence of either the internal or the international versions of the BSE. Given the fact that the literature about this issue is huge, it is impossible to summarize a representative sample of this literature. Globally, the studies investigating the validity of the Balassa-Samuelson hypothesis differs markedly in the used methods of estimation (OLS or its variants, GMM, panel data methods…), the nature of data used (time-series, cross section, panel data), and the privileged reduced form specifications to be tested to infirm or confirm the BSH. III. Testing Balassa-Samuelson effect in Kenya, Morocco, Nigeria, Senegal, South Africa, and Tanzania The Balassa-Samuelson hypothesis199 is testable using many econometric specifications. Recall that this model links the following variables: prices of traded-goods, prices of nontraded goods, productivity in the traded goods sector, Productivity in the nontraded goods sector200, and exchange rate. The main prediction of the Balassa-Samuelson model is the appreciation of the currency of country i with respect to the currency of country j when in country i tradable sector productivity increases faster compared to the nontradable sector productivity compared to what is going on in country j. 199  Epistemologically, we prefer to discuss about a model rather than a mere hypothesis, mechanism or effect because Balassa-Samuelson idea is an integrated model constituted around a stylized fact, some hypotheses, a proof, and many predictions. 200  If the assumption of constant returns to scale holds then marginal productivity of each factor (L or K) is proportional to its average productivity. For example, v EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 377 PARTIE III 1. Exchange rate Dynamics The six different countries, constituting our sample, present many differences due to their endowments and to what they experienced as economic policies in the last decades. The structural differences between their economies are significant. South Africa is an industrial country belonging to the BRICS. Morocco and Senegal are no-oil producing countries but have diversified economies compared to many African economies such as Algeria. Nigeria is a major oil producing country. Kenya and Tanzania are two eastern African countries. South Africa, Nigeria, and Morocco produced respectively 14.88%, 12.43%, and 5.31% of African GDP in 2011. Kenya, Tanzania, and Senegal produced respectively 2.16%, 1.74%, and 0.70% of the African GDP. These six different countries adopt different exchange rate regimes (Table 1). For more details about the evolution of the exchange regimes in these countries, it is fruitful to refer to (Ilzetzki et al., 2011). Tableau 1.Exchange regimes in the countries of our sample (2014) Country Exchange regime Senegal Conventional peg Tanzania Floating Morocco Conventional peg Nigeria managed float (dirty float) South Africa Floating Kenya Floating Source: International Monetary Fund, 2014, table 2, pp. 5-6 Temporal scrutinize of the dynamics of real exchange rate, with respect to the USD, in the six countries over the period 1970-2010 reveals contrasted evolutions. Indeed, the different currencies evolved differently over time. For Nigeria, the currency appreciated sharply over the period 1973-1984. Gelb (1988) estimated that this appreciation was about 187%. In the period 2003-2013, it seems that the South African Rand fluctuated more as response to market shocks rather to response to fundamental shifts. Indeed, higher interest rate and capital inflows resulted in a Rand exchange rate above its equilibrium level. The IMF reported that in 2010, Kenya was under a managed floating. This country passed from a fixed exchange rate in the 1960s and 1970s to a crawling peg regime in the 1980s. Since the adoption of the floating exchange rate in the beginning of the 1990s, the Kenyan currency depreciated sharply. Currently, the Moroccan dirham (MAD) is pegged to a basket including the Euro and the USD weighted accordingly to their weights in the Moroccan balance of payments operations. Senegal is member of the West African Economic and Monetary Union. The currency of this 378 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ union is the West African CFA franc. The Tanzania Shilling exchanged at a constant rate over the long period 1970-1980 due to the peg to the USD. Massive assistance inflows and huge restrictions on external transactions helped to sustain this peg (Bigsten and Danielsson, 1999). This stability ended in the beginning of the 1980s. The “Shilling price of the US dollar changed from 8.3 in 1981, to 51.7 in 1986 and to 670 in mid-1998” (Bigsten and Danielsson, 1999, p. 42). In the first decade of the millennium (2010s), the Tanzanian Shelling was overvalued and the Bank of Tanzania (BoT) frequently intervene to reduce the impact of inflows on the value of the TSH (nominal appreciation) via sterilization and other available tools. The World Bank program on international prices comparison provides valuable information about price levels and PPP based exchange rates in 2011 (World Bank, 2015). Table 2 provides for each of our six countries the price level and a measure (EPPP/E=Pd/E*PUSA) of the deviation of market exchange rate from the PPP exchange rate201 for the year 2011. Price index and the exchange rate are normalized with respect to the world price level and the USD. If this quantity is above unity then the currency is overvalued202. Data show that the Tanzanian currency is the most undervalued. The South African currency is the less undervalued. Tableau 2. Price index in African countries and EPPP/E -2011 (World price =100) Kenya Morocco Nigeria Senegal South Africa Tanzania Price index (World=100) 49.8 58.6 62.3 64.6 90 11.5 E /E PPP 0.386 0.54 0.483 0.50 0.657 0.33 Source: World Bank, 2015 2. Methodology, data and results The Balassa-Samuelson model predicts that a positive home to foreign country productivity differential (dual productivity gap or differential) produces an appreciation of the home currency. This appreciation is due to the faster increase of the price of nontraded goods compared to the price of traded goods. Currencies of emerging countries do appreciate according to this model, because their tradable sectors productivity growth exceeds the growth of the productivity of their 201  The PPP exchange rate is EPPP=P/P*. It is just the ratio of prices of a similar basket of goods in the domestic country P and in the foreign country P*. “Note that the overall PPP in 1997 was £1 = 326.3 yen, at a time when the exchange rate was £1=198.1 yen, suggesting a significantly undervalued Japanese yen” (discuss). “For example, the average price of a Big Mac in America in July 2015 was $4.79; in China it was only $2.74 at market exchange rates. So the “raw” Big Mac index says that the yuan was undervalued by 43% at that time” (the Economist, http://www.economist.com/content/big-mac-index). 202  Remark that in this logic the market or actual exchange rate is the benchmark. So if E<EPPP then the currency is overvalued. In the opposite case, the currency is undervalued. EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 379 PARTIE III nontradables sector. Analytically, the model links home to foreign country inflation differential with a weighted excess of traded goods sector productivity relative to non-traded goods sector productivity in the home country compared to the foreign country. The intercountry inflation differential is linked negatively with the real exchange rate EP*/P. We propose to test this main prediction for our sample of countries. Thus, we prefer to test these two reduced form specifications: PNt / PTt = α0 + α1 (yTt /yNt ) + εt internal version of BSE RERdt = β 0 + β 1 dualproductivitygap +ut external version of BSE We expect α1 to be positive and β1 to be negative. The first equation tests the internal BS effect. That is, if productivity in tradable sector increases faster compared to productivity in nontradable sector; then prices in nontradable sector will increase faster compared to prices of nontraded goods. The second regression serves to test the productivity bias hypothesis. RER is the real exchange rate of the domestic country regressed on dual productivity gap. The disturbance terms εt and νt are supposed to be white noises. We use as source of data the GGDC 10 sectors database203. As in Choudhri and Khan (2004), we consider manufacturing and agriculture as tradable sectors and all other sectors as nontradable. Productivity in each sector is measured by labor productivity calculated as constant value added (YT and YN) divided by the number of persons working in the considered sector (LT and LN). From the GDP’s current and constant values, we compute the price indices (deflators) of tradable and nontradable sectors (PT and PN). Table 3 provides the results of the test of the internal Balassa-Samuelson effect. In this table, we regress the log of internal relative prices (nontradables/tradables) on the log of internal relative productivity (tradables/ nontradables). Tableau 3. Internal version of BSE (1970-2010) Kenya Morocco Nigeria Senegal South Africa Tanzania linrelaprod1970100 0.614*** -0.014 0.431* -0.285*** 0.487*** -0.128** S.E. (0.04) (0.10) (0.21) (0.06) (0.04) (0.04) Constant -0.128*** -0.248*** 0.178* -0.104*** 0.051 -0.200*** S.E. (0.02) (0.02) (0.08) (0.01) (0.03) (0.03) * p < 0.05, ** p < 0.01, *** p < 0.001 (S. E: standard error) 203  “The Groningen Growth and Development Centre (GGDC) was founded in 1992 within the Economics Department of the University of Groningen. The GGDC 10-Sector database provides a long-run internationally comparable dataset on sectoral productivity performance in Asia, Europe, Latin America and the US. Variables covered in the data set are value added, output deflators, and persons employed for 10 broad sectors from 1950 onwards”. Visit http://www.rug.nl/research/ggdc/data/pwt/pwt-8.1 380 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ We search to uncover in each country the links between sectoral productivity ratio and sectoral price ratio. The internal version of the BSE effect states that higher productivity growth in tradable sector compared to nontradables sector will result in a faster growth of the nontradables price compared to tradables price. Table 3 summarizes the results. The results show that the internal version of the BSE is valid in Kenya, Nigeria, and in South Africa. It is rejected in Senegal and in Tanzania. In Morocco, the internal BSE is insignificantly rejected. Before performing our tests of exeternal version of BSE, it is important to perform the test performed by B. Balassa in his 1964 paper204. We regress for a cross section data, including in a first stage all countries and in a second stage only the African countries, a measure of the deviation of exchange rate from its PPP value on the per capita level for the year 2011. The data are from the ICP project (World Bank, 2015). The results are summarized in table 4. Tableau 4. Ratio of PPP to market exchange rate and level of development of a country World Africa Endogenous variable PPP/xr ln(ppp/xr) PPP/xr Ln(ppp/xr) E4 -0.71 -2.50 0.18 -1.31 Constant (-5.64) (-14.78) (1.82 (-5.76) 0.146 0.21 0.036 0.071 Ln(per capita expenditure at PPP) (10.79) (11.74) (2.84 (2.50) Number of observations 180 countries 50 countries R2 0.39 0.43 0.14 0.11 Source: Our calculations. In parenthesis the t-statistics. The results show that the level of development is linked positively with the deviation of PPP exchange rate from actual rate. For the whole sample, an increase of the per capita expenditure by 1% increases the EPPP/E by 0.21%. This means that richer a country gets, more appreciated its currency will became. The same is true for Africa but the elasticity of EPPP/E with respect to development level is lower (0.07). 3. Results and discussion In a second step, we focus on the links between the ratio of the domestic country to the 204  In its proper words, B. Balassa wrote in his 1964 paper that “if per capita incomes are taken as representative of levels of productivity, the ratio of purchasing power parity to the exchange rate will thus be an increasing function of income levels” (Ballassa, 1964, p. 586) EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 381 PARTIE III foreign country productivity and real exchange rate. Recall that for us RER is equal to the price of a basket of goods in the foreign country in terms of its price in the domestic country EP*/P. Thus, when RER increases this is synonymous of a real depreciation of the domestic currency. We will run two specifications to test the external version of the BSE. The endogenous variable is always the real exchange rate. Tables 5 and 6 provide results of each country with the two following specifications. The USA is used as the foreign country. ratioProductivitydTN is the tradable to nontradable sectors productivity differential in the domestic country. ratioProductivityUSATN is the same variable for USA. PercapitaGDPd and PercapitaGDPUSA are respectively per capita GDP in the countries of our sample and in the USA. The real exchange rate is EPUSA/Pd. The GDP deflators calculated from GGDC-10 sectors database are used to calculate the Real exchange rate. The use of the CPI to construct this index would provide the same results because the real exchange rate indices measured by the two price indices are highly correlated. For the six African countries, we have not observed a strong convergence of their per capita real GDP to the USA’s over the period of study. That means that these countries’ growth rates were below the growth rate of the USA (benchmark country). This is equivalent to a growth of productivity in each country that is lower compared to the growth of productivity in USA. The econometric results summarized in table 5 provide unsystematic evidence about the links between productivity and real exchange rate evolution when we use ydTN/yUSAT/N or yd/yUSA. When we focus on yd/yUSA, the BSE is valid in the cases of Kenya, Senegal, and Tanzania while in Morocco, Nigeria, and South Africa, this effect is not detected. Real exchange rate decreases when ydTN/yUSAT/N increases in Morocco. In the other five countries, it increases when ydTN/yUSAT/N increases. This means that the BSE is likely to operate in Morocco and is absent in the other countries of our sample if we use this second measure to gauge productivity evolution. 382 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ Tableau 5. Real exchange rate and dual productivity differential, 1970-2010 Morocco Kenya Nigeria Constant 2.29 11.95 141,02* -2.99 -8.68 8.99 yd/yUSA 53.78* -1621,24* 800.76* ydTN/yUSAT/N -6.64* 187 43.82 R 2 0.17 0.40 0.68 0.05 0.63 0.048 N 41 41 41 41 41 41 Senegal South Africa Tanzania Constant 870.85* 778.30* -955.92 109.31 1749,66* 986.88 yd/yUSA -8187.01* 4281.25** -20156,92* ydTN/yUSATN 1042.55* 13.62 1290.88 R2 0.47 0.34 0.08 0.00 0.60 0.057 N 41 41 41 41 41 41 *Significative at 5% level, ** significative at 10% level In table 6 we summarize results of regressing for each country the log of real exchange rate on the log of dual productivity gap (ldualrelprod). The results provide evidence that effectively the currencies of Morocco, Nigeria, Senegal, and South Africa do appreciate when the difference between their tradable sector productivity and their nontradable sector productivity is higher compared to the same difference in the USA. Tableau 6. Real exchange rate and dual productivity differential, 1970-2010 Kenya Morocco Nigeria Senegal South Africa Tanzania ldualrelprod1970100 0.828 -0.619*** -1.077*** -0.85*** -0.672** 2.491 S.E. (0.79) (0.11) (0.25) (0.17) (0.20) (2.03) Constant 5.1*** 4.527*** 4.335*** 4.437*** 4.538*** 5.235*** * p<0.05, ** p<0.01, *** p<0.001 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 383 PARTIE III Concluding remarks The dynamics of real exchange rate are linked to the competitiveness of a country, to the profitability of tradable and nontradable sectors, and to financial stability. The supply side factors such as intercountry productivity differentials are an important theoretical candidates in explaining these dynamics. This paper goal is to find out if the BSE is at work in a sample of countries in Africa. The crop of our paper includes three main results. First, the countries where a greater growth of productivity in tradable sector compared to nontradable sector induces an increase of the price of nontradable compared to the price of tradable are South Africa and Tanzania. In Nigeria and Senegal, this relationship is negative. However, in the four countries the relationship is insignificant. The second result is that the link between economic development and the price level is positive, statistically significant, and robust to the change of the sample. This confirms B. Ballassa’s 1964 result and previous results of many authors. The Third result concerns the presence or not of a BSE in each of the six countries. The use of two measures of productivity provides conflicting results. The use of per capita GDP as a measure of productivity shows that the BSE is valid in the cases of Kenya, Senegal, and Tanzania while in Morocco, Nigeria, and South Africa, this effect is not observed. The use of a disaggregated measure of productivity shows that only in Morocco a BSE is probably operating. These disappointing and conflicting results are probably due to the complexity of real exchange rate dynamics and imperfections in data. 384 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ References • Abeysinghe, T. and Yeok, T. L. (1998), “Exchange rate appreciation and export competitiveness. The case of Singapore”, Applied Economics, Vol. 30, pp. 51-55 • Asea P.K. and E.G. Mendoza, (1994), “The Balassa-Samuelson Model: A General Equilibrium Appraisal”. Review of International Economics, 2: 244-67. • Alessandria, G. and Kaboski, J., (2008), “Why are goods so cheap in some countries?”, Business Review Q 2, pp. 1-12 • Artus, P., (2012), « Quelle est l’ampleur véritable des déformations de la compétitivité- prix entre les pays de la zone euro ? », Flash économie, 03 mai 2012, n° 322 • Bahmani-Oskooee, M., (1996), “Decline of the Iranian Rial during the post-revolutionary period: productivity approach”, Economic Research Forum, Working paper 9615 • Balassa, B., (1964), “The purchasing-power parity doctrine: a reappraisal”, Journal of Political Economy, Vol. 72, No. 6, December, pp. 584-596 • Bigsten, A. and Danielsson, A., (1999), “Is Tanzania an emerging economy? A report for the OECD project “Emerging Africa” • Chinn, M. D., (2000), “The Usual Suspects? Productivity and Demand Shocks and Asia – Pacific Real Exchange Rates”, Review of International Economics, 8 (1), 20 – 43 • Choudhri, E. U. and Khan, M., (2004), “Real exchange rates in developing countries: are Balassa-Samuelson effects present?” IMF working paper, WP/04/188 • Connolly, M., (1983), “Exchange rates, real economic activity, and the balance of payements: evidence from the 1960s”, in E. Classen, and P. Salin, eds, Recent issues in the theory of the flexible exchange rates, Amesterdam: North-Holland • Coudert, V.  (2004), “Measuring the Balassa-Samuelson effect for the Countries of Central and Eastern Europe?” Banque de France Bulletin Digest – No. 122 – February pp. 23-43 • Dedu, V. and Dumitrescu, B. A., (2010), “The Balassa-Samuelson effect in Romania”, Romanian Journal of Economic Forecasting, 4/2010, pp. 44-53 • Faria, J. R. and M. Leon-Ledesma, (2000), “Testing the Balassa-Samuelson effect: implications for growth and PPP”, Department of economics Discussion Paper, University of Kent, No. 00.08 • Gelb, A. (ed.), (1988), Oil Windfalls: Blessing or Curse? Oxford University Press: Oxford • Genius, M.and Tzouvelekas, (2008), “The Balassa-Samuelson, productivity bias hypothesis: further evidence using panel data”, Agricultural Economics Review, vol. 9, No. 2, pp. 31-41 • Heston, A., Nuxoll, D.  A., and Summers, R., (1994), The differential-productivity hypothesis and purchasing-power parities: some new evidence”, Review of international economics, 2 (3), pp. 227-243 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 385 PARTIE III • International Monetary Fund, (2014), Annual Report on Exchange Arrangements and Exchange Restrictions, Washington, October 2014 • Ilzetzki, E., C.M. Reinhart, K.S. Rogoff, (20011), The Country Chronologies and Background Material to Exchange Rate Arrangements in the 21st century: Which Anchor Will Hold? • Ito, T., (1996), “Japan and the Asian Economies: A “Miracle” in Transition”, Brookings Papers on Economic Activity, 2: 1996, pp. 205-272 • Jonglez, O., (2008), “Real exchange rate appreciation in the emerging countries”, Trésor- Economics No. 35 • Kravis, I. B., and Lipsey, R. E. (1983), Toward an explanation of national price levels, Princeton Studies in International Finance, no. 52, November 1983 • Lee, j., and tang, M;-K., (2003), Does productivity growth lead to appreciation of the real exchange rate?, IMF working paper, WP/03/154 • Lopcu, K, Dülger, F., Burgaç, A., (2013), “Relative productivity increases and the appreciation of Turkish lira”, Economic modelling, 35, pp. 614-621 • Mihaljek, D. and Klau, M., (2003), The Balassa-Samuelson effect in central Europe: a disaggregated analysis, BIS working papers No. 143 • Okpanachi, U. M., (2013), “An assessment of monetary policy response to capital inflows in Nigeria”, CBN Journal of Applied Statistics Vol. 3 No. 2, pp. 75-98 • Peltonen, T. A. and Sager, M.  (2009), “Productivity shocks and real exchange rates a reappraisal”, Working paper series, No 1046, April  2009, http://www.ecb.europa.eu Network • Pesaran, M.H., Y. Shin, and R. J. Smith, (1999), “Bounds testing approaches to the analysis of long run relationships”, DEA Working paper No. 9907, University of Cambridge • Podkaminer, L., (2003), “Analytical notes on the Balassa-Samuelson effect”, BNL Quarterly Review, n° 226, pp. 207-221 • Rodrik, D., (2008), The real exchange rate and economic growth, Brookings Papers on Economic Activity, Issue 2, 2008, Pages 365-412 • Sekkat, K., (2015), “Exchange rate misalignment and export diversification in developing countries”, The Quarterly Review of Economics and Finance, Vol. • Zakaria, M.  and E. Ahmad (2009), “Productivity shocks and nominal exchange rate variability: a case study of Pakistan”, Journal of Economic Integration, 24 (1), March, pp. 175-189. • Samuelson P. (1964), “Theoretical Notes on Trade Problems”, Review of Economics and Statistics, 46: 145-54. • Trairatvoraku, P., (Policy panel: Regional challenges ahead – dealing with capital flows, prolonged exchange rate intervention and their consequences in Asia and the Pacific, BIS paper N. 66, pp. 384-386, url: www.bis.org/publ/bppdf/bispap66x.pdf • The Economist, July 16, 2014 386 EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE DÉSALIGNEMENT DU TAUX DE CHANGE ET COMPÉTITIVITÉ • World Bank, (2015), Purchasing power parities and the real size of World economies, A comprehensive report of the 2011 International Comparison Program. EQUILIBRES EXTERNES, COMPÉTITIVITÉ ET PROCESSUS DE TRANSFORMATION STRUCTURELLE DE L’ÉCONOMIE MAROCAINE 387