The Dutch Disease Revisited: Theory and Evidence 

Get Complete Project Material File(s) Now! »

The Dutch Disease Revisited: Theory and Evidence

Contrary to empirical evidence, the Dutch disease hypothesis, driven by Learning By Doing (LBD), does not predict the steady-state real exchange rate appreciation and economic growth deceleration due to a resource boom. To do so, I first represent a simple model to fill the theory’s gap, and then adopt a dynamic panel data approach for a sample of 132 countries over the period 1970-2014 to re-evaluate both symptoms of the hypothesis in systematic analysis. The main findings are threefold. First, a resource boom appreciates the real exchange rate. Second, the real exchange rate appreciation decelerates the rate of growth in both sectors such that the shrinkage is larger in the manufacturing sector than in the service sector. This, in turn, makes the relative output level of the manufacturing sector to the service sector be smaller and economic growth be slower. Third, these effects are more intensive in resource-rich countries than in resource-poor countries.

INTRODUCTION

Introduction

Why do natural resources countries tend to grow slower than countries without?, Why did Sierra Leone drop at an average rate of 37 % between 1971 and 1989 (Humphreys, J. Sachs, and Stiglitz, 2007)? and why has the income per capita in Nigeria stagnated over forty years (Sala-i-Martin and Subramanian, 2013)? A conventional interpretation that has attracted more attention of researchers refers to the Dutch disease hypothesis. The term of the Dutch disease was originally named in 1977 by T he Economist to describe the unfavorable repercussions of natural gas discoveries in the late 1950s on the Dutch manufacturing sector. The seminal work identified by Corden and Neary, 1982 was established based on a Salter-Swan framework to provide an explanation for the process of de-industrialization. The mechanism rests on two steps: 1) a resource boom appreciates the real exchange rate, and so 2) sectoral growth responds asymmetrically to the real exchange rate appreciation. It motivates us to make the question of whether there is strong empirical evidence to support the symptoms of the Dutch disease.
A useful starting point is to address the term “natural resources”. Over the past three decades, scholars have defined “natural resources” in dozens of ways. Natural resources are generally categorized in two classes of plantations (e.g. coffee, cocoa, rice) and minerals (e.g. oil, gas, coal). Adopting the terminology defined in Woolcock, Pritchett, and Isham, 2001, the former is called diffuse resources, while the latter is called point-source resources. Considering the type of natural resources, evidence shows that point-source resources are more valuable and provide a vulnerable influence on the economy. Hence, I only address point-source resources in this study1 (hereafter, the term « natural resource » is used to reference this specific subset of resources).
To shed light on the key question of why resource countries have usually failed to show better economic performance than others, I first clarify the mechanism of the Dutch disease. This can be illustrated in a two-sector small open economy framework2, as in M. Nkusu, 2004, in which the labor force is fully employed and can move freely across sectors. The literature on the Dutch disease highlights two different effects. The first one is the spending effect. Resource boom brings about an expansion in the total income of the economy and so increases the demand for both traded and non-traded goods. Whereas the price of traded goods is determined exogenously by the international market, the relative price of non-traded to traded goods must appreciate in order to confront the expanded demand for non-traded goods. The second one is called the resource movement effect.
1 Furthermore, given that the empirical database for the resource rent, collected from the World Bank, refers total natural resources rents to the mineral endowments, I ignore the former class of the terminology (i.e. plantation) to keep the consistency between the theory and empirical studies.
2In the standard Dutch disease model proposed by Corden and Neary, 1982, there are three sectors: the booming sector, lagging sectors producing traded goods and the non-traded sector producing services goods.
relative price appreciation will increase the real wage of labor employment in the non-trade sector, with respect to those in the traded sector. It makes a signal for labor forces to shift away from the traded sector and into the non-traded sector. Consequently, the traded sector shrinks and the non-traded sector expands (i.e. de-industrialization).
Although the framework could describe the mechanism of the Dutch disease in a short to medium-term, it would be interesting and more realistic to investigate the long-term dynamic response of a resource-dependent economy to a permanent increase in a resource rent. Evidence supports that the traded sector benefits more from learning by doing in the long-term (Ulku, 2004). Hence, the non-resource traded sector hit by worsening competitiveness is more likely to not fully recovered once the resource income runs out (Van der Ploeg, 2011b). One can be demonstrated this using a Salter-Swan model in which productivity growth is driven by learning by doing (LBD). As an initial attempt, Van Wijnbergen, 1984 studied a two-period and two-sector model in which future productivity in the traded sector depends increasingly on the current production of the traded sector. Later on, by an assumption that only labor employment in the traded sector contributes in generation of learning, Krugman, 1987 proposed a model of increasing returns to scale in the traded sector. While J. D. Sachs and Warner, 1995 and Gylfason, Herbertsson, and Zoega, 1999 made an assumption that learning generated by labor employment in the traded sector spills over perfectly to the non-traded sector. These later models clarify that the learning process induces the endogenous growth in both sectors: a natural resource boom reduces labor employment share in the traded sector, hampers learning by doing (LBD) and thus decelerates economic growth. In recent literature, Torvik, 2001 presents a general model in which both sectors can contribute to the learning process and there are imperfect learning spillovers between sectors. He demonstrates that within such a model a resource boom tends to depreciate the steady-state real exchange rate, while the steady-state economic growth is independent of a resource boom and the sectoral productivity growth depends on which one of the direct or the spillover effect is stronger. Bjørnland and Thorsrud, 2016 put forward Torvik, 2001 model so that the productivity spillovers between the booming resource sector and other domestic sectors. They show that a booming resource sector leads to the real exchange rate depreciation over the transition path and increasing the rate of growth in the economy and in both sectors, contrary to standard Dutch disease models. To investigate the empirical relevance of the theory model, they apply a Bayesian Dynamic Factor Model (BDF M) for Australia and Norway as representative cases studies. Their results are twofold: (1) a resource boom has significant and positive productivity spillovers on non-resource sectors, and (2) there is a two-speed transmission phase so that the non-traded sector expands faster than the traded sector. The influential works by J. D. Sachs and Warner, 1995; J. D. Sachs and Warner, 2001 and Rodriguez and J. D. Sachs, 1999 are representatives of a stream of empirical literature showing that natural resource dependence 3 decelerates economic growth. In particular, in a cross-section of countries during 1970–90 J. D. Sachs and Warner, 2001 show that a 10 % increase in the ratio of natural resource exports (% of GDP) is associated with as much as 0.4–0.7 % lower annual GDP per capita growth. In recent studies, researchers have applied panel data rather than cross-section approach to avoid the problem of omitted variables bias. A group of these studies has found that a natural resource boom retards the institutional development and this, in turn, hampers economic growth (e.g. S Mansoob Murshed, 2004; Collier and Hoeffler, 2005; Mehlum, Moene, and Torvik, 2006b). Furthermore, a study by Manzano and Rigobon, 2001 dismisses the curse once one controls for fixed effects in panel data estimation. Using annual data for 81 manufacturing sectors in 90 countries over the period 1977-2004, Ismail, 2010 put forward that a 10 % increase in oil price slows down, on average, the manufacturing growth rate by 3.4 %. Further, it shows that this negative effect is stronger in countries that have a more open capital market to foreign investment and also sectors that are less capital intensive. Consistent with the former work, a recent study for 41 resource exporter countries over the period 1970-2006 shows that price movement is negatively correlated with the manufacturing value-added (Harding and Venables, 2016). More evidence for 135 countries for the period 1975-2007 shows that a resource windfall income induces savings of about 30 %, shrinks non-resource exports by 35-70 %, and expands non-resource imports by 0-35 % (Harding and Venables, 2010). Furthermore, a number of studies (e.g. Rajan and Subramanian, 2008) consider a foreign aid rather than the resource rent and find that the aid leads to slower growth in the manufacturing sector relative to the service sector.
Out of this context, a growing number of studies examines only one of the symptoms of the Dutch disease. Strong evidence for a positive effect on the real exchange rate from commodity price appreciation (first step of the symptom) has been documented by Koranchelian, 2005 for Algeria, Zalduendo, 2006 for Venezuela, Oomes and Kalcheva, 2007 for Russia, and Beine, Bos, and Coulombe, 2012 for Canada. Furthermore, Cashin, Céspedes, and Sahay, 2004 for a panel of 58 commodity-exporting countries over the period 1980-2002, Korhonen and Juurikkala, 2009 for a panel of 12 oil-exporting countries over the period 1975-2005 and Ricci, Milesi-Ferretti, and J. Lee, 2013 for a panel of 27 developing & 21 developed countries over the period 1980-2004 report a positive correlation in the commodity prices–exchange rate nexus. In contrast, others have studied the impact of the substantial exchange rate overvaluation on growth (second step of the symptom). Empirical evidence on this subject suggests that the real exchange rate appreciation decelerates
3Two different criteria used to classify the economies depending on the natural resource are (1) resource dependence referring to the value of the natural resource as a share of GDP or total national wealth and (2) resource abundance referring to per capita value of the stock of natural resource wealth. Empirical studies across a comprehensive sample of countries show that natural resource abundance has a positive effect on economic performance (see. Brunnschweiler and Bulte, 2008; Alexeev and Conrad, 2009; Esfahani, Mohaddes, and Pesaran, 2013; Cavalcanti, Mohaddes, and Raissi, 2011).
growth. Perhaps among well-known of these studies is that of Rodrik, 2008; Aguirre and Calderón, 2005. Other studies include those of Eichengreen, 2007; Williamson, 2009; Habib, Mileva, and Stracca, 2017). In terms of the sectoral performance, Sekkat and Varoudakis, 2000 assessed this nexus for a panel of major Sub-Saharan African countries during the period 1970-1992. Their findings indicate that the real exchange rate depreciation fosters manufacturing exports’ performance.
Among a few limited studies concerning both symptoms of the Dutch disease hypothesis, Javaid and Riazuddin, 2009 adopted a static and dynamic panel data technique to structurally analyze the hypothesis in a sample of 6 selected South-East Asian economies over the period 1981-2007. They first investigate the response of the real exchange rate to foreign inflows and second the impact of a change in the real exchange rate on growth rate in the manufacturing and service sectors. The findings confirm an appreciation in the real exchange rate, a contraction in the manufacturing sector and an expansion in the service sector as a foreign aid inflows.
In the same vein, Lartey, Mandelman, and Acosta, 2012 studied the Dutch disease effect of remittances under different exchange rate regimes in an unbalanced panel data set of 109 countries over the period 1990-2003. They pursued a dynamic estimation procedure to estimate a regression model in which the relative output of the traded to the non-traded sector was captured as the dependent variable and remittance (% GDP) was their explanatory variable of interest. Their findings show that an increase in remittances leads to an expansion in the share of the service sector and a shrinkage in the share of the manufacturing sector. The paper also suggests that the resource movement effect is stronger under fixed nominal exchange rate regimes.
To sum up, in the context of the natural resource curse, on the one hand, the theoretical models, as in Torvik, 2001; Bjørnland and Thorsrud, 2016, predict that the (steady-state) real exchange rate depreciates, and the rate of growth in the economy is constant or increases due to a resource boom, contrary to the empirical evidence. On the other hand, researches have estimated the impact on the real exchange rate from a commodity price rather than a resource-dependence proxy or they have studied the direct impact of a resource-dependence proxy on economic growth rather than the impact of the real exchange rate on sectoral growth. Out of the resource curse’s context, the literature examines the relationship between the real exchange rate and economic growth rather than sectoral growth or they have captured this nexus in a different empirical technique and/or specification from the present paper. Hence, these reasons motivate us to fill these gaps through developing a simple dynamic theory that leads to the real exchange rate appreciation and economic growth deceleration in the long-term and also through a comprehensive systematical empirical analysis that investigates both symptoms of the Dutch disease hypothesis.
Motivated by the literature, I first develop a simple theory, driven by LBD, in which the productivity is generated in both manufacturing and service sectors, while there are an imperfect spillover from the manufacturing to the service sector and a technology spillover from the resource sector to the manufacturing sector. The model describes the standard Dutch disease mechanism, the same as Torvik, 2001. While it shows that a resource boom results in a conditional appreciation in the (steady-state) real exchange rate and also an unconditional depreciation in the rate of growth in the economy, contrary to Torvik, 2001; Bjørnland and Thorsrud, 2016. I then collect an unbalanced panel data set of 132 countries from the period 1970 to 2014 to revisit the Dutch disease symptoms. Using the Generalized Method of Moments (GM M) technique, I adopt a dynamic estimation procedure to avoid the autocorrelation problem and the problem of reverse causality between variables of interest (endogeneity problem). Estimated results illustrate some clear evidence in supporting the positive relationship between a resource-dependence proxy and the real exchange rate. They also demonstrate that the real exchange rate appreciation causes the sectoral growth to shrink more in the manufacturing sector than in the service sector and so the economic growth to decelerate. Finally, the empirical approach suggests that these adverse effects are more intensive in resource-rich than in resource-poor countries.

READ  Are There Any Industry-Related Differences in why SMEs Engage in CSR?

A model of the Dutch disease

I extend Torvik, 2001 model to make theoretical predictions consistent with empirical results. Consider a two-sector economy: manufacturing and service, indexed by M and S respectively. Assume there are no assets and capital accumulation and the labor force is the only production factor. Labor, inelastically supplied by a continuum of symmetric-identical households, can move freely across sectors.

Table of contents :

0 Introduction – version Française 
0.1 Les faits stylisés de la malédiction des ressources
0.2 Les explications de la malédiction des ressources naturelles
0.2.1 Problèmes économiques
0.2.2 Problèmes politico-économiques
0.3 But de l’étude
0.3.1 Chapitre 2: La maladie hollandaise revisitée: théorie et preuves
0.3.2 Chapitre 3: Les inégalités de revenus alimentent-elles la maladie hollandaise?
0.3.3 Chapitre 4: Capacité d’absorption et malédiction des ressources naturelles
1 Introduction – English version
1.1 The stylized facts of the resource curse
1.2 The explanations of the natural resource curse
1.2.1 Economics issues
1.2.2 Political-economics issues
1.3 Purpose of the study
1.3.1 Chapter 2: The Dutch Disease Revisited: Theory and Evidence
1.3.2 Chapter 3: Does income inequality feed the Dutch disease?
1.3.3 Chapter 4: Absorption capacity and Natural Resource Curse
2 The Dutch Disease Revisited: Theory and Evidence 
2.1 Introduction
2.2 A model of the Dutch disease
2.3 Theory meets empirical model
2.3.1 Data and Methodology
2.4 Econometric Results
2.4.1 Resource-dependence and the real exchange rate
2.4.2 Real exchange rate and relative sectoral output
2.4.3 Real exchange rate and sectoral growth
2.4.4 Real exchange rate and economic growth
2.5 Conclusion
2.A Steady state response of a resource boom
2.B Commodity Price index: Data Description, Sources and Methodology
2.C The net foreign assets and resource-dependence
2.D List of Countries
3 Does income inequality feed the Dutch disease?61
3.1 Introduction
3.2 The model
3.2.1 Short run analysis (static model)
3.2.2 Long run analysis (dynamic model)
3.2.3 Dutch disease dynamics
3.2.4 Absolute Productivity Growth
3.3 Empirical approach
3.3.1 Data and Empirical methodology
3.3.2 Impact of the windfall income on the real exchange rate
3.3.3 Income inequality response
3.3.4 The intensity of the Dutch disease
3.3.5 The sectoral growth
3.4 Conclusion
3.A Proof of Proposition 3
3.B Proof of Proposition 4
3.C Proof of Proposition 7
3.D List of countries included in the sample database
4 Absorption capacity and Natural Resource Curse
4.1 Introduction
4.2 Implication of Absorption capacity
4.3 Stylized Facts
4.4 Empirical approach
4.4.1 Methodology and Data
4.4.2 Estimation Results
4.4.3 Robustness Tests
4.5 The Model
4.5.1 Consumer problem
4.5.2 Economic response to the windfall income
4.5.3 Imperfect financial market and changing the demand composition
4.6 Conclusion
4.A Statistical cross-country estimation
4.B List of countries included in the sample database
4.C Windfall income and social value of wealth
4.D Solution of the dynamic system
4.E The real exchange rate and the capacity constraint
5 References

GET THE COMPLETE PROJECT

Related Posts