Disaggregated analysis of the curse of natural resources in most natural resource-abundant countries

Highlights

• This study employs ARDL bounds testing to explore the resource curse hypothesis.

• The sample consists of ten resource-abundant countries.

• The study examines the impact of natural resource rents separately on the income level.

• Empirical results provide heterogeneity across natural resource rents.

Abstract

A vast body of literature examining the natural resource curse hypothesis (NCRH), the studies, by summing all types of natural resources, uses a single measure for the natural resource. To fill this gap, the study aims at testing the NRCH hypothesis by using the ARDL method and use the data distinguishing five different types of natural resources (coal, forests, minerals, natural gas, and oil) and covering the period from 1990 to 2017 for ten countries with different level of economic development, being measured by Human Development Index. The study finds mixed results. Firstly, the study finds very little evidence favoring adverse growth effects across each type of commodity. The results show no positive or negative impacts of natural resources on economic growth for developed economies (countries with high HDI scores). For countries with moderate and low HDI scores, we found mixed results. In particular, we find some evidence supporting NRCH for point-source resources for these two group countries (moderate and poor).

Introduction

Intuitively, resource-rich countries have the potential to accelerate their economic development by exporting these natural resources. Suppose the lack of financial resources, which has been one of the most important problems in developing countries, is considered a reference. In that case, the revenues accrued from the natural resources can be used to finance private and public investments. Furthermore, considering the foreign exchange scarcity in poor countries, the foreign currency earnings will facilitate the import of technology, intermediate, and capital goods, which speed up the economic growth (Badeeb et al., 2016). Nevertheless, during the last forty years, some empirical studies in this subject found precisely the opposite.

According to the literature, this paradoxical phenomenon, labeled as the ‘natural resource curse hypothesis' (NRCH), refers to the inverse relation between the natural resource abundance/dependence and economic growth and other development outcomes. This pessimistic view has attracted much attention, particularly from the empirical domain (Auty, 1993; Sachs and Warner, 1995; Van Der Ploeg and Poelhekke, 2017).

Although a vast literature exists in examining the link between natural resource abundance/dependence and economic growth, the subject continues to attract scholars to examine NRCH. We believe that the subject will retain its popularity in the academic circuits for some reasons. Firstly, the empirical results in the NRCH are not conclusive. In their meta-analysis on the links between natural resources and economic growth, Havranek et al. (2016) show that 40% of empirical studies find a negative economic effect of natural resources, 40% of studies concluded no effect, 20% studies show a positive relationship. Secondly, we have observed significant shifts in global demand for natural resources, and some pieces of evidence show that the transformation dynamics are still in operation. After the 2000s, global transformations in the form of population growth, human mobilization, rapid urbanization, rapid per capita income increases in Asia and Africa have caused a significant change in the landscape of the resource markets and the expected transformations to remain intact in the future. For example, due to the COVID-19 outbreak, several resource-dependent countries have experienced severe economic contractions due to the plummeting prices. The geopolitical competition between the USA and China is also expected to influence the landscape in years to come.

Moreover, the World Economic Forum for 2020 (WEF, 2020) underlines that climate and environmental concerns dominate the top five long-term global risks. Therefore in the future, The supply side of the subject is also important. According to the IMF (2012), 51 countries, home to 1.4 billion people, are classified as resource-rich countries (Venables, 2016). In other words, around %20 of the global population's welfare depends on the size of the natural resource revenues and the way these resources will be utilized.

As highlighted by Paprakis (2017), the studies collectively emphasize the complexities and conditionalities of the ‘curse’ – its presence/intensity is context-specific mainly, depending on the type of resources, sociopolitical institutions, and linkages with the rest of the economy. The manifestation of the resource curse for a particular country depends on at least five sets of interrelated factors. The first one is the price channel or international channel (also known as Dutch disease view; Corden and Neary, 1982; Corden, 1984). Secondly, resource windfall may cause over-optimism about the future, which would distort incentives and motivation mechanisms for some variables, including human capital, savings, investments (i.e., Gylfason and Zoega, 2006; Van Der Ploeg and Poelhekke, 2017). The third factor is the linkages (input-output, spill-over) between the natural resource industry with other sectors and actors. The fourth one is the factors related to the political configuration (including corruption, rent-seeking, and institutions' quality. The final one is the type of resources. Different natural resources may have a different impact on economic growth. As Lay and Mahmoud (2004) point out: Even if all the resources negatively impact economic performance, the oil curse may well function differently from the banana curse.

A large portion of empirical studies in NRCH literature uses all the natural resources being summed up to generate one measure for resource abundance (Torvik, 2009). Therefore those studies are based on the assumption that all types of natural resources have the same impact on economic growth. Empirical studies distinguishing types of resource are minimal, but the specification has received attention recently (Isham, 2005; Jović et al. (2016); Boschini et al., 2007; Jović et al. (2016); Prljic et al., 2018; Huang et al., 2020; Zuo and Zhong, 2020). The vast literature in the field indicates that the resource curse is an extremely complex phenomenon. Incorporating different resource types into econometric analysis may, at least to some extent, alleviate some of these complexities. Firstly, differentiating the natural resource sectors is a novel effort in addressing NRCH. Secondly, an appropriate investigation of the ‘curse’ hypothesis requires a more specific or narrow perspective. One way of doing so is to distinguish different types of resources in econometric analysis. Thus, differentiating the natural resource types will allow us to test the curse hypothesis for each natural resource type. Thirdly and more prominently, incorporating different resource types into analysis would allow in determining the degree of importance for each type of resource in influencing economic growth.

By distinguishing the five different types of resources (coal, forest, mineral resources, natural gas, and oil), this study focuses on examining the possible link between natural resource abundance and economic growth for 10 research-rich countries (Algeria, Australia, Brazil, Canada, Chile, India, Indonesia, Mexico, Nigeria, and the US) for a period from 1990 to 2017. To fulfill this objective, the study uses the autoregressive distributed lag model in the augmented specification of the production function.

Although the manifestation of the resource curse is closely linked with the type of natural resources, the number of studies in line with this point is rare (for example, Isham et al., 2005; Boschini et al., 2007; Boschini et al., 2013; Jović et al. (2016); Prljic et al., 2018; Huang et al., 2020; Zuo and Zhong, 2020). Compared to other studies with similar treatment, our study has some differences that lead the current research to contribute to the existing literature. Firstly, the data span and the countries covered in this research are different from the previous studies. For example, the data in Isham et al. (2005) and Boschini et al. (2007) covers from the mid-1970s to mid-1990s. The data Boschini et al. (2007) used in the study covers from 1965 to 2005. Since the global trade has displayed a considerable transformation due to favorable climate stemming from globalization after the early 2000s, Isham et al. (2005) and Boschini et al. (2007) studies cannot account for the effects of these transformations. Moreover, although the data used by Boschini et al. (2007) covers some parts of globalization; however, the methodologies used in the study (pooled regression and various IV regression) are static, and therefore methodological tool utilized in the study is not compatible entirely in capturing the dynamic effects. They also attempted to use panel regression with the fixed effect specification (to account for the dynamic effects), but the results they reported were statistically weak. Moreover, the recent studies by Jović et al. (2016), Prljic et al. (2018), and Huang et al. (2020) focus on Asian countries, whereas Zuo and Zhong (2020) focus on the provinces of China.

Unlike previous studies with similar treatment, our study followed a simple procedure with two elements: availability of data and balanced distribution in economic development. In our study, the countries are distributed evenly for development measured by Human Development Index (HDI), released yearly by the United Nations. The countries covered in this study fall into three categories based on HDI: (a) countries with spectacular performance -the top segment of HDI ranking (Australia, Canada, USA), (b) countries with moderate performance, i.e., medium-range in HDI ranking (Algeria, Brazil, Mexico, Chile) and (c) countries with poor performance (India, Indonesia, Nigeria). The current study offers another useful perspective on the resource curse literature by taking the level of economic development into account. Secondly, the definition of resource type in our study is different from the previous studies. In our study, exports of five different natural resources are treated separately, whereas Isham et al. (2005), Boschini et al. (2007), and Boschini et al. (2013) classify the commodities into four categories. However, their classification is considerably different from ours. In their analysis, Zuo and Zhong (2020) focus on the resource curse in China's provinces by taking three types of resources.

The methodology used in our study is also different. Zuo and Zhong (2020) use panel data and use the Instrumental Variable (IV) framework. Both Isham et al. (2005); Boschini et al. (2007) employ the static regression models. Prljic et al. (2018) use the machine learning approach and artificial neural network as methodological tools in addressing the question. In our study, we use the ARDL bounds testing approach. The existing literature uses different econometric techniques to test the links between economic growth and natural resources. However, the ARDL bounds testing technique offers some advantages. In the ARDL bound testing approach, the order of integration of the series does not matter. This property is beneficial when the data points are limited as compared to other techniques.

The study's structure is as follows: The second part of the study reviews the existing literature from theoretical and empirical domains. The third part introduces the data and econometric methodology. The fourth part presents the empirical results, and the last part concludes the study.