Table 1 emphasises the important econometric studies of the RC, the dependent variable, and independent variables. There are two innovations of the S-W approach. The first is the breath of analysis: it includes evidence of a large number of developed and developing countries. The second is that S-W conclude that resources (% of total exports) fail to harness economic development. One criticism levelled at the S-W (2001) study is that cross-section data can not observe booms or collapses for decades as accurately as time series approaches do. The latter method allows observation of the effect of resource booms after 10 or 20 years. B-B (2008b) fails to find evidence for the RC and suggests that resource abundance, not rents, is conducive to economic growth via institutions.

In recent work on the RC both Van Der Ploeg and Poelhekke (2010) and Collier and Goderis (2007) use panel and cross country data. The former study corrects for endogeneity in the modes of the RC. Both studies report strong evidence for RC. The latter work tests for the RC using co-integration methods which allows for short and long run effects of resource dependence and economic growth. Those authors identify four channels through which the effect operates the most important are commodity prices with high volatility, among others. Collier and Goderis (2007) test the weight of each of these factors using cointegration techniques rather than a levels approach in their econometric specification. Other advanced techniques based on panel data find conclusive evidence against the RC, however, when using cross country data the evidence for the RC is strong (Manzano and Rigobon, 2001).

The theory of RC is partly based on the Salter Swan (SS) model which explains three features of the RC. First it describes the so-called ‘Dutch disease’ effect following a windfall in resource revenues. The effect occurs after the extra wealth generated by the sale of natural resources induces appreciation of the exchange rate and an ensuing contraction of the traded sector (Van Der Ploeg, 2011), this is discussed in section 5. Second the SS model explains the changes that both resource rents and exports have on relative prices of non traded goods (non manufacturing goods), and effects on employment when the economy relies on resource rents. Third, employment in non traded goods rises after an increase in exchange rate; while, a lower exchange rate is associated to higher demand for non traded goods and thus with more workers employed in the non traded sector. The SS model describes the increase in resource dependence so long as relative prices move positively, holding the other variables constant. We define the following variables of the SS model as:

Q is the world price of natural resources, E export volume of natural resources, Ct consumption of traded goods, C _n consumption of non traded goods, L _T Employment in the traded sector, P real exchange rate or relative prices between traded and non traded goods. F and G stand for coefficients of employment (traded and non traded sectors, respectively). ε is elasticity of substitution between traded and non traded goods.

In Van Der Ploeg (2011) the following identities explain the SS model:

H stands for productivity, H ≡ H T H n , of the traded and resource sectors relative to that of the non traded sector; exports of resources equals net imports of traded goods is described by     H T Q E = c t - H t F L t ; and the output of the non traded sector described by H n × G L n , and, output of traded sector in H T × F L H T . National income is given in Y ≡ P H N G ( L N ) + H T × F L H T + H T Q E . The variable P   ∈ - 1 is defined as the real exchange rate elasticity of non traded goods with respect to traded ones.

Using the SS framework, Van Der Ploeg (2011) describes the causal relationships for the equilibrium in the markets for non traded goods (‘subscripts “T” denotes the traded sector, “N” the non-traded).

The equilibrium in the markets for non traded goods is achieved in,

H N G L N = C N = Y 1 + P   ∈ - 1 (1)

= [ P H N G L N + H T F L T + H T   Q E   ] P + P   ∈ (2)

Recall ϵ stands for the substitution elasticity between traded and non traded goods and “P” for real exchange rate of traded versus non traded goods.

The ratio of national income to relative prices in eq. 2 yields,

P   ∈ = H F 1 - L n + Q E G L N (3)

Solving for QE/G(L _N ) in Equation 3, allows to capture resource dependence (Equation 4),

Q E G L N = P t + H [ F 1 - L N (4)

Equation (1) assumes the economy achieves equilibrium for the market for non tradable goods and equation (2) and equation (3) are written in their reduced form. In the short run, a boom in natural resource revenues QE increases national income, aggregate demand and leads to exchange rate appreciation. The same boom produces a decline in the traded sector and an expansion of the non traded sector.

A commonality of many studies (Table 1) is their use of the definition of governance used in S-W (2001) and in Collier and Goderis (2007). A gap in the literature is that the majority of studies of the RC do not distinguish between the effects of resource abundance and resource rents. The latter metric is narrow because it does not allow for wider effects of the resources-economy space over time that is captured in subsoil wealth differences among countries.

Papers that use methodological improvements [Sala I Martin & Subramanian (2003) for Nigeria; Caselli & Michaels (2009) and Medeiros Costa-Dos Santos (2013) for Brazil; Holden for Norway (2013); Alcott and Keniston (2018) for the USA] focus on a single nation breaking the long practice of bundling numerous countries for analysis of resource dependence. Previous papers of the RC include Frankel (2012) and Van Der Ploeg (2011). The papers included in those two reviews are largely technical ones, however, theoretical aspects are explored in Van Der Ploeg (2011).

Table 1

The institutional school focuses on the quality effect of political institutions on economic growth and some argue that parliamentary governments are better suited to mitigate the negative effects on economic growth of resource rents more effectively than presidential ones do, claiming “long run ability to deal with natural resources depends largely on country specific constitutional arrangements” Andersen and Aslaksen, (2006). Gylfason (1999) extends the institutional analysis to that of the strength of civil liberties and finds supportive evidence for the RC by using natural capital instead of resource rents. Le Billon (2010) argues that there are four channels through which the RC effects (in the case of conflict) operate:

how the RC influences institutional quality and economic performance and resource conflicts,

These studies are based on time-series data. Harvey et al. (2010) find evidence of long run decline in commodity prices (includes oil and coal) which explains long run declining economic growth or the Prebisch thesis (explained below). The studies on the effects of commodity prices use a different methodology (Deaton and Miller, 2006) and rely on the vector auto regression system (VAR). That system relies on data for a single variable that is observed over time. The latter have, however, arrived to conclusions close to those of S-W regarding the direction of impact of the resource dependence on GDP. The VAR approach, however, differs from the S-W methods because the former does not use structural relationships of the economy.

This fourth wave of studies of the literature on RC emerges in 1950s through the 70s including Frank, (1967), Prebisch (1952), and Lewis (1952). The same wave includes Bulmer Thomas (1994, 2013). The Prebisch hypothesis holds that a secular decline in the terms of trade in Latin America affects economic growth. A long term decline in the commodity NBTT (net barter terms of trade) means that countries should develop their own industries (infant industry argument) in order to avoid poor economic performance (Prebisch, 1952). Bulmer-Thomas (2013) finds evidence that the NBTT and the ITT (income terms of trade) indicators point to strong dependence on natural resources for the Latin American region.

There are six weaknesses in the studies of the modern school approach, 1) Missing data and measurement error; 2) the literature is dominated by the use of dummy variables; 3) the use of natural resource variables which are endogenous to growth and to war (Le Billon, 2010) biasing the quantitative results. 4) Studies also suffer from reverse causality issues, spurious correlation and robustness (B-B, 2008b). Besides this, 5) the literature relies on a narrow definition of the resource curse and it has assumed that natural resources are correctly priced producing miscalculated indicators of resource dependence. 6) Further problems includes scant or no analysis of how access to international credit markets influences how severe the RC is. Nations of abundant natural resource, however, are likely to be given access to global credit markets.

The resource literature offers six explanations for the RC effect: Dutch disease, governance, conflict, excessive borrowing in global credit markets, inequality and volatility of commodity prices. Many of the above papers reflect incremental improvements for modelling the effect of resources abundance on the economy.

Data for Explanatory variables are sourced from various publicly available data banks. Table 2 gives definitions of and sources of variables and Table 3 reports standard deviation and observation samples and mean variable values. Table 3 shows the description of data for empirical analysis.

Table 2

ln Natural Logarithm

Table 3

The values for the variables have been calculated for the countries are listed in Table 3; we make some observations. Resource rents (Rents) average 9.6% of GDP in our sample of countries between 1970-2011 with Uzbekistan and Iceland having the largest and lowest shares at 49% and much below 1%, respectively. In the same period, GDP growth rates per capita (GDPpop) average 5.5% per year; GDPpop growth for Bhutan shows the highest growth rates (17.1% per yr.) and Liberia (Africa) records the lowest average growth (1.6% per yr.). Lebanon records the lowest annual growth rate in resource rent (0.0168% of GDP) in the same period; and Uzbekistan the highest (53.8% of GDP). Between 2000-2011 the terms of trade (TT), on average, expanded by 17%, (base year, 2000 =100) supported by strong growth in natural resource prices. The index for institutions (Democ) averages 0.67, -0.10 for rule of law (Law), -0.08 for government effectiveness (GovEff). Saudi Arabia and Bhutan records the lowest Democ index and Norway the highest. The highest value of the Law index is 2 (Sweden) and the lowest -2. (Afghanistan). Somalia registers the lowest index (-2.24) of government effectiveness and Finland and Singapore the highest with 2.23 and 2.24 respectively.

Subsoil wealth per capita (SubAssets in US$, 2006) averages US$3817 in our sample; Kuwait shows the largest level of subsoil assets and Bangladesh the lowest. The share of mineral exports (MinXp) averages 1.1% of GDP with Papua New Guinea and Hong Kong (China) having the largest and the smallest shares at 18% and much below 1% (1970-2012), respectively. The most volatile variables are TT, Law, and MinXp.

Next we inspect correlation data (Appendix 1 and 2). Resource rents and GDP per capita are negatively correlated (Fig. 2), while mineral rents and GDP per capita (not graphed) show negative correlations. Subsoil wealth (SubAssets) and GDP per capita show weak correlations (Fig. 5A). A negative association between the Law and Subsoil wealth can be seen (Fig. 6A). Law and subsoil wealth shows points positive correlations but the correlation is not so strong (Fig. 6A, Appendix 1). Law and Rents (Fig. 7A, App. 2) variables also show negative associations, particularly if rents lie around 10% of GDP. As we approach the origin of the graph, increasingly larger gains in “Law” are needed to cut the ratio of Rents. The most notable correlation is the link between the rule of law (Law) and GDPpop which is positive (Fig. 8A).

Our econometric models extend the SS model and we validate empirically the SS framework as follows. The variable Rents’ for the ith country enters both the economic growth and the institutions model (equations 4 and 5). We assume that the GDP share of resources is ≥ QE/G(Ln) of eq. 3. We build two models to establish the strength of the RC: 1) resource dependence and institutions; 2) resource abundance and economic growth. The expected direction of impact of resource rents (our main hypothesis) is shown in parenthesis below equations 4 and 5, and follows from the theory discussion above.

To test for the transmission mechanism of the RC a model of institutions is introduced, the model uses institutional quality and abundance. It uses proxies for resource dependence (GDP as a share of resource rents) and resource abundance (subsoil asset wealth). For democracy we include three metrics: an index of the rule of law (Law), effective Government (GovEff) and for the change in democracy in 500 years (Democ), “ln” stands for natural logarithms of all variables. For examining the relationship between institution quality and abundance we use,

and for economic growth we use,

Of particular interest are two parameters in equation 5, namely (₁ ln (Rents) which will be less than zero if the RC occurs; and (₂ will be > 0 if the resources foster economic growth. Estimation results of equations 4 and 5, using ordinary least squares technique are reported in Tables 4-5.

We derive five key results using the institutions model (eq. 4) which are shown in Table 4. Natural capital, subsoil wealth, and mineral rents are positively associated to high quality institutions confirming that these sort of institutions can manage more effectively natural resource rents than weak institutions do. (Table 4, columns 2 and 5). Except for mineral rents (MinXp), these are long term effects.

Stronger institutions are seen to be associated to bigger resource abundance (SubAssets); and Weaker institutions are associated to bigger resource rents - this provides indirect evidence of RC effects, but only in the short run (Table 4, columns 2, 3, and 5). Amongst other characteristics, the institution quality (Law) measures contract enforcement which is particularly prominent in the petroleum and minerals sectors. Resource rents (Rents) are negatively associated to our three proxies of institutional quality which in turn are associated to lower GDP growth (Table 4, columns 2, 5, and 7). Resources are associated to two opposing effects on GDP growth via institutions (Law): these are positively correlated to GDP growth (1970-2011) but this is not the case every year.¹ Below a ratio of 10% of Rents in GDP a much larger gain in democracy is needed to cut that ratio for the entire 1970-2012 period).

Table 4

The change in democracy (Democ) is boosted by subsoil wealth and natural capital (Table 4, column 5) but the negative effect of Rents on Democ is larger than its effect on the rule of law (contemporary index). This is because the 500 year democracy variable includes early institutions and current institutions and this is associated to decreasing rents which gives evidence of RC effects. The negative effect of Rents on Democ reflects over reliance of governments on resource revenue that is volatile, alternatively a gain in Democ is associated to lower Rents. The impact chain of institutions is the following: falling rents to better institutional quality (or Democ) and to higher economic growth.

The estimated growth model (Equation 5) shows both high economic and statistical significance (Table 5) with five main outcomes emerging. These results are complemented by cross-country evidence of the non-linearity of resource dependence and economic growth.

Economic growth is associated to resource abundance and Law which explains a larger part of the story on economic growth. The magnitude of both coefficients is high, particularly that of Sub-Assets (Table 5, columns 5 and 6).

Economic growth is retarded by resource rents in the short run (Table 5, columns 4, 5, and 6). This is the case even after controlling for subsoil assets, institutional quality and productivity growth in the initial year (1970). But the effect is exactly the opposite following a boost to resource abundance. Two reasons stand out for the latter. First, regions with abundant resources attract capital inflows from abroad which in turn stimulates economic growth via an income effect (Table 5, columns 2, 3, 5, and 6). Secondly, countries with resources and good institutions attract drilling activities for hydrocarbon production which stimulate economic growth. The puzzle is that the two resource metrics are correlated positively using cross-country data on 32 nations: higher abundance is associated to resource rents.

Table 5

Economic growth can be predicted by current and past institutions (Table 5, column 2, 7). The variable Democ includes a period that predates the recent GDP growth and it shows the expected sign: higher Democ is associated to higher GDP growth, but is not statistically significant. The positive change in Law supports economic growth but the Law variable captures a contemporaneous effect.

Figure 2 shows that resources (Rents) do both harness economic growth (1970-2011) and depress economic growth depending on the country.

Figure 2 Source: the authors, based on data of World bank (2015)

The checkered lines represent the median value (GDPpop: 5.5% pa. and Rents: 9.64 %, 5% for meeting the definition of resource dependence) and countries lying below the median are deemed underperformers. Russian and Venezuela are underperformers (Lower GDP growth). The best performers include mainly the EU-28 or the OECD nations.

The boost to GDPpop of resource abundance runs counter to the RC hypothesis. The models also show that poor rates of growth in GDPpop lead to exports of natural resources and thus to greater resource dependence (Table 5).

Many resource poor countries match the growth rates of those with high resource dependence (share of, mainly oil, resources as a percentage of total exports). The RC hypothesis can be confirmed in its soft form. However, a broader metric of resource dependence shows exactly the opposite impact i.e. economic growth is supported by natural resource abundance. This is explained by financial speculation and currency appreciation as foreign investors take notice of abundant underground resources which leads to economic growth. A rise in the TT index results from an increase in the price of mainly energy exports since the countries experience a resource boom (prices and volume). The increase in TT lowers economic growth confirming the SS theory: The price of non-oil tradable goods should increase at the expense of non-tradable goods which allocates more capital and labour to the non-tradable goods sector. As S-W (2001) show, a bigger TT is associated to resource dependence. The TT effect boosts the RC, however, the transmission mechanism between the variable TT and resource dependence is the change in the exchange rate.

Mineral resource rents are associated to poor economic growth. The effect reflects the impact of high resource rents and the sign is consistently negative although not statistically significant. Such rents are geographically dispersed, unlike for oil and gas, and so the effect on growth is not strong.

To demonstrate resource dependence we use data for oil revenues. Oil revenue is the largest component for total resource revenues. Figure 3-4 emphasises the role of manufacturing exports as a function of oil rents for 1966 to 2013 for several geographical regions. Manufacturing exports represent the traded goods sector. Figure 3-4 also demonstrate that manufacturing exports can be crowded out by oil rents in many countries, following currency appreciation. This in turn reacts to capital inflows from abroad creating the so-called ’Dutch disease’. The overall relationship between energy/fuel exports and manufacturing exports is negative in most of the period considered. However, some developing countries temporarily escape the RC; for example Mexico (Fig. 3a), China and Malaysia (Fig. 4a). China shows a stronger decline in oil dependence (reduction in RC effects) and higher manufacturing export growth. This is a consistent trend since 1984. By contrast Indonesia and many other countries show strong RC effects (4a). Exceptions to the RC include Japan, China and the Korea Republic (Fig. 4a). India shows RC effects in the 2000s when fuel exports rose dramatically. Throughout the 1970s India´s energy dependence is too insignificant to curtail exports of manufacturing. The evidence suggests that China is escaping the RC effects more quickly than India or Indonesia (Fig. 4a). Figure 4b reveals that countries with low resource dependence (below 5 % Share of GDP) are either resource poor or technology advanced. European countries stand out, but some also suffer from RC effects e.g. Canada, Norway and U.S. (2012). In short potential GDP may have been much higher in the absence of resource exports (4b).

Latin America’s energy dependence (Fig. 3a) has grown at the expense of manufacturing exports and the misallocation of capital and labour; Asia´s dependence shows some parallels: it reduced energy dependence while improving export performance of manufacturing goods (Fig. 4a). Africa´s dependence (Fig. 3b) reveals that most countries remain locked in and path-dependent on an export-led growth model based on oil and other resources, harming GDP growth. The Middle East region continues to have low levels of manufacturing and high rents (Fig. 5).

Considering the institutions model and the growth together, our results show RC effects are not as strong as previously thought. The resource abundance (SubAssets) effect is much more important for growth. Despite the fact the resource abundance is not exogenous enough (Equation 5) we can argue that in some cases underground resources will be associated to development and economic growth. In terms of coefficient size, the resource rent effect will be the biggest effect in the institutions models and for the economic growth models (Table 4 and Table 5) the resource abundance effect is the dominant one.

Figure 3 The arrows indicate from older to newer data. Arrows pointing to the upper left indicate countries escaping from the RC effect, whilst arrows: down or to the bottom right indicate that the RC deepens: a) Latin America, b) Africa Source: authors, based on data in World Bank (2015)

Figure 4 The arrows indicate from older to newer data. Arrows pointing to the upper left indicate countries escaping from the RC effect, whilst arrows: down or to the bottom right indicate that the RC deepens: a) Asia, b) selected OECD economies Source: authors, based on data in World Bank (2015)

Figure 5 Region: Middle East, selected countries

The policy measures proposed in this section stem from 1) the econometric analysis and 2) from overall analysis of the evidence reviewed in the current work. In uncovering the key characteristics of the nexus among resource rents, subsoil wealth, institutional quality and economic growth over time, we are able to suggest some impacts of potential risks arising from institutional activity (or lack of), assess the impact, and recommend responses (Table 6). To be considered as a low risk for international development the political and socio-legal systems (or institution quality) need to be incorruptible and stable: this is borne out by the econometric results which show the importance of institutions and democracy for economic growth (section 4.1 and tables 4-6). We consider the role of institutions and other factors in the light of Collier (2008), who proposes a chain of five key decision points aimed at harnessing economic growth through: 1) changes in how the resource extraction contract is negotiated, 2) improving design features of the contract, 3) improving both transparency in revenues and 4) maximising the aggregate savings decision, and 5) opening up the process on how the public investment decision is taken.

Table 6

An area of concern are the rules regarding exploration and production (extraction). Our models show that production (Rent) are associated to both weaker rule of law (Law) and the democratic experience (Democ). To mitigate this association the legal system should be fair and overseen by an independent judiciary. Furthermore, a lack of rule of law leads to multi-level corruption. An impartial judiciary relies on a policing service free from corruption. For companies to invest in resource extraction, they need to be certain that any contract signed with another company, the Government, or Government agency will be upheld and disputes resolved fairly in the court system. Alongside this requirement is the need for internationally acceptable ethical standards of behaviour (employment, legal, fiscal, etc.) in public life.

The legislature (‘Government effectiveness’) sets the policy environment and the regulatory framework, which should be enforced equally for all actors. Two measures can be adopted. First, improvement in the transparency of the rules governing how exploration and production rights or ownership are granted. Secondly, open data systems (EITI, 2017) should be implemented. Together these will allow the citizenry to monitor the quality of the management of the nation’s resources.

The taxation system is an important arm of the Government and should be predictable and transparent. Transparency means total disclosure of company payments and extraction rents (EITI, 2017), monitoring the sale of (oil) resource concessions, infrastructure provision and barter arrangements. One way to improve transparency is ‘The publish what you pay’ campaign which forces companies to reveal financial information with governments (Collier, 2011).

The final impact of the lack of effective government relates to the economy. The implications for economic growth are clear: countries with a stronger rule of law, more effective government, and better democratic experience are more likely to register GDP growth. How oil revenues are allocated for expenditure on public infrastructure requires open data practices allowing civil society to understand how revenues are recorded in the national budget including revenue management and expenditure. Adherence to the system of national accounts (United Nations et al., 2008) is one such step.