Natural Resource Booms and Economic Growth

Introduction

Economic development is “the process of improving the quality of human lives and capabilities by raising people’s levels of living, self-esteem, and freedom” (Todaro & Smith, 2015). For most developing nations, discovering large reserves of oil and gas is an exciting prospect that could greatly contribute to their economic development. Many Middle Eastern countries for example, have experienced rapid economic growth over the past number of decades, fuelled largely by revenues from natural resource exports. However, a rise in the price of oil and gas in the 1970s led to discussion on the implications of a natural resource boom on the long-term health of an economy. The Economist coined the term “Dutch disease” in 1977. Following the discovery of the Groningen gas fields in the Netherlands in the early 1970s and the subsequent surge in Dutch gas exports, the Netherlands faced a decline in export competitiveness in its other industries. The various mechanisms that caused these effects are investigated in further depth in this report. The primary focus of this paper is to investigate the Dutch disease (DD) syndrome in a resource-rich developing nation, in order to better understand the effects of a natural resource boom on an economy’s development. In particular, the aim is to break down how an oil boom affects the sectoral composition of a developing economy and the implications this has on economic development. There is similar literature that exists but it mostly focuses on an older time period. We hope to provide an updated, relevant analysis with empirical evidence. This paper includes an OLS regression model with data from Nigeria, Cameroon and Ghana to examine the presence of DD in these countries.

The main focus of the empirical analysis is on Nigeria. According to the IMF, Nigeria was Africa’s largest economy and the world’s 29th largest economy in 2018. Nigeria is also a member of the Organisation of Petroleum Exporting Countries (OPEC) due to its status as a major global crude oil producer. As the richest country in the poorest continent and the country with the 11th largest oil reserves in the world (BP Statistical review of World Energy 2017), it is undeniable that Nigeria possesses immense significance as a developing country to the global economy. Economist Jim O’Neil listed Nigeria in a group of four developing countries with particularly strong economic prospects over the next few decades (Business Insider, 2013). This combination of vast economic potential and the abundance of natural resources Nigeria possesses make the country an ideal subject of this study. Cameroon and Ghana were also included in this study to provide context and compare with the results found from Nigeria. Both Cameroon and Ghana are also included in the list of Africa’s 15 largest economies. Cameroon, like Nigeria, is also an oil exporting developing country whereas Ghana is not a major global oil and gas producer. These similarities and differences, along with the geographic proximity of the three nations, provide depth to any discussion of results from the regression analysis. Other countries from West Africa that may share a border with Nigeria, such as Benin or Niger, are not included in the study due to their significantly lower GDP which could distort results and make any comparison drawn between the nations less meaningful.

2. Literature Review

2.1. Causes of Dutch Disease

The theory of Dutch Disease at its core refers to the appreciation of a country’s currency caused by an influx of foreign revenue, which causes disruptions in the domestic economy. One of the reasons for this is that when a country’s exchange rate appreciates, exports become more expensive, imports conversely become cheaper and the country suffers a decline in international competitiveness. To further our understanding of the mechanisms that lead to DD, we will first consider an economy that experiences an oil boom; the country has discovered large reserves of crude oil and has become a net exporter of oil. As Corden and Neary (1982) did in their paper, let us assume a framework with a small open economy that has a traded goods sector and a non-traded goods sector. The non traded sector refers to ‘services’, industries such as construction, finance and public goods. In this framework, the traded goods sector can be broken down into two sub-sectors: manufacturing and energy (the domestic oil industry). We can refer to the oil industry as the “booming sector” and the manufacturing sector as the “non-booming sector”. According to Corden and Neary (1982), the effects of a resource export boom can be decomposed into two segments: the “resource movement effect” and the “spending effect”. As the oil industry grows and becomes more profitable, there is an increased demand for resources, such as labour and capital, from the sector. If we assume labour to be a mobile factor, the oil industry will attract workers from the non booming sector, or “lagging sector”, and drive up wages. The lagging sector loses

resources and struggles to keep up with these wage increases and suffers as a result. The oil industry has therefore benefitted at the expense of the manufacturing industry (Fardmanesh, 1991). This is the resource movement effect (Corden, 1982). However, it is important to note here that the resource movement effect is only valid if the booming sector requires vast amount of resources that can only be extracted from other industries in the economy. This is not always the case. For example, gas extraction in the Dutch energy industry in the 1960s was capital intensive and did not lead to much job creation (The Economist, 2014), meaning that other industries weren’t pressurised and the resource movement effect was arguably negligible (Kremers, 1986). Another example of the resource movement effect being rendered less significant is in Cameroon where the resources used by the oil industry are primarily foreign (Benjamin, 1987) so other sectors’ resources are relatively unaffected. In instances such as these, the DD implications of the natural resource boom would occur because of the spending effect (Corden, 1982) rather than the resource movement effect. The boom causes a rise in real income and this increased wealth leads to greater consumption in the economy. The extra consumption generated can happen in either the traded goods sector or the non-traded goods sector. We assume that the price of traded goods are determined in world markets and therefore remains unaffected by the rise in consumption. However, the non-traded goods sector is affected and the increased demand causes a rise in its price; i.e. a real appreciation. The revenues generated from the oil sector are also used to purchase more imports to meet the increase in demand for traded goods (Benjamin, 1987). These mechanisms are classified under the “spending effect” and lead to adjustments throughout the economy.

Fig. 1 can be used to better demonstrate how the real exchange rate fluctuates as a result of the oil boom. The diagram shows the production possibilities curves between traded goods (manufacturing and oil can be aggregated here into a single traded good as the terms of trade are fixed) on the vertical axis and non-traded services on the horizontal axis. The diagram also shows the highest attainable indifference curves, given the production possibility frontier, to represent the aggregate demands.

Before the boom, the price of services, which can also be considered as the real exchange rate, is given at point a, the shared tangent between the two curves. TS is the pre-boom production possibility curve and I0 is the corresponding highest possible indifference curve. Once the boom occurs, a number of adjustments take place throughout the economy which can be categorised into the resource movement effect and the spending effect. First, the boom increases the maximum output of traded goods from 0T to 0T’ causing the production possibility frontier to shift out to T’S but leaves the production capacity of services unchanged. The point of production has changed from a to b which shows a fall in the output of services caused by labour moving out of the services sector to the booming sector which we know to be the resource movement effect. At this point we are isolating the resource movement effect and not considering the spending effect, so we assume the income-elasticity of demand for services is zero, indicating that the income-consumption curve is a vertical straight line through a which intersects T’S at j. Hence, an excess of demand for services now exists between the production point b and j and this excess demand leads to a rise in the price of services, i.e. a real appreciation to restore equilibrium. However, this adjustment will lead to a new equilibrium that lies on T’S between b and j which means that despite the real appreciation, the resource movement effect has still lessened the output of services to a lower level than it was before the boom. Let us now isolate the spending effect and consider it independently of the resource movement effect, meaning that we assume the oil industry does not use any labour. The boom will cause the production possibilities curve to shift vertically upwards and the new point of production b lies vertically above a. We now assume services are a normal good and the increase in income from the boom raises demand for services. Demand moves along the income-consumption curve of 0n and intersects T’S at point c. The spending effect has therefore also caused excess demand for services and the price of services must rise, i.e. a real appreciation has occurred again. Except this time, the adjustment leads to a new equilibrium that lies between point c and j meaning the output of services has now risen compared to the pre-boom situation. Both the resource movement effect and the spending effect lead to a real appreciation but the impact of both effects combined upon the output of services is ambiguous; the resource movement effect decreases the output whereas the spending effect increases the output. Fig.1 shows a scenario where the spending effect is stronger as g lies on the right side of j. In either case, the boom has raised the equilibrium to point g where the relative price of services is higher than it initially was at a. So far we have considered DD in the context of a natural resource boom, where discovery and subsequent exploitation of a resource leads to an influx of foreign currency and a real appreciation. However, there are other ways in which a country could suffer from similar ailments. Foreign aid, for example, could cause very similar issues. The purpose of foreign aid is to promote growth and tackle poverty in developing nations (Mavrotas, 2010). The aid is often largely allocated to non-traded services within the economy such as healthcare, education and construction. The services sector then experiences an increase in wages which draws skilled labour from the traded sector. The traded sector has therefore experienced a reduction in profitability and competitiveness and this has led to a decline in exports (Rajan, 2011). This is the resource movement effect. Similarly to the scenario borne from an oil boom, the increase in income from the aid will also have a spending effect. The additional income is spent domestically and raises the price of non-traded goods relative to traded goods, further damaging the competitiveness of the traded sector.

2.2. Consequences of Dutch Disease

We have established that DD tends to damage a country’s non-booming traded sectors, such as manufacturing or agriculture. A country with an oil boom for example, could derive the majority of its income from energy exports. The pertinent question then is, why is this an issue? Is there anything wrong with a country relying on raw-material revenues in the long-run to facilitate its economic development? Enders and Herburg (1983) write that despite the label of “disease” associated with a natural resource boom, the overall economic situation of the country is indeed improved in both the short run and the long run. Despite the side effects, the real income of the nation increases and neo-classical trade theory suggests there are welfare gains from the movements of factors of production. There is also the possibility that the booming sector does not require large inputs of factors of production, be it labour or capital. In this instance, the revenue generated from the sector has a large element of economic rent. “The country gains additional income without having to work for it” (Enders, 1983). However, many authors have discussed the problems with relying on economic rent for income rather than having to work for it. Kaldor (1981, p. 8) wrote that a country that relies on rent for income would miss out on the benefits that only a strong manufacturing sector could bring. These benefits include “cultural, technical and intellectual development … and the associated urbanisation”. Ellman (1981, p. that the source of income could cease to exist. For example, a country could use up their natural resource reserves or the global demand for the resource could diminish. In this instance, if the country has missed out on a period of industrial development due to over-reliance on the rental income, it would then struggle to generate income. Another important consideration to note is the volatile nature of commodity markets. Oil markets for example, are renowned for being particularly unpredictable due to the international nature of trading and the political instability of some of the world’s biggest oil producers. Fig. 2 shows the price fluctuations of Brent crude oil, an international benchmark for crude that is widely traded, over the past five years (2014 – 2019). The price of Brent at its highest was well over $100/barrel before it suffered a dramatic collapse and fell to below $40/barrel around a year later. A country that relies heavily on oil revenues is at the mercy of internationally determined prices and is vulnerable to such swings in the oil price and could find its economy suffering as a result.

2.3. Developed Nations

The Economist first coined the term “Dutch Disease” in 1977, in reference to the Netherlands’ economic distress caused by the increase in Dutch gas exports. Following the discovery of large gas reserves, Dutch energy exports soared. The rise in international demand for Dutch gas led to an appreciation of the real exchange rate which then caused a contraction in the non-energy exports sector and inflation in the non-traded sectors (Benjamin et al, 1989). Hutchinson’s (1994) work focused on the prevalence of DD in three Western European, developed economies: The Netherlands, the United Kingdom and Norway. Both the UK and Norway experienced a rapid increase in oil-exporting revenues following discoveries in the North Sea. As noted above, Kremers (1986) discussed the nature of the Dutch gas industry and its lack of direct impact upon the wages of other sectors in the domestic economy or the absence of the resource movement effect. This is due to the capital intensive nature of gas production. However, Hutchinson (1994) reported that the gas sector did in fact have an indirect effect upon the wages of other industries due to higher levels of productivity. The idea here being that wider wage negotiations in the economy were influenced by the aggregate indicators of labour productivity used by the Dutch government in determining wages of certain industries. The aggregate productivity figures were often distorted because of the gas sector and led to an increase in wages in other sectors. Hutchinson (1994) predicted in his paper that “an energy boom will cause a contraction of the manufacturing sector both through resource movement effects and spending effects” in Norway, the Netherlands and the UK. His findings supported this hypothesis as all countries experienced an increase in natural resource revenue over a similar period in which they experienced a decline in their manufacturing industries. However, after performing cointegration analysis, Hutchinson concluded that there was no clear relationship between the development of the energy sector and the manufacturing sector. This view is further supported by Gylfason (2001) who found in his work that concerns of deindustrialisation in the Netherlands were ultimately overstated. The Netherlands suffered from symptoms of DD in the early 1960s but quickly recovered not long after and has since consistently had impressive total export figures. An important point to note here is that there was a wider trend of deindustrialisation amongst developed countries in the late 20th century and the manufacturing sector in the developed countries experienced a contraction. This can be explained by other factors including a global trend of the manufacturing industry moving to developing countries to take advantage of cheaper factors of production. Therefore it would be a reasonable assumption to make that there is a lack of conclusive evidence of DD persisting in developed nations.

2.4. Developing Nations

Benjamin, Devarajan and Weiner’s (1987) paper on DD in Cameroon discusses the natural resource boom in a developing country and provides insight into how this might differ from a developed country. A key difference highlighted is the relationship between imported and domestic goods in developing countries. Domestically manufactured goods tend to be imperfect substitutes for imported goods. This is in contrast to developed nations where consumers are more likely to be able substitute between domestic and imported goods based on preferences and endowments. This is an important point to note because it diminishes the impact of the spending effect. A real appreciation and the additional income generated from the booming sector would likely cause increased demand for imports and a contraction in the consumption of domestically produced goods. However, imperfect substitutability would mean that consumers may not react this way and consumption of domestic goods will not be affected. Another important difference highlighted by Benjamin et al. (1987) is the industries affected by DD in the respective nations. The agricultural industry in developing nations tends to contract following a natural resource boom, instead of the manufacturing sector like in developed countries. Following the boom in oil prices in the 1970s, the manufacturing sectors in developing economies actually expanded (Fardmanesh, 1991). Agricultural sectors on the other hand, were plagued by a number of issues including falling productivity, stagnating output and migration of workers to urban centres (Nyatepe-Coo, 1994). This is because the agricultural sector is more likely to be the dominant sector in developing countries and therefore more likely to be affected by the negative consequences of DD. The final important distinction in a developing country is that the domestic energy industry is more likely to be an “enclave” that mostly uses imported labour and materials. A developing country is less likely to have the level of high-skilled labour or extensive infrastructure and machinery required for oil production for example and would have to use foreign sources instead. The effects of DD would therefore be primarily caused by the spending effect instead of the resource movement effect. Benjamin et al. (1987) reached the conclusions that i) the spending effect alone can have similar effects to the resource movement effect; ii) the agricultural exports of the developing country are most likely to be hurt by the booming sector; iii) imperfect substitutability actually causes some sectors to expand; iv) the income gap between rural and urban consumers widens.

2.5. Cures for Dutch Disease

Enders (1983) discusses a number of possible policy options that are designed to isolate and tackle specific aspects of DD. In particular, he lists seven options:

1. Deficit-financed public demand for manufactured goods

2. Wage freeze

3. Devaluation

4. Production subsidies

5. Import tariffs

6. Improving labour mobility

7. Investment abroad

Enders (1983) concludes that policy options 1-5 are “calmatives” rather than cures of DD; they provide only partial relief in the short and long run. Enders (1983) also states that only options 6 and 7 “can be relied upon as being definitely successful”. As established above, a resource boom leads to the resource movement and spending effects causing inflationary pressures as wages and spending rise. Consequently, the non-booming traded sector contracts with output and employment of this sector falling. Enders (1983) argues that the impact of these effects can be mitigated by increased mobility of factors of production, particularly labour, between sectors in the economy. Greater labour mobility would mean that those who suffer as a result of the contraction in the non-booming sector can instead seek work in the booming or services sector, diminishing the magnitude and duration of adverse adjustments made to the economy. After all, Enders (1983) writes that in the long run, full employment is only restored by workers from the non-booming sector moving to the booming sector. Another significant factor to consider is the additional income generated by the booming sector leading to the spending effect. Enders (1983) discusses how lowering the surge in spending would moderate the disruptive effects of the resource boom and “buys time for a slower and less erratic adjustment process”. Investing revenues from resource exports abroad also has the advantage of allowing the nation to reap the returns of the investment even after the resource boom comes to an end. An important caveat to consider here is that a prolonged resource boom could then lead to a secondary boom in the future caused by the returns from the foreign investments. Therefore, an investment policy must be chosen that allows for fluctuations in revenue and a “steady, or smoothly changing, spending pattern”. For example, Norberg and Blomström (1993) argue that by keeping diamond market revenues in foreign capital markets rather than consuming them domestically, Botswana has managed to largely avoid the spending effect and as a result, there is little evidence that the country’s agriculture or manufacturing industry has suffered from the effects of DD.

3. Data Analysis

The data chosen in this paper was taken from the World Bank Open Data source. The variables included in the model are the share of agricultural output in nonoil GDP, the share of manufacturing output in nonoil GDP, oil revenues as a share of GDP and the global level of imports from Sub-Saharan Africa. Time-series annual data over a 35-year period from 1981 to 2016 was used for Nigeria, Cameroon and Ghana. This time period was selected as 1981 is the first complete year of data for all the variables and 2016 is the latest complete year for the variables openly available from the World Bank online data source. It is important to note that DD is characterised as a syndrome that affects the health of an economy in the short and long-run. By including the maximum number of years openly available, the aim is to provide the most comprehensive results possible that are also relevant for the present time period.

3.1. Key Variables

“Agriculture, forestry, and fishing, value added (% of GDP)” is the name given to the data point by the World Bank. This variable refers to agricultural output as a share of GDP. It is a proxy for the size and health of the domestic agricultural industry in the country. As discussed in chapter 2, the agricultural traded goods industry is the sector most likely to contract as a consequence of DD in a developing country. It is the primary non-booming traded sector. This variable is therefore one of the dependent variables in the model as we hope to examine its relationship with the booming sector. This variable is represented by the notation A.

The other dependent variable chosen is manufacturing output as a share of GDP which represents the size and health of the domestic manufacturing industry. This variable is represented by the variable M. The effect upon the manufacturing traded goods sector in a developing country that has experienced an oil boom is expected to be different to the agricultural sector. This is why the agricultural and manufactured traded goods sectors are modelled independently rather than using the overall GDP of the economy as a whole as a single dependent variable. We hope to highlight the adjustments caused by a resource boom on each industry and therefore provide a more thorough analysis of the various impacts on the economy. An important note here is the adjustment made to the dependent variables. The output of agriculture and manufacturing are both included in the model as a share of nonoil GDP rather than just real GDP. Nyatepe-Coo (1994) emphasises the importance of this in his empirical model. If the variables are not adjusted accordingly, the share of the non-booming sector in the economy would decline even if it was growing, as oil and GDP grow at a faster rate. Therefore, in order to truly isolate the non-booming traded sector’s progress over time, only its contribution to nonoil GDP must be considered. This is done in practice by reducing real GDP, in constant 2010 US$, by the oil revenues as a percentage of GDP. The share of output of the non-booming sectors is then taken of the new nonoil GDP value. The World Bank provides oil revenues as a percentage of GDP and defines “oil rents” as the “difference between the value of crude oil production at regional prices and total costs of production”. This is one of the independent variables used in our empirical model and is denoted by O. This variable is included to represent the nation’s reliance on natural resource exports for revenue. A country with a large share of their GDP being derived from oil revenues would be considered to be heavily dependent on their domestic oil industry and international oil markets. The choice was made to focus on oil revenues, rather than say gas for example, as both Nigeria and Cameroon have experienced considerable oil booms. The other explanatory variable used in the model is global merchandise imports from low- and middle-income economies in Sub-Saharan Africa as a percentage of total merchandise imports, denoted by IM. This variable was included in order to provide a proxy for relevant global trends in trade that may have influenced the traded industries in West African countries. For example, a significant global shift in increasing or decreasing imports from sub-Saharan African countries may better explain the fluctuations in growth of the non-booming traded sectors than revenue from the booming sector. A drawback here is that only regional data for sub-Saharan Africa was found on the World Bank database. Sub-Saharan Africa covers many more countries than just West Africa and hence could lead to unreliable results as other parts of sub Saharan Africa could have different patterns in trade that would influence the overall results. Regional data specific to West Africa would be more useful and relevant.

3.2. Descriptive Statistics

To provide further insight into the data, we examine some descriptive statistics of key variables, the table of this is given in Appendix A. Nigeria’s mean real GDP is more than ten times the size of Cameroon and Ghana’s. This could be expected due to Nigeria’s status as the largest economy in Africa and a top 30 global economy. However the magnitude of this difference is striking. A difference of this size should help us better understand just how much bigger Nigeria’s economy is and this is a factor that should be taken into consideration when we analyse the results of the model. Cameroon and Ghana’s economies are quite similar sizes; Ghana’s real GDP is slightly higher, with a difference of less than US$1bn. Ghana’s standard deviation of its GDP is relatively the highest, with it being over half of the mean. Nigeria’s standard deviation is similar but is slightly smaller relatively at just over half of the mean implying the data for both countries is quite spread out. Cameroon’s standard deviation is relatively smaller than the other two at less than a third of the mean indicating that Cameroon’s data is less spread out, implying homogeneity. We can deduct from this that Cameroon’s economic performance has been relatively more stable from 1981 to 2016 than Ghana’s and Nigeria’s. Moving onto oil rents as a percentage of GDP, we see considerable differences in the maximum and minimum values. Nigeria’s maximum is 54.11 meaning that at the Nigerian oil industry’s peak, it accounted for over half of Nigeria’s GDP indicating just how heavy Nigeria’s reliance was on oil at one point. However, the minimum is much smaller at 3.02 and this figure occurred in 2015. The figure in 2016 was also 3% which implies that Nigeria has gone a long way in diversifying its economy and reducing its reliance on oil revenue. Cameroon’s maximum figure is 12.62, a considerable amount but nowhere near as high as Nigeria’s maximum. With a mean of 0.76 and a minimum of 0.01, Ghana’s economy is by far the least dependent on oil revenues. As for the size of the agricultural sector in the countries, Nigeria and Cameroon’s mean sizes are 23 and 18.67 respectively meaning that the agricultural industries contributed to around 20% of GDP; a sizeable chunk. Both have almost identical standard deviations too of around 4.8 which is considerably smaller than their means. With regards to the dispersion of data, this implies the size of the agricultural sector has not fluctuated too much over the past few decades. Ghana, on the other hand, has quite different figures for its domestic agricultural industry. Ghana’s mean is 37.39 and its maximum is 59.73, indicative of Ghana’s heavy reliance on agriculture. The high figures for agricultural contribution to GDP for all three countries is in line with our expectations as they are all developing countries. As for the manufacturing sectors contribution to GDP, the mean value for Nigeria and Cameroon is again very similar at 14.72 and 14.63 respectively. Ghana’s mean value is lower at 8.72. Nigeria and Cameroon also have similar maximum values at 21.1 and 18.23 whereas Ghana’s value is again lower at 11.75. These figures could reveal a striking difference in the levels of economic development between the countries. Healthy manufacturing industries are often considered as a positive sign of economic development for developing countries who previously relied on agriculture. The manufacturing sector tends to require more technology, higher skilled workers and a better quality of capital than the agricultural sector. Nigeria and Cameroon having larger manufacturing industries and smaller agricultural industries than Ghana could indicate their economies have progressed and developed quicker than Ghana’s by going through periods of industrialisation.

It is also interesting then to note that both Nigeria and Cameroon are considered to have experienced oil booms too whereas Ghana has not. A possible conclusion to be drawn from this then is that whilst a country experiencing an oil boom may face the ailments of DD, it is still better off as a result of the oil boom than it would have been had it not experienced an oil boom at all. However, this data is not comprehensive enough to decisively reach this conclusion.

3.3. Correlation Analysis

A potential issue that should be addressed is the presence of collinearity within the regression model. Collinearity would exist if the independent variables within our model are highly correlated. Although the reliability of the model is unaffected by this phenomenon, the standard errors will be increased. This is an issue because it could cause a variable to appear significant when it is not or vice versa.

There is a degree of collinearity between the independent variables for all three countries but not to a level where it is an issue that needs to be addressed. Had any variables shown a correlation of ±0.75 or greater, the variable would have to be omitted.

4. Methodology

The empirical model used in this paper is an ordinary least squares (OLS) regression analysis model using time-series annual data. OLS models minimise the sum of squared residuals. In a regression analysis, the term “residuals” refers to the differences in the actual observed values and the estimated values given by the model. A regression analysis was used in this paper to try and determine a possible link between the development of non-booming sectors in an economy and the booming sector.

4.1. Empirical Model

The model presented in this paper is based on the model produced by Fardmanesh (1991), in a paper that was written for the academic journal “World Development”. Fardmanesh (1991) analyses the effects of an oil boom on five developing OPEC countries using a reduced-form three-sector model. The share of the agricultural, manufacturing and non-traded output to nonoil GDP all take the form of the dependent variable in separate relations with the ratio of oil revenues to GDP and an index of world relative price of manufactured goods to agricultural products for less developed countries making up the two explanatory variables in each relation. The model used in this paper is similar but has some key differences. We use a reduced two-sector model, omitting the non-traded services sector as a dependent variable. This is because this data could not be found for the countries used in our model. We find that even without this variable, our model remained relevant as the effects upon the traded non-booming sectors are still analysed. The suffering of traditional industries in an economy like agriculture and manufacturing is arguably the primary negative consequence of DD so it is important that this part of the analysis was kept intact. Where Fardmanesh (1991) concentrated on five oil-exporting countries from various geographic regions, we take a different approach by instead focusing on three countries with differing oil industries but from the same region. This modified method was taken because we hope to clarify the effect of an oil boom on a country’s economic development, rather than simply testing the effects of an oil boom on an oil-exporting country. It is our belief that this can be done more effectively by comparing countries with similar economic characteristics, such as geography and culture, and using the oil boom as an independent variable to compare the difference in results. The model in this paper also uses time-series annual data but for the period 1981- 2016 whereas Fardmanesh (1991) focuses on the period 1966-86. This is done to create a more relevant, up-to-date model for today’s realities. The number of years observed is also increased in our model to 36 years compared with 21 years in Fardmanesh’s (1991) model. The sample size was increased to try and provide more reliable results as DD is a syndrome that affects an economy’s health over the long-term. For further discussion on why these specific years were chosen, refer back to chapter 3. The final difference between the two models is in the second independent variable. Fardmanesh (1991) uses an index of world relative price of manufactured goods to agricultural products for less developed countries. This variable is deemed inadequate for our model and is replaced with the share of global merchandise imports from low- and middle-income economies in Sub-Saharan Africa of total merchandise imports.

As our model focuses on countries from West Africa, we feel a region-specific variable that identifies possible relevant trends in global trade is more relevant in determining the state of traded sectors in the region’s economies. The variable used in Fardmanesh’s (1991) model is for less developed countries globally, a flawed variable when considering countries specific to a region. The two relations of our model estimated are:

A = α0 + α1O + α2IM + e1

M = β0 + β1O + β2IM + e2

Where β0 and ε0 denote the constant terms, αi and βi (i = 1, 2) are the estimated coefficients of the two independent variables and ei (i = 1, 2) is the error term.

A: Share of agricultural output in nonoil GDP

M: Share of manufacturing output in nonoil GDP

O: Ratio of oil revenues to GDP

IM: Share of global merchandise imports from low- and middle-income economies in Sub-Saharan Africa of total global merchandise imports

5. Results

Six regressions were run in total, all simple OLS. Both the agricultural and manufacturing sectors were modelled for each country with oil revenues and regional trade as the independent variables. The results of our model are presented in Table 4. The full tables generated by the statistical software Eviews are presented in Appendix B.

5.1. Nigeria

Nigeria’s agricultural sector showed a negative but insignificant relationship with oil revenues. Specifically a unit increase in the share of oil revenue in GDP would contract the agricultural sector by 0.09 units. A negative relationship was expected as empirical literature suggests an oil boom in a developing country tends to cause the agricultural sector to contract. The insignificance of this relationship is surprising however, especially considering the scale of the Nigerian oil industry. However it is worth noting that in Fardmenesh’s (1991) model, he found a similarly negative relationship of -1.36 at a 1% significance level using data from 1966-86. He also mentions that Nigeria’s agricultural sector contracted by a whopping 43.25% in the period 1970-82 when oil prices fell sharply after rising for years prior. It is therefore a possibility that the Nigerian agricultural industry had already suffered the worst of the effects of DD by 1981 – the starting point of our model, which helps explain the lack of significance. Table 4 also shows that the Nigerian manufacturing industry had a positive, significant relationship with oil revenues. A unit increase in oil revenues as a percentage of GDP would expand the manufacturing sector by 0.21 units, at a 5% significance level. This result supports the hypothesis that an increase in oil revenues actually boosts the manufacturing sector in a developing economy. As discussed in chapter 2.4, this is primarily due to imperfect substitutability between domestically produced manufactured goods and imports in a developing economy. In this instance, the revenues generated from oil exports could support the advancement of the country’s manufacturing sector. With regards to the differences in time periods, Anjande (2017) presents a Dutch disease model in Nigeria from 1981-2014, a similar time period to the one used in this study. Anjande (2017) concludes that Dutch disease does exist in Nigeria in the above time period. However a key difference between Anjande’s (2017) model and the model used in this paper is that Anjande (2017) measures the impact of an oil boom on overall GDP rather than looking at the effect on sectoral composition as we have done with agriculture and manufacturing. Therefore we can’t conclusively state whether our findings are completely in line with Anjande (2017) in terms of overall GDP. However, if we understand Dutch disease in the context of a developing country specifically as a phenomenon that adversely affects the agricultural industry and supports the manufacturing industry, as described by Benjamin (1987), then our results are in line with Anjande’s (2017) as we also find evidence of this definition of Dutch disease in Nigeria.

Another noteworthy Nigerian result is the agricultural sector’s relationship with global sub-Saharan imports. The model estimated a negative relationship, with a relatively large coefficient of -7.26, at a 1% significance level. This implies that global imports from sub-Saharan Africa increasing causes the agricultural sector in Nigeria to contract. The reasoning behind this is unclear. It is possible that the Nigerian agricultural sector has forfeited some of its comparative advantage to regional rivals as a result of exogenous factors. Eyo (2008) suggests that macroeconomic policies in Nigeria have possibly caused an environment that was not agricultural sector friendly. In particular, Eyo (2008) cites ineffective exchange rate policy, management of inflation and a lack of access to credit as key determinants of the state of the Nigerian agricultural sector. These findings could potentially help explain the significant negative relationship our regression analysis produced for the agricultural sector and global imports from sub-Saharan Africa.

5.2. Cameroon

Moving onto Cameroon, the results of the model suggest the agricultural sector has enjoyed a positive relationship with oil revenues over the past number of decades. Our findings suggest a unit increase in the share of oil revenues in GDP would cause the agricultural sector to expand by 0.91 units at a 5% significance level. This is contradictory to our understanding that the effects of Dutch disease mean an oil boom causes the agricultural sector to contract in a developing country. Benjamin et al. (1987) explain that Cameroon actually implemented policies that helped to offset the harm to the non-booming sector. First of all, a large proportion of the revenues generated from oil exports were saved abroad, tackling the spending effect. Secondly, the oil revenues that were spent domestically by the government raised “the producer price of cash crops, lowering the effective export tax … counteracting the cost increase from the real exchange rate appreciation”. This could help to explain the positive relationship between the agricultural sector and oil revenues found in our model. Similar to the Nigerian agricultural sector, the Cameroonian manufacturing sector also proved to have a negative relationship with global imports from sub-Saharan Africa at a 1% significance level. Again, the reasoning behind this is unclear and a comprehensive investigation of this is beyond the scope of this paper. However, Söderling (2000) finds that the manufacturing sector in Cameroon has contracted considerably since the 1980s, largely because of the effects of Dutch disease and “inward-looking policies for the manufacturing sector”. It can therefore be reasonably hypothesised that Cameroon suffered a decline in competitiveness to regional export rivals, and as a result, experienced a negative relationship between its manufacturing sector and global imports from sub-Saharan Africa.

5.3. Ghana

With regards to a relationship with global merchandise imports from sub-Saharan Africa, Ghana’s manufacturing industry suffers similarly to Cameroon’s according to the results of our model. The agricultural industry however, enjoys a positive relationship. Our model estimates a coefficient of 8.67, the highest of any variable in all of the regressions and the only positive relationship with global imports from sub-Saharan Africa. This would indicate that unlike the other sectors in the other countries, Ghana’s agricultural sector benefits significantly from a shift in global trade trends towards sub-Saharan Africa, implying that Ghana has some level of comparative advantage over its regional production rivals. Indeed to support this point, Wolter (2009) comments that there are indications of an emerging modern agricultural sector in Ghana. As for Ghana’s sectoral relationship with oil revenues, the agricultural and manufacturing results are both in line with our expectations, based on the empirical literature. Agricultural output is negatively related to oil revenues, with a coefficient of -4.92 at a 1% significance level. This suggests that increasing oil revenues leads to a contraction in the agricultural sector, a result that we expect of a developing country. The manufacturing relation is also as expected of a developing nation; positive. However this is at a 10% significance level, which we deem not an acceptable level of significance to definitively draw a conclusion from. It is interesting to note however, that the Ghanaian agricultural sector had the most pronounced adverse effects from an increase in oil revenues, a relationship that we’d expect from a developing country, according to empirical literature. This is a surprising result nonetheless, seeing as Ghana has the smallest oil industry out of the three countries and is not considered to have experienced an oil boom. A possible explanation of this is that both Nigeria and Cameroon’s governments have actively implemented policies since 1981 to try and counteract the effects of Dutch disease whereas the Ghanaian government has not, due to the lack of a domestic booming oil industry. This would make the non-booming traded Ghanaian sectors more susceptible to the Dutch disease ailments.

6. Conclusion

The primary focus of this paper was to better understand the effects of an oil boom on a developing country’s economic development and in particular, use empirical evidence to provide further insight into how Dutch disease affects the sectoral composition of an economy. The results were mixed but still partly in line with findings of prior research. Rising oil revenues cause the domestic agricultural sector in a developing country to contract as the spending effect and resource movement effect lead to a number of adjustments throughout the economy. Also, due to imperfect substitutability between domestically produced manufactured goods and imports in a developing nation, an oil boom can actually lead to the manufacturing sector to expand as the increased revenues are spent on industrialisation. This was supported by the estimated effect of oil revenues on the Nigerian manufacturing industry in our model. The unexpected results of Cameroon, along with the explanation from Benjamin et al. (1987) seemed to suggest that the worst of the adverse effects of Dutch disease can be avoided by implementing effective macroeconomic policy, such as investing a proportion of the revenues generated from oil exports abroad to offset the spending effect. This view is potentially further supported by Ghana’s results which showed some of the strongest adverse effects to oil revenues, a possible consequence of a lack of government policy addressing Dutch disease due to Ghana’s relatively insignificant oil industry.

There are possible limitations of this research, specifically to do with the model presented. Firstly, by choosing 1981 as the starting point, much of the effects of the severe fluctuations in oil prices of the 1960s and 1970s are absent from the model. Although one of our objectives was to provide an updated, relevant analysis of previous work done, some of the clearer Dutch disease adjustments may have been omitted using the time period that was used. For example, Fardmanesh (1991) found that Nigeria’s agricultural sector had contracted by 43.25% from 1970 to 1982, indicating that Nigeria had already suffered the worst of the Dutch disease effects by 1981, the starting point of our model. Ultimately, the purpose of this research is to contribute to the wider literature available on developmental economics. We hope to have helped improve understanding of the role of natural resource exports in the advancement of a developing economy. The idea of discovering large reserves of oil and gas may not be the ‘jackpot’ it was once considered, but with proper management of increased revenue and shrewd government policy, a developing country could avoid a contracting agricultural sector and further the advancement of its domestic manufacturing sector, contributing to rapid economic development.

7. References

Anjande, G. (2017). An Economic Analysis of Dutch Disease in Nigeria. International Journal of Management Studies, Business & Entrepreneurship Research, 2(1).

Benjamin, N.C., Devarajan, S. and Weiner, R.J. (1989) 'The ‘Dutch’ disease in a developing country: Oil reserves in Cameroon', Journal of Development Economics

Ellman, Michael, "Natural Gas, Restructuring and Re-industrialisation: The Dutch Experience of Industrial Policy". In: Terry Barker and Vladimir Brailovsky (Eds.), Oil or Industry? London 1981

Enders, K. and Herberg, H. (1983) 'The Dutch Disease: Causes, Consequences, Cures and Calmatives', Weltwirtschaftliches Archiv,.

Eyo, E. (2008). Macroeconomic Environment and Agricultural Sector Growth in Nigeria. World Journal of Agricultural Sciences.

Gylfason, T. (2001). Lessons from the Dutch disease: Causes, treatment, and cures. Institute of Economic Studies.

Hutchison, M.M. (1994) 'Manufacturing Sector Resiliency to Energy Booms: Empirical Evidence from Norway, the Netherlands, and the United Kingdom', Oxford Economic Papers

Kremers, Jeroen J.M. "The Dutch Disease in the Netherlands." in Natural Resources and the Macroeconomy. eds. J. Peter Neary and Sweder Van Wijnbergen. The MIT Press: Cambridge, MA, 1986

Mavrotas, G. (2010) Foreign Aid for Development: Issues, Challenges, and the New Agenda. Oxford University Press.

Neary, J.P. and Corden, W.M. (1982) 'Booming Sector and De-Industrialisation in a Small Open Economy', Economic Journal

Norberg, H., Blomström, 1993. Dutch disease and management of windfall gains in Botswana. In Blomström, M., Lundahl, M. (Eds.), Economic Crisis in Africa: Perspectives on Policy Responses.

Nyatepe-Coo, A.A. and Nyatepe-Coo, A.A. (1994) 'Dutch disease, government policy and import demand in Nigeria', Applied Economics

Söderling, L. (2000) 'Dynamics of export performance, productivity and real effective exchange rate in manufacturing: The case of Cameroon', Journal of African Economies

Todaro, M.P. and Smith, S.C. (2015) Economic development. Twelfth edn. Harlow, England: Pearson.

Wolter, D. (2009) 'Ghana: Agriculture is Becoming a Business', OECD Journal: General Papers

World Economic Outlook. (2019). International Monetary Fund.