The Dutch disease : natural recourses a curse or a blessing?

1

Bachelor Economics and Business

University of Amsterdam

Under guidance of C.W. Haasnoot

26-06-2018

The Dutch Disease:

Natural recourses a curse or a blessing?

Mink Perrée

10781722

Bachelor thesis

Abstract

According to the Dutch Disease theory, sudden windfalls in natural recourses can lead to declining manufacturing industries. This research attempts to confirm the existence of the Dutch Disease and focusses on the role of institutional quality in deciding whether recourse abundancy is a curse or a blessing. I use regression analysis to find a relationship between manufacturing exports, natural recourses rents and institutional quality. The results show no evidence for the existence of the Dutch Disease. Furthermore, no evidence is found that institutional quality effects manufacturing exports in case of a sudden recourse boom.

2 Statement of Originality

This document is written by Mink Perrée who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

3

Contents

4

Introduction

On the 29th _{of May 1959 a field of gas was discovered in Groningen. All of a sudden The Netherlands}

were rich because this gas was worth hundreds of billions of Gilders. The government almost directly began to exploit this gas. However, despite the natural resource abundancy, the economy was in bad shape during the 1970’s with industries declining and unemployment increasing. This phenomena was named the ‘Dutch Disease’ by an article in the Economist in 1978. This article concluded that because of the export of natural gas, the exchange rate appreciated. This resulted in higher imports and lower price competitiveness for exports. The result was higher unemployment and decreasing firm profits in the manufacturing sector. Unemployment increased from 1.1% in 1970 to 5.1% in 1978.

In the literature different causes are being analyzed in causing the Dutch Disease. Over the recent years, much focus is being laid on the role of institutional quality. Boschini (2007) came to the conclusion that institutional quality is playing an important role as to whether recourse abundancy is positive or negative for a country. Furthermore he discovered that the type of natural recourse is also decisive. Mehlum (2006) concluded in his research that resource-abundancy can be a curse or a blessing depending on weather institutions are grabber friendly. National income will decrease when institutions are grabber-friendly and increase when they are producer-friendly. Murshed (2008) concludes that historically recourse abundancy is a blessing. However, recourse abundancy can become a curse

because of two reasons. First, macroeconomic mechanisms like the spending effect caused by excessive spending and a relative price effect due to the appreciation of the real exchange rate. Second,

abundancy can become a curse because of an adverse political economy, for example, bad institutional setting and corruption.

Earlier research mainly focused on the role of the exchange rate when there was a windfall in natural recourses. Corden and Neary (1982) found in their research that because of this windfall in natural recourses the exchange rate appreciated. This meant a decline in demand for domestic products from foreign countries and therefore a decrease in the manufacturing industries. Harding and Venables (2010) investigated the relationship between recourses exports and non-recourses exports. They found that for each euro extra recourses-export means a 65 cents decline in non- recourses exports. Stijns (2003) did an empirical research on the Dutch Disease and made use of a gravity model. He discovered that there is a negative relationship between the energy exports and manufacturing exports, as well between the world energy price and manufacturing export.

5 The role of institutions is an important part of my thesis. First of all, I want to investigate

whether the Dutch Disease exists. Second, I want to study whether the institutional quality of a country is playing a significant role in explaining the Dutch Disease. Therefore my research question will be: Is resource abundancy a curse and what is the role of institutions on the Dutch Disease? To answer this question I will make use of empirical data that I will use for a regression. I will be using manufacturing exports as my dependent variable and natural recourses rents and institutional quality as explanatory variables.

My first hypothesis is that I expect a negative effect of natural recourses rents on the

manufacturing exports. My second hypothesis is that I expect the combined effect of natural recourses rents and institutional quality to be negative. In the next section I will be discussing the literature on which my research will be built on. I will then move on by discussing the methodology and data. Next, I will show the results of my empirical research followed by a discussion on these results. Finally, I will conclude my thesis. The results show that there is no significant evidence that supports either of my hypotheses.

Literature review

Since the Dutch Disease was introduced by The Economist in 1978, many research has been done on the Dutch Disease. In this section I will start by using literature that explains what the Dutch Disease is. Next, I will discuss a modelling framework that this paper will be based on. I will then move onto empirical research on the resource curse and finally I will focus on literature that links the Dutch Disease and institutional quality.

Wierts and Schotten (2008) explain in their work that the Dutch Disease was caused for two reasons. First, because of exporting the natural gas from Groningen the Guilder appreciated. The result was, as stated before, diminishing firm profits and high unemployment. The other reason is that politicians have a tendency to increase government spending to increase welfare to an extent that it becomes unsustainable once the natural recourses dry up (Gylfason and Zoega, 2002; Van der Ploeg and Pelhekke, 2009). This unsustainable government spending occurred in The Netherlands during the 70’s. A sound way to go about with the revenues from the natural gas would have been to put the revenues

6 in a fund and only spent the return. The government decided to use this money however for

government spending, for example pensions and unemployment benefits.

The model of Corden and Neary (1982) explains the Dutch Disease. The model contains three goods in a small open economy. Goods ‘energy’ (XE) and ‘manufacturing’ (XM) are tradable and good

‘services’ (XS) is non-tradable. XE and XM are determined by global supply and demand, XS by domestic

supply and demand.

In this model the exchange rate is being defined as the relative price between non-tradable and tradable goods. Furthermore, three important assumptions are being made. First, the markets

of the two production factors (labor and capital) are perfect. Second, labor is the only mobile factor and each good has its own capital factor. Third, the model uses real terms meaning national output is equal to national expenditure.

According to the model there are two effects: the resource movement effect and the spending effect. The spending effect means an increase in demand on XS which is the result of an increase of the

real income that is caused by a windfall in natural recourses. Prices of services increases as a result of this increased demand. Since the exchange rate is defined as the relative price between non-tradable and tradable goods, this means an appreciation of the exchange rate. The recourse movement effect means that mobile recourses are reallocated because of an increase in the marginal product of the factors in the sector that is booming.

7 Total labor supply is represented by de distance of OSOT. The distance from OS is the labor input for services and the distance from OT is the labor supply for total traded goods. Labor demand for services is represented by the LS curve, labor demand for total tradable goods is represented by LT and labor supply for manufacturing is represented by LM.

Figure 1 illustrates the effects of a boom in the energy industry. First we take a look at the recourse movement. Because of the boom, the demand for labor in the energy sector goes up. This is depicted by the shift LT to LT’. A new equilibrium is formed in point B. Therefore, labor has shifted from manufacturing goods to energy goods and has gone from m to m’. The decrease in labor input means lower services output and thus an excess demand for services. To reach a new equilibrium, the exchange rate must appreciate (exchange rate = relative price between non-tradable and tradable goods, see figure 2). Since the relative price of services has increased, demand for labor in services increases creating a new equilibrium in G. Labor input for manufacturing exports is further decreased to m’’ and thus an even further deindustrialization.

Next the spending effect is being analyzed. Because of the boom, the production frontier is expanded from TS to T’S (see figure 2). Demand for services is depicted by curve ON at the initial

exchange rate. Point b is the highest possible indifference curve that can be reached after the boom. ON and T’S intersect at the right of point b meaning that there is an excess demand for services. Therefore, prices of services increases. This leads to an appreciation of the exchange rate in order to restore the equilibrium.

In conclusion, output of services is decreased by the recourse movement effect resulting in excess demand for services and thus an appreciation of the exchange rate. Demand for services is increased by the spending effect again leading to excess demand and thus an appreciation of the exchange rate. Both effects contribute to deindustrialization. In my thesis I will use manufacturing exports as my dependent variable instead of the exchange rate. I will elaborate on this further in the methodology and data section.

8 As mentioned before, there have been several empirical studies on the Dutch Disease. Chen and Rogoff (2003) did empirical research on the relationship between the real exchange rate and natural recourse booms. They tested the relationship between the real exchange rate and world commodity prices. The countries tested were Canada, Australia and New Zealand. They used these countries because of a large share of commodity exports and relatively small size so that they don’t effect the world price of commodities. A significant relationship between the real exchange rate and world commodity prices is found for New Zealand and Australia. When the trend of declining world prices is being left out of the relationship, the relationship is not significant for Canada. What can be concluded therefore is that there is a positive relationship between commodity price shocks and the real exchange rate. And thus the Dutch Disease exists. However the research was only being done for three developed countries so it isn’t very useful in analyzing developing countries which are the countries that most often are affected by the Dutch Disease.

To find more general evidence on the the Dutch Disease, Stijns (2003) uses a gravity model of trade to study the recourse curse. In his paper Stijns tests four hypotheses. Exchange rate appreciation, increase in non-traded output, decrease in manufacturing sector production and a decrease in

manufacturing exports. The first hypotheses has been tested by Chen and Rogoff (2003). For the second and third hypotheses there is not enough data to do testing. Therefore, only the last hypotheses is being

9 tested in this paper. To test the fourth hypotheses, Stijns uses world energy prices because he makes the assumption that world energy prices are exogenous to manufacturing trade and net energy exports can be endogenous. Furthermore, Stijns has three different ways to test the fourth hypotheses. He uses world energy prices, net energy exports and net energy exports combined with world energy prices as independent variables.

His results confirm that both world energy prices and net energy exports have a negative relationship with manufacturing exports for all three approaches. Therefore, convincing evidence has been found that support the Dutch Disease. He finds that a one percent increase in net energy exports will, ceteris paribus, decrease a net energy exporter’s real manufacturing exports by on eight of a percent. Furthermore, a one percent increase in the price of energy will, ceteris paribus, decrease a net energy exporter’s real manufacturing exports by half a percent.

Another way of looking at the Dutch Disease is to see how a sudden foreign exchange windfall effects the balance of payment, as was studied by Harding and Venables (2010). The accounts that were studied are non-recourses exports, total imports and net accumulation of foreign assets. They compared two types of foreign exchange windfalls; net recourses exports and foreign aid. They used data from 133 countries over a time period of 1960 to 2000. In their results they found that every dollar or recourses exports leads to a decrease of 50 cents for non-recourses exports, imports are increased by 15 cents and savings are increase by 35 cents. For every dollar in foreign aid, exports remain almost the same and imports are increased by 40 cents.

The question remains as to what causes the recourse curse to occur. In his work, Ranis (2000) uses six different factors that may stand economic development in its way due to recourse abundancy. First of all, he thinks there is too little focus on human development in countries that have excessive recourses rents. Second, recourse abundancy leads to rent-seeking activity at the cost of production activity. Third, industrialization is substituted by import. Fourth, the rents of recourses are not equally divided because of the interests of the elite which Mehlum (2006) sees as the result of grabber friendly institutions. Fifth, there is a lack of export diversification leading to decreasing economic growth since the volatility of energy prices are much higher than manufacturing good prices. Finally, the

competitiveness of non-mining tradable sectors is weakened because of the Dutch Disease effects. Auty (2001) criticizes Ranis because he does not give an explanation as to why recourse abundant countries are underperforming. Furthermore, he finds that in other Dutch Disease literature there is little emphasis on the role of political and institutional structure. In his work, Auty (2001) comes

10 with three factors that link economic performance and recourse abundancy. These factors are the dominant system of landholding, the type of political state and the choice of development strategy.

In response to Auty’s criticism of there being too little emphasis on the role of institutions in the Dutch Disease, Mehlum (2006) introduced the concept of grabber-friendly institutions. According to his theory, a country consists of grabbers and producers. In case of a bad institutional quality, a windfall in natural recourses will result in grabbers seizing all the benefits from the windfall resulting in an unequal distribution of recourses revenues. In case of perfect institutional quality, those institutions will equally divide the recourses benefits between the grabbers and the producers. To test the hypothesis that institutional quality plays a decisive role in the recourse curse, Mehlum (2006) makes use of a regression. He uses GDP growth as a dependent variable and among some control variables he uses recourse abundancy and institutional quality as explanatory variables. The institutional quality index is an unweighted average of five indexes of Political Risk Services from Keefer and Knack (2002): Rule of law, bureaucratic quality, corruption in government, risk of expropriation and government repudiation of contracts. These characteristics are various aspects that indicate whether institutions are more grabber or producer friendly. The indexes are on a scale of zero to one with the higher the higher the value the better the institutional quality. The results show that the recourse curse is weaker the higher is institutional quality.

Boschini (2007) also confirms that institutional quality is vital in deciding whether recourse abundancy is a curse or a blessing, but wants to investigate whether the type of recourse also plays a role. He did a regression using the average yearly growth rate as the dependent variable and measures of recourse wealth and institutional quality as independents variables, including an interaction variable. Four different measures of natural recourses are being used by Boschini (2007). The broadest measure is the share of primary exports to GNP from Sachs and Warner (1995) which includes everything from meat to precious metals. The second broadest is a measure including the share of exports of ores and metals to GDP. The third broadest measure is the share of mineral production in GNP which excludes ores and is a production measure instead of an export measure. Finally, the fourth measure isolates the most appropriable recourses. It uses the value of production of gold, silver and diamonds as a share of GDP. For institutional quality, the average of the Political Risk Services from Keefer and Knack (2002) are being used. The results show that indeed institutional quality has a significant effect. Better institutions have a positive effect on the average growth rate and vice-versa. Furthermore, he finds significant evidence that the effect of institutional quality (either positive or negative) on the average growth rate is stronger for recourses that are easier to appropriate, like diamond, gold and silver.

11

Methodology

To investigate whether the Dutch Disease exists I will make use of a regression. I want to show a

relationship between manufacturing exports and natural recourse wealth. For the regression I will make use of panel data since it is more suitable than cross-sectional data or single time series (Verbeek, 2008) for data that varies over multiple dimensions, in this instance time and countries.

ME= α + β1NaRecRe + β2InstiQua+ β3NaRecReInstiQua + µ

The dependent variable ME is the manufacturing exports as a percentage of the total

merchandise exports. Merchandise exports are exports of goods and exclude services. NaRecRe are the natural recourses rents as a percentage of the GDP as estimated by the World Bank (2018). Natural recourses rents are the sum of oil rents, natural gas rents, coal rents (hard and soft), mineral rents and forest rents. For the measure of institutional quality I will use six categories as constructed by the World Bank (2018). Each category has been given a value between -2.5 and 2.5 with the higher the value the better the institutional quality. The categories used are control of corruption, government effectiveness, political stability and absence of violence/terrorism , regulatory quality, rule of law and voice and accountability. I have added up these values to construct one variable for institutional quality which is InstiQua. I have added interaction variable NaRecReInstiQua between natural recourses rents and institutional quality to capture the combined effect on the manufacturing exports. µ gives the error term and α is the constant.

If the Dutch Disease does exist, β1 should be negative since a windfall in natural recourses should lead to a decrease in manufacturing exports. β2 could be positive since better institutional quality would normally mean that a country has better economic conditions and therefore would have a positive influence on manufacturing exports. It could however also be negative because countries that have good institutional quality are often richer countries that have low merchandise exports. For example, in 2016 The Netherlands had 67.3 manufacturing exports as a share of total merchandise exports and has a rating between 0.9 and 1.98 for each institutional quality parameter as calculated by the World Bank (2018). In contrast, China had 93.8 manufacturing exports as a share of total

merchandise exports and has a rating between -1.62 and 0.36 for each institutional quality parameter. The interaction variable could be either positive or negative since I expect the effect of natural recourses on manufacturing exports to be negative and the effect of institutional quality on manufacturing exports

12 could either be positive or negative. Nonetheless, for either a positive and a negative effect of

institutional quality I expect the coefficient of the interaction variable to be negative.

Data

In order to investigate the relationship between manufacturing exports and natural recourse wealth and institutional quality, the framework used in the literature review has to fit into a model that consists of measurable variables. The model of Corden and Neary (1982) and the work of Stijns (2003) propose multiple possible dependent variables that can be used. For example, in both their works, an increase in the real exchange rate is observed. They define the exchange rate as the relative price of non-traded goods to traded goods. A variable could be constructed for the exchange rate by dividing a price index for non-traded goods by a price index for traded-goods.

Next manufacturing output is considered as a dependent variable. Tradable goods output could be a suitable variable for this. Another choice as a dependent variable could be non-traded output per country. Non-traded output should increase according to the theory in case of a natural recourse windfall. However I could not find enough data to use this as my dependent variable. Especially the developing countries often lacked data while these are often the countries that are recourse-abundant.

However the most commonly used dependent variable are non-recourses exports (Stijns, 2003), (Harding and Venables, 2010) because the availability is widely available. A downside of using non-recourses exports is that its results might be misinterpreted. A decrease in exports doesn’t have

to mean that output is decreased while an increase in exports must lead to an increase in output. This is because the rise in domestic demand could have been more than the decrease in exports. Stijns (2003) however finds that this is unlikely to occur since the effect of the decrease in manufacturing exports outweighs the effect of the rise in domestic demand.

Taking all of these arguments in consideration I have chosen to take manufacturing exports as my dependent variable. The dataset I will use for my regression is extracted from the World Bank (2018) and are defined as the manufacturing exports as a percentage of total merchandise exports.

13 (chemicals), 6 (basic manufactures), 7 (machinery and transport equipment) and 8 (miscellaneous manufactured goods) and exclude 68 (non-ferrous metals).

Considering which independent variables to use to explain manufacturing exports three

different types of recourse booms have to be taken into account (Corden and Neary, 1984). First of all, a recourse boom can be due to a once and for all exogenous technical improvement. Second, a windfall discovery of new recourses can cause a recourse boom. Third, a recourse boom can be caused by an exogenous rise in the world price of natural recourses relative to the import price. For the third option a proxy is often being used for recourse and development expenditures. This data is however hard to find. Technological improvements in a specific sector is hard to measure since data on technological

improvement overall is already hard to measure.

Net energy exports and world energy prices are being used by Stijns (2003) and its data is available widely. Since developing countries often discover new volumes of natural recourse, windfall discoveries could be a fit approach. Examples of these sudden windfall discoveries can be found in Uganda (2009) and Ghana (2007) and The Netherlands (1959). Natural recourses exports as a percentage of GDP is used by most studies to account for natural recourse, for example Sachs and Warner (1997) and Arzeki and van der Ploeg (2010).

An alternative could be the value of natural capital as calculated by the World Bank (2018). Natural capital is calculated by considering natural gas, oil, coal, minerals, metals, forests, agricultural lands and pasture lands. Bond and Malik (2009) and Brunschweiler and Bulte (2009) among others choose this approach in their research on the recourse curse. The upside of using the value of natural capital as calculated by the World Bank (2018) is that it is more accurate compared to natural recourse exports since the value of natural capital are the endowments of a country and natural recourse exports are only what we sell abroad. The downside is that there is only data on 1997, 2006 and 2011. As an alternative, I choose to use recourses rents as a percentage of a countries GDP (estimated by the World Bank) as an independent variable. Total natural resources rents are the sum of oil rents, natural gas rents, coal rents (hard and soft), mineral rents, and forest rents. Natural recourses rents

are the difference between the price of a commodity and the average cost of production. The price of units of specific commodities are estimated and subtracted by estimates of the average costs of

extraction of harvesting per unit (including a normal return on capital). Next, unit rents are multiplied by the physical quantities countries extract or harvest to calculate the rents per commodity as a share of gross domestic product (GDP). This variable covers exports as well as domestic revenues and has widely available data.

14 When plotting a scatter between manufacturing exports and natural recourses rents we find one value that seems odd. The measure for manufacturing exports is a percentage of the total merchandise exports. Therefore a value of above 100% is impossible. Thus, I will remove this value for manufacturing exports from Iran in 2005 before doing regression analysis.

In order to investigate the role of institutions on the recourse curse, I will make use of data gathered by the World Bank (2018). As mentioned before the World Bank (2018) uses six indicators of institutional quality on a country based level. Those are control of corruption, government effectiveness, political stability and absence of violence/terrorism, regulatory quality, rule of law and voice and

accountability. Each indicator has been given a value between -2.5 and 2.5. -2.5 is the lowest rating the World Bank (2018) gives for each indicator meaning institutions are poor and 2.5 is the highest value meaning institutions are very good. I have chosen to first use those indicators as separate variables to determine for each category if there exists a significant relationship on manufacturing exports. I then continue testing by adding up the six indicators to form one variable for institutional quality in order to make an interaction variable between natural recourses rents and institutional quality and to account for multicollinearity. Data on institutional quality is available for the period 1996-2016, excluding 1997, 1999 and 2001. Therefore, I choose to do my regression on this time period (so excluding 1997, 1999 and 2001). 0 1 0 0 2 0 0 3 0 0 4 0 0 M a n u E x p 0 20 40 60 80 NaRecRen Graph 1

15

Results

In this section I will show the results of my empirical research. First I will present a scatterplot to give a graphical illustration of the relationship of manufacturing exports and recourses rents. The scatterplot has manufacturing exports on the y-axis and total natural recourses rents on the x-axis. I took a sample from the year 2006 since using my entire data set wouldn’t have given a clear scatterplot as a result of the large amount of observations in my dataset. What is shown from the graph contradicts my hypothesis. I expected a negative relationship between manufacturing exports and recourses rents. However this graph shows a positive relationship. This would suggest the Dutch Disease does not exist. We will examine the relationship further by looking at regression analysis.

0 2 0 4 0 6 0 8 0 1 0 0 0 20 40 60 80

Total natural resources rents (% of GDP)

Manufactures exports (% of merchandise exports) Fitted values

16

Table 1

The regression is depicted in table 1. The first model includes only natural recourses rents as a

dependent variable. For the models 2 to 7, we run a regression between manufacturing exports, natural recourses rent and one of six parameters for institutional quality as calculated by the World Bank (2018). We see that natural recourses rents has a significant positive effect on manufacturing exports which goes against my expectation of a negative relationship. Government effectiveness, regulatory quality, rule of law and voice and accountability have a significant effect on the manufacturing exports. The regression is in line with the scatterplot of 2006 above, since the parameter of natural recourses rents is positive (contrary what I expected). The beta’s of government effectiveness, regulatory quality, rule of law and voice and accountability have a positive value meaning that the better these parameters are the higher are the manufacturing exports.

There are a few adjustments to be made to the regression to give more reliable results. First of all, I expect the regression to suffer from high multicollinearity since I expect the parameters for

institutional quality to be highly correlated with each other. Therefore I constructed a correlation matrix to see if this is indeed the case. From the correlation matrix we can clearly see high correlations

between the variables of institutional quality. For example, the correlation between control of corruption and rule of law is 0.9407 and are thus close to being perfectly correlated. Therefore the regression will suffer from high multicollinearity. To solve this problem I have added up the values of the different parameters to construct one variable for institutional quality.

17

Table 2

My research question focusses on the combined effect of institutional quality and recourse abundancy. This graph only shows the individual effect of natural recourses rents and the institutional quality parameters on manufacturing exports. In order to test the combined effect, I will make use of an interaction variable between recourses rents and the newly constructed variable for institutional quality. The new interaction variable is NaRecRenInstiQua2. Running the new regression gives table 3. The results show that both natural recourses rents and the overall institutional quality have a significant positive effect on manufacturing exports, however, the combined effect is not significant. Instead of a negative relationship as I expected, table 3 shows a positive relationship between manufacturing exports and natural recourses rents. The R-squared is 1.2% so the model explains a small fraction of the variation. Table 3 VoiAcc -0.5143 0.7641 0.7471 0.6698 0.7677 0.8174 1.0000 RuleLaw -0.4184 0.9407 0.9309 0.7719 0.8994 1.0000 RegQua -0.4157 0.8655 0.9303 0.6395 1.0000 PolStab -0.3084 0.7323 0.6860 1.0000 GovernEff -0.4054 0.9257 1.0000 CntrlCorr -0.3892 1.0000 NaRecRen 1.0000 NaRecRen CntrlC~r Govern~f PolStab RegQua RuleLaw VoiAcc

18 Next, we have to look at which type of model should be used. The random effects model contains variation in the true correlation that is being estimated for each study, for the fixed model the true correlation of each study is the same. Therefore the observed variation in estimated correlations is assumed to only be due to the effect of random sampling in the fixed effects model. We run the Hausman test to decide whether a random or a fixed model is more appropriate. The results are shown in table 4. Our null-hypothesis is that the random effects model and the fixed effect is appropriate. If the hypothesis is rejected, this means only the fixed effects model is appropriate.

Table 4

The p-value is very small and therefore we reject the null hypothesis. We will thus use the fixed effects model to account for invariant effects from the model. The results are depicted in table 5. The evidence shows that recourses rents and the combined effect of natural recourses and institutional quality have no significant effect on manufacturing exports. Institutional quality has a significant negative effect on manufacturing exports. The variables explain 2.5% of the model.

19

Table 5

There are many very small countries in the dataset like Malta, Andorra and San Marino. I want to investigate whether the results would be different if I would exclude those countries from the data. To do this, I run a fixed effects regression only for countries with a population of above 1 million inhabitants. In the results in table 6 natural recourses rents and the interaction variable are still insignificant but now also institutional quality has become insignificant. The explanatory value has decreased to 0.006.

20

Discussion

My research focusses on trying to find evidence for the existence of a negative relationship between manufacturing exports and recourses rents and investigating what the role of institutions is on the recourse curse. After accounting for invariant effects, outliers and excluding countries with less than one million inhabitants, both of my hypotheses are rejected. No significant evidence is found that supports that natural recourses rents have a negative effect on manufacturing exports. Therefore my thesis does not confirm the existence of the Dutch Disease. Furthermore, I found evidence for a negative

relationship between institutional quality and manufacturing exports. Finally, no evidence is found that supports my hypothesis that the combined effect of natural recourses rents and institutional quality have a negative relationship with manufacturing exports.

A possible explanation for not finding significant results for recourses rents and the interaction variable might be that there is not enough variation in the data. Since there was no data on institutional quality from before 1996, I only used data on a small period of time. Only in eight instances there was an increase of more than 5% in natural recourses rents. In comparison, recourses rents increased with 300% between 1973 and 1974 in the Netherlands after the gas field in Groningen was discovered. The same problem arises with institutions since improving or declining institutional quality often happens over a longer period of time so only small changes are being observed.

What could explain the negative relationship between institutional quality and manufacturing exports is the fact that richer countries tend to have better institutional quality compared to poorer countries but at the same time rich countries often import manufacturing goods instead of producing and exporting them, also discussed in the methodology section.

One of the issues with doing regression analysis is that the regression could suffer from omitted variable bias. Important variables can be left out that are correlated with other independent variables. In my analysis, I did not use any control variables which led to a low explanatory value. Part of the omitted variable bias problem is solved by using a fixed effects model since this accounts for the invariant time effects in my regression.

Analyzing institutional quality, the variable that I used could have been constructed in a more sophisticated way. I simply added up each parameter to form one overall variable for institutional quality while a better technique would have been principal component analysis. What this does is reducing a set of possibly correlated variables to one uncorrelated variable that is the principal component.

21

Conclusion

My thesis tries to find evidence for the existence of the Dutch Disease. What distinguishes my thesis from earlier studies is that I tried to find a relationship between manufacturing exports, natural

recourses rents and institutional quality to explain the Dutch Disease. Most studies use natural capital or energy exports as their explanatory variable. In contrast to energy exports, recourses rents captures both exports and domestic revenues so it is more inclusive. Furthermore, its data is much more widely available than natural capital.

The results show no evidence for the existence of the Dutch Disease. As discussed, a cause for recourses rents and the combined effect of recourses rents and institutional quality not to be significant could be a lack of variation in the explanatory variables. Further studies could focus on trying to find data for a longer periods to have more variation in the independent variables or to focus on countries that experienced recourse windfalls in case studies. Also, a more sophisticated technique could be used to create the overall effect of institutional quality on manufacturing exports. Furthermore, I found a significant negative relationship between manufacturing exports and institutional quality. Finally, my regression doesn’t contain control variables. Further research could be focusing on adding these variables to improve the regression to account for the possibility of omitted variable bias.

This thesis shows no indication that there is a relationship between manufacturing exports and recourses rents and as a result this paper does not confirm the Dutch Disease. Therefore, no evidence is found that recourse abundancy is curse. Policy makers that have concerns over shrinking manufacturing exports when exploiting recourses in case of sudden windfall might want to reconsider their

standpoints. Furthermore this paper shows no evidence that improving institutional quality has an effect on manufacturing exports in case of a sudden recourse boom. However, in order for policy makers to make policy decisions when an abrupt discovery of recourses occurs, further research should be done. Using a dataset over a longer period of time could capture more variation between both natural recourses rents and institutional quality and might give different results. Alternatively, further analysis could focus on doing case studies on countries that experienced natural recourse windfalls.

22

Reference list

Arezki, R. and Ploeg, van der, F. (2010), Trade policies, institutions and the natural resource curse. Applied Economic Letters, 17, 1443-1451.

Auty, R.M. (2001a) 'The Political Economy of Resource-Driven Growth', European Economic Review 45(4): 839-846.

Auty, R. (2001b). “Why resource endowments can undermine economic development: concepts and case studies”, Paper Prepared for the BP-Amoco Seminar, Lincoln College Oxford University, November 28-29.

Auty, R. (2001c). (Ed.) “Resource abundance and economic development”,Oxford: Oxford University Press.Economic Review 45(4): 839-846.

Auty, R.M. and A.H. Gelb (2001d) 'Political Economy of Resource-Abundant States', Resource abundance and economic development: 126-144.

Bond, S.R. and Malik, A. (2009), Natural resources, exports structure, and investment. Oxford Economic Papers, 61, 675-702.

Boschini, A.D., Petterson, J. and Roine, J. (2007), Resource Curse or Not: A Question of Appropriability. Scandinavian Journal of Economics, 109 (3), 593-617.

Brunnschweiler, C.N. and Bulte, E.H. (2009), Natural resources and violent conflict: resource abundance, dependence, and the onset of civil wars. Oxford Economic Papers, 61, 651-674.

Chen, Y. and Rogoff, K. (2003), Commodity Currencies. Journal of International Economics, 60, 133- 160. Corden, W.M. and Neary, J.P. (1982), Booming Sector and De-Industrialization in a Small Open Economy. The Economic Journal, 92 (368), 826-841.

Corden, W.M. (1984), Booming Sector and Dutch Disease Economics: Survey and Consolidation. Oxford Economic Papers, New Series, 36 (3), 360.

23 Economist, the. 1977, November 26th, The Dutch Disease. 82-83.Gylfason and Zoega, 2002

Harding, T. and Venables, A.J. (2010), Foreign Exchange Windfalls, Imports and Exports. Working Paper, University of Oxford, Department of Economics.

Keefer, P., & Knack, S. (2002). Polarization, Politics and Property Rights: Links between Inequality and Growth. Public Choice, 111(1/2), 127-154.

Mehlum, H., Moene, K. and Torvik, R. (2006), Intstitutions and the Resource Curse. The Economic Journal, 116, 1-20.

Murshed, S.M. (2008) 'Development Despite Modest Growth in the Middle East’, Review of Middle East Economics and Finance 4(3): 1-31.Ranis (2000)

Ploeg, van der, F. and Poelhekke, S. (2009), Volatility and the Natural Resource Curse. Oxford Economic Papers, 61, 727-760.

Ranis, G., F. Stewart and A. Ramirez (2000) 'Economic Growth and Human Development', World Development 28(2): 197-219.

Sachs, J.D. and Warner, A.M. (1997), Natural Resource Abundance and Economic Growth. Working Paper, Center for International Development and Harvard Institute for International Development. Stijns, J. (2003), An Empirical Test of the Dutch Disease Hypothesis Using a Gravity Model of Trade. Working Paper, University of California, Berkeley, Department of Economics.

Verbeek, M. (2008), A guide to modern Econometrics, Third Edition. John Wiley and Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, England

Wierts, P. and G. Schotten (2008) , Dutch natural gas revenues and fiscal policy: Theory versus practice . World Bank (2018) World Development Indicators. http://databank.worldbank.org/ ddp/ home.do