The Effects of Natural Resource Rents on FDI Inflows

Transcription

1 University of Colorado, Boulder CU Scholar Undergraduate Honors Theses Honors Program Spring 2018 The Effects of Natural Resource Rents on FDI Inflows Rodrigo Paton Micale Follow this and additional works at: Part of the Growth and Development Commons, and the International Economics Commons Recommended Citation Paton Micale, Rodrigo, "The Effects of Natural Resource Rents on FDI Inflows" (2018). Undergraduate Honors Theses This Thesis is brought to you for free and open access by Honors Program at CU Scholar. It has been accepted for inclusion in Undergraduate Honors Theses by an authorized administrator of CU Scholar. For more information, please contact

2 The Effects of Natural Resource Rents on FDI Inflows Rodrigo Paton Advisor: Sergey Nigai University of Colorado Boulder Department of Economics Undergraduate Honors Thesis Spring

3 Abstract This paper examines the effects of natural resource rents on Foreign Direct Investment Inflows. I analyze the role that institutional quality has in mitigating or exacerbating these effects, depending on a country s development status. For that, I run a panel data regression at the country level using a data set for 106 countries for the period between 1984 and I use specific categories of natural resource rents (oil, mineral, gas, coal, forest rents) instead of a composite measure to obtain more detailed results. I create interaction terms between different measures of institutional quality and natural resource rents to analyze whether a country s institutional quality can weaken or accentuate effects caused by natural resource rents. The evidence suggests that the relationship between the natural resource rents and the inflow of FDI differ by a country s development status. Further on, the role played by country s institutional quality indicators in weakening or accentuating this effects depends on the type of rents studied and the country s development status. 2

4 Introduction The United Nations Conference on Trade and Development (UNCTAD) has repeatedly emphasized the importance of Foreign Direct Investment (FDI) inflows for developing economies. There is a widespread belief regarding FDI s ability to contribute to sustained economic growth in the long run. This idea originates in FDI s potential for positive spillover effects, including knowledge and technology transfers, job creation, productivity boost and competitive enhancement. The World Bank has also repeatedly encouraged countries to pursue policies that make their markets more attractive to FDI. Governments of certain countries have responded accordingly, with 147 policy changes directed at making their environments more favorable to FDI in 2006, 74% of which happened in developing countries (UNCTAD, 2007). This has encouraged a strand of the literature focusing on the determinants of Foreign Direct Investment. In the literature, several studies try to quantify the effect of natural resource abundance on FDI, but the results are inconclusive (For example see: Bokpin et al. 2015; Poelhekke, S & Van der Ploeg, R, 2012). Two theories have motivated research regarding the existence of an FDI-resource curse. The first argues that resource extraction is a capital intensive activity that requires high levels of capital investment. According to this theory, possible spillover effects, such as job creation and technology transfer to other industries, do not necessarily take place because natural-resource-rich countries tend to devote resources to that same industry. Additionally, this could create an appreciation of the currency and make the non-resource and manufacturing sectors less competitive. This diversion of resources to the natural resource sector creates a crowding out effect relative to the other sectors and is known as the Dutch Disease. The second theory is known as the Resource Curse, which states that natural resource rich countries will have worse economic growth outcomes than the rest of the world. Mehlum, 3

5 Moene, & Torvik (2006) find that the economic growth rates of resource-rich countries depend on whether their institutions are grabber-friendly or producer-friendly. Producer-friendly institutions are distinguished as those where rent seeking and production are complementary activities, and grabber-friendly institutions, where rent seeking and production are competing activities. There is an evident gap in the literature in terms of empirically testing the Resource Curse which is surprising as it may have adverse effects for economic growth, conflict, war, and social conditions. This work aims at answering the following set of questions: (1) How do natural resource rents affect FDI? (2) Can institution quality mitigate/exacerbate this effect? (3) Do these results depend on a country s development status? I answer the first question by running an empirical model based on panel data at the country-year level in which FDI inflows represent the main dependent variable, and natural resource rents represent the main explanatory variable. I control for macroeconomic characteristics of each country by using several macroeconomic and institutional variables suggested by the literature. To answer the second question, I use four specific institutional quality variables. This is novel relative to the literature as previous studies have found no significant effects of institutions when using composite measures. I create interaction terms for three different measures of natural resource rents and three different measures of institutional quality and add it to the baseline regression. Finally, I create a dummy variable that differentiates between developed and developing countries and compare the results for the previous questions. I contribute to the literature on the relationship between FDI and natural resources in several ways. First, I use an extensive data set covering 106 developed and developing countries 4

6 and differentiate results by each group. Poelhekke, S & Van der Ploeg, R (2012) use firm level data from multinational corporations in the Netherlands, Asiedu (2013) limited her data set to developing countries exclusively and Bokpin et al., (2015) focus their study on the African region. Secondly, I use six different variables to measure natural resource rents: total natural resource rents, coal rents, oil rents, natural gas rents, mineral rents, and forest rents. Bokpin et al. (2015) use three different decomposed measures of natural resource rents (oil, mineral, and forest) and found each type had different impacts on FDI inflows and interacted differently with each of the control variables. I now run a similar test using a broader data set and more measures of natural resource rents. Finally, I look at how each of the institutional variables interacts with each individual type of rent. Asiedu (2013) studied the interaction between oil rents and specific institutional quality variables; I extend this study to cover the interaction between each specific natural resource rent and each institutional quality variable, as well as differentiating between each country s development status. Literature Review Literature on the adverse effects that natural resources can create for different countries is broad. This section begins by introducing the main theories regarding natural resources adverse effects on economic outcomes. The Resource Curse is a term given to the negative correlation between a country s growth performance and the abundance of natural resources. Mehlum (2006) finds that this negative correlation is affected by how a country s institutions manage natural resource rents. The author distinguishes between producer-friendly institutions, where rent seeking and production are complementary activities, and grabber-friendly institutions, where rent seeking and production are competing activities. Grabbers are entrepreneurs that target natural resource rents and use all their capacity to seize the greatest amount possible of 5

7 them. They do this through specialization in unproductive influence activities, and the extent of their success depends on the quality of a country s institutions. The author concludes that countries with resource abundance and grabber-friendly institutions have lower GDP growth and thus suffer from the Resource Curse. A closely related term, the Dutch Disease, refers to a situation in which an export-led boom, such as the one created by the discovery of natural resources or a sudden rise in a resource s price levels, creates enough extra wealth to raise the real appreciation of a country s currency and generate a contraction of other tradable activities. There is scarce literature and no consensus on the relationship between FDI inflows and a recipient country s export performance; however, scholars in the field agree that FDI can create real currency appreciation (Froot & Stein, 1991). Kojo (2015) explains that in Long Run economic models, capital and labor are assumed to be perfectly mobile internationally, so the real exchange rate is not affected by an export boom. Because of this long-run equilibrium condition, I am required to carry this study in a short-run framework, but it is important to mention that long-run studies, such as Blonigen & Piger's (2014) provide robust evidence for the inclusion of oil in studies concerning FDI determinants. Research regarding the role of natural resource production in determining FDI in the short-run is scarce. Doytch et al. (2015) study the impact of mining FDI on FDI to the manufacturing, financial, and non-financial sectors. The authors find that it varies across sectors as well as the income group and region that the countries are a part of. Mining FDI seems to crowd in manufacturing FDI in lower middle-income countries, while crowding out total services FDI in upper middle-income and high-income countries. Poelhekke, S & Van der Ploeg (2012) study the effects of natural resources on FDI to the resource sector and the non-resource 6

8 sector. They use firm-level data from multinational corporations (MNC s) in the Netherlands and capture natural resources by using subsoil assets of oil. They find that natural resources attract FDI in the Resource sector but crowd out FDI in the non-resource sector. This crowding out effect is found to be stronger for countries that were not resource producers in the past, but in general, the contractions of non-resource FDI outweigh the gains from resource FDI. They also find institutional quality to have a positive effect on resource FDI. Asiedu (2013) examines the effects of oil exports and oil rents on aggregate FDI inflows. Her paper puts a greater emphasis on the role of institutions by using specific measures of institutional quality instead of an aggregate measure. It also studies whether each of the institutional measures can mitigate the effects caused by natural resources by adding an interaction term to the regression. The author finds that natural resources have an adverse effect on FDI and that better quality institutions can mitigate the effect but not neutralize it. I expand on this research by analyzing the possible mitigation/exacerbation effects of three institutional quality variables on three different types of natural resource rents. Additionally, the results are distinguished by a countries development level. Another study by Asiedu & Lien (2011) examined the effects of Democracy on FDI and found that democracy promotes FDI only if the value of the share of mineral and oil in total exports is low. On the other hand, countries whose exports are dominated by these natural resources will experience a negative correlation between democracy and FDI. Bokpin et al. (2015) focused on the effect of natural resources on FDI in Africa by using data from 49 countries of that region. Their paper uses decomposed measures of natural resource rents, identifying between oil rents, mineral rents, and forest rents. This paper finds that different 7

9 measures of natural resource rents can have different impacts on FDI inflows and interact differently with both macroeconomic and institutional variables. Three theoretical models give us a framework for the inclusion of variables representative of market size, transport, and trade cost, and political and institutional determinants. The OLI (Ownership, Locational and Internalization) framework, developed by Dunning (1980), studies the motivation of transnational companies to invest abroad rather than export their products and distinguishes them by three groups of advantages. This model encouraged analytical FDI theory based on two different types of motivations for FDI: Horizontal and Vertical. According to Markusen (1984), Horizontal FDI takes place when multinationals invest in production in a foreign country instead of increasing production at home and exporting the produced goods. Reasons to engage in this type of FDI include a reduction of trade costs and evasion of trade restrictions. On the other hand, Helpman (1984) describes Vertical FDI as taking advantage of lower factor prices in other economies. Firms could be interested in moving unskilled laborintensive activities to countries with low wages. Markusen & Maskus (2001) developed a knowledge-capital model of multinational enterprises that includes both Horizontal and Vertical motivations. According to their model, horizontal multinationals are attracted to countries with similar size and high total demand (large market), while vertical multinationals are attracted to countries with different sizes and endowments of their own. Data The dependent variable examined is annual net Foreign Direct Investment (FDI) inflows as a percent of GDP. While it would have been ideal to distinguish between resource and non- 8

10 resource FDI and/or sectoral FDI such data is not generally available. Data on net FDI inflows was extracted from the World Bank Development Indicators. I use a number of different measures for the main explanatory variable Natural Resource Rents. First, there is total natural resource rents as a share of GDP, measured as the sum of oil rents, natural gas rents, coal rents (hard and soft), mineral rents, and forest rents. I also include measures for each of the specific sectors rents mentioned above, where each one is measured as the difference between the value of the resource s production at world prices and the total cost of production. The only exception to this are forest rents, which are measured as round wood harvest multiplied by the product of average prices and a region-specific rental rate. Natural resource rents represent the profit generated by extraction activities and are interpreted as the relative importance of resource extraction in a country s economy. Data on Natural Resource rents is extracted from the World Bank Development Indicators. I use four macroeconomic control variables: GDP growth, GDP per capita, inflation and trade openness. GDP growth measures a country s market potential and it is measured as percent annual growth. GDP per capita is intended to represent average domestic income and it is measured in current US dollars. Inflation is measured as the percent change in consumer prices and it is intended to control for macroeconomic stability. Trade Openness is measured as total trade as percent of GDP and it represents locational advantages like reduction of trade cost and avoidance of trade barriers. All of the macroeconomic control variables are extracted from the World Bank Development Indicators. Variables representing specific aspects of institutional quality are used as institutional controls. Previous studies that used composite measures of institutional quality did not find significant effects while studies using specific measures have. I use four different variables: Law 9

11 and Order, Corruption, Government Stability and Investment Profile. These variables were extracted from the PRS s International Country Risk Guide. Law and Order considers the strength and impartiality of the legal system and the popular observance of the law. Corruption can distort the economic and financial environment by reducing the efficiency of government and business through excessive patronage, nepotism, favor-for-favor and secret party funding. The measure emphasizes on these types of corruption, although it also takes into account financial corruption. Both of these measures take values ranging from 0 to 6, where 0 represents the lowest level of institutional performance and 6 the highest. Government stability refers to a government s ability to carrying out its declared programs and its ability to stay in office. Finally, Investment Profile considers factors affecting investment risks that are not covered by political, economic and financial risk components. These factors are: contract viability, risk of expropriation, profits repatriation and payments delays. Both of these measures take values ranging from 0 to 12, where 0 represents the lowest level of institutional performance and 12 the highest. Methodology The dataset examined contains 106 developed and developing countries for years 1984 through To carry out my analysis, I run a panel data regression at the country-year level. To determine the impact of natural resource rents on FDI I will use the following benchmark regression: / / 7 FDI $,& = α * T $& +.01 β. N.,$,& γ 4 x 4,$,& δ z 8 8,$,& +μ & + η $ + ε $,& Where FDI $,& is the main dependent variable, logged foreign direct investment inflows as a percent of GDP in year t and country i. The variable T $& stands for total natural resource rents in year t and country i. The main explanatory variable N.,$,&, represents the different shares in total 10

12 rents for each source of resource rents on year t and for country i: oil rents, coal rents, mineral rents and gas rents with forest rents as reference category. The values for total natural resources are logged. The variable x 4,$,& represents a vector of macroeconomic control variables for each country and year: log GDP growth, log GDP per capita, inflation and trade openness. The variable z 8,$,& represents a vector of institutional control variables: law and order, government stability and investment profile. 1 The terms μ & and η $ represent time and country fixed effects respectively, while the term ε $,& the error term. The equation above is used to test my first hypothesis: that α. <0 when using oil rents as a measure of natural resource rents. Based on previous research I expect the other measures of natural resource rents to have a positive correlation with FDI inflows, that is α. > 0. I expect the macroeconomic variables to have a positive effect on FDI, that is β 4 >0. The only exception to this is inflation, since higher levels of inflation should discourage investors. Finally, I expect better institutional quality to have a positive correlation with FDI, that is γ 8 >0. The second specification model is: / / / 7 FDI $,& = α * T $& +.01 β. N.,$,& γ 4 x 4,$,& δ z 8 8,$,& ε(n. Z 8 ) + μ & + A η $ + ε $,& A In this model, I add the interaction term ε.,8 (N. Z 8 ) to measure any possible mitigation/exacerbation effects that institutional variables could have on the effects of natural resource rents on FDI. One disadvantage to this model is possible collinearity between the different measures for institutional quality. Based on previous results, the institutional variables are expected to mitigate any negative effects created by natural resource rents but not eliminate 7 1 I decided to drop the measure for corruption control due to its correlation with the measure for law and order. 11

13 them completely. If the rents create a positive effect, we expect better intuitional quality to exacerbate it. Finally, a dummy variable is added to the model, which differentiates between developed and developing economies. Countries are classified as developing if their GDP per capita levels are bellow $ and as developed if they are above said level. The dummy variable is added to the regression specifications that exclude the interaction terms and the ones that include them. Descriptive Statistics Table 2 provides summary statistics for developed countries, whereas Table 3 reports summary statistics for developing countries. Comparing the mean values of key variables reflects the importance of differentiating the results across the regression models by a country s development status. First, GDP per capita levels have seen lower growth in developing countries than in developed, as shown by the log values of GDP per capita (Tables 2-3). According to the literature on the determinants of FDI, lower levels of GDP per capita should be an incentive for vertical investment and a counter incentive for horizontal investment. Secondly, mean inflation levels in developed countries are 2.85% and show little variation across the sample but are 67.3% and show high levels of variation in developing countries (Table 2). Higher levels of inflation reflect market instability and are a counter incentive for FDI. Lastly, all of the institutional quality measures have lower mean and minimum values in developing countries then in developed countries. Results In this section I present the regression results of the four measures of natural resource rents used across seven different specifications (Table 4). The results are organized as follows: (i) Column 1 shows the results when including macroeconomic controls, country and time fixed 12

14 effects, (ii)column 2 shows the results when adding institutional control variables, (iii) they are differentiated between developed countries (Column 3) and (iv/) developing countries (Column 4), (v)columns 5 through 7 show the regression results when including all the interaction terms and differentiating by country s development status. The coefficients for each interaction term are found in Table 5. The coefficient for log total natural resource rents is positive and significant across all the specifications of the regression, indicating the relevance of natural resource rents in attracting FDI. All of the macroeconomic controls behave as predicted by the literature. Better institutional quality is positively correlated with FDI across most of the specifications, but some exceptions apply (Table 4). The coefficient for government stability turns negative and significant for developed countries, indicating that a one-unit increase in government stability creates a -.07% decrease in FDI inflows. The main results of this research paper are now presented by testing each of the three hypotheses on each of the specific measures for natural resource rents. Oil Rents The coefficient for oil rents is positive and insignificant when controlling for fixed effects and macroeconomic indicators (Table 4, column 1). This continues to be the case when controlling for institutional quality measures (Column 2), when differentiating between developed and developing countries (Columns 3-4). When adding the interaction terms to the regression (Column 5), the coefficient for oil rents continues being positive and insignificant but some of the interaction terms themselves do take on significant levels. In column 5, the coefficient for oil rents takes the positive value of.336, the interaction term for law and order and oil rents is positive and significant at the 1 percent level while the interaction terms for government stability, or investment profile, and oil rents take negative and significant values at 13

15 the 10 percent level (law x oil=.332, gstab x oil= -.107, invest x oil= ). According to this result, the average country in our sample will see an increase of.1732 percent in FDI. When the variable for law and order increases from its average value of to 6 this effect increases to.386 percent, while if the variable law and order decreases its value to 1 this effect turns negative, indicating a percent decrease in FDI. This indicates that oil rents have a positive effect on FDI and that better quality of law and order will exacerbate this effect while better quality of government stability and investment profile will mitigate this effect (Figure 1). When differentiating between country s development status, some of the interaction terms turn significant for the sample of developing countries (Table 5, Column 7). In Column 7, the coefficient for oil rents takes the positive value of.182, the interaction term for law and order and oil rents is positive and significant at the 1 percent level, the interaction term for government stability and oil rents is negative and significant at the 10 percent level and the interaction term for investment profile and oil rents takes a negative and insignificant value (law x oil=.395, gstab x oil= -.119, invest x oil= ). According to this result, the average developing country in our sample will see a decrease of -.02 percent in FDI. When the variables of law and order decreases from its average value of to 1 this negative effect increases to percent, while if the variable for law and order increases from its average value to 5 this effect turns positive, indicating a.2 percent increase in FDI. This indicates that, for developing countries, oil rents have a positive effect on FDI and that better quality of law and order will exacerbate this effect while better quality of government stability and investment profile will mitigate this effect and ultimately turn it into a negative one (Figure 2). Gas Rents 14

16 The coefficient for gas rents is positive and insignificant when controlling for fixed effects and macroeconomic indicators (Column 1) and turn negative and insignificant when controlling for institutional quality measures. The coefficient for gas rents turns negative and significant at the 10 percent level for developed countries (Column 3), indicating that a 1 percent increase in gas rents generates a percent decrease in FDI. The contrary happens for developing countries, where the coefficient turns positive and significant at the 5 percent level (Column 4), indicating that a 1 percent increase in gas rents generates a percent increase in FDI. When adding the interaction terms (Column 5), the coefficient for gas rents turns positive and insignificant with a value of.999, but some of the interaction terms do take significant levels (Table 5, Column 5). The interaction term for law and order and gas rents is negative and significant at the 1 percent level with a value of -.891, the interaction term for investment profile and gas rents is positive and significant at the 1 percent level with a value of.221 and the interaction term for government stability and gas rents is positive and insignificant with a value of (Table 5, Column 5). According to this result, the average country in our sample will see an increase of.003 percent in FDI. When the variable for law and order decreases from its average value of to 1 this effect increases to.222 percent, while if the variable law and order increases its value to 6 this effect turns negative, indicating a percent decrease in FDI. This indicates that gas rents have a positive effect on FDI and that this effect is exacerbated by better quality of investment profile while better quality of law and order mitigates this effect and can ultimately turn it into a negative one (Figure 3). When adding the interaction terms, the coefficient for gas rents in developing countries turns negative and insignificant with a value of (Table 4, Column 7), but all the coefficients 15

17 for the interaction terms turn significant (Table 5, Column 7). The coefficient of the interaction term for law and order and gas rents takes a negative and significant at the 1 percent level value of , the interaction term for government stability and gas rents is positive and significant at the 10 percent level with a value of.279 and the interaction term for investment profile and gas rents has a positive and significant at the 1 percent level value of.371. According to this result, the average developing country in our sample will see an increase in FDI of.026 percent. When the variable law and order decreases its value from the average of to 1, this effect continues being positive but takes a higher value, indicating a.18 percent increase in FDI. When the variable law and order increases it value to 6, the effect turns negative, indicating a -.18 percent decrease in FDI. This indicates that for developing countries, gas rents have a positive effect on FDI and better quality of law and order can mitigate this effect and ultimately make it negative (Figure 4), while better quality of investment profile and government stability will exacerbate this effect. Mineral Rents The coefficient for mineral rents is negative and insignificant when controlling for fixed effects and macroeconomic indicators. (Table 4, Column 1). This continues to be the case when adding the institutional quality controls (Column 2). The coefficient continues taking a negative and insignificant value for developed countries (Column 3) but turns positive and insignificant for developing countries (Column 4). When adding the interaction terms, the coefficient for mineral rents turn positive and significant at the 5 percent level for developed countries, taking a value of The coefficients of the interactions terms for government stability, or investment profile, and mineral rents take small negative and insignificant values, while the coefficient of the interaction term for law and order and mineral rents takes a negative and significant at the 5 16

18 percent level value of (Table 5, Column 1). According to this result, the average developed country in our sample will see an increase in FDI of.0256 percent. When the variable law and order increases from its average value of to a value of 6, this effect turns negative, indicating a percent decrease in FDI. When the variable law and order decreases to a value of 1 the effect continues being positive, indicating a.64 percent increase in FDI. This result indicates that mineral rents have a positive effect on FDI in developed countries but that better quality of law and order can mitigate this effect and ultimately make it negative (Figure 5). Coal Rents The coefficient for coal rents is negative but insignificant when controlling for fixed effects, macroeconomic indicators and institutional quality measures (Table 4, Columns 1-2). The coefficient turns negative and significant at the 5 percent level for developed countries (Column 3), indicating that a 1 percent increase in coal rents generate a percent decrease in FDI. The opposite happens for developing countries, where the coefficient turns positive and significant at the 5 percent level, indicating that a 1 percent increase in coal rents generates a percent increase in FDI. When adding the interaction terms, the coefficient for coal rents turns negative and insignificant for all countries (Column 5) but negative and significant for both developed and developing countries. For developed countries (Column 6), the coefficient for coal rents takes a negative and significant at the 10 percent level value of , the coefficient of the interaction term for law and order and coal rents takes a positive and significant at the 1 percent level value of 2.019, the interaction term for government stability and coal rents takes a negative and insignificant value of and the interaction term for investment profile and coal rents takes a positive and insignificant value of.142. According to this result, the average developed country 17

19 in our sample will see an increase of.003 percent in FDI. If the variable law and order increases its value from its average of to 6, the positive effect increases to.085 percent. If the variable law and order decreases its value to 1, this effect turns negative, indicating a percent decrease in FDI. This result indicates that for developed countries, coal rents have a negative effect on FDI but better quality of law and order can mitigate this effect and ultimately turn it into a positive one (Figure 6). For developing countries (Column 7), the coefficient for coal rents takes a negative and significant at the 10 percent level value of , the coefficient of the interaction term for law and order and coal rents takes a positive and significant at the 1 percent level value of and the interaction terms for government stability, or investment profile, and coal rents take small positive but insignificant values (gstab x coal=.0196, invest x coal=.129). According to this result, the average developing country will see a.02 percent increase in FDI. If the variable law and order increases its value from an average of 3.16 to 6 this positive effect on FDI increases to.17 percent. If the variable law and order decreases its value to 1 this effect turns negative, indicating a percent decrease in FDI. This result indicates that, for developing countries, coal rents have a negative effect on FDI but greater quality of law and order can mitigate this effect and ultimately turn it into a positive one (Figure 7). Robustness Regressions I run two different tests to check the robustness of the results with respect to the interaction terms as they represent the main variables of interest. In the first test (Table 6, Columns 1-3), we run individual regressions for each one of the interaction terms analyzed. In the second test (Table 6, Columns 4-6), we run individual regressions for each one of the interaction terms analyzed while dropping the institutional quality variables not being analyzed 18

20 by the interaction term of that individual regression. I now report the robustness tests by category of natural resource rent. Oil Rents The robustness tests confirm most of the results obtained when analyzing all countries in the sample (Columns 1 and 4). The directions of the coefficients for the interaction terms remained unchanged in both test. The interaction term for government stability and oil rents increased its significance from the 10 percent to the 5 percent level in the first test (Column 1), but it changed back to its original significance level in the second test (Column 4). The interaction term for investment profile and oil rents lost its 10 percent significance in both test, suggesting that this institutional measure does not impact the relationship between oil rents and FDI. The robustness tests also confirm the results obtained when analyzing the developing countries in the sample (Columns 3 and 6). The direction of the coefficients for the interaction terms remain unchanged. The interaction term for government stability and oil rents lost its 10 percent significance in the first robustness test (Column 3) but regained it in the second one (Column 6). Gas Rents The robustness tests confirm the results obtained when analyzing all countries in the sample (Columns 1 and 4). The directions of the coefficients for the interaction terms remained unchanged in both tests. The interaction term for the government stability and gas rents increased its significance from the 10 percent to the 5 percent level in the first test (Column 1) but returned to its original level in the second test (Column 4). The interaction term for investment profile and gas rents lowered its significance level from 5 percent to 10 percent across both test. 19

21 The robustness tests also confirm the results obtained when analyzing the developing countries in the sample (Columns 3 and 6). All of the coefficients for the interaction terms remained unchanged in the first test with the exception of that for government stability and gas rents, which increased its significance from the 10 percent to 1 percent level. In the second test, the significance levels of the interaction terms coefficients for government stability, or investment profile, and gas rents decreased from the 1 percent to the 5 percent level. Mineral Rents The robustness tests confirmed the results obtained in the previous section of the paper. The coefficient of the interaction term for law and order and mineral rents was negative and significant at the 5 percent level when analyzing all the countries in the sample (Columns 1 and 4) and developed countries exclusively (Columns 2 and 5). Coal Rents The robustness tests confirm the results obtained when analyzing the developed countries in the sample (Columns 2 and 5). The coefficient of the interaction term for law and order and coal rents continues taking a positive and significant value at the 1 percent level across both test. The coefficient of the interaction term for the measure government stability and coal rents turns negative and significant at the 10 percent level in both test, suggesting that for developed countries, greater government stability will mitigate the positive effect that coal rents have on FDI inflows. The first robustness test confirmed the results obtained when analyzing the developing countries in the sample (Column 3). The coefficient of the interaction term for the measure law and order and coal rents continues taking a positive and significant value at the 1 percent level. This changes in the second test (Column 6), where it is lowered to the 10 percent level. 20

22 Robustness Test 3 Both the main dependent variable, FDI inflows, and main independent variable, natural resource rents, are measured as a share of GDP, causing the results to be dependent on the countries economic size. The main results derived are not dependent on the size of the economy, however to assure the main results hold across different econometric models, I run an additional robustness test in which I repeat my original analysis using the levels of both FDI and total natural resource rents. To create my measures for FDI levels and total natural resource rents levels, I multiply each of the variables by GDP measured in current US$ and log the resulting value. FDI Level $,& = ln (FDI $,& GDP $,& ) The robustness test confirmed the majority of previously obtained results. The measure for log GDP per capita turned positive and significant at the 1 percent level across all specifications, confirming that countries with greater market size see greater amounts of FDI inflows (Table 7). The interaction term for oil rents and government stability turned significant at the 10 per cent level and continued taking a negative value in developing countries (Table 8, Column 3). The interaction term for gas rents and investment profile increased its significance level from 10 to 5 percent in developed countries (Table 8, Column 2) and from 5 to 1 percent in developing (Table 8, Column 3). Finally, the interaction term between mineral rents and law and order decreased its significance level from 5 to 10 percent in developed countries. Conclusion In this paper, I analyzed the effects of different categories of natural resource rents on FDI inflows. I also analyzed the link between institutional quality, natural resource rents, and FDI inflows while differentiating the results by a countries development status. To obtain my 21

23 results, I ran a panel data regression at the country level using a data set for 106 countries for the period between 1984 and I also used specific categories of natural resource rents (oil, mineral, gas and coal) instead of a composite measure and created interaction terms for each different measure of institutional quality and natural resource rents. I find that the effects that natural resource rents have on FDI vary across the resource category of the rent and the countries development status. I also find that the relationship between the institutional quality variables, natural resource rents, and FDI varies across rent category and the countries development status. In developing countries, oil rents and gas rents created a positive effect on FDI inflow, while mineral rents showed no significant effect and coal rents created a negative effect on FDI. The results suggest that better quality of law and order will be beneficial to developing countries rich in oil and coal while greater quality of government stability and investment profile will be beneficial to developing countries rich in gas rents. In developed countries, gas rents and coal rents created a negative effect on FDI inflows, while oil rents showed no significant effects and mineral rents showed a positive effect. The results suggest that better quality of law and order will be beneficial for developed countries rich in coal rents while better quality of investment profile will be beneficial for countries rich in gas rents. The previous results hold with the hypothesis that better institutional quality benefits countries rich in natural resources; however, not all of the results in the analysis suggest beneficial effects from greater institutional quality. Greater quality of law and order will hurt developing countries that are rich in gas rents. The results also suggest that greater quality of law and order will hurt developed countries that are rich in mineral rents. These results are counter intuitive and are likely due to the inaccurate measurement of certain institutional quality variables in certain countries. 22

24 A natural extension to this paper would carry a similar analysis using sectorial FDI inflows. However, this data is not directly available and goes beyond the scope of this paper. 23

25 Bibliography Asiedu, E., & Lien, D. (2011). Democracy, foreign direct investment and natural resources. Journal of International Economics, 84(1), Asiedu, Elizabeth. (2013). Foreign Direct Investment, Natural Resources and Institutions. Retrieved November 14, 2017, from Blonigen, B. A., & Piger, J. (2014). Determinants of foreign direct investment. Canadian Journal of Economics/Revue Canadienne d économique, 47(3), Bokpin, G. A., Mensah, Lord, & Asamoah, M. E. (2015). Foreign direct investment and natural resources in Africa. Journal of Economic Studies, 42(4), Doytch, N., U Mendoza, R., & Siriban, C. I. (2015). Does Mining FDI Crowd in Other Investments? Investigation of FDI Intersectoral Linkages. Comparative Economic Studies; New Brunswick, 57(2), Dunning, J. H. (1980). Towards an Eclectic Theory of International Production: Some Empirical Tests. Journal of International Business Studies, 11(1), Froot, K. A., & Stein, J. C. (1991). Exchange Rates and Foreign Direct Investment: An Imperfect Capital Markets Approach. The Quarterly Journal of Economics, 106(4),

26 Helpman, E. (1984). A Simple Theory of International Trade with Multinational Corporations. Retrieved from Kojo, N. C. (2015). Demystifying Dutch Disease. Journal of International Commerce, Economics & Policy, 6(2), 1. Markusen, J. (1984). Multinationals, multi-plant economies, and the gains from trade. Journal of International Economics, 16(3 4), Markusen, J. R., & Maskus, K. E. (2001). Estimating the knowledge-capital model of the multinational enterprise. The American Economic Review; Nashville, 91(3), Mehlum, H., Moene, K., & Torvik, R. (2006). Institutions and the Resource Curse. The Economic Journal, 116(508), Poelhekke, S, & Van der Ploeg, R. (2012). Do Natural Resources Attract Non-Resource FDI? Retrieved November 13, 2017, from f UNCTAD. (2007). World Investment Report: Transnational Corporations, Extractive Industries and Development. 25

27 Appendix Table 1 Summary statistics for all observations Variable Mean Std. Dev Min Max Ln FDI Ln TotNat Oil Rents Gas Rents Min Rents Coal Rents Ln GDP gr Inflation Ln GDP pc Trade Openness Law and Order Government Stability Investment Profile

28 Table 2 Summary statistics for Developed countries Variable Mean Std. Dev Min Max Ln FDI Ln TotNat Oil Rents Gas Rents Min Rents Coal Rents Ln GDP gr Inflation Ln GDP pc Trade Openness Law and Order Government Stability Investment Profile

29 Table 3 Summary statistics for Developing Countries Variable Mean Std. Dev Min Max Ln FDI Ln TotNat Oil Rents Gas Rents Min Rents Coal Rents Ln GDP gr Inflation Ln GDP pc Trade Openness Law and Order Government Stability Investment Profile

30 Table 4 - Regression Results (1) Controls and Fixed Effects (2) Institutional Quality (3) Developed (4) Developing (5) Interaction Terms (6) Developed (7) Developing Ln TotNat 0.227*** 0.221*** 0.181* 0.239*** 0.252*** *** (0.0349) (0.0347) (0.0725) (0.0410) (0.0352) (0.0791) (0.0418) Oil Rents (0.247) (0.245) (0.590) (0.280) (0.439) (2.016) (0.469) Gas Rents * 1.069** (0.265) (0.265) (0.455) (0.362) (0.894) (2.697) (1.330) Mineral Rents ** (0.172) (0.171) (0.339) (0.209) (0.441) (2.509) (0.461) Coal Rents ** 1.463** * * (0.382) (0.378) (0.638) (0.513) (1.537) (4.049) (1.683) Ln GDP gr *** *** ** *** * (0.0247) (0.0246) (0.0477) (0.0285) (0.0242) (0.0468) (0.0279) Inflation * ** ** ** ** ( ) ( ) (0.0196) ( ) ( ) (0.0194) ( ) Ln GDP pc *** *** (0.0772) (0.0776) (0.291) (0.0874) (0.0773) (0.296) (0.0868) Trade *** *** *** *** *** Openness ( ) ( ) ( ) ( ) ( ) ( ) ( ) Law and order 0.216*** *** 0.186*** (0.0332) (0.107) (0.0363) (0.0526) (0.186) (0.0599) Government * stability (0.0161) (0.0382) (0.0188) (0.0284) (0.0691) (0.0321) Investment *** Profile (0.0157) (0.0417) (0.0188) (0.0265) (0.0565) (0.0336) constant 1.738* ** (0.791) (0.787) (2.947) (1.311) (0.787) (3.273) (1.231) Observations adj. R- sq Standard errors in parentheses * p<0.1 ** p<0.05 *** p<

31 Table 5 Regression results for interaction terms (Specifications 5 through 7) (5) Interaction Terms (6) Developed (7) Developing Ln_fdi Ln_fdi Ln_fdi law x Oil 0.332*** *** (0.0868) (0.329) (0.0975) gstab x Oil * * (0.0428) (0.0913) (0.0477) invest x Oil * (0.0393) (0.0693) (0.0516) law x Gas *** *** (0.152) (0.485) (0.241) gstab x Gas * (0.0710) (0.110) (0.112) invest x Gas 0.221*** 0.233** 0.371*** (0.0607) (0.0835) (0.111) law x Mineral ** ** (0.0878) (0.403) (0.108) gstab x Mineral (0.0511) (0.120) (0.0588) invest x Mineral (0.0454) (0.0976) (0.0568) law x Coal 0.705** 2.019*** 1.085*** (0.239) (0.546) (0.276) gstab x Coal (0.103) (0.220) (0.116) invest x Coal * (0.0832) (0.172) (0.108) Observations adj. R- sq Standard errors in parentheses * p<0.1 ** p<0.05 *** p<

32 Table 6 Robustness test (1 and 2) for interaction terms (1) Interaction terms (2) Developed (3) Developing (4) Interaction terms (5) Developed (6) Developing law x Oil *** *** *** *** (0.0772) (0.3056) (0.0854) (0.0771) (0.3033) (0.086) gstab x Oil ** * * (0.034) (0.0762) ) (0.0342) (0.0757) (0.0372) invest x Oil * (0.0319) (0.059) (0.0408) (0.0321) (0.0588) (0.0409) law x Gas *** *** *** *** (.1428) (0.4572) (0.2357) (0.1417) (0.4554) (0.2369) gstab x Gas ** *** * ** (0.0657) (0.0958) (0.105) (0.0663) (0.0956) (0.107) invest x Gas * ** *** * ** ** (0.0541) (0.073) (0.103) (0.0539) (0.0727) (0.1033) law x ** ** ** ** Mineral (0.077) (0.394) (0.0962) (0.0767) (0.393) (0.0969) gstab x Mineral (0.0382) (0.1034) (0.0408) (0.0385) (0.1033) (0.0411) invest x Mineral (0.0325) (0.086) (0.0391) (0.0328) (0.087) (0.0394) law x Coal ** 2.039*** *** 0.644** 1.99*** * (0.2381) (0.506) (0.2724) (0.2373) (0.507) (.2716) gstab x Coal * * * (0.1006) (.210) (0.1123) (0.1014) (0.272) (0.1137) invest x Coal * ** (0.0797) (0.1686) (0.0998) (0.0797) (0.1674) (0.0999) Standard errors in parentheses * p<0.1 ** p<0.05 *** p<