Unbundling the Resource Curse∗ (Preliminary and incomplete, please do not quote without permission)

Anne D. Boschini†, Jan Pettersson‡ and Jesper Roine§ November 2008

Abstract This paper decomposes the resource curse and its potential reversal in three important dimensions; type of resource, type of institutional measure, and time. Studying components of primary exports we find that the curse is (to various degrees) present for all components but that its possible reversal is mainly driven by the interaction between institutions and ores and metals. With respect to institutional measures results are similar for outcome based measures and more durable rules. Changing sample starting dates gives qualitatively similar results, despite the fact that there is great variability in the importance of resources over time. Finally, we address the issue of institutions being determined by our resource measures and conclude that some resources certainly seem to influence institutional development, but that this is unlikely to drive our results. Our results also hold when instrumenting institutional quality. Keywords: Natural Resources, Resource Curse, Property Rights, Institutions, Economic Growth, Development JEL: O40, O57, P16, O13, N50 ∗

Paper submitted for the DEGIT-XIII conference in Manila, November 2008. Financial support from Sida (grant SWE-2005-329) is gratefully acknowledged. † Corresponding author: Anne Boschini, Department of Economics, Stockholm University, S-106 91 Stockholm, Sweden. E-mail: [email protected] ‡ Department of Economics, Stockholm University. E-mail: [email protected] § SITE, Stockholm School of Economics, P.O. Box 6501, SE-113 83, Stockholm, Sweden. E-mail: [email protected]

1

1

Introduction

While it seems clear that there is a robust negative relationship between a country’s share of primary exports in GDP and its subsequent economic growth, it seems equally clear that there are plenty of exceptions to this general pattern.1 In the recent past natural resources have been positive for economic growth in countries such as Australia, Botswana, Canada, and Norway, and historically there are also many examples of resource led growth.2 As Frederick van der Ploeg (2007) notes in a recent overview, ”the interesting question is why some resource rich economies [...] are successful while others [...] perform badly despite their immense natural wealth”. Recent work by Mehlum, Moene and Torvik (2006) suggest that the answer lies in differences in institutional arrangements across countries. When institutions are ”grabber friendly” resources push agregate income down, while resources under ”producer friendly” institutions raise income. Similarly, Boschini, Pettersson and Roine (2007) propose that the extent to which natural resources are good or bad for growth depends on their ”appropriability” in two dimensions. First, natural resources do not by themselves harm growth, but become a problem in the absence of good institutions (institutional appropriability) and second, for some types of resources this problem is bigger than for others (technical appropriability). Both these studies find empirical support for the basic idea that resources can have positive effects on growth given that institutions are good enough, emphasizing the interaction effect between these variables.3 In this paper we analyze the interaction effect and its possibility to re1

The negative relationship between the primary export share and subsequent growth was first established in a cross-section in Sachs and Warner (1995), and its robustness has been confirmed in, for example, Gylfason, Herbertsson and Zoega (1999), Leite and Weidmann (1999), Sachs and Warner (2001) and Sala-i-Martin and Subramanian (2003). Doppelhoefer, Miller and Sala-i-Martin (2004) find that the fraction of primary exports in total exports is negatively related to growth and one of 11 variables which is robust when estimates are constructed as weighted averages of every possible combinations of included variables. 2 The point that resources have contributed positively to growth in the past has forcefully been argued by Wright (1990), David and Wright (1997) and Findley and Lundahl (1999). See also Wright and Czelusta (2004). 3 Note that the interaction effect introduced in these studies is not the same as controlling for institutional quality. This has been done in many previous studies, including – as pointed out by Mehlum et al. (2006) – the study by Sachs and Warner (1995), without changing the negative relationship between primary exports and growth.

2

verse the resource curse in grater detail. Starting from a basic regression of the type used in Mehlum et al. (2006), which focuses on the effect of interacting the broad resource measure of primary exports in GDP with a composite measure of institutional quality, we study to what extent we can add precision to their argument by decomposing the result with respect to the type of resource and to the measure of institutional quality used. The reasons for doing so are based on previous research. With respect to types of resources it has been argued that so called ”point-source” resources, such as plantation crops, minerals and fuels, are more problematic than ”diffused” ones.4 Characterized by being more ”centrally controlled”, point-source resources lead to more societal division and weaker institutions, which in turn are associated with lower growth.5 We therefore decompose primary exports in its four main components; exports of agricultural raw materials, food exports, fuel exports and exports of ores and metals. We find ores and metals to stand out as the resource class having both a resource curse and a positive reversal through stronger institutions. Regarding the measure of institutions, Acemoglu and Johnson (2005) distinguish between property rights institutions (rules and regulations to protect citizens against the power of the government and elites) and contracting institutions (rules and regulations to enable and facilitate contracting between citizens). This adds to the distinction made by Glaeser et al. (2004) between outcome based measures of institutions (assessed as a function of past events) and rules based measures (where laws and rules are defined ex ante in order to constrain the political body’s behaviour, such as constitutions). They ar4

E.g. Auty (1997), Woolcock, Pritchett and Isham (2001), Isham, Pritchett, Woolcock, and Busby (2005), and Bulte (2005). Leite and Weidmann (1999) also find different effects from components of primary exports on growth (through their different effects on corruption) and Ross (1999) find important differences in that fuels and ores and metals have negative effects on democratic development while food and agriculture, if anything, are positive. Sachs and Warner (2001), on the other hand, argue that the distinction is not very important. 5 Engermann and Sokoloff (2000) have a similar argument suggesting that American colonies, which were either dominated by plantation agriculture or mining (Carribbean and South America) or mainly non-plantation agriculture (North America), developed different institutional environments as a result. High concentrations of wealth in the South lead to institutions favouring the elite, while more dispersed wealth in the North lead to institutions providing more political influence and more economic opportunity to the broader population. These developments, in turn, have led to very different long-term economic development to the advantage of the North.

3

gue that most indicators of institutional quality that is used in the literature are outcome measures (reflecting recent elections or governments past policy choices) and as such are not measures of institutions in terms of durable and binding constraints on policy choices. Different institutions should therefore be expected to have a different impact on how resources affect growth. Assume that the institution ”law and order” is of main importance in order to ensure productive extraction of resources, then another measure of institutions, say the degree of democracy, may work poorly. This could partly be explained by democracy being a less precise measure of the important institutional mechanism (equal democracies having different levels of law and order would imply that law and order captures the cross country variation better), but most importantly these two measures are not conceptual substitutes - law and order may be effectively enforced even in autocracies. This paper employs two main measures of property rights institutions: ”institutional quality” constructed from the ICRG data base and the polity-variable from the Polity IV data base. Institutional quality is a composite measure of how risky investments are, the extent of corruption, the transparency and strength of the legal system and the strength and quality of the bureaucracy. The polity index is also composed from underlying indices and aims at measuring the degree of democracy along a 21-point scale. However, polity puts more constraints on the legeslitative power the higher up the index while ICRG is a fully history dependent coding of the outcome of institutional workings. Hence, polity is to a much larger extent rules based than is ”institutional quality”. Our general results indicate that the polity measure yields less precise estimates of the relation between institutions and growth, but there are differences between specifications and time periods. Moreover, individual countries differ in their resource intensity over time. The choice of startyear should therefore be expected to influence results. Therefore, we systematically run our regressions on four different time spans using a homogenous country sample. All time periods end in 2005, the startyears are, respectively, 1965, 1970, 1975 and 1980. While we do obtain different point estimates for different time periods, we do not find any pattern in the differences and the general results seems robust to the choice of startyear. The article proceeds as follows. Section 2 presents the data we are using and Section 3 specifies the empirical model. Results are presented in Section 4 efter which we address the problem of institutions being endogenous to growth as well as the potential endogeneity between institutions 4

and resources. Influencial observations are considered in Section 6. Finally, section 7 concludes.

2

Data

All data in this article comes from the World Bank online data set World Development Indicators. As the broadest measure of resources we use the share of primary exports to GDP, P rimExp. This is the sum of the exports of agricultural raw materials, food, fuels and ores and metals. This measure is the same as the one used in Sachs and Warner (1995), by them labelled SXP . Hence, in terms of appropriability this measure includes everything from meat to precious metals. We then employ the four different measures of natural resources to explore potential differences between them.

3

Empirical model

Our first econometric specification for testing the proposed effects of resources and institutions in country i becomes growthi = Xi′ α + β1 N Rij + β2 Insti + β3 (N Rij × Insti ) + εi ,

(3.1)

where growth is the average yearly growth rate of per capita GDP between varying startyears and 2005, X is a vector of controls including initial GDP per capita level, period averages of openness and investment ratios, dummy variables for Sub Saharan Africa and Latin America respectively and a constant. N R is a measure(s) of natural resource wealth where j = 1 or j = 1, 2, 3, 4 depending on specification (i.e. primary exports or its four components). Inst is our measure of institutional quality. N R × Inst is the interaction(s) between natural resources and institutional quality. The existence of a resource curse and its reversal implies that β1 is negative (the standard resource curse finding), β2 is positive (the standard finding that good institutional quality is beneficial for growth) and that β3 , the coefficient for the interaction between natural resources, is positive and - if it is to reverse the resource curse - have an absolute value larger than β1 .6 This 6

The fact that our measures of institutional quality has been rescaled to a 0-1 measure allows us to directly compare the coefficients.

5

would mean that as long as the institutional quality is good enough, natural resources will have a positive net effect on economic growth. However, we do expect results to differ between types of resources. Furthermore, we hypothesize that the results should be different depending on the used measure on institutions. Since resource intensity for our different resource measures differ between 1965-1980, we also expect results to vary between choice of startyear.

4

Results

Our first results, using two different measures for institutions (Institutional quality, constructed from ICRG, and Polity2 from the PolityIV data set) and four different start periods (1965, 1970, 1975 and 1980) are presented in Table 1. We find varying results regarding the potential for institutional quality to reverse the resource curse (upper panel). Focusing on ICRG (columns 1-4), the signs are as expected in every regression. The interaction term however does not enter significantly and the negative effect from primary exports loses in statistical significance as the time span narrows. For our more rules-based measure, polity, results suggest both the existence of a resource curse and its reversal. Contrary to the ICRG measure, these results are more pronounced for 1975 and 1980 than for earlier startyears. When disaggregating primary exports into its four components (agricultural raw materials, food, fuels, and ores and metals, lower panel), we find that, for a majority of time periods and for a majority of resource-sectors, the direct negative effect from resources and its interaction with institutions do not enter significantly. Ores and metal exports stands out as the resource-type that have both a significant direct negative effect as well as a possibility of growth-reversal through good quality institutions. Our preferred empirical model is thus specified as: growthi = Xi′ α + β1 Insti + β2 agrigdpi + β3 f oodgdpi + β4 f uelsgdpi +β5 ores metgdpi + β6 (ores metgdpi × Insti ) + εi (4.1) Results are presented in Table 2. In each and every regression, each resource measure enters with a negative sign although significance levels are low in general. The exception is exports of ores and metals that enters significantly at the five per cent level at least in all regressions. The beneficial growtheffect from institutions does not prove robust over time spans when using 6

polity to measure institutional quality. Moreover, the interaction with ores and metals are not significant for startyears 1965 and 1970. Our results this far can thus be summarized: Primary exports tend to have a negative though statistically weak influence on economic growth. Exports of ores and metals however have both an economic and statistical sizeable negative effect on growth. This effect however, can be reversed, leading ores and metals exports to have a positive effect on growth. In general, these effects are less statistically significant the shorter is the time span and the choice of institutional variable is important such that, in general, the rulesbased polity variable is producing smaller and weaker coefficient estimates than does the ICRG measure.

5

Endogenous institutions

In our empirical model, we suspect that institutions for various reasons (omitted variables, errors-in-variables, and, in particular, a potential simultaneous causality between institutional quality and economic growth) are correlated with the error term. In particular, growth may be a driver for institutional development (think for example of rapidly growing countries spending more resources to protect property rights). The causal effect from institutions on growth may be estimated using an instrumental variable technique. This possibility depends crucially on whether the instrument set used is valid. In order for an instrument to be valid, it needs to fulfill both the criteria of instrument relevance (that it is sufficiently correlated with the endogenous variable, institutions in our case) and of exogeneity (that the instrument is uncorrelated with the error term, i.e. the instrument has no partial effect on growth once institutions are controlled for). In this context, it is important to note two things. First, even if an instrument has been considered ”good” in general, it is not certain that this is the case in a particular sample (that is, possible violations of instrument validity always need to be considered). Second, the validity of an instrument is likely to change depending on specification chosen. While some Z may be a valid instrument for X (say, institutions) when analyzing its effect on Y1 (say, log GDP per capita), the validity considering the effect from X on Y2 (say, economic growth) may be quite different. The exogeneity of the instrument may change depending on whether the different omitted variables (i.e. the error term) are related to the instrument or not. This relates to our finding above that institutions affect 7

different outcome variables (GDP growth in different time-spans) differently. Hence, we would not be surprised if our results varied between specifications also when instrumenting for institutions. A nice example of varying effects of this kind is Bardhan (2005, Table 5, p. 508) who in a levels specification (institutions and income level) uses four different dependent variables and three different country samples. Glaeser et al. (2004) shows credibly that the potential reverse causality (i.e. growth influencing institutions) is something that need to be addressed. Likewise, Chong and Calder´on (2000) performing Granger causality tests, find evidence for two-way causality between economic growth and institutions (BERI and ICRG), and in particular that ”economic growth causes institutional quality in a much higher percentage than the opposite” (p. 78). A similar conclusion is drawn by Paldam and Gundlach (2008) who find that measures of democracy as well as corruption supports the ’Grand Transition’-view (income positively effecting institutions) over the ’Primacy of institutions’-view (institutions positively affecting growth in income). As instruments for institutions (and the interaction of institutions and resources) we employ the often used instruments latitude and eurfrac (the share of population speaking an European language at home).7 We also add to the instrument set the interaction of latitude and resources. We run regressions both on our first specification (using primary exports as the resource measure) and our preferred specification (using four sectors of natural resources and the interaction of ores and metals with institutions). Results are presented in Tables 3 and 4. Qualitatively we obtain the same results as in our OLS regressions albeit the precision of our estimates is not satisfactory. Looking at our first-stage results (the lower panels of Tables 3 and 4), this is not surprising. While in many cases the instrument set passes the Hanson J-test for overidentifying restrictions (Ovid) and the instrument set do enter jointly significant in the first stage regressions, the F-value is low in each and every specification and the Shea partial R-squared is very low, suggesting that our instruments are weak. It is therefore not surprising that our institutional variables lose in significance in the second-stage regressions. OrM etExp, retains its expected properties, though. Another concern, related to the endogeneity discussion above, is that natural resources preceeds institutions, which in turn drives economic devel7

These instruments were first used by Hall and Jones (1999).

8

opment.8 Either the extraction of resources or the knowledge of the existence of extractable resources affect the quality of institutions. Alternatively, institutional quality may determine the extent to which resources are searched for and extracted. In any of these cases our model would be misspecified, and our institutions measure would capture part of the resource effect, or vice versa. Even if we agree with the general idea that resources (or more generally geographic conditions) have been an important determinant in shaping institutions historically, the severity of the problem depends partly on the sample at hand but mainly on which resources are studied. Furthermore, resources which are important today (or have been in the past decades) have in many cases become major exports rather recently, and in some cases they are even based on recent discoveries. This means that many resources do not fit the description of given, exogenous endowments which have shaped institutions over the very long run as in e.g. Engerman and Sokoloff (2002).9 Figure 1 shows, for respective resource measure in our data, the institutional development (polity) before and after 1975 for four groups: democracies as of 1975 that were rich and poor in natural resources in 1975, and autocracies as of 1975 that were rich and poor in natural resources in 1975.10 If resources affect institutions, we would expect different developments for the high and low resources groups. For agricultural raw materials (a) and ores and metals (d), the pattern is largely similar within the two autocratic groups as well as within the two democratic groups. For fuels (c) however, it seems that oil extraction among autocracies has indeed made institutions deteriorate: high-oil exporters democratized significantly slower post-1975 than did the group of no-oil and low-oil exporters. This results are supported by the findings in e.g. Ross (2001), Tsui (2005) and Vicente (2008). The other resource for which we observe divergence post 1975 is food ex8

This has been hypothesized by Engerman and Sokoloff (2002) and Isham et al. (2005), for example. Sala-i-Martin and Subramanian (2003) also suggest a similar mechanism and present empirical evidence for this hypothesis in a different setting. 9 As an example, Norway and Ecuador discovered oil in the late sixties (at a time when institutional differences between these countries were already large). The same is true for diamonds in Australia and Botswana. Looking at resource dependence in many African countries the resources which dominated exports in the 1970s and 1980s had a zero or close to zero contribution to exports in 1960 (see e.g. the listing in Boahen, 1987). 10 For some resources, however, there are not many resource rich democracies in 1975, making the development quite dependent on individual countries. Note also that the country coverage is smaller than 85 prior to 1965. Figures using instead the startyears 1965, 1970 and 1980 looks largely similar. Results are available from the autors.

9

ports (b). However, the direction of change is here opposite from that of fuels: autocracies low in food exports have remained more autocratic than their autocratic high-food counterparts.

6

Influential observations

Some countries are more endowed with certain types of natural resources than other countries. Since our resource measures are related to the overall size of the economy (as shares of GDP), a country can obtain a high measure either due to its absolute affluence of a certain resource or due to its poor performance in other sectors (i.e. a low GDP). We therefore would expect some countries to have a large effect on our results. Figures 2 to 5 present, in raw form, respective countries relative abundance of resources for our four time periods. Eye-balling the figures, our assessment is that the following countries are likely to be influential for respective resource: For agrigdp: Liberia (and maybe Malaysia and Congo-Kinshasa); for foodgdp: Liberia, Congo-Kinshasa (and maybe Gambia); for fuelgdp: Liberia, CongoKinshasa (and maybe Kuwait); for ores metgdp: Liberia, Congo-Kinshasa, Zambia (and maybe Botswana). We therefore run our preferred specification without Liberia, Congo-Kinshasa and Zambia and thereafter we also take away Malaysia, Gambia, Kuwait and Botswana. Results are presented in Table 5. Focusing on the estimates for ores and metals and its interaction, results seem fairly robust to these exclusions when using icrg as institutional measure. The large exception being the 78-country sample for the period 1980-2005 when neither ores and metals nor the interaction remain significant. When instead using polity as institutional measure, signs and coefficient sizes, though not significance levels, are in parity with the original estimates. When deleting the additional four countries, the interaction term in particular loses its economic and statistical significance. As a more formalized alternative procedure, we do not focus on resources per see, but instead check for influential observations using the DFITS index when estimating equation (4.1). Observations with a DFITS index larger than the p absolute value of 2 k/n (where k is the number of independent variables, including the constant, and n the number of observations), are excluded from the sample.11 Table 6 reports the results and the countries excluded from 11

DF IT Si = ri

p

hi /(1 − hi ), where ri are the studentised residuals and hi the leverage.

10

the sample for each time period. The outcome varies to a surprisingly small extent when excluding outliers based on DFITS.

7

Conclusions

In this article, we systematically study whether the relation between resources, institutions and economic growth varies over three dimensions: i) different kinds of natural resources ii) different kinds of institutions and iii) different time periods. Decomposing primary exports in its four main components; exports of agricultural raw materials, food exports, fuel exports and exports of ores and metals, we find ores and metals to stand out as the resource class having both a resource curse and a positive reversal through stronger institutions. Regarding the institutional dimension, we employ two main measures of property rights institutions: ”institutional quality” constructed from the ICRG data base and the polity-variable from the Polity IV data base. Our general results indicate that the polity measure yields less precise estimates of the relation between institutions and growth, but there are differences between specifications and time periods. When changing the startyear in our sample, we do obtain different point estimates for different time periods. However, we do not find any pattern in the differences and the general results seems robust to the choice of startyear.

11

Tables Table 1: Time, Resources and Institutions. (N=85) inst primexpgdp primX inst agrigdp agriX foodgdp foodX fuelgdp fuelX ores metgdp oresX

(1) 196505 3.118*** (1.012) -5.520** (2.654) 4.151 (4.408) 4.159*** (1.333) 1.697 (8.581) -18.582 (14.131) -8.005** (3.959) 4.015 (6.242) 1.915 (3.270) -8.264 (5.842) -13.201*** (3.095) 26.510** (11.493)

ICRG (2) (3) 197005 197505 2.616** 3.035** (1.014) (1.223) -6.413*** -5.153* (2.148) (2.824) 7.550* 6.074 (3.829) (5.238) 3.856** 3.404** (1.461) (1.542) 3.110 6.259 (10.206) (14.120) -19.936 -27.088 (16.706) (21.736) -6.963 -2.666 (4.726) (4.246) 2.401 2.455 (8.534) (8.075) -0.003 4.469 (2.615) (4.446) -3.710 -12.271 (4.903) (8.053) -14.494*** -17.584*** (3.588) (3.141) 33.998*** 47.461*** (12.839) (9.229)

(4) 198005 2.339* (1.382) -5.718 (4.044) 7.244 (7.474) 2.843* (1.692) -6.619 (14.362) -6.147 (23.363) -2.991 (5.085) 2.175 (9.562) 5.673* (3.279) -15.324** (6.973) -15.228*** (4.485) 33.580*** (8.550)

(5) 196505 0.341 (0.797) -5.604*** (2.002) 2.123 (3.521) 0.360 (0.793) -7.122 (6.902) 4.088 (9.039) -8.080 (5.433) 3.462 (5.857) -4.157** (1.835) 2.203 (3.591) -12.452*** (4.400) 8.111 (8.540)

Polity (6) (7) 197005 197505 1.214* 0.008 (0.702) (0.710) -3.295** -4.765*** (1.647) (1.742) -0.614 5.938** (3.999) (2.475) 0.985 -0.257 (0.628) (0.923) -3.867 -15.591* (7.229) (8.355) 8.582 20.838 (15.100) (13.823) -8.600* -6.930 (4.327) (4.777) 3.328 6.193 (5.452) (5.869) -1.846 -3.862** (1.447) (1.806) -1.248 4.657* (4.107) (2.730) -12.998*** -10.092*** (3.744) (3.455) 12.607 20.591* (8.711) (11.558)

(8) 198005 -0.476 (0.756) -4.975** (2.215) 7.789** (3.390) -0.645 (0.885) -20.969* (11.071) 19.548 (16.908) -7.103** (3.212) 10.526* (5.419) -3.738* (2.169) 3.765 (3.260) -9.166*** (3.148) 22.255*** (4.759)

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). See text for details.

12

Table 2: Time, Resources and Institutions. Interaction term for ores and metals (N=85) inst agrigdp foodgdp fuelgdp ores metgdp oresX

(1) 196505 3.580*** (1.011) -7.307** (3.408) -6.082** (2.528) -2.306** (1.129) -12.072*** (3.128) 21.960* (11.818)

ICRG (2) (3) 197005 197505 3.321*** 2.417** (1.031) (1.110) -7.001* -7.900* (3.530) (4.193) -5.732** -1.670 (2.863) (2.265) -1.942* -1.742 (1.022) (1.307) -13.284*** -16.108*** (3.009) (2.725) 29.622** 43.698*** (11.224) (9.564)

(4) 198005 2.160* (1.167) -10.008** (4.543) -2.339 (2.829) -2.011 (1.557) -15.281*** (3.727) 33.338*** (7.933)

(5) 196505 0.793 (0.499) -5.267 (3.301) -6.068** (2.795) -3.270** (1.390) -12.382*** (3.087) 7.907 (5.867)

Polity (6) (7) 197005 197505 1.211*** 1.022* (0.454) (0.586) -0.971 -6.927 (3.326) (5.258) -6.698** -4.958* (2.951) (2.857) -2.054* -2.099 (1.134) (1.347) -13.309*** -9.435*** (3.091) (3.553) 13.451* 17.259 (7.376) (11.793)

(8) 198005 0.548 (0.519) -11.550* (6.258) -3.128 (2.571) -2.295 (1.530) -9.699** (3.978) 22.108*** (5.870)

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). See text for details.

13

Table 3: Instrumenting for institutions, primary exports specification

inst primX lngdppc inv mopen ssa lac primexpgdp Constant Observations R2 Jpval ShInst ShResInst Ovid latitude latX eurfrac Fval IVpval latitude latX eurfrac Fval IVpval

(1) 196505 5.179* (3.073) 10.530 (15.964) -1.062** (0.489) 0.139*** (0.031) 0.012** (0.006) -2.398*** (0.442) 0.181 (0.805) -7.208 (5.454) 3.812* (2.202) 85 0.674 0.219 0.134 0.149 0.701 0.003 (0.002) -0.005 (0.011) 0.124** (0.048) 3.27 0.026 -0.001* (0.001) 0.006 (0.004) 0.003 (0.010) 1.34 0.269

ICRG Polity (2) (3) (4) (5) (6) (7) 197005 197505 198005 196505 197005 197505 6.645 8.235* 8.107* -6.006 -13.338 -32.710 (4.228) (4.430) (4.713) (7.161) (20.805) (69.577) 16.217 15.617 19.764 29.045 59.804 120.607 (16.913) (15.864) (15.685) (30.322) (90.980) (248.148) -1.250* -1.444** -1.422** 0.304 1.007 3.140 (0.629) (0.674) (0.695) (0.773) (1.999) (7.278) 0.135*** 0.110** 0.113* 0.125* 0.025 -0.089 (0.039) (0.048) (0.057) (0.064) (0.216) (0.469) 0.008 0.008 0.007 0.008 0.012 -0.010 (0.007) (0.007) (0.009) (0.009) (0.022) (0.057) -2.553*** -3.111*** -3.224*** -2.147*** -2.904 -5.372 (0.569) (0.708) (0.935) (0.706) (1.797) (6.777) 0.524 0.555 0.468 -1.930* -3.885 -9.468 (0.979) (0.994) (0.938) (1.137) (4.574) (17.679) -7.903 -7.209 -8.998* -18.194 -28.525 -53.810 (5.030) (5.092) (5.352) (14.402) (37.558) (106.419) 4.360 5.576 5.527 1.003 1.507 1.025 (2.746) (3.385) (3.847) (1.844) (3.745) (8.333) 85 85 85 85 85 85 0.557 0.418 0.330 0.189 -1.816 -9.894 0.291 0.183 0.186 0.603 0.806 0.885 0.101 0.128 0.157 0.021 0.007 0.002 0.133 0.150 0.251 0.024 0.010 0.003 0.504 0.429 0.719 0.748 0.995 0.807 First stage Institutions regression 0.003 0.003 0.002 0.010* 0.009 0.010* (0.002) (0.002) (0.002) (0.005) (0.005) (0.005) -0.005 -0.004 0.001 -0.052** -0.038** -0.041** (0.010) (0.008) (0.009) (0.022) (0.018) (0.016) 0.118** 0.120** 0.128*** 0.120 0.181 0.201** (0.048) (0.047) (0.046) (0.143) (0.141) (0.095) 2.95 3.30 3.27 2.54 2.63 4.50 0.038 0.025 0.026 0.063 0.056 0.006 First stage Institutions*Primary Exports regression -0.001* -0.001 -0.001* 0.002 0.002* 0.003* (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) 0.006 0.006 0.008** -0.009 -0.007 -0.011 (0.004) (0.004) (0.004) (0.008) (0.005) (0.007) 0.000 0.001 0.001 0.048** 0.052** 0.057** (0.011) (0.011) (0.010) (0.023) (0.023) (0.025) 1.30 1.14 1.78 1.94 2.52 2.01 0.281 0.340 0.158 0.130 0.064 0.120

(8) 198005 -6.046 (9.296) 57.793 (58.460) 0.253 (0.938) 0.055 (0.115) -0.010 (0.022) -2.635** (1.188) -3.521 (2.692) -17.227 (18.322) 2.221 (3.740) 85 -0.964 0.454 0.057 0.022 0.765 -0.000 (0.006) -0.010 (0.021) 0.307*** (0.089) 4.53 0.006 0.000 (0.001) 0.001 (0.007) 0.048** (0.022) 1.65 0.184

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). See text for details.

14

Table 4: Instrumenting for institutions, ores and metals exports specification

inst oresX lngdppc inv mopen ssa lac agrigdp foodgdp fuelgdp ores metgdp Constant Observations R2 Jpval ShInst ShResInst Ovid latitude latX eurfrac Fval IVpval latitude latX eurfrac Fval IVpval

(1) 196505 6.431** (2.606) 17.227 (19.189) -1.202*** (0.311) 0.128*** (0.032) 0.017*** (0.005) -2.321*** (0.391) 0.358 (0.555) -6.772 (4.180) -6.859*** (2.177) -0.908 (1.746) -9.021 (6.233) 4.384*** (1.214) 85 0.722 0.002 0.099 0.363 0.698 0.002 (0.002) 0.042** (0.019) 0.125*** (0.043) 4.11 0.010 -0.000** (0.000) 0.011*** (0.002) 0.001 (0.002) 9.75 0.000

ICRG Polity (2) (3) (4) (5) (6) (7) 197005 197505 198005 196505 197005 197505 7.173** 6.262 7.396 3.718 3.550 2.586 (3.110) (3.784) (4.628) (4.541) (3.850) (3.358) 26.166 38.395* 31.671* 14.007 16.318 96.358 (20.242) (19.341) (16.298) (25.265) (26.799) (86.478) -1.287*** -1.136** -1.235** -0.874 -0.784 -0.876* (0.374) (0.468) (0.598) (0.621) (0.483) (0.501) 0.119*** 0.078** 0.079* 0.169*** 0.165*** -0.006 (0.035) (0.038) (0.042) (0.055) (0.060) (0.111) 0.016*** 0.013** 0.016** 0.011 0.007 -0.003 (0.005) (0.006) (0.006) (0.011) (0.008) (0.016) -2.436*** -3.111*** -3.246*** -1.222 -1.203* -2.398** (0.417) (0.384) (0.487) (0.806) (0.695) (1.024) 0.550 -0.018 0.078 -0.586 -0.275 -0.986 (0.648) (0.784) (0.878) (0.641) (0.824) (1.041) -7.218 -8.252 -11.057* -4.135 2.562 -7.775 (4.874) (5.007) (6.575) (5.221) (3.697) (10.977) -7.143** -2.896 -4.781 -8.276** -8.342** -7.382 (2.989) (3.257) (4.122) (4.152) (3.856) (6.471) -0.226 -0.502 -0.150 0.201 0.884 0.908 (1.571) (1.546) (2.104) (5.347) (4.408) (2.987) -10.269* -13.206** -13.605* -15.643 -14.400 -26.188 (6.127) (6.081) (7.338) (12.541) (10.307) (16.966) 4.691*** 5.097*** 5.193** 3.009 2.362* 8.113* (1.439) (1.619) (2.300) (2.172) (1.308) (4.449) 85 85 85 85 85 85 0.680 0.662 0.596 0.442 0.565 -0.084 0.000 0.000 0.000 0.100 0.000 0.253 0.089 0.079 0.069 0.012 0.016 0.031 0.401 0.284 0.241 0.145 0.122 0.097 0.935 0.898 0.746 0.269 0.266 0.530 First stage Institutions regression 0.002 0.002 0.001 -0.001 -0.001 0.000 (0.002) (0.002) (0.002) (0.005) (0.005) (0.005) 0.050*** 0.057*** 0.074*** 0.065 0.122*** 0.028 (0.015) (0.016) (0.013) (0.057) (0.040) (0.062) 0.121*** 0.121*** 0.103** 0.144 0.180 0.185* (0.043) (0.043) (0.042) (0.139) (0.143) (0.104) 6.95 8.52 17.40 0.77 3.63 1.11 0.000 0.000 0.000 0.516 0.017 0.349 First stage Institutions*Ores and Metals Exports regression -0.000* -0.000 -0.000*** -0.000 -0.000 0.000 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) 0.012*** 0.015*** 0.024*** 0.019** 0.019*** 0.007 (0.001) (0.003) (0.004) (0.007) (0.005) (0.008) 0.002 0.003 0.001 0.009** 0.009*** 0.005 (0.002) (0.002) (0.002) (0.004) (0.003) (0.004) 58.28 8.07 10.66 4.42 12.95 1.03 0.000 0.000 0.000 0.007 0.000 0.384

(8) 198005 2.168 (1.664) 39.559** (16.825) -0.676** (0.292) 0.038 (0.059) 0.013** (0.006) -2.709*** (0.469) -1.359*** (0.455) -12.214* (6.234) -4.578 (3.665) -0.323 (2.188) -14.849*** (5.259) 5.015** (2.297) 85 0.530 0.017 0.077 0.391 0.215 -0.006 (0.005) 0.091 (0.059) 0.227** (0.091) 3.18 0.029 -0.000 (0.000) 0.029*** (0.010) 0.000 (0.004) 3.05 0.034

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). See text for details.

15

Table 5: Outliers (1) 196505 inst agrigdp foodgdp fuelgdp ores metgdp oresX R2 inst agrigdp foodgdp fuelgdp ores metgdp oresX R2

3.155*** (0.977) -7.499* (3.816) -6.611** (2.603) -2.732** (1.153) -16.466*** (4.428) 24.847* (13.020) 0.704 2.251** (0.875) -6.470 (4.251) -5.267* (2.737) -2.238* (1.162) -17.874*** (4.212) 36.743*** (11.737) 0.672

ICRG Polity (2) (3) (4) (5) (6) (7) 197005 197505 198005 196505 197005 197505 Table 2 excluding LBR, ZAR and ZMB. N=82 3.166*** 2.049* 1.755 0.809 1.246*** 0.949 (1.046) (1.095) (1.164) (0.499) (0.447) (0.607) -6.685 -8.740* -5.290 -5.557 -2.008 -4.042 (4.044) (4.757) (5.310) (3.560) (3.365) (4.970) -6.490** -2.369 -2.919 -6.889** -7.323** -5.688** (2.933) (2.215) (2.882) (2.815) (2.888) (2.655) -2.151** -1.996 -2.201 -3.552** -2.244* -2.062 (1.037) (1.316) (1.540) (1.410) (1.149) (1.355) -11.336** -17.004*** -13.556** -11.805* -10.300 -8.800 (4.318) (5.309) (5.464) (6.319) (6.545) (8.592) 24.774* 44.240*** 29.963*** 2.303 6.953 20.650 (12.825) (11.021) (9.488) (11.657) (13.723) (14.222) 0.657 0.631 0.603 0.640 0.631 0.580 Excluding also MYS, GMB, KWT and BWA. N=78 2.272** 2.084* 2.679** 0.336 0.890** 1.094* (0.938) (1.137) (1.163) (0.482) (0.417) (0.628) -5.201 -10.495 -4.070 -6.619 -4.981 -9.587 (4.968) (8.161) (9.027) (4.316) (4.705) (7.553) -4.282 -1.996 -4.417 -4.570 -3.912 -2.135 (2.633) (2.655) (2.941) (2.844) (2.591) (2.563) -1.792* -1.962 -2.221 -2.902** -1.845 -1.952 (1.065) (1.513) (1.776) (1.360) (1.148) (1.463) -10.529*** -12.953** -5.097 -8.294 -6.943 -2.317 (3.836) (5.508) (5.485) (8.025) (5.805) (8.836) 24.060* 24.363 -3.271 0.728 1.765 -11.644 (12.840) (21.580) (18.091) (15.149) (13.675) (16.732) 0.615 0.583 0.582 0.604 0.594 0.567

(8) 198005 0.606 (0.540) -2.379 (4.902) -3.941 (2.474) -2.163 (1.516) -7.526 (5.017) 20.151*** (6.755) 0.609 0.970 (0.617) -3.516 (8.373) -4.184 (2.552) -2.195 (1.673) -5.260 (5.249) -1.542 (10.188) 0.577

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). See text for details.

16

Table 6: Outliers, using DFITS

inst agrigdp foodgdp fuelgdp ores metgdp oresX Observations R2

(1) 196505 3.116*** (0.753) -2.672 (2.818) -3.425** (1.394) -2.567** (1.105) -15.349*** (2.814) 30.380*** (10.902) 77 0.813

ICRG (2) (3) 197005 197505 3.074*** 2.210** (0.837) (0.924) -4.108 -6.380** (3.031) (2.954) -3.651* -1.505 (1.914) (1.796) -1.552 -1.544 (0.965) (1.221) -12.909*** -14.394*** (3.076) (3.734) 28.754** 35.903** (11.765) (16.278) 80 75 0.784 0.781

(4) 198005 2.314** (1.006) -7.207 (5.502) -1.810 (2.998) -1.264 (0.995) -14.856** (7.054) 36.894 (26.466) 75 0.699

(5) 196505 0.368 (0.330) -3.471 (3.117) -2.338 (2.042) -3.243** (1.275) -10.620*** (2.544) 8.598 (10.719) 75 0.758

Polity (6) (7) 197005 197505 1.093*** 0.735 (0.281) (0.508) -0.364 -10.034 (2.403) (6.569) -3.424 -2.344 (2.109) (2.084) -1.826** -2.160* (0.744) (1.290) -13.505*** -14.563*** (2.114) (4.384) 19.078*** 32.531*** (5.091) (9.233) 78 74 0.823 0.695

(8) 198005 1.100*** (0.397) -3.328 (6.813) -2.666 (2.131) -1.086 (0.846) -8.172*** (2.829) 4.149 (8.003) 75 0.716

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). The regressions drops influential observations as identifyed by DFITS. Dropped countries are, in turn: Column (1): BWA, DZA, LBY, MYS, NER, NIC, OMN, ZMB; col (2): BWA, DZA, MYS, NER, NIC; col (3): BWA, DZA, IRN, KOR, LBY, NER, NIC, SAU, TTO, ZAR; col (4): BWA, DZA, HTI, KOR, LBR, LBY, NER, OMN, SAU, ZMB; col (5): BWA, DZA, GUY, NIC, OMN, PAN, TTO, VEN, ZAR, ZMB; col (6): BWA, DZA, GUY, JAM, NIC, OMN, VEN; col (7): BWA, CHL, GUY, IRN, JAM, KOR, LBR, MYS, TTO, ZAR, ZMB; col (8): BWA, CHL, HTI, KOR, LBR, MYS, OMN, SAU, ZAR, ZMB. See text for details.

17

Table 7: Time, Resources and Institutions, unrestricted sample.

Obs: inst primexpgdp primX R2 inst agrigdp agriX foodgdp foodX fuelgdp fuelX ores metgdp oresX R2 inst agrigdp foodgdp fuelgdp ores metgdp oresX R2

(1) 196505 94 2.512** (1.057) -5.763** (2.387) 4.988 (3.970) 0.707 3.575*** (1.256) 0.279 (8.215) -15.560 (13.131) -7.626** (3.263) 4.368 (5.216) 1.601 (3.361) -7.478 (5.998) -13.883*** (1.776) 30.047*** (4.460) 0.747 3.121*** (0.925) -7.528*** (2.840) -5.357*** (1.914) -2.258** (1.049) -13.277*** (1.717) 27.708*** (4.455) 0.742

ICRG (2) (3) 197005 197505 100 102 3.052*** 3.159*** (1.119) (1.173) -4.114* -3.935 (2.464) (2.491) 2.716 1.769 (4.605) (3.756) 0.694 0.631 3.885*** 3.516** (1.350) (1.348) 4.884 1.525 (10.165) (14.254) -17.452 -11.903 (15.709) (22.115) -5.913 -3.172 (3.866) (3.762) 3.062 -0.705 (6.923) (7.309) 2.949 1.469 (2.182) (3.216) -9.049*** -6.990 (3.217) (4.476) -11.898*** -11.993** (2.457) (4.765) 25.514*** 23.466 (6.228) (17.508) 0.729 0.671 3.158*** 2.693*** (0.965) (0.946) -4.258 -4.802 (3.080) (4.062) -4.211* -3.604* (2.150) (1.998) -2.164* -2.617*** (1.090) (0.991) -11.592*** -11.865*** (1.972) (3.951) 24.080*** 22.781 (5.018) (15.681) 0.719 0.661

(4) 198005 102 3.815*** (1.334) -2.384 (3.720) -0.347 (5.716) 0.564 3.682** (1.572) -9.006 (13.833) 7.033 (21.614) -1.108 (4.753) -3.419 (8.228) 4.975** (2.398) -12.001*** (3.004) -12.540** (5.198) 24.093* (12.820) 0.637 2.803** (1.264) -5.875 (4.789) -3.308 (2.606) -2.287* (1.270) -13.848*** (4.742) 26.090** (11.819) 0.609

(5) 196505 92 0.300 (0.753) -5.858*** (1.938) 3.129 (3.324) 0.672 0.333 (0.751) -9.906* (5.362) 7.052 (7.406) -8.132 (4.906) 5.550 (5.641) -3.985** (1.727) 1.970 (3.475) -9.838*** (2.971) 3.529 (5.982) 0.708 0.963** (0.475) -6.927** (3.134) -4.648* (2.676) -3.040** (1.332) -9.699*** (2.621) 3.240 (5.024) 0.700

Polity (6) (7) 197005 197505 100 105 1.195* 0.125 (0.620) (0.622) -2.996* -4.591*** (1.595) (1.503) 1.201 4.786* (3.733) (2.688) 0.670 0.572 1.043* 0.293 (0.604) (0.789) -5.333 -12.075 (6.031) (7.934) 10.377 14.003 (14.241) (15.128) -6.841* -5.230 (4.007) (3.610) 6.785 4.617 (6.156) (5.439) -1.049 -3.759** (1.370) (1.629) -1.858 3.935 (3.987) (2.624) -7.865*** -6.432** (2.880) (2.452) 3.461 3.133 (7.017) (12.273) 0.711 0.590 1.487*** 1.287** (0.453) (0.522) -2.484 -5.893 (3.537) (4.909) -2.181 -3.100 (3.521) (2.875) -1.182 -2.294* (1.127) (1.263) -7.607*** -5.847** (2.842) (2.549) 3.119 0.508 (7.085) (11.700) 0.701 0.576

(8) 198005 103 0.740 (0.750) -3.191 (2.122) 2.193 (3.544) 0.502 0.869 (0.884) -16.507 (10.440) 13.949 (17.758) -5.264* (3.098) 2.310 (7.148) -1.745 (2.029) 0.119 (3.287) -7.403** (3.025) 9.894 (10.215) 0.545 1.147** (0.540) -8.949 (5.904) -3.474 (2.585) -1.523 (1.400) -7.946** (3.607) 10.839 (10.249) 0.541

Notes: Dependent variable is growth. Robust standard errors in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions include the controls listed in text (not shown). See text for details.

18

1 0

.2

.2

.4

.4

.6

.6

.8

.8

1

Figures

1945

1955

1965 1970 1975 1980 year demhagri1975 authagri1975

1990

2000

2010

1945

1955

demlagri1975 autlagri1975

1965 1970 1975 1980 year demhfood1975 authfood1975

2000

2010

demlfood1975 autlfood1975

.8 .6 .4 .2 0

0

.2

.4

.6

.8

1

(b) Food

1

(a) Agriculture

1990

1945

1955

1965 1970 1975 1980 year demhfuel1975 authfuel1975

1990

2000

2010

demlfuel1975 autlfuel1975

1945

1955

1965 1970 1975 1980 year demhores_met1975 authores_met1975

(c) Fuels

1990

2000

2010

demlores_met1975 autlores_met1975

(d) Ores and metals

Figure 1: Development of polity measure for democracies and autocracies (defined in 1975) when dividing the sample in groups of below-the-average and above-the-average resources in 1975.

19

10

10

BWA

BWA

0

CYP IRL

MYS

GAB TZA NIC

ZAR

.05

.1 agrigdp

.15

.2

GAB

CIV

LBR

0

.05

(a) agrigdp1965

.15

.2

5

KOR BWA THA IRL IND IDN LKA OMN CYP CHL TUR PAK EGY NOR TUN ESP PRT GBR FIN DOM USAISR JPN AUS DNK AUT BFASWE CAN SDN MAR ITA FRA NLD LBY GRC NZL CRI PAN JAM COL TTO IRN URY CHE GHA GUY MEX SYRHND ECU SLV COG PHL DZA MLI BRA SEN KWT ARG PER GTM ETH BOLZAF GMB KEN CMR PRY VEN MWI NIC GAB ZMB TZA SLE TGO MDG NER SAU CIV HTI

THA MYS

avgr

0

IDN LKA EGY

MYS

ZAR

NIC CIV

−5

OMN TUNPRT PAKNOR FIN JPN DOM GBR USA AUT ESP TUR LBY TTO DNK ITA AUS MAR NLD CANSWE FRA GRC ISR CRI COLBFA URY SDN MEX PAN HND NZL SYR CHE BRA ECU PHL PRY DZA COG MLI GTM GHA CMR ARG GUYETH KEN JAM PER SLV GMB SEN BOLZAF IRN MWI TZA KWT VEN TGO SLE NER ZMB GAB SAU MDG HTI

.1 agrigdp

(b) agrigdp1970

BWA

KOR

5

SDN

−5

−5 0

avgr 0

MYS

ZAR LBR

CYP IRL CHL IND

THA IDN

TUN NOR LKA EGY IND DOM CHLPRT OMN FIN ESP JPN AUT SYR TUR PAK GBR GRC ISR ITA USA TTO FRA BRA CAN MAR DNKNLD CRI COL SWE BFA AUS MEX ECU ETH PRY PAN URY NZL COG PHL DZA HND KEN GTM CHE CMR MLI IRN GUY SAU ARG PER BOL GMB MWI SLV LBY JAM ZAF TZA SEN GHA TGO VEN SLE NIC ZMB HTI MDG KWT NER

avgr

avgr

THA CYP IRL IDN JPNPRT LKA DOM TUN NOR IND EGY ESP FIN CHL GRC AUT ITA ISR PAK BRA FRA TUR NLD CRI GBR MAR USA AUS DNKCAN SWE TTO SYR COL PAN MEX IRN LBY ECU PRY BFA DZA COG PHL KENURY NZL ETH GTM CHE SDN GUYMLI HND SAU ARG CMRSLV ZAF GMB JAM MWIPER BOL GHA SEN TGO SLE CIV VEN MDG ZMB HTI KWT NER

5

KOR

0

5

KOR OMN

LBR

−10

−5

ZAR

LBR

0

.05

.1 agrigdp

.15

.2

0

.05

.1

.15

agrigdp

(c) agrigdp1975

(d) agrigdp1980

Figure 2: Growth and Agricultural raw material exports as share of GDP for different periods

20

10

10 BWA

BWA

5

KOR

0

IDN

GMB

THA MYS

ZAR

GUY

GMB

ZAR

−5

−5

LBR

0

.1

.2 foodgdp

.3

.4

LBR

0

.1

5

5

BWA

CYP IRL LKA

GUY

avgr

0

NOR OMN TUN PAK PRT FIN JPN DOM GBR USA AUT ESPTTO TUR LBY ITA AUS DNK NLD CAN FRA GRC ISR MAR COL BFA SWE CRI URYPAN MEX SDN HND SYR BRA PRY NZL CHE PHL ECU DZA COG MLI GTM ETH GHA CMR ARG KEN PER JAM SLV ZAF SEN IRNBOL MWI TZA KWT VEN SLE TGO ZMB NER GAB SAU MDG HTI NIC CIV

GMB

KOR

THA IND IDN MYS OMN CHL CYP EGY TUR PAK NOR TUN ESP PRT GBR FIN DOM USA JPN AUS DNK SDN AUT NLD BFA SWE ISR CAN MAR ITA FRA LBY GRC NZL PAN COL JAM TTO IRN URY ECU CHE MEX SYR SLV COG DZA MLI PHL ARG BRA SENETH KWT PER GTM BOL PRY ZAF KEN CMR VEN GAB ZMB TZA SLE TGO MDG SAU NER

THA MYS

.3

(b) foodgdp1970

BWA KOR

IDN CHL IND EGY

.2 foodgdp

(a) foodgdp1965

IRL LKA

CRI GHA

HND

GUY

GMB NIC MWI CIV

HTI ZAR

−5

avgr 0

CYP IRL

LKA TUN INDEGY PRTNOR DOM CHL OMN FIN ESPSYR JPN AUT TUR PAK GBR GRC ITA USA TTO MAR FRA BRA CANISR DNK NLD CRI AUS COL SWE MEXBFASDN ECU ETH PRYPAN COGDZAURY NZL PHL HND KEN GTM CHE CMR MLI IRN GAB SAU ARG PER BOL SLV MWI LBY JAM ZAF SENTZAGHA TGO VEN SLE NIC CIV ZMB HTI MDG KWT NER

avgr

avgr

THA CYP IRL IDN MYS JPN PRT LKA DOM TUN NOR IND EGY ESP FIN CHL GRC AUT ITA ISR PAK BRA FRA TUR NLD CRI GBR MAR DNK USA AUS CAN TTO SYR SWE COL PAN IRNMEX LBY ECU PRY BFAURY DZA COG PHLETH KEN GAB GTM CHE HND SDN NZL GUY SAU MLI ARG CMR ZAF PER JAM SLV MWI TZAGHA BOL TGO SEN CIV VEN SLE NIC MDG ZMB HTI KWT NER

0

5

KOR OMN

LBR

−10

−5

ZAR

LBR

0

.1

.2 foodgdp

.3

.4

0

.1

.2

.3

foodgdp

(c) foodgdp1975

(d) foodgdp1980

Figure 3: Growth and Food exports as share of GDP for different periods

21

10

10 BWA

BWA

THA CYP IRL MYS IDN

TTO

IRN

LKA EGY IND NORTUN PRT DOM CHL FIN ESP JPN AUT SYR TUR PAK GBR GRC ISR ITA USA FRA BRA CANNLD MAR DNK CRI AUS COL SWE BFA MEX ECU ETH SDN PRY PAN URY COG NZL PHL HND KEN GTM CHE CMR MLI GUY ARG PER BOL GMB SLV MWI JAM ZAF TZA GHA SEN TGO SLE NIC CIV ZMB HTI MDG NER

avgr

avgr 0

5

KOR

THA CYP IRL IDN MYS JPN LKA PRT DOM TUN NOR IND EGY ESP FIN CHL GRC AUT ITA ISR PAK BRA FRA TUR NLD CRI GBR MAR USA AUS CAN DNK SYR SWE COL PAN MEX ECU PRY BFA COG URY PHL KEN ETH GAB DZA GTM CHE HND SDN NZL GUY MLI ARG CMR ZAF MWI PER GMB JAM SLV TZA BOL GHA SEN TGO SLE CIV NIC MDG ZMB HTI

LBY SAU

0

5

KOR OMN

VEN KWT

NER ZAR

TTO DZA

GABIRN

−5

−5 0

KWT

.2

.4

.6

LBR

0

.2

.4

fuelgdp

(b) fuelgdp1970

KOR BWA THA IRL IND LKA CHL CYP

5

5

BWA KOR THA CYP IRL

IDN OMN LBY

0

TTO

avgr

DZA IRN KWT

VEN GAB

MYS

IDN

TUR PAKGBR EGY NOR TUN ESP PRT FIN DOM USA JPN AUS DNK SDN AUT NLD BFA SWE ISR CAN MAR ITA FRA GRC NZL CRI PAN JAM COL IRN URY CHE GHA MEX GUY SYR HND ECU SLV PHL MLI BRA SEN ARG PER GTM ETH BOL ZAF KEN GMB CMR PRY MWI NIC ZMB TZA SLE MDGTGO NER CIV

OMN

TTO DZA COG

LBY KWT

VEN

GAB SAU

HTI

SAU

ZAR

−5

PAK PRTNORTUN FIN JPN DOM GBR USA AUT ESP TUR DNK ITA AUS MAR CANNLD FRA GRC ISR BFA SWE CRI COL SDN URY MEX HNDPANSYR NZL CHE BRA PRY PHL ECUCOG MLI GTM ETH GHA CMR ARG GUY JAMKEN PER SLV GMB ZAF SEN BOL MWI TZA TGO SLE NER ZMB MDG HTI NIC CIV

.6

fuelgdp

(a) fuelgdp1965

avgr 0

LBY

VEN

ZAR

LBR

CHL MYS LKA IND EGY

OMN

SAU

LBR

−10

−5

ZAR

LBR

0

.2

.4 fuelgdp

.6

.8

0

.2

.4 fuelgdp

(c) fuelgdp1975

(d) fuelgdp1980

Figure 4: Growth and Fuel exports as share of GDP for different periods

22

.6

10

10 BWA

BWA KOR

5

5

KOR OMN

THA CYP IRL IDN MYS

avgr

LKA TUN NOR EGY IND PRT DOM CHL OMN FIN ESP JPN AUT SYR TUR PAK GBR GRC ISR ITA USA TTO FRA BRA CAN MAR DNK NLD CRI AUS COL SWE BFA MEX ECU ETH SDN PRY PAN URY COG NZL PHL DZA HND KEN GTM CHE CMR MLI IRN GAB SAU ARG PER GMB SLV MWI LBY JAM ZAF TZA GHA SEN TGO VEN SLE NIC CIV HTI MDG KWT NER

GUY

0

0

avgr

THA CYP IRL IDN MYS JPN LKA PRT DOM TUN NOR IND EGY ESP FIN CHL GRC AUT ITA ISR PAK BRA FRA TUR NLD CRI GBR MAR USA AUS CAN DNK TTO SYR SWE COL PAN MEX IRN LBY ECU PRY BFA DZA COG URY PHL KEN ETH GAB GTM CHE HND SDN NZL SAU MLI ARG ZAF MWI GMBCMR PER JAM SLV TZA BOL GHA SENSLE CIV VEN TGO NIC MDG HTI KWT NER

ZMB

ZAR

ZMB

ZAR

−5

−5

LBR

0

.1

.2

.3

.4

.5

LBR

0

.1

.2

ores_metgdp

5

KOR

TZA KWT VEN SLE NER GAB SAU MDG HTI NIC CIV

0

CHL

OMN PAK PRTTUN NOR FIN JPN DOM GBR USA AUT ESP TUR LBY TTO DNK ITA AUS MAR NLD CAN FRA GRC ISR BFA SWE CRI COL SDN URY MEX PAN HND SYR NZL BRA CHE PRY PHL ECU DZA MLI COG GTM ETH GHA CMR ARG KEN JAM PER SLV SEN GMB ZAF IRN MWI

.5

GUY BOL

BWA

THA IRL IND IDN MYS LKA OMN CYP EGY TUR PAK NOR TUN ESP PRT GBR FINAUS DOM USA JPN DNK SDN AUT NLD BFA SWE ISR CAN MAR ITA FRA LBY GRC NZL CRI PAN JAM COL TTO IRN URY CHE GHA MEX SYR HND ECU SLV COG PHL DZA MLI BRA SEN KWT ARG PER GTM ETH GMB KEN CMR ZAF PRY VEN MWI NIC GAB TZA SLE MDG SAU CIV HTI

avgr

5

THA IRLCYP MYS IDN LKA IND EGY

.4

(b) ores metgdp1970

BWA

KOR

.3 ores_metgdp

(a) ores metgdp1965

avgr 0

GUY

BOL

CHL

GUY BOL TGO

ZMB NER

TGO ZMB

−5

ZAR

LBR

LBR

0

.1

.2

.3 ores_metgdp

.4

.5

−10

−5

ZAR

0

(c) ores metgdp1975

.1

.2 ores_metgdp

.3

.4

(d) ores metgdp1980

Figure 5: Growth and Ores and metal exports as share of GDP for different periods

23

References Daron Acemoglu and Simon Johnson. Unbundling Institutions. Journal of Political Economy, 113(5):949–995, 2005. Pranab Bardhan. Institutions matter, but which ones? Economics of Transition, 13(3):499–532, 2005. A. A. Boahen. General History of Africa VII: Africa under Colonial Domination 1880-1935. UNESCO/Heinemann/University of California, Paris, 1987. Alberto Chong and C´esar Calder´on. Causality and feedback between institutional measures and economic growth. Economics and Politics, 12(1): 69–81, March 2000. Stanley L. Engerman and Kenneth L. Sokoloff. Factor Endowments, Inequality, and Paths of Development among New World Economies. Economia, 3(1):41–109, 2002. Edward L. Glaeser, Rafael La Porta, Florencio Lopez de Silanes, and Andrei Shleifer. Do Institutions Cause Growth? Journal of Economic Growth, 9 (3):271–303, September 2004. Robert E. Hall and Charles I. Jones. Why Do Some Countries Produce so Much More Output per Worker than Others? The Quarterly Journal of Economics, 114(1):83–116, February 1999. Jonathan Isham, Michael Woolcook, Lant Pritchett, and Gwen Busby. The Varieties of Resource Experience: Natural Resource Export Structures and the Political Economy of Economic Growth. The World Bank Economic Review, 19(2):141–174, 2005. Martin Paldam and Erich Gundlach. Two views on institutions and development: The grand transition vs the primacy of institutions. Kyklos, 61 (1):65–100, 2008. Michael L. Ross. Does Oil Hinder Democracy? World Politics, 53(3):325– 361, 2001. Jeffrey D. Sachs and Andrew M. Warner. Natural Resource Abundance and Economic Growth. NBER Working Paper 5398, 1995. 24

Xavier Sala-i-Martin and Arvind Subramanian. Addressing the Natural Resource Curse: An Illustration from Nigeria. NBER Working Paper 9804, 2003. Kevin K. Tsui. More oil, less democracy? theory and evidence from crude oil discoveries. Manuscript, University of Chicago, November 2005. Pedro C. Vicente. Does oil corrupt? evidence from a natural experiment in west africa. August 2008.

25