Natural Resources and Economic Outcomes Forthcoming in Economic Evolution and Revolutions in Historical Time

1. Introduction Why do some countries grow while others do not? This question is one of the most important in economics. Literally thousands of papers have been published on the topic. And it is of enormous policy significance, because the current understanding of the answer to this question affects the behavior of a variety of organizations including the World Bank, the International Monetary Fund, the Inter-American Development Bank and the Gates Foundation. Economists have recently begun to recognize that growth may well be path dependent. If it is, then understanding the current state of economic development is not sufficient to make policy. One needs to understand the historical evolution of the economy or economies in question. Only then can appropriate policies be formulated. During the 1980s the United States experienced declining industrial competitiveness in the world economy. The reasons for the decline were hotly debated. As a means of informing the debate, Wright (1990) sought to understand the basis for American industrial success up to 1940. Economists tended to assume that America's advantage had always been technological. Among economic historians, some of whom had thought more carefully about the issue, the conventional wisdom was slightly more nuanced. Chandler (1959, 1977) argued that the build out of the transportation network during the nineteenth century in the United States allowed the creation of 2 a single national market and that this market enabled the rise of large-scale manufacturing enterprises. Indeed, Atack, Haines, and Margo (in this volume) show that the building of railroads was not merely associated with the growth in manufacturing from 1850 to 1870 – it was causal. Goldin (1998) and Goldin and Katz (2008) show that the expansion of human capital also played an important role in the development of technology and industrial competitiveness more broadly. Technology was important, but it was intimately related to the emergence of national markets and the expansion of human capital.D 1 Wright (1990) showed that the conventional wisdom missed an important part of the story. " The most distinctive characteristic of U.S. manufacturing (from 1879 to 1940) was intensity in nonreproducible natural resources. " D 2 D Technology was certainly important, but it was a technology built on national markets and a specific class of natural resources. Wright addressed a second issue: " whether resource abundance reflected geological endowment or greater exploitation of geological potential. " D 3 D He argued that abundance …

a single national market and that this market enabled the rise of large-scale manufacturing enterprises. Indeed, Atack, Haines, and Margo (in this volume) show that the building of railroads was not merely associated with the growth in manufacturing from 1850 to 1870it was causal. Goldin (1998) and Goldin and Katz (2008) show that the expansion of human capital also played an important role in the development of technology and industrial competitiveness more broadly. Technology was important, but it was intimately related to the emergence of national markets and the expansion of human capital.D 1  showed that the conventional wisdom missed an important part of the story.
"The most distinctive characteristic of U.S. manufacturing (from 1879 to 1940) was intensity in nonreproducible natural resources."D 2 D Technology was certainly important, but it was a technology built on national markets and a specific class of natural resources. Wright addressed a second issue: "whether resource abundance reflected geological endowment or greater exploitation of geological potential." D 3 North Carolina. South Carolina appropriated $500 for Lardner Vanuxem, a professor at South Carolina College, to make "a geological and mineralogical tour during the recess of the college."D 8 D The sums appropriated for later surveys appear to have been substantially higher than the amounts appropriated in North Carolina and South Carolina. In addition to expenses, later appropriations often covered salaries and printing of reports. The printing of reportsand more generally the dissemination of the geological information in the reportswas greatly aided by the fall in the cost of printing.D 9 D State geological surveys served multiple purposes. Hendrickson (1961) finds: "In all states a major purpose was to locate, describe, and publicize such natural resources as salt and mineral springs, building stones, shales, clays, slates, coal, and ores. … In some states-Maryland, Although bigger in scope, federal surveys tended to serve similar purposes. As Atack, Haines, and Margo (in this volume) discuss, the federal government made 59 surveys of roads, canals, and railroads between 1824 and 1838. Geological information was ancillary benefit of these surveys. During the 1830s federal surveys examined the mineral lands of Iowa, Wisconsin and Illinois. After a state survey in Michigan identified rich copper deposits on Chippewa land, the federal government purchased the land. Authorization of federal geological surveys of the Chippewa land and the area around Lake Superior followed shortly thereafter. In 1853 Congress appropriated $150,000 for surveys to determine the best route from the Mississippi to the Pacific Ocean. The Corps of Topographical Engineers examined four potential routes and collected considerable scientific information. Transportation was also the impetus for the Geological Exploration of the Fortieth Parallel (the route of the Pacific railroad).
Oddly, individuals involved in oil exploration tended to dismiss geological surveys as only being relevant for minerals. This changed suddenly in 1911 with the discovery of oil in the Cushing Field in Oklahoma. The Cushing Field vividly demonstrated that oil was associated with anticlines, and in the west, locating the anticlines was a key to discovering new oil. Knowles (1959) writes: "The majority of discoveries during the next fifteen years resulted from the industry's sudden acceptance of petroleum geology as a science. … Most of the large companies established permanent geological departments."D 12 American universities quickly saw the utility of establishing mining-related courses. In many universities, this built on existing programs in civil and other types of engineering.D 13 D Civil engineering had long been part of the curriculum in military academies and a variety of other private and public universities. With the increasing importance of bridges, roads, canals, dams, and buildings during the nineteenth century, the value of systematic training in civil engineering became evident. The Morrill Acts of 1862 and 1890 facilitated the expansion of training of engineers by providing federal funding through grants of land to states to support the creation of new colleges and universities. Together the acts aided the establishment of more than 70 land grant colleges It was explicitly part of the colleges' mandate "to teach such branches of learning as are related to agriculture and the mechanic arts."D 14 D By 1882-1892, the largest share of graduates and non-graduates (46 and 39 percent) of land grant universities were in engineering.
The share had shrunk slightly to 33 and 34 percent by 1919-1922, but  Successful exploitation of mineral resources required more than just courses aimed at getting resources out of the ground. For the resulting oil and minerals to be put to use, they had to be translated into products by relatively skilled workers.D 19 D Some skills were self-taught or learned on the job, but many skills were developed or refined in universities. American universitiesboth public and privateoffered courses in areas such as accounting, business, chemistry, metallurgy, civil engineering and mechanical engineering. This relatively sophisticated practical education was suited, by design, for big business. Although the numbers of university-trained students involved were small in absolute terms, their impact was large. Graduates with business backgrounds managed hundreds or even thousands of workers, while graduates with technical backgrounds made critical breakthroughs in products and processes. David and Wright (1997) attributed the differences in exploration among the United States, Great Britain, and Australia largely to differences in beliefs. Americans believed that minerals were there, while the British, and to a greater extent the Australians, did not believe that One of the key questions was: How secure were property rights? Since investment is related to the security of property rights, security has implications both for understanding the early stage of miningwhich focused on placer miningand later stages of miningwhich focused on quartz mining. During the placer mining phase, levels of investment were generally low. Men progressed from panning for gold to more involved methods such as long toms and cradles.
Long toms and cradles required some investment, but the investment could be protected by having one of the miners on site most of the time.
Property rights were insecure if miners were absent from their claims for extended periods of time. This often happened if miners left to prospect for gold elsewhere. Claim jumpingseizing of a claim that was not currently in productionwas a perennial issue. As population roseand it rose extremely rapidly throughout 1849 and 1850newly arrived miners wanted to mine these claims. Strikingly, claim jumping was institutionalized in mining district constitutions through requirements on how claims must be marked, how many claims could be held, and how often the claim had to be worked for the use right to be maintained. Exceptions were often made for sickness and lack of water. However, to a first approximation, if the marking and work requirements were not met, others could begin mining the land. Conflict was inevitable, since newcomers often seized a claim first and asked questions later. Previous occupiers, having not met the work requirements, would also sometimes return and try to assert their claim by force. An unattractive variant on this was Americans simply using force to drive Mexican, Chinese, or other foreign-born miners off valuable claims.
Insecurity arose for somewhat different reasons also in riskier, larger scale projects.
Damming and river turning projects were common, because they allowed direct access to placer gold in the river bottom. Here, one set of risks arose from the fact that other groups upstream or downstream would engage in similar projects, rendering the existing project infeasible. There was sometimes collateral damage for nearby placer claims, which were either flooded or lost access to customary sources of water. A second set of risks emerged from weather and engineering. Poor weather, particularly excessive rainfall, could sweep away partially completed projects or prevent their completion. Even under normal weather conditions, dams and other projects could fail due to poor design.
Quartz regions had their own issues. Claim jumping was less of an issue, because much of the work was done by corporations. But the apex rule and the possessory nature of claims still led to conflict. In her discussion of one of the most famous early quartz companies, the Mariposa Mining Company, Maureen Jung notes that "Like so many early mining companies, it was also plagued by lawsuits. Mariposa's legal disputes over mineral and water rights spanned a number of years and a variety of succeeding companies."D 21

D
The conflict was evident more generally in the number of cases reaching the California Supreme Court from 1850 to 1866. Most of these cases were brought by large quartz or water companies. Conflict of this type was not confined to California. According to Goldman (1981) in the Comstock, "Between 1860 and 1865, twelve major mining companies were involved in 245 different lawsuits, generating ten million dollars in litigation fees alone."D 22 The surprise is not that property rights were insecure, but that semi-secure property rights emerged in the midst of the chaos. As Zerbe and Anderson (2001) argued, cultural concepts of fairness led to the establishment of norms that were acceptable to American miners. These norms were often written down in mining district constitutions. The existence of culturally acceptable norms greatly facilitated the operation of the private-order institutions that governed property rights, by guaranteeing that miners would engage in first-party and second-party enforcement of property rights. In this context, first-party enforcement meant that many miners would choose to obey the norms. Second-party enforcement meant that miners would resolve disputes among themselves, through a combination of reference to norms and threats of punishment (usually violence) for violation of the norms.
However, norms in and of themselves were not sufficient to guarantee the successful operation of an institution. Most private-order institutionsincluding those that supported property rights in the gold rushrequire the threat of third-party enforcement of property rights.
That is, individuals who were not directly involved in the conflict (third parties) need to participate in the punishment in order for incentives to be maintained. Diaries, letters, and other historical accounts from the period offer some examples of third-party enforcement. In some instance, third-party enforcement was simply neighbors helping neighbors. In more complicated disputes, there would be a mining-district meeting or a smaller meeting in which the alcalde or some respected person within the mining district heard the two parties' arguments. Once a decision was made, the parties were expected to abide by the decision. If the loser(s) did not, third-party enforcement would, in principle, ensure compliance. In practice, the assembled miners or alcalde might well side with the stronger party, irrespective of the merits of the other party's claims. Or the decision simply might not be enforced. The question is why miners who were not directly involved in the conflict would take costly actions such as taking time away from mining to attend a camp meeting or risking physical injury by helping force recalcitrant miners off of a claim. Information transmission limited the functioning of third-party enforcement.
Taken together, the combination of public knowledge, education, the ethos of exploration, and incentives proved extremely powerful. Key stakeholders saw the value of investing in federal and state geological surveys and the education of mining engineers. Americans believed that minerals were around and so invested in finding them. Incentives were strong in the sense that federal and state governments did little to regulate, tax, or otherwise impede private exploration and development. All of these contributed to the early exploitation of American mineral resources.

Natural Resources and Growth
Were natural resources good or bad for economic growth?  argued that they were good for growth in industrial exports, while Sachs and Warner (1995) argued that they were bad for economic growth. Figure 2 illustrates Sachs and Warner's basic empirical relationship.
Natural resource exports as a share of GDP in 1970 was negatively related to growth over the next two decades in a large sample of countries. This relationship holds in a regression framework across a variety of specifications.
Sachs and Warner evaluated a number of hypotheses for the negative effect of resources on growth, including rent seeking and corruption, protectionism, the pricing of tradable and nontradable goods, and shifts in labor demand from learning by doing sectors. They found limited evidence of natural resources affecting growth through bureaucratic quality. The evidence was stronger for protectionism. In most economies, resource abundance negatively affects the manufacturing sector, which leads to a protectionist response. In a few oil countries, this does not hold. Overall, they found evidence of a U-shaped effect. Other effects were small.
One might wonder whether the United States was in some way unique, but evidence suggests that states have also suffered from resource curses. Goldberg, Wibbels, and Mvukiyehe (2008) conduct an analysis similar to Sachs and Warner for the American states for 1929-2002. They find negative relationships between natural resources and per capita income, a measure of very long run growth, and between natural resources and growth. Figure  questions remain as to why these papers reach such very different conclusions. I will return to these differences at the end of the paper.
In light of this conflict and to better understand the resource curse, subsequent papers investigated the mechanisms through which natural resources acted on growth. Mehlum, Moene, and Torvik (2006) studied growth over a longer period 1965-1990 using the sample of 87 countries from Sachs and Warner (1997b). They found that the effect of resources on growth was mediated by institutions. In countries where resource exports represented more than 10 percent of GDP, the resource curse only held for the countries with low-quality institutions. In countries with low-quality institutions, rent seeking and production were competing activities, while in countries with the high-quality institutions, rent seeking and production were complementary activities. Robinson, Torvik, and Verdier (2006) modeled the interaction of natural resource growth with political policies. Based on the model and empirical analysis, they found that institutions were important mediators of the effects of resource booms. Subsequent work by Yang (2008) argued that policies and not institutions affected outcomes. Countries with good institutions could have bad policies and the reverse.
Other papers questioned Sachs and Warner's measure of natural resources. The measure used by Sachs and Warner (1995) and later scholars was a measure of export dependency and not endowment or production. New measures of resource endowments from  suggest that endowments and growth are positively related. The basic empirical relationships between endowments and growth are illustrated in Figure 4.
The literature has a number of interrelated deficits, including the short period over which growth is typically measured, the reliance on cross-sectional identification of effects, and the cursory examination of institutions and policies. Ideally, the period of study would extend back into the nineteenth century. With a panel data set, time-series and cross-sectional estimates of the effects could be constructed and compared. One could also test whether the relationships between natural resources and economic growth have shifted over time. For example, the observed relationships could be examined for the period 1879-1928, the period in which  found strong positive effects of natural resources on American exports. This paper is not the first to remark on the short periods over which growth economists examine outcomes. In his 1996 paper, U Why, Indeed, in America?U Paul Romer noted the lack of dialog between the economic history literature and the growth literature and the short horizon that most growth economists consider. More than ten years later the situation remains the same.
A more detailed examination of political institutions and policies related to public knowledge, education, incentives and technology can clarify their importance. Drawing on a number of current and historical case studies, Wright and Czelusta (2004) argued that policies are critical determinants of the effects of natural resources on growth. Countries with policies that have focused on exploration, technology, and knowledge-related investments have been successful. Similarly bad policies or institutions can lead to undesirable outcomes such as rentseeking and corruption. "If minerals are conceived as fixed stocks, and mineral abundance as a "windfall" unconnected to past investment, then the problem becomes one of divvying up the bounty rather than creating more bounty."D 23 D Subsequent work by Brunnschweiler and Bulte (2008) explored these themes further. They conclude "Contrary to the paradoxical result that resource abundant countries tend to invite rent seeking and therefore suffer from worse institutions, we find that countries with certain institutional designs may fail to industrializeand failing to develop significant non-resource sectors may make them dependent on primary sector extraction."D 24 D This line of work represents a promising start, but there is all too little of it. Ross (2001) began a line of research on natural resources and economic outcomes that very much parallels the work in economics. Using data for in 113 countries over the period 1971-1997, he examined the relationship between oil exports as a percentage of GDP the Polity scores of countries.D 25 D Polity scores capture the degree to which regimes are authoritarian and democratic, with higher values indicating greater levels of democracy. He found that countries with high levels of natural resource exports were less likely to be democratic. Goldberg, Wibbels, and Mvukiyehe (2008) took a very similar approach, but examined the American states. They showed that natural resource dependence was associated with less competitive state politics. Their work also suggests a mechanism through which resource dependence acts on politics. Because tax revenue is easier to generate in states with high levels of resource dependence, political elites can take actions that preserve their political power at lower cost. Thus, natural resources have a conservative influence on the political structure.

Natural Resources and Political Outcomes
Later authors found that the sign of the effect depended on a variety of factors. For example, Jones Luong and Weinthal (2006) found that private ownership of the oil industry was good for democracy and state ownership was bad for democracy. Analysis by Dunning (2008) showed that the effect was dependent on the size of the economic rents from the natural resource and the structure of the economic elite. Herb (2005) used a different measure of natural resourcesthe percentage of government revenue from resources. This weakened the relationship between resources and democracy. Other plausible measures such as windfall profits from oil or minerals as a fraction of GDP or per capita generate no association at all (Ross 2006). And as Haber (2006) notes, one needs to be concerned about reverse causality. Authoritarian regimes may be bad for GDP, which would account for the negative association in any measure of natural resources that is computed as a share of GDP.
In contrast to the economics literature, the political science literature has recently begun to move beyond cross-sectional identification. Haber and Menaldo (2009) built a data set that allows for time-series identification of the effects of natural resources on political outcomes as measured by the Polity score.D 26 D Using a variety of archival and contemporary sources, they constructed extended series of the percentage of government revenue from oil or minerals for seventeen countries. Like the use of endowments rather the exports in the growth context, government revenue from natural resources appears to be a more appropriate independent variable than natural resource exports as a share of GDP in the political context. All seventeen countries had average natural resource exports as a share of GDP of at least 6 percent, and so were significant exporters. Haber and Menaldo find no relationship between the percentage of government revenue from resources and the Polity score. They also re-analyze the conventional data sets using time-series techniques. No effect is found there either. What they do find is that the conventional cross-sectional relationships are sensitive to a variety of factors including how one deals with outliers.

Understanding Natural Resources, Growth and Political Outcomes
I want to advance a historical framework within which the empirical patterns for natural resources, growth and political outcomes can be reconciled. Suppose countries with strong political institutions were the first to develop their natural resources. And for these countries, natural resources facilitated growth. Although some countries may have experienced more success than others for a variety of reasons, natural resources were often a complement to manufacturing. One reason was that high transportation costs and limited functioning of world markets made exporting most natural resources without further processing or value added unattractive. Countries without strong political institutions and some manufacturing or processing capability simply did not develop their natural resource base. Eventually, particularly after World War II, transportation costs fell and world commodity markets expanded. This made it profitable for many more countries to develop their natural resources, often with the aid of large multinational corporations. But most of the countries with large and relatively untapped natural resource bases were precisely those countries with weak political institutions or limited manufacturing or processing capability. To be sure, there were some exceptions. North Sea Oil discoveries in Britain and Norway would certainly be examples. But the vast majority of countries were negatively selected. This did not necessarily doom them. Some countries began to invest in education, technology, and processing. Many other countries experienced turmoil, because newfound wealth, volatility in world commodity markets, and weak institutions proved to be an unfortunate combination. Hence, we observe a resource curse.
A similar relationship may hold for the American states, although possibly for slightly different reasons. Western states may or may not have had weaker political institutions initially, but they certainly had fewer settlers. Low density and late settlement led to later discovery and development of resources. By the time they were discovered, transportation costs within the United States had already fallen dramatically. As a result, resources were transported to eastern and midwestern states that already had manufacturing capability. Thus, resources may have been a curse at the state level, yet simultaneously have been a benefit at the national level.
This hypothesis is in principle testable at both the country and state level. While testing it is beyond the scope of this paper, the hypothesis suggests that bringing together natural resource, growth, and political data over a long time span would be a fruitful exercise. One would need to regime, the competitiveness of elections, a lack of corruption or perhaps all of the above.
Measures of some of these variables exist for many countries from 1800 on and most American states beginning around 1870. Similarly data on growth and natural resources exist or could be constructed for a substantial number of countries back to 1900 or earlier. Analysis of these panels could inform current debates regarding the interrelation among politics, resources, and growth. Given the importance of the issue, such an endeavor seems worthwhile. Sources: Data from Bairoch (1982) reprinted in O'Brien (1998), Table 4, p. 12. Sources: Created from original data for Sachs and Warner (1997a) available at H U http://www.cid.harvard.edu/ciddata/ciddata.htmlU H .  Sources: From Hendrickson (1961), footnote 18 and Merrill (1906). All dates are the beginning of the first state survey. Cost estimates are from Merrill (1906).