Chapter 5. Metal Prices Signal Global Economic Shifts
- Rabah Arezki, and Akito Matsumoto
- Published Date:
- December 2017
Metal prices have been declining since 2011 after a long upward trend that began in the early 2000s (Figure 5.1). Some analysts consider this to be a signal that we are nearing the end of the so-called commodities supercycle. Although that is difficult to ascertain with confidence, the prolonged fall in metal prices is consistent with a more typical commodity boom-and-bust cycle. Indeed, after a period of high metal prices during the 2000s, investment and capacity in the sector increased substantially. At the same time, high prices led to downward adjustments in demand. Those adjustments contributed to a gradual decline in metal prices after 2011, which in turn lowered profit expectations and reduced investment in the sector, especially in high-cost mines. The decline in investment will eventually reduce capacity, and lower production should eventually lead to a rebound in metal prices and, in turn, an upturn in investment. In fact, prices did rebound to some degree in 2016.
Figure 5.1.Metal Price Indices
Sources: IMF, Primary Commodity Price System; and IMF staff calculations.
Understanding the evolution of metal markets is important for at least two reasons. First, metals are at the heart of the world economy because they are key intermediate inputs to industrial production and construction. Metal markets are thus shaped by shifts in the volume and composition of global demand and supply, and transformations in metal markets also signal important changes in the world economy. Second, for some countries, metal exports are a large portion of total exports, and fluctuations in metal prices can have important macroeconomic consequences. This chapter addresses the following questions:
What are metals?
Where are the primary centers of metal production and consumption?
How have metal markets evolved?
What lies ahead?
What are Metals?
Metals are mineral bodies that come in a variety of forms. “Base metals” are those that oxidize or corrode relatively easily. Among the base metals, a distinction is made between ferrous and nonferrous metals. Ferrous metals, typically iron, tend to be heavy and relatively abundant. Nonferrous metals, which are generally more expensive than ferrous metals, do not contain iron in significant amounts and have desirable properties such as low weight (for example, aluminum), higher conductivity (for example, copper), or nonmagnetic properties or resistance to corrosion (for example, zinc and nickel). In contrast to base metals, “noble metals” are resistant to corrosion or oxidation. These include the precious metals—so called because of their perceived scarcity—such as gold, platinum, silver, rhodium, iridium, and palladium. Chemically, precious metals are less reactive than most elements and have high luster and high electrical conductivity.
Unless otherwise indicated, this chapter focuses on four main base metals: iron ore,1 copper, aluminum, and nickel. All four experienced price declines since 2011, although to varying degrees (Figure 5.1). These metals are used for many purposes but especially for construction and machinery because of their ductility and malleability.
Where are the Primary Centers of Metal Production and Consumption?
Metal production and metal consumption are concentrated in a few countries, but the locations often overlap (Figures 5.2 and 5.3). China is a primary center for both consumption and production, which reflects its importance in global industrial production. A few individual entities, including some multinational and state-owned corporations, control large market shares in the production and refining of some key metals. This high degree of concentration at times causes concern about the potential for market manipulation and collusion, for example, through output restrictions, export bans, and/or stock accumulations (see Rausser and Stuermer 2014 for an analysis of collusion in the copper market).
Figure 5.2.Metal Consumption in 2015
Sources: Bloomberg Finance L.P.; World Bureau of Metal Statistics; and IMF staff calculations.
Figure 5.3.Top Five Companies Producing More Than 2 Percent of World Production
Sources: Bloomberg Finance L.P.; and IMF staff calculations.
From an economic point of view, iron ore is by far the most important base metal, with about $225 billion in global sales.2 Nearly all iron ore goes toward production of steel, which is used for construction, transportation equipment, and machinery. Iron ore prices were previously determined largely through negotiations between Japanese steelmakers and Japanese iron ore producers. The market has recently become more transparent, and the price on delivery at Chinese ports is now used as the benchmark price. Because mining iron ore is capital intensive, production is concentrated among a few producers (Figure 5.4), and production levels depend crucially on the level of investment, which has declined in recent years. The top iron-ore-producing country, China, accounts for about half of global production, followed by Australia and Brazil.3 The demand for iron ore comes primarily from steel-producing countries such as China, which consumes more than half of world production. In turn, half of world steel production is used for construction. In advanced economies, the use of scrap metal is becoming more important, reducing the demand for iron ore.
Figure 5.4.Metal Production in 2015
Sources: Bloomberg Finance L.P.; World Bureau of Metal Statistics; and IMF staff calculations.
1 Mine production for China is based on crude ore, rather than usable ore, which is reported for the other countries. 2Overseas department of France.
Copper is the second most important base metal by value, at roughly $130 billion annually.4 Copper is used for construction and electrical wire. Chile is the largest producer, followed by China and Peru. Relatively few companies are involved in copper production; Chile’s Codelco is the largest. Copper prices have been more transparent than those for iron ore because copper futures markets and London Metal Exchange settlements are used as benchmarks. China consumes about half of all refined copper.
The third most important base metal by value is aluminum, at $90 billion annually.5 Aluminum is used in the aerospace industry as well as other industries requiring light metal. Aluminum is slightly different from other base metals because it requires refining, typically from bauxite which is quite abundant. That refining process is very energy intensive, and as a result, large producers of aluminum are located where electricity is cheap. The largest producer is China, followed by Russia, Canada, and the United Arab Emirates. Aluminum prices are more stable than those of other metals because electricity prices are heavily regulated in most countries. Recycling has become an important part of aluminum production because recycling is much less energy intensive than producing primary aluminum. China consumes about half the world’s production of primary aluminum. In contrast, developed economies rely more on recycling and in turn have less influence over primary aluminum prices.
The fourth most important base metal is nickel, at about $40 billion annually.6 Nickel is used in alloys such as stainless steel. The Brazilian mining company Vale and Russia-based Norilsk Nickel are the top two producers, and together they account for 23 percent of global production. Conventional roasting and reduction processes are used to extract nickel metal from ore, typically at purity levels greater than 75 percent. China consumes about half the world’s smelted and refined nickel; the next largest consumer is Japan.
Nickel markets have been affected by the policies of producing countries because producers sometimes seek to take advantage of the oligopolistic nature of these markets. Indonesia, which produced 27 percent of global output in 2012, imposed an export ban on nickel ore in January 2014 to increase incentives for domestic processing. The Philippines and New Caledonia (a dependent territory of France in Oceania) have sought to use the opportunity created by that ban to increase their own market share, but they may be unable to meet the portion of Chinese demand that previously relied on Indonesian production. On the other hand, the global inventory of refined nickel has been increasing, suggesting a supply glut.
How Have Metal Markets Evolved?
In recent decades there have been dramatic shifts in the volume and the structure of both demand for and supply of major metals.7 Global production has increased for most metals owing to the rapid investment in capacity that occurred during the 2000s (Figure 5.5, panel 1). The concentration of demand has shifted from advanced economies toward emerging market and developing economies and from the western hemisphere and Europe toward Asia—especially China because of its rapid growth (see Figure 5.3; Figure 5.5, panel 2; Figure 5.6, panel 1). On the supply side, the so-called frontier of extraction of nonferrous metals, including precious metals such as gold, has shifted advanced economies to emerging market and developing economies because of the rapid improvement in the investment climate first in Latin America and then in sub-Saharan Africa (see the Annex for further detail on this dramatic shift in global metal supplies). High-income member countries of the Organisation for Economic Co-operation and Development (OECD) accounted for close to half the discoveries of major mines between 1950 and 1990. Since 1990, sub-Saharan Africa and Latin America and the Caribbean have doubled their shares of such discoveries, although the actual level has fallen to only about half that in the period running from 1950 to 1990. The pattern of global trade in metals has radically changed as a result of this shift in the location of major discoveries. It should be noted that for steel and aluminum, production tends to occur in countries that not only have combined deposits of iron ore or bauxite—which are abundant worldwide—but that also have port facilities, easy access to energy, and proximity to markets.
Figure 5.5.Evolution of the Metal Market
Sources: Bloomberg Finance L.P.; World Bureau of Metal Statistics; and IMF staff calculations.
Note: The figures reported for iron ore production in China are in crude terms relative to what other countries report. Iron ore production data should thus be interpreted with caution: production figures for iron ore are not consistent with those for consumption because the latter are based on effectively usable iron ore.
Figure 5.6.Development of the Metal Market
Sources: Bloomberg Finance L.P.; IMF Primary Commodities Price System; World Bureau of Metal Statistics; and IMF staff calculations.
Note: Investments are deflated by the mining price index and oilfield machinery, rebased to 2000 = 100. Total investment is the sum of capital expenditures for Anglo American PLC, BHP Billiton Ltd., Corp. Nacional del Cobre de Chi, Freeport-McMoRan Copper & Gold, Glencore PLC, Grupo Mexico SAB de CV, Mitsubishi Corp, Mitsui & Co. Ltd., Rio Tinto PLC, and Vale SA.
On the demand side, growth has been the driving force behind global metal consumption since the early 2000s (see Figure 5.6), and the growth of Chinese demand largely explains the shift in global demand toward Asia. In fact, China is now the main consumption center for most metals. Metal consumption in India, Russia, and Korea has also increased but still lags far behind China’s, whereas consumption in Japan has stagnated somewhat. The rapid rise in demand from emerging markets has been a key factor in determining the price levels of metal and other commodities (for systematic evidence on the importance of China and emerging markets in driving metal and oil prices, see Gauvin and Rebillard 2015; Aastveit, Bjørnland, and Thorsrud 2015).
On the supply side, investment in the metal sector has been on the decline, although this trend should reverse in the wake of the price rebound that began in 2016. Indeed, available data on investment by major iron-ore-producing companies suggest that the rapid increase in investment during the period of high metal prices in the early 2000s was followed by a gradual decline since 2011 that closely followed the trajectory of metal prices (Figure 5.6, panel 3). For ferrous metals, investment is a good indicator of future supply capacity, as mentioned. For non-ferrous metals, a much more relevant indicator of supply is the actual quantity available from mineral deposits. A unique data set on new discoveries of mineral deposits is used here to assess the emergence of new frontiers for metal extraction, and that assessment indicates that prices played little role in driving discoveries of mineral deposits (see the Annex). Instead, rapid improvements in institutions in Latin America and Africa, including those related to property rights and political stability, led to a gradual increase in the number of major discoveries of metals in those regions since the 1990s. These findings have important implications both for the welfare of individual countries and for our global understanding of the balance of forces shaping metal markets and the pattern of global trade in metals. The overall pattern of global metal trade in recent decades has been characterized, as noted, by a shift in the major destination countries from the western hemisphere and Europe to Asia and a shift in the source countries from advanced economies to emerging market and developing economies. In 2002, metals were exported mainly from Canada and Russia to the United States or from Australia to Japan, Korea, and China. By 2014 almost half of metal exports were going from Australia, Brazil, and Chile to China. China has become the largest importer of metals, with its share increasing from less than 10 percent in 2002 to 46 percent in 2014 (Table 5.1).
|1. Bilateral Metal Trade, 2002|
|2. Bilateral Metal Trade, 2014|
Many developing economies depend heavily on metal exports (Table 5.2). For Chile, Mauritania, and Niger, for example, metals now account for more than half of their total exports of goods. Countries whose metal exports as a share of GDP have risen dramatically are vulnerable to fluctuations in metal prices. Since 2002, the discovery of new metal deposits has dramatically changed the list of leading exporters (as a percentage of GDP), adding to the list of resource-dependent countries that face new challenges in terms of macroeconomic management.
|Papua New Guinea||7.07|
China has recently attempted to rebalance its economy away from investment-led growth toward growth supported by more domestic consumption. Metal use is intensive in machinery, construction, transportation equipment, and manufacturing industries, and so the declining growth in these sectors has slowed the growth of Chinese demand for metal since 2010 (Figure 5.7). The global metal price index has decreased correspondingly. On one hand, with increased domestic consumption, the share of the services sector in the Chinese economy will increase, and this should also slow the consumption of metals. On the other hand, infrastructure and housing needs in China remain high, and strong construction growth can increase metal demand, as seen during 2016. Even so, despite the dramatic increase in Chinese metal imports, these represent less than 2 percent of China’s GDP (Figure 5.8).
Figure 5.7.China: Composition of Metal Use, Growth Rates by Sector
Sources: National Bureau of Statistics of China; World Input-Output Database; and IMF staff calculations.
Note: The growth rates of total demand for metals are calculated as the weighted sum of output growth rates for each sector, with weights being the shares of metal input in the individual sector in the total economy. The share of metal input for each sector is based on the World Input-Output Database. The value of the share of metal input in the most recent year (2011) was chosen, given that the share of metal input has been quite stable over the years. Because output data for China are not available at the sector level, profit data by sector were used as a proxy for most of the industries. For nonindustry sectors, GDP data by industrial classification were used.
Figure 5.8.China: Metal Imports
Sources: IMF World Economic Outlook Database; UN Comtrade; and IMF staff calculations.
The slower pace of investment in Chinese manufacturing and the ample global supply of metals have both exerted downward pressure on metal prices in recent years. However, the decline in metal prices started much earlier, and it therefore makes sense to explore what may lie ahead. It is helpful in this regard to go beyond the price outlooks generated by the behavior of futures markets and instead to review the forces that underpin the demand and supply of metals.
On the demand side, Chinese economic growth is projected to slow further, albeit gradually, but with considerable uncertainty around the timing and the nature of the shift. In broad terms, however, the effect of slower growth in China will be to lower metal prices (Figure 5.9).8 In addition, a slower pace of growth in China’s industrial production could produce further metal price declines.
Figure 5.9.Growth Rates of Metal Price Index
Sources: IMF, Primary Commodity Price System and Global Data Source; and IMF staff calculations.
Note: The figure shows the actual and fitted annual growth rate of the metal price index. The fitted growth rate is based on a regression of the annual growth rate of the metal price index on the annual growth rate of China’s industrial production.
Outside China, a number of advanced economies have prioritized infrastructure spending, including the United States, and such spending is sometimes associated with stronger metal demand. However, overall metal consumption by advanced economies is lower than in emerging markets, and advanced economies also rely more heavily on recycled metals, both of which would limit the increased metal demand likely to result from increased infrastructure spending.
On the supply side, declining investment in the metal sector is unlikely to lead to a substantial price rebound in the near future, although temporary outages or the closure or exiting of large mines would help prices recover. Low energy prices have in fact helped keep down or reduce mining and refining costs, including for copper, steel, and aluminum. High-cost or high-pollutant mines would certainly close first, considering that current metal prices may be close to the breakeven point for these high-cost mines. However, SNL Metals & Mining research suggests that metal prices will have to fall much further to trigger significant reduction in capacity due to plant closures. that prices would need to fall further before substantial capacity becomes vulnerable to closure (SNL Metals & Mining 2015). Moreover, the expansion of metal extraction in Latin America and Africa as a result of an improved investment climate is unlikely to be reversed to any great extent; to the contrary, the investment climate in those regions can be expected to steadily improve. As a result, ample global supply will likely continue to push down metal prices.
The interplay between weaker demand and a steadily increasing supply, given the existing cost structure in global metal markets, points to a continued glut, leading to a low-for-long price scenario. In turn, the risk associated with such a scenario is that investment will continue to falter and lead to a sharp increase in prices down the road.
The fundamental factors that underpin demand for primary commodities, including metals, garner a great deal of attention, but supply-side factors do not. As described in the main part of this chapter, the center of gravity for global metal demand has shifted from the western hemisphere toward Asia as a result of the high growth in emerging markets—especially China—over the past two decades. And while demand for metals emanating from emerging markets has been a key driver of recent global market developments, progress in the quality of institutions has helped to increase the supply of metals and to shift its composition. In fact, developments on the supply side have been perhaps just as dramatic as on the demand side, particularly the discoveries of major metal deposits that signal new potential for a further expansion of global supply. This analysis shows how the frontiers of metal exploration and extraction have shifted from advanced to emerging and developing economies.1
Metal Deposit Discoveries
Available data suggest that developing economies have substantial deposits of metals that have yet to be discovered. There is an estimated $130,000 of known subsoil assets beneath the average square kilometer of the member countries of the OECD, which contrasts with only about $25,000 for African countries (Collier 2011 and McKinsey Global Institute 2013). It is unlikely that those differences represent actual variations in the geological formations in advanced and developing economies. Instead, they can be attributed to institutional differences, specifically the quality of property rights and the stability of political institutions, that can dampen exploration efforts in developing economies. There were rapid and significant improvements in the institutional environments of many developing economies during the 1990s, however, and a cursory look at the data on political risk seems to indicate that the timing of these improvements coincides with an increase in the share of metal discoveries in Latin America and Africa (Figure 5.1.1).
Figure 5.1.1.Discoveries in Latin America and the Caribbean and Sub-Saharan Africa
Sources: International Country Risk Guide; MinEx Consulting; and IMF staff calculations.
Figure 5.1.2 shows how the frontier of metal exploitation has gradually moved from advanced to developing economies. Even as the total number of discoveries remained broadly constant, the distribution changed significantly. High-income OECD member countries accounted for 37 to 50 percent of all discoveries during 1950–89, but only 26 percent during 2000–09, whereas the shares of sub-Saharan Africa and Latin America and the Caribbean doubled. Latin America was home to the most discoveries of metal deposits since 1990.
Figure 5.1.2.Number of Mine Discoveries by Region and Decade
Source: MinEx Consulting.
Note: OECD = Organisation for Economic Co-operation and Development.
What Factors Drive Discoveries?
Investments in exploration and extraction activities involve sunk costs and are thus subject to the so-called hold-up problem—when two parties may both benefit by cooperating but refrain from doing so because they fear ceding to the other increased bargaining power or other advantages.2 For an investment to be profitable, there must be a stable political environment, a low risk of expropriation, and a favorable investment climate (Acemoglu, Johnson, and Robinson 2001; Bohn and Deacon 2000). Cust and Harding (2014) provide evidence that the quality of the institutional environment substantially affects oil and gas exploration.3 Mining operations could be considered more “expropriatable” than oil facilities because mining outputs do not move through pipelines but instead must be transported exclusively on land.
To assess the importance of institutional factors in the discovery of metal deposits, this analysis uses a three-way panel data set, a zero-inflated Poisson model with the number of mine discoveries by country, year, and type of metal as the dependent variable.4Nitm denotes the number of mines discovered in country i at time t and for a specific metal m. Nitm is assumed to follow a Poisson distribution.
The main explanatory variable of interest is a country’s political risk rating, obtained from the Political Risk Index in the International Country Risk Guide (ICRG), which reflects property rights and political stability. Because metals differ in their abundance and location, metal fixed effects are included in the regressions. Also included are country fixed effects to capture time-invariant country characteristics that are hard to observe, such as actual geology and year fixed effects to control for technology and other global shocks. In addition, price changes are controlled for over the past five years. The baseline specification is as follows:
in which X includes d, f, and g, which are time, country, and metal fixed effects, respectively; and the key covariates, which are lagged changes in prices for specific metals, ∆pricet-1,m; and the measure of political risk, ICRG. The key coefficients of interest are γ and β.
It should be noted that the quality of institutions may be endogenous to metal discoveries in that these discoveries may, for instance, trigger conflicts over resources and erode institutions (Ross 2001, 2013). Any such endogeneity will tend to bias the coefficient associated with institutions toward zero, and as such, that coefficient should be interpreted as presenting a lower bound. To somewhat alleviate issues of reverse causality, the political risk rating is included with a one-year lag. In addition, lagged discoveries are controlled for, to account for the clustering of discoveries. The interactions between ICRG and metal price and between price and fixed effects are also explored. Other robustness checks consist of adding controls such as GDP per capita and the initial capital stock and using price levels instead of changes. The main results remain unchanged.
The ICRG’s PRR Political Risk Rating is found to be statistically and economically significant (Table 5.1.1). The results indicate that a 1 standard deviation improvement in the PRR Political Risk Rating in a particular country—which corresponds to a move from the conditions in, for example, Mali to those in South Africa, or South Africa to Chile, or Chile to Canada—would lead to 1.2 times as many metal discoveries in that country. A thought experiment can further convey the relevant magnitude: if the median quality of property rights in Latin America and sub-Saharan Africa were to suddenly improve to equal those of the most advanced economies in each of these regions (Chile and Botswana, respectively), there would be a 15 percent increase in the number of metal discoveries worldwide, all else equal. The increase in the number of discoveries increases to 25 percent if instead the quality of Latin American and sub-Saharan African property rights were to suddenly rise to the same level as in the United States, again all else equal.
|Variables||(I) Number of Discoveries||(II) Number of Discoveries||(III) Number of Discoveries||(IV) Number of Discoveries|
|Political Risk Rating, Lagged||0.0216***|
|Polity2 Score, Lagged||0.0128|
|Stock of Discoveries, Lagged||0.0161***|
|Political Risk Rating × Change in Metal Price||−0.00635|
|Log Change in Metal Price||−0.449|
|Log Change in Metal Price, Lagged||−0.334|
|Country Fixed Effects||YES||YES||YES||YES|
|Year Fixed Effects||YES||YES||YES||YES|
|Metal Fixed Effects||YES||YES||YES||YES|
This analysis indicates that the quality of a country’s institutions is an important driver of exploration for and ultimately discovery of metal deposits. Institutions affect discoveries through a variety of channels, not only on the perceptions of risk by potential foreign investors. For instance, better institutions could affect the adoption of better technologies or improve the quality of the labor force. The analysis here does not attempt to separate such additional channels.
The results also suggest that movements in metal prices during the past five years are not statistically significant in explaining the number of discoveries. Instead, the likelihood of additional discoveries appears to increase with the number of previous discoveries, as would be expected given the reduced risk of exploring close to a known deposit.
What Are the Implications?
The shift in the frontier of metal exploitation from advanced economies to emerging market economies will likely have important consequences for the individual countries with newly found metal deposits, especially in Latin America and Africa. Indeed, these discoveries expand the list of resource-rich countries. New mines mean more investment and jobs and increased government revenues. There are new trade routes from Latin America and Africa to emerging Asia. There are also, however, new challenges facing newly resource-rich countries in the conduct of macroeconomic policy over both the short and the long term. A future steady increase in the quality of institutions if coupled with a slowdown in demand could lead to excess supply and exercise further downward pressure on prices.
2001. “The Colonial Origins of Comparative Development: An Empirical Investigation.” American Economic Review91(5): 1369–401..
2016. “Shifting Frontiers in Global Resource Wealth: The Role of Policies and Institutions.” CEPR Discussion Paper 11553, Centre for Economic Policy Research, London..
2015. “What Drives Oil Prices? Emerging versus Developed Economies.” Journal of Applied Econometrics, 30(7): 1013–28..
2000. “Ownership Risk, Investment, and the Use of Natural Resources.” American Economic Review90 (3): 526–49..
2011. The Plundered Planet: Why We Must—and How We Can—Manage Nature for Global Prosperity. Oxford, UK: Oxford University Press..
2014. “Institutions and the Location of Oil Exploration.” OxCarre Research Paper 127, Department of Economics, Oxford Center for the Analysis of Resource Rich Economies, University of Oxford, Oxford, UK..
2015. “Towards Recoupling? Assessing the Global Impact of a Chinese Hard Landing through Trade and Commodity Price Channels.” Unpublished. Banque de France and International Monetary Fund..
McKinsey Global Institute. 2013. “Reverse the Curse: Maximizing the Potential of Resource-Driven Economies.” London: McKinsey & Company.
2014. “Collusion in the Copper Commodity Market: A Long-Run Perspective.” Unpublished. University of California at Berkeley..
2013. The Oil Curse: How Petroleum Wealth Shapes the Development of Nations. Princeton, New Jersey: Princeton University Press.
2006. “The Log of Gravity.” Review of Economics and Statistics88 (4): 641–58..
Prepared by Rabah Arezki (team leader), Rachel Yuting Fan, Akito Matsumoto, and Hongyan Zhao, with contributions from Frederik Giancarlo Toscani and research assistance from Vanessa Diaz Montelongo.
Iron ore is rock from which iron metal can be economically extracted.
World production of iron ore is currently 3 billion metric tons with its metal content weighing about 1.4 billion tons, according to the U.S. Geological Survey. The price of iron ore with 62 percent iron content was evaluated at $100 a metric ton, close to the average price in 2014
China’s share, however, is much smaller when the ore’s metal content is taken into consideration. Iron ore is also important for individual countries, such as Ukraine, which relies on coal and iron ore to produce steel.
World mine production was 18.7 million metric tons in 2014. It is evaluated at $7,000 a metric ton, close to the average price in 2014.
World primary aluminum production was about 50 million metric tons, and the associated price was $1,900 a metric ton.
Nickel mine production was 2.4 million tons in 2014, and the price of refined nickel was roughly $17,000 a metric ton.
Metals include aluminum, copper, iron ore, lead, nickel, tin, uranium, and zinc.
This conclusion is the result of a basic econometric exercise using historical data and regressing the annual change of the logarithm of China’s industrial production as an independent variable and the annual change of the logarithm of the IMF’s metal price index as the dependent variable. The exercise shows that 60 percent of the variance in metal prices is explained by fluctuations in China’s industrial production.
The data used in this annex are from MinEx Consulting. The list of metals used in the analysis is comprehensive and includes precious metals and rare earth elements. The data set excludes iron ore and bauxite, which tend to be relatively more abundant than other metals and require for their exploitation proximity to port facilities in the case of the former and substantial energy availability for the latter.
The results presented in this section are also robust to an array of checks including additional controls and estimators. Arezki, Toscani, and van der Ploeg (2016) present extensive technical details and an in-depth discussion of endogeneity.
Their identification strategy relies on exploiting variations in institutions and oil deposits sitting on both sides of a border.
Large numbers of zeros and the heteroscedasticity of errors may imply that ordinary least squares results will be biased and inconsistent. Silva and Tenreyro (2006) suggest the Poisson pseudo–maximum likelihood estimator to address this issue. This analysis follows this suggestion and uses zero-inflated Poisson models. The count data are modeled as a Poisson count model, and a logit model is used to predict zeros.