Measuring Power, Power Cycles, and the Risk of Great-Power War in the 21St Century

Home , Declinism, Great Power, Hard power, National power, Power politics, Power transition theory

MEASURING POWER, POWER CYCLES, AND THE RISK OF GREAT-POWER WAR IN THE 21ST CENTURY

JACOB L. HEIM | BENJAMIN M. MILLER Cover design Peter Soriano Cover image: Adobe Stock/kras99

Limited Print and Electronic Distribution Rights This document and trademark(s) contained herein are protected by law. This representation of RAND intellectual property is provided for noncommercial use only. Unauthorized posting of this publication online is prohibited. Permission is given to duplicate this document for personal use only, as long as it is unaltered and complete. Permission is required from RAND to reproduce, or reuse in another form, any of our research documents for commercial use. For information on reprint and linking permissions, please visit www.rand.org/pubs/permissions.html. RAND’s publications do not necessarily reflect the opinions of its research clients and sponsors. R® is a registered trademark.

For more information on this publication, visit www.rand.org/t/RR2989. Library of Congress Cataloging-in-Publication Data is available for this publication. ISBN: 978-1-9774-0462-6 © Copyright 2020 RAND Corporation Measuring Global Power great-power wars are rare, their consequences are dramatic. The combination of full mobilization and The Trump administration’s 2018 National Defense intense competition between powerful states leads to Strategy signals that great-power competition has very high death tolls and widespread destruction. For supplanted terrorism as the primary focus of U.S. example, Figure 1 illustrates that battle deaths during defense planning.1 This reflects a growing concern in great-power wars dwarf those from other interstate recent years that U.S. power, particularly its military conflicts.4 power, has been declining relative to the growing The consequences of major wars between great global power of Russia and China.2 This concern powers go beyond the deaths of soldiers and civilians. renews longstanding questions about how we should Wars of this magnitude can lead to major changes in measure international power, which nations have the international system. WWI led to the breakup of the most power, which states are gaining and losing the Ottoman and Austro-Hungarian Empires and the power, and when such shifts in relative or perceived creation of new states in the Middle East and Eastern power might portend conflict between major pow- Europe, for example. World War II (WWII) led to the ers.3 In this report, we explore these questions, illustrating a quantitative, scenario-based approach for policymakers who are interested in measuring the interstate balance of power, assessing the impact of Abbreviations shocks on the balance of power, and identifying periods during which shifts in the balance of power could potentially portend conflict between major powers. AIIB Asian Infrastructure Investment There are different theories that describe conditions Bank under which the balance of power could increase the CINC Composite Index of National threat of conflict. We use one such theory—power Capability cycle theory—to illustrate how the approach pre- GDP gross domestic product sented in this analysis can be used to quantitatively GPI Global Power Index assess the balance of power and associated risks in a systematic way across a variety of potential future IF International Futures scenarios. Policymakers could use this approach to assess which nations have the most power and then IISS International Institute of Strategic Studies identify—and attempt to avoid—risky scenarios. The approach we illustrate in this report, as well NIC National Intelligence Council as the theories on which it is built, focus on great- OECD Organisation for Economic Co- power wars: conflicts that involve intense and direct operation and Development combat between major powers. The ability to identify PLA People’s Liberation Army and avoid these conflicts is particularly important because the stakes of these wars are high. They PPP purchasing power parity involve significant potential changes to the global RCP Relative Concentration Pathway order, and participants mobilize their societies to bring all elements of power to bear. These wars are R&D research and development not simply fought with the weapons on hand when they start. Great-power wars, fortunately, are rare SSP shared socioeconomic pathway events. It has been more than 70 years since the last UN United Nations great-power war ended in 1945. There have been even longer periods between great-power wars; there WWI World War I were 99 years between the end of the Napoleonic WWII World War II Wars and the start of World War I (WWI). While

1 FIGURE 1 Battle Deaths from Interstate Wars, 1816–2007

WWII 12

10 About 75 percent of all interstate battle deaths between 1816 and 2007 WWI 8 occurred in WWI or WWII.

2 Vietnam Iraq-Iran

Number of battle deaths (millions) Korean War War War 0 1823 1846 1849 1853 1855 1859 1862 1864 1866 1876 1879 1884 1894 1898 1904 1907 1911 1913 1915 1917 1919 1929 1932 1935 1938 1940 1943 1945 1948 1951 1956 1958 1965 1967 1969 1971 1974 1977 1979 1982 1987 1991 1993 1998 2001 2007 SOURCE: Data from Sarkees and Wayman, 2010. creation of the United Nations (UN) and a U.S.-led the Soviets winning the Space Race. The energy crises economic order based on trade and economic liberal- of the 1970s did not spell the end of American pros- ization. Defeat in these wars can have grave conse- perity. Predictions of Japan as Number One evapo- quences; the Ottoman Empire disintegrated because rated after the collapse of Japan’s asset bubble in the of WWI, Germany was partitioned and occupied for mid-1990s was followed by two decades of economic decades after WWII, and Japan shifted from imperial stagnation. Claims in the 1990s that the United States rule to a constitutional monarchy. had led the world into a new era of stability collapsed After a post–Cold War hiatus, there has been along with the World Trade Center in 2001. Today, growing concern in the past decade that competition the United States is experiencing record prosperity between the great powers could lead to a major war yet is once again uneasy about its place in the world. in the next decade or two.5 This concern underscores China’s sustained economic and military growth has the importance of finding ways to anticipate—and raised concerns about whether the United States can avoid—great-power conflict. This analysis explores maintain its military and economic dominance. The one approach to this problem based on the balance of simultaneous deterioration of the U.S.-Russia relation- power. ship has also raised concerns about a costly military Political scientists, economists, and interna- competition. The reemergence of power politics and tional relations theorists have yet to agree on the spheres of influence harkens back to dangerous eras of most meaningful metrics for understanding—let history when major powers engaged in brinksmanship alone forecasting—the balance of power.6 Competing and overt and proxy conflicts on multiple continents.8 views on how to define, measure, and interpret global A particular focal point of the concern over power are based on different assumptions that lead renewed great-power competition relates to Sino- to different conclusions about the implications of the American competition and whether China’s rise global balance of power. Over the past 70 years, the means that the two states are destined for war.9 United States has wrestled with competing prophecies Despite this focus on the rise of China, however, we of declinism and triumphalism.7 These assessments do not have a reliable way to measure China’s power were generally wrong. Sputnik did not, in fact, presage relative to U.S. power. As this report shows, the

2 perceived balance of power depends on assumptions contrast, the United States plods along at a steady rate about how components of power are measured and of growth, with a noticeable impact of the financial assumptions about how power is defined in the first crisis in 2009. Through this lens, China’s GDP passed place. For this reason, there is much to be gained that of the United States in 2013 and continues to from a framework that not only assesses power rocket away. using a clear set of well-defined assumptions but Figure 2 tells a story of relentless Chinese growth that allows researchers the flexibility to compare the that leaves the United States lagging behind. But implications of trading one set of assumptions for Figure 2 does not come close to telling the full eco- another set. This research develops such a frame- nomic story, in part because of how it measures GDP. work and applies it to three alternative assumptions When comparing two economies that use different about future economic, demographic, and climate currencies, one must develop a conversion factor. PPP outcomes. measures use a common basket of goods to develop a relative price index. This is useful for comparing the standard of living between two countries. But The Role of Perspective American unease over China’s growth has less to To illustrate the challenge, we present three ways of do with the rising standard of living of the Chinese comparing the size of the U.S. and Chinese econo- people and more to do with the potential of China mies, showing that each perspective tells a different channeling that economic growth into international story. Figure 2 shows what many may think of when influence, including through military force.10 they hear that China will soon eclipse the United The common basket of goods researchers States. It shows the gross domestic product (GDP) generally use to calculate PPP measures may be less of each country, measured using purchasing power relevant to questions of latent military power.11 A parity (PPP). This choice of metric shows the expo- haircut may be cheaper in Beijing than in Manhattan, nential growth of China’s economy. Its GDP doubled but is an advanced fighter jet cheaper for the People’s between 2003 and 2008, and the global financial Liberation Army (PLA) than for the U.S. Department crisis of 2008 barely slowed China’s growth. In of Defense? Fighter jets are not in the basket of goods

FIGURE 2 U.S. and China GDP at PPP Exchange Rates, 1990–2016

25 China United States

Trillions of current international dollars (PPP) 0 1990 1992 19941996 1998 2000 2002 20042006 2008 2010 2012 2014 2016

SOURCE: Data from International Monetary Fund, “World Economic Outlook Database,” database, April 2018.

3 used by the PPP, and the PLA still relies on foreign have researchers attempted to build frameworks sources for some key high-tech components, such as to do so. One recent approach by a team from UN jet engines.12 University has begun funding research into mea- Figure 3 shows how the relative position of the suring what it calls “inclusive wealth,” an attempt to United States and China appears through the lens of quantify a country’s human, physical, and natural GDP compared using market exchange rates rather capital.13 Figure 4 shows how the relative position than PPP. This tells a very different story from of the United States and China appears from this Figure 2. Through this lens, China still had a period perspective. of exponential growth in the early 2000s, but the Through this lens, we would conclude that not magnitude of this growth was significantly less. At only does the United States have a massive lead, but market exchange rates, Chinese GDP grew faster this lead grew in absolute terms between 1990 and than U.S. GDP between 2005 and 2015, but it did not 2010. In 1990, the United States had roughly $84 tril- catch U.S. GDP, nor is it clear whether China can lion more inclusive wealth than China, using maintain this rapid growth. the inclusive wealth metric; in 2010, it had roughly Even if the story of U.S. and Chinese GDP did $111 trillion more. not depend on the assumptions of the storyteller, no These three measures of the economic balance single statistic provides a complete basis for under- between the United States and China lead to vastly standing the relative economic power of two large different interpretations of economic power. One and complex economies. GDP is a measure of flow; it might ask which of these stories is true. The answer is concerns the value of a country’s annual production that all of them are. Each is accurate within its scope; but not its stock of wealth. We might think of GDP each illustrates a different dimension of a complex as a person’s salary. If we asked who had the high- and multidimensional component of global power. est salary in a given year, we’d get a very different One might argue that none of the metrics answer than if we asked who had the most wealth presented above fully captures the complexity of a in that year. It is even more complicated to measure modern economy. For example, all three illustrations a country’s wealth than its GDP, and only recently take Chinese-reported GDP figures at face value.

FIGURE 3 U.S. and China Real GDP at Market Exchange Rates, 1990–2017

25 China United States

10 U.S. dollars (trillions) 5

0 1990 1992 19941996 1998 2000 2002 20042006 2008 2010 2012 2014 2016

SOURCE: Data from International Monetary Fund, 2018.

4 FIGURE 4 depth of the resources that a nation could bring Inclusive Wealth for the United States to bear in instances of militarized disputes.” The and China, 1990–2010 Correlates of War project has gathered data for the CINC from 1816 through 2012 for historical research on interstate wars. Consequently, the CINC has an 160 China industrial-age focus, using such indicators as energy United States consumption and iron and steel production to assess 120 a nation’s industrial might that could, in wartime, be translated into military power. The CINC is perhaps the most widely used method of measuring 80 national power; by one estimate, over 1,000 studies have employed it.18 However, analysts have noted that

40 the CINC may not accurately reflect the current and U.S. dollars (trillions) future balance of power in the postindustrial age.19 One might argue that future power hinges on new 0 1990 1995 20002005 2010 factors, such as supercomputing power or prowess in research and development. The focus of the CINC on SOURCE: Data from United Nations University International Human Dimensions Programme and United Nations Environment Programme, traditional and tangible measures of military power Inclusive Wealth Report 2014: Measuring Progress Toward Sustain- tends to neglect the important roles that technology ability, Cambridge, United Kingdom: Cambridge University Press, 2014. and human capital play in contemporary military capabilities.20 Figure 5 illustrates how the three There are many reasons to doubt the accuracy of decades between 1980 and 2012 appear through the official Chinese economic statistics, which could lead lens of the CINC and suggests why the CINC may to yet another perspective on the relative economic not accurately reflect the balance of power in the sizes of China and the United States.14 And a nation’s postindustrial age. According to the CINC, the Soviet economy represents just one element of global power. Union was the most powerful state until 1989 and Diagnosing whether the United States is ahead of or China surpassed the United States in 1995, neither of behind China, and whether its position is stable or which seems plausible. eroding, depends very much on the assumptions that To respond to the challenge of measuring power go into the diagnosis. in the postindustrial or information age, researchers have developed new ways of measuring global power Global Power Metrics that extend industrial-age measures, such as the CINC. One such measure is the Global Power Index Researchers in various countries have proposed many (GPI), developed under the auspices of the National potential metrics for measuring global power. In Intelligence Council (NIC).21 The GPI includes China, a small academic industry is now devoted to measures of the military, economic, technological, building measures of what it calls “comprehensive political, and demographic capacity of nations. national power.”15 Some Chinese indices include Unlike industrial-age indices like the CINC, the GPI dozens of variables.16 A Western example of a global includes nuclear weapons as a factor in the military power metric can be found in the Correlates of capacity of a state, trade as a factor in the economic War Project, which in 1963 developed a six-factor capacity of a state, research and development (R&D) Composite Index of National Capability (CINC) that expenditures as a measure of technological capacity, includes (1) total population, (2) urban population, government revenues as a measure of political capac- (3) military personnel, (4) military expenditures, ity, and working-age population (rather than total (5) primary energy consumption, and (6) iron and population) as a measure of labor capacity. steel production.17 These demographic, industrial, The full GPI has six distinct periods during and military indicators reflect the “breadth and which the elements of power and the weights on those

5 FIGURE 5 Balance of Power, 1980–2012, Measured Using CINC

0.25

0.20

China France German Federal 0.15 Republic (before 1990) Germany India 0.10 Japan Russia United Kingdom United States 0.05 Share of G20 power measured using CINC

0 1980 1984 1988 1992 1996 20002004 2008 2012

SOURCE: Data from Singer, Bremer, and Stuckey, 1972. elements change. It also measures a state’s share of as the denominator, and scenarios built by aggregating power relative to the rest of the world (a denomina- the individual elements of our modified GPI measures. tor of 186 countries). The International Futures (IF) Table 1 summarizes the elements and weights of the model maintained at the University of Denver’s Pardee modified GPI. While distinct from the full GPI, our Center for International Futures has incorporated modified GPI is more consistent with standard views the GPI into its modeling framework and used the of the distribution of power in the 21st century than is IF model to create forecasts to 2050.22 The IF model the main alternative methodology (the CINC). While includes feedback loops between disparate factors we used independent data sources and spreadsheets to in an attempt to reflect systemic effects; all of these aggregate data, the resulting picture is closer to the full feedback loops involve input parameters that mod- GPI view than the CINC is and seems broadly more elers must set to try to match reality. Figure 6 illus- consistent with our understanding of the past 30 years trates how the period between 1980 and 2012 appears of history—certainly more consistent than the picture through this lens. Here we see a more intuitive and offered by the CINC. plausible story. During the last decade of the Cold When considering how to measure a nation’s War, the Soviet Union was the second-most-powerful power, researchers must decide whether to focus on state after the United States, until the fall of the Soviet absolute or relative capabilities. The proper modeling Union in 1991 caused it to become less powerful than choice depends on the purpose of the model. When Japan. Japan remained the second-most-powerful state considering inherently competitive questions, such as until 2004, when China passed it. the risk of warfare between states, relative measures For this report’s illustrative effort, we wanted a of power are particularly informative. In other cases, more transparent way to create alternative scenar- such as when considering nations’ abilities to provide ios, which prevented us from using the full GPI. The health and economic opportunities to their citizens, modified GPI presented here uses a subset of the full absolute levels are more informative than relative GPI’s measures and weights, a smaller set of countries ones. Investigations of the balance of power between

6 FIGURE 6 Balance of Power, 1980–2012, Measured Using Full GPI

20 China France Germany 15 India Japan Russia

Full GPI score United Kingdom 10 United States

0 1980 1984 1988 1992 1996 20002004 2008 2012

SOURCE: Data from Moyer and Markle, 2017.

Table 1 Elements and Weights of the Modified GPI Used in This Report

Dimension Indicator Weight Measure Source

Military capacity Nuclear weapons 5 Logarithm of number of Federation of American warheads Scientists Nuclear Notebook, 2019

Military power 20 Military expenditures Stockholm International Peace Research Institute, 2019a

Economic capacity GDP 25 GDP (PPP) Maddison Project Database, 2018

Trade 15 Total trade (exports plus imports) World Trade Organization historical trade data, 2019

Technological capacity Innovation 10 Research and development UN Educational, Scientific and expenditure Cultural Organization, 2019

Political capacity Governance 15 Government revenues International Centre for Tax and Development, 2019; World Trade Organization, 2019b

Human capacity Population 10 Working-age populationc UN Population Division, 2019 a We used data from the Correlates of War Project National Material Capabilities to fill in gaps for Russia and China pre-1988. b Start dates vary by country, with most countries' data beginning in 1980. We use imputed missing revenues using linear regression analysis based on country fixed effects and an interaction between country fixed effects and changes in GDP. c Population between 15 and 64 years old.

7 major powers are typically focused on competitive challenges, so metrics like the GPI measure a state’s FIGURE 7 power relative to other states. The full GPI compares Balance of Power in 2015 of the G20, each state’s power to that of the world as a whole (a Measured Using Modified GPI total of 186 countries). For our illustration in this report, we chose to use a smaller set. As an initial South Africa Argentina screening measure, we calculated the modified GPI Australia for all 19 state members of the G20. We then focused Turkey Mexico our analysis on the eight countries with the highest Canada GPI scores. Figure 7 shows the share of power among Indonesia the states of the G20 in 2015. The United States had Italy about 25 percent of the share of power, driven by its Saudi Arabia United States large economy and military spending. China follows, South Korea with about 20 percent of the share of power. These Brazil two states, combined with India, Russia, Germany, United Japan, France, and the United Kingdom, have had Kingdom more than 75 percent of the total power of the cur- France rent G20 countries throughout the period of 1980 Japan to 2017—again, measured by the modified GPI. China Figure 8 shows that, in all three measures of power (the full GPI, modified GPI, and CINC), these eight Germany Russia India countries have consistently represented a majority of power. Through all three lenses, these eight great

FIGURE 8 Combined Share of Global Power for United States, China, India, USSR/Russia, Germany, Japan, France, and the United Kingdom, Measured Using Full GPI, Modified GPI, and CINC, 1980–2010

100

Share of full GPI 20 Percentage of global power Share of CINC Share of modified GPI

0 1980 1985 19901995 2000 2010 2010

SOURCES: Correlates of War National Material Capabilities v5.0 (J. David Singer, Stuart Bremer, and John Stuckey, “Capability Distribution, Uncertainty, and Major Power War, 1820–1965,” v.5, data set, in Bruce Russett, ed., Peace, War, and Numbers, Beverly Hills, Calif.: Sage, 1972); Moyer and Markle, 2017.

8 powers possess the preponderance of power in the France, and the United Kingdom) has shifted from international system. 1980 through 2017.23 If we look at 2015, we can see We use GPI as the metric of global power in our similar results to the snapshot provided in Figure 7; illustrative examples because of its pedigree (it is used at that point in time, the United States had about by the NIC in some of its Global Trends reports) and 25 percent of the G20’s power, followed by China because we assess that it is a more accurate measure with about 20 percent. Comparing this view of the of state capacity than are alternative methodolo- balance of power using our modified GPI with that gies. This modeling choice—an assumption—could provided by the full GPI in Figure 6, one can see that certainly be replaced with an alternative or modified this is a larger share of power than China had in the index. For example, one might consider adding other full GPI (about 12 percent in 2012). This is because factors, such as metrics related to energy produc- the modified GPI we use here omits some of the tion, ability to project military force, or resilience to parameters included in the full GPI, on which China climate change. is relatively weaker.24 This once again illustrates how one’s choice of lens can affect perceptions of relative power and why we recommend explicitly examin- Changes in Global Power over Time ing how a different lens can change the narrative. Figure 7 provides a snapshot estimate of the bal- The methodology presented in this analysis could ance of power at a single point in time. To inform use either measure of power (and could even do an discussions regarding how the balance of power has explicit sensitivity analysis comparing the two), but changed in the past and might change in the future, the rest of this report will use the modified GPI for it can help to examine how these relative shares purposes of illustration. of power shift over time. Figure 9 shows how the From this vantage point, the United States is still relative share of power for the eight great powers (the the most powerful country, but its edge has eroded. United States, China, India, Russia, Germany, Japan, Looking over the past several decades, we can see

FIGURE 9 Share of G20’s Modified GPI of the United States, China, India, Russia, Germany, Japan, France, and United Kingdom (1980–2017)

0.35

0.30

0.25

China 0.20 France Germany India Japan 0.15 Russia United Kingdom United States 0.10

0.05

Share of G20 power (measured using modified GPI) 0 1980 1985 1990 1995 2000 2005 2010 2015

9 bout of declinism. In some elements of power, the Even though some United States has been eclipsed, and the margin between the United States and the next-most- scholars may view powerful state is the smallest it has been in at least four decades.25 Historical fluctuations in the distribu- indicators of global tion of relative power can shape current perspectives. Expectations of the future can also cast shadows power like the GPI onto the present, coloring how we perceive current events. If leaders feel that everything is going their as incomplete, these way, then they may become more risk-averse, lest they accidentally alter their expected positive tra- measures can provide jectory. Conversely, if leaders feel that they face an useful insight into impending decline in power, then they might become more risk-acceptant and try to find a way to prevent situations in which or slow that decline. In the most extreme setting, research suggests that significant shifts in global military force is the power might create heightened tensions and sensitiv- ities that increase the risk of major interstate war.26 ultimate arbiter. While policymakers and commentators debate the scope and nature of American interests, most would agree that U.S. core interests include managing great- the impact of some major events, such as the precip- power competition and preventing major war. Even itous decline of Russian power with the collapse of though some scholars may view indicators of global the Soviet Union. We can also see America’s relative power like the GPI as incomplete, these measures dominance during the 1990s (what some scholars can provide useful insight into situations in which called its “unipolar moment”) and then China’s dra- military force is the ultimate arbiter. matic rise during the first 15 years of the 21st century. On a smaller scale, we can see Japan’s relative rise and fall and India’s climb into the No. 3 position Global Power Dynamics and circa 2007. In 2017, the United States remained the Global Conflict most powerful country, but by the smallest margin during this period. This perspective also shows that There are many models that link the distribution of Russia’s overall share of power has remained quite global power to the prospects for major interstate stable and relatively small over the past 20 years. war, according to different theories of why wars Short-term crises, such as the oil price spike in 2008, occur.27 Put broadly, when assessing whether one did not dramatically shift the long-term balance of scenario is more stable than another, analysts apply power. a model (ranging from a heuristic to a formal model) The modified GPI perspective in Figure 9 echoes to assess the prospects for crisis or war under differ- the relative market exchange rate GDP perspective ent distributions of global power. One such approach displayed in Figure 3, even though the GPI includes would be to use a quantitative metric (such as the many more factors. China had a rapid period of rela- GPI) within a theoretical model that evaluates the tive growth in the 2000s, but that growth has begun likelihood of a war erupting under different distribu- to slow. The factors that contribute to the United tions of global power. States retaining an edge over China include its advan- There are many theoretical models that an tage in nuclear weapons and military expenditures. analyst could use for this purpose.28 Among these This perspective suggests part of the reason why models, power cycle theory represents an intriguing the United States may be currently experiencing a option due to its quantitative nature and its ability

10 to operate on aggregated metrics, such as the GPI. some broad characteristics because they emphasize Power cycle theory relates the relative distribution of long-term causes of war. In these theories, uneven power in the international system to the likelihood of rates of growth among states play an important role major wars—that is, large wars that will reorder the in creating systemic disequilibria. Theories differ international system.29 For this reason, it focuses on on which rates of growth matter most; some focus latent indicators of military power. The theory con- exclusively on economic growth, while others focus cerns long-term shifts in power that take place over on broader indexes that include population. Theories decades, rather than the year-to-year fluctuations in also differ on what configurations of powers are the military capabilities that arise as states actualize their most dangerous; for example, transition theories latent power by fielding new weapon systems, testing focus on when a rising power’s capabilities approach new technologies, or training their militaries in new those of the leading power in absolute terms, while concepts of operation. Power cycle theory posits that power cycle theory focuses on when the trend in the largest wars—measured by duration and num- a nation’s growth changes (peaks, bottoms out, or ber of casualties—tend to occur when multiple great reaches an inflection point). All theories generally powers simultaneously experience critical points at accept the argument that a discrepancy between a which their relative rates of growth fundamentally state’s perceived status and its desired status influ- shift. These major wars are also sometimes called ences its behavior. We use power cycle theory in this extensive wars because they involve multiple major report because of its unambiguous and quantifiable powers that fully mobilize, leading to a large number character (the theory leads to specific predictions tied of casualties and restructuring of the international to quantitative conditions). Although we apply power system. Scholars have found confirming evidence cycle theory, we do so mainly as an illustration of for the theory when testing it against the historical how one can combine international relations theory record as a whole and when examining case studies with future balance-of-power scenarios to consider in specific major wars (such as WWI).30 By focusing which ones may be more unstable than others; we on fundamental elements of national power and encourage strategists to consider many lenses when the risk of wars that could reorder the international evaluating scenarios. system, these sorts of frameworks can help strategists step back and look for structural shifts in power that Power Cycle Theory can destabilize the international system. Of course, destabilizing shifts represent only one concern out of To answer questions about whether a balance of many that national security strategists confront on a power in a given scenario makes a major war more daily basis—from terrorism and power vacuums to or less likely, one needs to apply a theory that relates nuclear proliferation and transnational crime—but certain configurations of power to predictions about they are a necessary concern that requires attention stability. Power transition theory, for example, might and foresight. Power cycle theory, like all models, is a focus on the period around 2023, when China’s simplification of reality, but we judge that it has value modified GPI score surpasses that of the United in helping analysts understand the balance of power States. Power cycle theory, however, suggests that and prospects for major wars in a systematic and the risk of war is higher when several major pow- quantifiable way. We do not view it as a replacement ers go through critical points at similar times—not for critical thinking, the study of history, regional when their shares of global power cross each other. expertise, or other methods. We consider it to be As mentioned earlier, critical points occur when the a valuable tool to add to the larger toolbox used by direction or acceleration of a state’s relative growth national security analysts and those concerned about trend changes, such as when a state’s power falls after how future trends could affect great-power competi- reaching its zenith or rises after reaching its nadir. tion and war. Critical points may also occur when the rate There are many theories of warfare involving of growth or decline accelerates or decelerates. For cycles.31 While each differs in particulars, they share example, in our baseline scenario, Chinese relative

11 growth experiences an inflection point around 2011. Leaders start to make hard choices about whether it Before 2011, Chinese relative growth was accelerat- is time to start pursuing their goals more assertively ing, in line with the economic trends that we dis- before the window of opportunity closes. In China’s cussed in the opening. After 2011, however, Chinese case, we have witnessed growing evidence since 2011 relative growth decelerates. While it is still growing that China has abandoned Deng’s patient strategy in absolute and relative terms, its rate of relative and begun to act more assertively.33 Some in China growth slowed. Figure 10 highlights that, between argue that what has changed is the world’s reaction to 1990 and 2010, China’s relative power growth rate China rather than China’s behavior.34 However, many accelerated. After 2010, its relative power growth rate analysts observed that, by 2010, China had begun to began decelerating. In the baseline scenario, its rela- debate the hide and bide strategy.35 Since then, China tive rate of growth continues to slow, but it does not has become significantly more assertive in pursuing peak. The point where China’s relative power growth its foreign policy goals. In 2013, China began a large- rate stops accelerating and begins to decelerate scale campaign to build artificial islands in the South (marked with the black dot in Figure 10) is a partic- China Sea to bolster its ability to enforce its claims ular type of critical point called an inflection point, there.36 In 2015, the China-led Asian Infrastructure and it has special significance for a rising power. Investment Bank (AIIB) began operation after several Before the inflection point, time is on the side of years of proposals and organization. The AIIB is seen the rising power. Rising powers have unrealized goals by many as a potential rival to the World Bank and and aspirations. Before the first inflection point, the International Monetary Fund, which China views however, the rising power has many incentives to be as dominated by American, European, and Japanese patient in achieving those goals and realizing its aspi- interests. In 2016, China announced the creation of rations. If it waits, it will have more relative power its first overseas military base in Djibouti. During in the future and therefore more bargaining leverage a 2017 speech to the Chinese Communist Party to achieve its goals. The strategy espoused by Deng Congress, President Xi broke with Deng’s strategy, Xiaoping, which is sometimes translated as hide and saying “It is time for us to take center stage in the bide, reflects this sort of patient strategy.32 world” and calling China a “great power.” The first inflection point changes these incen- Power cycle theory argues that it is not coinci- tives. When relative power growth begins to decel- dence that a state’s foreign policy behavior would erate, the state’s leaders begin to experience the first change as it passes through this sort of critical point. reminders that their country will not rise indefinitely. Other types of critical points are when its relative power peaks (an upper turning point), when a declin- FIGURE 10 ing power’s rate of decline slows (the second inflection Chinese Power Growth Relative to G20, point), and when its relative power bottoms out and Baseline Scenario, 1965–2040 begins to rebound (the lower turning point). The theory argues that each type of critical point pres- ents challenges for national leaders as they assess the 0.30 country’s position and trajectory in the international 0.25 system and decide how to adjust their policies to

0.20 achieve their goals in a new environment. Rising states tend to have role deficits as inter- 0.15 national institutions are slow to accommodate their 0.10 growing power (for example, China’s complaints about an underrepresentation in the World Bank 0.05 Share of G20 power (measured using GPI) and International Monetary Fund or India’s push for 0 a permanent seat on the UN Security Council) and 1965 1975 1985 1995 2005 2015 2025 2035 give them commensurate influence; states on the decline tend to have capability deficits as their power

12 erodes despite their maintaining prominent roles in to speed up. Several major powers going through institutions such as the UN Security Council (such such periods at the same time does not guarantee a as France and the United Kingdom retaining perma- war between these countries but increases the risk of nent seats on the UN Security Council).37 Research such a conflict. by Doran and Parsons suggests that when multiple To be clear, power cycle theory does not predict great powers pass through critical points, the risk how individual leaders will respond to these criti- of miscalculation increases, and the systemwide cal periods. Rather, it suggests that critical periods risk of a major war increases.38 For example, Figure are important environmental factors that shape the 11 shows that, in the years leading up to WWI, five perceptions, incentives, and policy choices of indi- major powers (Germany, Italy, Austria-Hungary, the viduals. These factors can affect leaders, as well as United States, and France) all passed through critical their constituencies. Research suggests that, during points, and four were in their critical window—the periods around a critical point, it is more difficult for period from three years before to seven years after leaders to make accurate assessments of the relative the critical point. balance of power, increasing the risk of miscalcu- Power cycle theory does not claim that these lation, including the risk of underestimating an overlaps cause conflict but rather that they create opponent’s willingness to fight.39 The combination of conditions ripe for major conflict, similar to the this uncertainty and a heightened sense of urgency buildup of dry kindling increasing the risk of a contributes to the increased risk of armed conflict. major forest fire. The key idea is about expectations; As we noted earlier, power cycle theory is just countries are more likely to reassess their role in the one lens that analysts can use to assess the risk of war global order or be unusually sensitive to challengers in the alternative scenarios. An alternative lens would when the trajectory of their share of global power be power transition theory. One perspective on power changes: Their share of global power starts to decline, transition theory posits that the risk of war increases their share of global power starts to rise, their rate of when the ratio of the established power’s capacity global power increase (or decrease) starts to slow, or (in this case, the United States) to the rising power’s their rate of global power increase (or decrease) starts capacity (in this case, China) falls below two.40 In the

FIGURE 11 States in Critical Window, by Year, During the Lead-Up to World War I, 1900–1920

United States

United Kingdom

Austria- Hungary

Germany World War I Russia

France

Italy

1900 1901 1902 1903 190419051906 1907 1908 1909 1910 19111912 19131914 1915 1916 1917 1918 1919 1920

SOURCE: Data from Tessman and Chan, 2004, and Doran, 1991.

13 baseline case of the balance of power using the mod- executives recognize a fundamental change before ified GPI summarized in Figure 9, China breached their competitors, enabling a quicker and more this threshold in 2007, so by that measure the power effective response. This example highlights another transition had already begun, and therefore this benefit of scenario thinking: It does not require per- example does not meaningfully diverge from the fect foresight or prediction. baseline case.41 A full scenario planning exercise is beyond the scope of this report. Instead, we designed examples that explicitly show the flexibility of our framework Scenario Thinking to support a broader scenario planning exercise. As discussed earlier, it is difficult to diagnose the We begin by constructing a baseline scenario of a current balance of power. As we consider future future balance of power, measured using the modified scenarios, these difficulties compound. Despite these GPI. For each element of the GPI, we constructed a challenges, there are potential benefits to considering baseline projection out to 2040.45 Table 2 describes how future scenarios. A major benefit of scenario think- we projected each element of the GPI to construct this ing is to help analysts and policymakers quickly rec- baseline scenario. After describing this baseline sce- ognize when fundamental conditions have changed, nario, we illustrate how alternative projections may or requiring a different policy response. Another may not alter conclusions about the risks identified by prominent use of scenarios in policy analysis involves power cycle theory. Specifically, for completeness, we developing adaptive strategies to handle a wide range illustrate the impact of altering the projection of one of potential futures.42 Scenario thinking is also a measure for one country, one measure for all coun- fundamental way to handle uncertainty in defense tries, and multiple measures for multiple countries. analysis.43 Figure 12 summarizes how the balance of power A classic example of how scenario-based plan- shifts in the baseline scenario. We can see that the ning can enable quicker recognition of fundamental baseline assumptions, which include sustained changes and more perceptive reactions to them can Chinese GDP growth that gradually moderates, lead to be found in Shell Oil’s experience in the 1960s and China steadily continuing to gain power relative to the 1970s.44 In the 1960s, Shell engaged in a scenario United States. This scenario means that China shakes planning exercise in which executives considered a off the temporary pause (circa 2015) in its relative rise range of alternative futures. In one of these futures, that we noted when we introduced the modified GPI oil supply suddenly contracted. The executives in Figure 9. While China continues to gain power rel- thought through the second-order effects of this ative to the United States in this scenario, it does so at shock, including how the resulting price spike would a slower rate than it did during the meteoric Chinese lead to demand destruction in the medium term, expansion from 2000 to 2015. We can also see that this meaning that oil refining capacity would not need scenario includes a steady growth in Indian power, to expand as quickly as it had since the early 1950s. with India emerging as the clear third-most-power- When the 1973 oil embargo hit, Shell executives ful state. Japan’s relative power remains stable, and remembered the scenario planning exercise and Russia’s relative power continues to erode. the second-order implications of the price spike. Shell moved to quickly reduce its refining capacity, enabling it to avoid costly excess capacity when oil Assessing Future Scenarios demand dropped in the following years. In contrast, A benefit of this framework is that we can compare the oil industry as a whole kept adding refining the risk of conflict predicted in the baseline scenario capacity at the same rate for two years, and industry- with the risk of conflict predicted in alternative sce- wide refining capacity did not stop growing narios. We constructed three illustrative examples by until 1980. To be clear, the Shell scenario planning making different assumptions about what the future exercise did not predict the Yom Kippur War or the might look like. As long as alternative assumptions following 1973 oil crisis. The exercise did help Shell

14 TABLE 2 Forecast Methods for Each Element of the GPI

Dimension Indicator Measure Baseline Projection Method Military capacity Nuclear weapons Logarithm of number of Beginning in 2015, for China and India we warheads assume the number of warheads increases following a linear OLS projection based on a rolling window from the previous eight years. We assume other countries’ count of warheads remains stable at 2015 levels. Military power Military expenditures Beginning in 2018, we assume military expenditures grow at the same rate as the country’s average annual growth rate for military expenditures from 2005 to 2017. Economic capacity GDP GDP (PPP) Beginning in 2017, we applied the GDP growth rates projections from the Organisation for Economic Co-operation and Development (OECD) to our GDP data where available. For Argentina and Saudi Arabia, we applied GDP growth rate projections from World Economic Outlook and IF to our GDP data. Trade Total trade (exports plus Beginning in 2018, we applied a linear forecast imports) based on the prior 17 years. Technological capacity Innovation R&D expenditure We interpolate missing data on the percentage of GDP spent on R&D, then assume countries maintain their most recently reported percentage of GDP spent on R&D.a Political capacity Governance Government revenues We use a linear OLS regression with country fixed effects and a country-specific coefficient on GDP to project government revenue as a percentage of GDP. We then multiply this value by the projected GDP.b Human capacity Population Working-age populationc We use the UN World Population Prospects 2017 “Medium Variant” as our baseline. NOTES: OLS = ordinary least squares a For most countries, 2014 or 2015 is the last reporting year for which data on the percentage of GDP spent on R&D is available. Total spending on R&D will still vary as the country’s GDP rises or falls. b Start dates vary by country, with most country’s data beginning in 1980 and ending in 2014 or 2015. c Number of people between 15 and 64 years old. can be articulated in ways that have quantifiable by climate change, called the shared socioeconomic implications for nations’ share of global power, we pathways (SSPs), that results in changes to GDP and can use power cycle theory to describe the risk of population for many countries. In particular, we use conflict in alternative scenarios in comparison with the second SSP, known as SSP2 or the “Middle of the the baseline scenario. Road–Intermediate Challenges” scenario. These illus- We illustrate the flexibility of this approach trative examples were chosen in part for their political using three illustrative examples, as listed in Figure and economic relevance and in part for their ability to 13. The first example, “Chinese Lost Decade,” uses demonstrate the flexibility of this methodology. an alternative assumption about a single GPI factor, GDP, for a single country, China. The second example, Illustrative Example 1: Chinese Lost “Rapid Global Population Increase,” uses an alternative assumption about a single GPI factor, popula- Decade tion, for many countries. The third example, “SSP2 The Chinese Lost Decade example considers the Climate Change,” looks at one of five commonly used implications of China experiencing a decade of slower collections of assumptions about the future implied economic growth than is currently forecast. Some

15 FIGURE 12 Balance of Power (Measured Using Modified GPI) in Baseline Scenario (1980–2040)

0.35

0.30

0.25

China 0.20 France Germany India Japan 0.15 Russia United Kingdom United States 0.10

0.05

Share of G20 power (measured using modified GPI) 0 1980 1990 2000 2010 2020 2030 2040

revenue and R&D spending are forecast as a fraction FIGURE 13 of GDP, the changes in GDP also change the total Three Illustrative Examples amount of government revenue and R&D spending. Alternative assumptions involving feedback loops Illustrative Example Change Relative between Chinese GDP and other elements of China’s Name to Baseline GPI could be implemented but are not necessary for our illustrative purposes. This illustration also does Chinese Lost Decade not include any feedback loops between the change in Chinese economic growth and other countries’ eco- Rapid Global Population nomic growth, which is another factor that should be considered in a full application of this approach. SSP2 Climate Change Figure 14 shows how GPI shifts under this alternative assumption for the eight major powers.47 Unsurprisingly, China’s power falls, and, since GPI scholars argue that expectations for China’s future is a relative measure, other countries’ share of global growth are overstated and that events will eventually power rises. These gains are most notable for the transpire to cause a drastic slowdown in the Chinese United States and India. economy, followed by a period of more-modest Is the risk of global conflict greater in the growth.46 We can examine how such an example Chinese Lost Decade example or the baseline sce- differs from the baseline scenario by replacing baseline nario? If one uses the lens of power cycle theory, GDP forecasts of China’s economic growth (which are then multiple major powers going through critical based on OECD forecasts) with a 1-percent growth points around the same time increases the risk of rate from 2020 to 2030 and a 3-percent growth rate global conflict, and the location of countries’ critical from 2031 to 2040. Other elements of the GPI are points can shift under different assumptions about not changed directly, although because government the future. Figures 15 and 16 show the alignment of

16 FIGURE 14 How GPI Shifts Under Chinese Lost Decade Scenario

0.35

China 0.30 China baseline France France baseline 0.25 Germany Germany baseline India 0.20 India baseline Japan Japan baseline 0.15 Russia Russia baseline United Kingdom Share of G20 power 0.10 United Kingdom baseline United States 0.05 United States baseline

0 1980 1990 2000 2010 2020 2030 2040

FIGURE 15 FIGURE 16 States in Critical Window, by Year, in States in Critical Window, by Year, Baseline Scenario (2015–2040) in Chinese Lost Decade Example (2015–2040)

United States United States China China India India Russia Russia Germany Germany Japan Japan 2015 2020 2025 2030 2035 2040 2015 2020 2025 2030 2035 2040 critical points in the baseline and in the Chinese Lost Decade example. For each critical point, we consider which the state must wrestle with the challenges of a window starting three years prior to the year of the changing trends and expectations created by the the critical point and seven years after the year of critical point. The size of the window is based on the critical point, for a total of an 11-year (inclusive) power cycle theory research, but alternative assump- window.48 This window represents a period during tions could easily be applied.

17 First, the figures remind us that critical points critical windows overlap slightly, creating a period are rare events. They represent identity crises on a in which the two largest powers are both reassessing national scale—hence, not something that happens their role in the world, increasing the risk of global regularly. The baseline scenario shows no sign of conflict. critical points on the horizon (to 2040) for any of the eight largest powers. China and Japan are Illustrative Example 2: Rapid Global both emerging from the end of their critical point windows—China’s caused by a realization that its Population Increase rapid growth is slowing, and Japan’s caused by sta- The Rapid Global Population Increase example bilization after years of relative decline. Otherwise, considers the implications of different assumptions the baseline scenario could be viewed as showing a about future fertility rates. The baseline scenario “business as usual” future. uses the “medium variant” of the UN’s forecast There are a few notable changes in the Chinese for population ages 15 to 64.49 In this example, we Lost Decade example in Figure 14. The United States replace the UN’s medium variant population fore- experiences a critical point in the early 2020s, around cast with assumptions of a higher global population the start of the Chinese economic crisis, as China’s growth rate—the UN’s “high variant” population decline slows the decline in U.S. relative power. As forecast. Specifically, for all countries, we replace the for China, the economic slowdown itself does not median population forecast for ages 15 to 64 with trigger a critical point because the Chinese economy the upper 95th percentile from the same UN fore- had already been slowing (albeit at a calmer rate). cast. Although the magnitude varies slightly from However, China does experience a critical point as it country to country, all countries see an increase emerges from the economic crisis because the rate at in forecast working-age population relative to the which China gains relative power begins to increase medium variant in the baseline. Figure 17 shows the again. In the assumptions of this example, the two resulting modified GPI scores; we find in this case

FIGURE 17 How GPI Shifts Under Rapid Global Population Increase

0.35

0.30

0.25

China 0.20 France Germany India Japan 0.15 Russia United Kingdom Share of G20 power United States 0.10

0.05

0 1980 1990 2000 2010 2020 2030 2040

18 that this alternative assumption about global fertility this because we want changes in the probability of rates makes no noticeable difference in the relative conflict to be the output of our model, not an input. GPI. The same story is true if we use the bottom 95th Note that we choose to focus on SSP2, but the same percentile of the population forecast. This is a largely approach could easily be used to examine other SSP intuitive result—because GPI is a relative measure, assumptions. there is no noticeable change when all countries face Different SSP scenarios are assumptions about a similar shock that moves all the countries in the how societies will behave. They are complemented same direction. If one country, such as the United by another set of assumptions known as Relative States, maintained higher population growth while Concentration Pathways (RCPs). RCPs measure (in the other major powers suffered population declines, watts per square meter, W/m2) the amount of addi- then we would expect to see a shift in the GPI. This is tional energy the Earth will absorb rather than return a useful example because it reminds us that there are to space due to the greenhouse effect. This in turn cases where, at least under the assumptions of power causes the outcomes associated with climate change cycle theory, alternative assumptions do not signifi- temperature changes, sea-level rise, and changes in cantly alter the probability of major-power conflict. the frequency and severity of extreme weather events. Even extreme assumptions about global demographic There are four RCPs that make different assumptions trends make little difference in this assessment of risk about the amount of additional energy the Earth will because GPI is a relative measure. absorb: 2.6 W/m2, 4.5 W/m2, 6.0 W/m2, and 8.5 W/m2.

Illustrative Example 3: SSP2 Climate FIGURE 18 Change Temperature Changes Associated with RCP4.5 and RCP8.5 The SSP2 Climate Change example considers the implications of replacing the GDP and population Mean RCP4.5 temperature 2021–2040 minus 1986–2005 January–December AR5 CMIP5 subset projections from the baseline scenario with GDP and population projections associated with standard climate change assumptions. Projections of how GDP and other outcomes will be affected by climate change depend on assumptions about future deci- Celsius sions by governments, institutions, and the general 11.0 9.0 population, including assumptions about the prob- 7.0 ability of future conflict. There are five commonly 5.0 4.0 used collections of assumptions about these future 3.0 decisions, called the SSPs. These SSPs were con- Mean RCP8.5 temperature 2021–2040 minus 2.0 1.5 structed by a global consortium of climate scientists 1986–2005 January–December AR5 CMIP5 subset 1.0 to provide a common set of assumptions for discuss- 0.5 ing and researching the impacts of climate change.50 0.0 –0.5 SSPs are the global standard for this purpose and –1.0 are being incorporated into the Intergovernmental –1.5 Panel on Climate Change’s sixth assessment report.51 –2.0 We choose one of these pathways, known as SSP2 or the “Middle of the Road–Intermediate Challenges” scenario, because it makes no assumptions about radical changes in the probability of conflict. In SOURCE: Koninklijk Nederlands Meteorologisch Instituut, SSP2, “social, economic, and technological trends do “KNMI Climate Explorer,” database, undated. not shift markedly from historical patterns.”52 We do NOTE: AR5 = 5th Assessment Report; CMIP5 = Coupled Model Intercomparison Project Phase 5.

19 RCP2.6 is associated with the least increase in global Figure 19 shows the GPI under the SSP2 Climate temperatures and other elements of climate change, Change example and RCP8.5. Relative to baseline, while RCP4.5, RCP6.0, and RCP8.5 are associated China in this example sees significant gains in GPI, with increasingly large changes in global tempera- the United States and India see moderate declines, tures and other elements of climate change. As shown and Russia sees a slight gain. in Figure 18, the impacts associated with any RCP Figures 20 and 21 show clearly why this is the vary across the globe, foreshadowing and underpin- case. Under SSP2 and RCP8.5, global GDP contin- ning this example’s varying impact on power across ues to increase but by significantly less than in the countries. baseline scenario. Figure 20 shows the GDP of India The economic, demographic, and climatic and Russia as examples. India takes the largest hit impacts associated with different SSP and RCP relative to baseline, with its GDP in 2040 being more assumptions involve large and complex modeling than $10 trillion lower than it would have been under efforts. Our SSP2 Climate Change example is com- baseline scenario assumptions. Russia is largely paratively simplistic: We make two specific changes unaffected, and therefore gains power relative to to our baseline scenario. First, we replace our base- India. Figure 21 shows the growth in GDP from 2017 line forecasts of countries’ populations ages 15 to 64 to 2040 for all eight countries. China, Russia, and with SSP2’s population assumptions. Second, we rep- Germany see increases in GDP under less drastic licate the methodology of Burke, Hsiang, and Miguel RCP assumptions, and Japan is largely unaffected. (2015) to map the temperature changes associated India, the United Kingdom, the United States, and with all four RCPs and the population assumptions France miss out on a large amount of GDP growth associated with SSP2 to changes in countries’ GDP relative to baseline. Another element we notice from growth rates.53 In this way, our SSP2 Climate Change Figure 20 is that, for many countries, the difference example involves changes to both population and in GDP growth is driven largely by switching from GDP for all countries. baseline assumptions to growth rates based on SSP2 assumptions and the Burke, Hsiang, and Miguel

FIGURE 19 How GPI Shifts Under SSP2 Climate Change

0.35

0 1980 1990 2000 2010 2020 2030 2040

20 FIGURE 20 (2015) model, with variation among the different GDP of India and Russia in Baseline RCPs being relatively minor. This helps illustrate Scenario and in SSP2 and RCP8.5 the flexibility of this approach, showing how the methodology can easily incorporate a wide variety of assumptions. 30 Despite the shifts in outcomes, we again see no new critical points arise under any RCP assumptions 25 India baseline as the future GPI forecasts of major powers remain India RCP8.5 smooth. Under the most extreme RCP8.5 assump- Russia baseline 20 Russia RCP8.5 tion, we do see minor changes in the timing of earlier critical points, which shifts the end date 15 of preexisting critical point windows (Figure 22). Because critical points are calculated by identifying 10 where a smoothed measure of GPI experiences a minimum, maximum, or inflection point, a change

2011 U.S. dollars (trillions) 5 in the future trajectory of a country’s GPI can sometimes shift the location of critical points in recent 0 years. 1980 1990 2000 2010 2020 2030 2040

Insights In this work, we present a methodology for quantitatively comparing the risk of major-power conflict across a variety of future scenarios. We used three illustrative examples to show the potential of this methodology: Chinese Lost Decade, Rapid Global

FIGURE 21 GDP Growth from 2017 to 2040

300

Baseline 250 RCP2.6, SSP2 RCP4.5, SSP2 200 RCP6.0, SSP2 RCP8.5, SSP2

150

100

50 Percentage growth in GDP from 2017 to 2040 0 India China United RussiaUnited FranceJapan Germany Kingdom States

21 FIGURE 22 power among the largest states. Such shocks might Number of States in Critical Window, by still affect the risk of conflict for reasons unrelated to Year, in SSP2 Climate Change Example the balance of power. (RCP8.5) This Methodology Enables Comparison of United Different Assumptions States One of the major benefits of this methodology is China its versatility. If a researcher finds the assumptions behind a particular GPI forecast unpalatable or India prefers an alternative measure of national power, the implications of those alternative assumptions Russia can be evaluated as long as they can be described quantitatively. Thus, this methodology provides Germany flexibility on two levels. First, it provides a quantitative method for assessing the risk of future conflict Japan between global powers under alternative scenarios. Second, the methodology provides a framework for 2015 2020 2025 2030 2035 2040 systematically comparing the implications of different assumptions in determining the global balance of Population Increase, and SSP2 Climate Change. The power and the associated risk of conflict. An import- results provide several important insights. ant limitation is that the framework does not assess risks of great-power conflict that are unrelated to the balance of power. Insights from This Exercise The Balance of Power Can Be Sticky The Believability of the Outputs Depends on The different assumptions in our three illustrations the Believability of the Inputs did not result in drastically different forecasts of In highlighting this framework’s capability to flexibly the balance of power. Even assumptions that might incorporate any assumptions, it is important to appear extreme in the moment don’t radically alter remember that the believability of the risk assessment countries’ share of global power. In other words, output by the model depends on the believability of global power is “sticky.” This makes sense because the assumptions that are inputs. Because the model the elements of power, such as size of economy and is agnostic to the believability of the input assump- military spending, take time to evolve. The examples tions, it provides a common ground for comparing in this report illustrate that GPI is not easily shifted alternative assumptions. That does not mean that all over short periods of time. Thinking in this manner possible scenarios are equally likely, and the model can help policymakers focus on truly game-changing user will need to assess separately the likelihood of problems rather than day-to-day crises. That said, the scenario inputs. As noted in our first insight, “sticky” does not mean “immutable.” our illustrative exercises remind users that the risk of global conflict is not likely to radically change Not All Shocks Are Equal overnight. Because our model treats global power as a relative concept, the types of shocks that matter for deter- Insights This Approach Could Provide mining the balance of power are the types that create “winners” and “losers.” Shocks that shift elements A Tool to Help Avoid World War III of GPI up or down about the same amount for all There are many reasons that great powers might countries don’t change the relative distribution of end up engaged in war. No theory can predict with

22 certainty when conflict between great powers will A Tool That Can Help Decisionmakers Identify and will not happen. Instead, theories such as power and Avoid Emerging Threats cycle theory identify conditions that correspond with Another benefit of this model is that it enables us to increased risk of global power conflict. The method- think about the importance of changes in the future. ology presented in this report illustrates a quantita- While power can be “sticky,” when it does shift, that tive and flexible approach to assessing the future risk shift can create serious consequences. Therefore, of great-power conflict in hundreds or thousands of there is value in aids that help distinguish between potential future scenarios. day-to-day crises and the rare but important shifts Although the baseline and three examples we in the global balance of power. Future researchers examined do not describe conditions that are ripe could use this model to examine a variety of differ- for a conflict between major powers, we emphasize ent scenarios and see which ones create conditions that these examples were picked for illustrative pur- of heightened risk. Armed with that knowledge, poses and do not represent a systematic assessment decisionmakers can then work to avoid those paths of potential future scenarios. Alternative scenarios of heightened risk or put extra precautions in place might produce greater risk, and this methodology should they occur. The Chinese Lost Decade example does not capture risks unrelated to global power. reminds decisionmakers that if such an economic Further, this methodology provides neither a prob- decline occurs in China, it could elevate tensions abilistic forecast nor a diagnosis of the problem. between the United States and China, creating major Rather, the approach is akin to the “check engine” risks for both. Decisionmakers, armed with strategic light on a car’s dashboard: If it indicates an increased context, could then seek to avoid provoking such a risk of a major war, then policymakers and analysts conflict or proceed with additional caution should should pause and examine why before continuing to a similar scenario occur. Though this framework is drive forward. illustrated with a small set of illustrative examples, the approach is designed to be fully quantified so that it could be used to generate, assess, and compare thousands of alternative scenarios.

Appendix: Mathematical Details of Calculating Critical Points In line with other power cycle analyses, this analysis Ft()=+ b b** t + b t2 + b * t3 used the following asymmetric logistic growth func- 12 3 4 . tion to model each state’s power cycle.* k For each country, the coefficients b1 through b4 Pt()=+ d 1exp+ []Ft() are estimated using a least squares regression, which leads to a function that estimates a state’s relative power at time t. To calculate the upper and lower The term P represents a country’s relative power turning points of the state’s power cycle, we set the at time t. The terms d and k are related to the min- first derivative equal to zero (P’(t) = 0) and solve for t. imum and maximum scores for the state’s relative To calculate the first and second inflection points, we power during the time frame. These values can be de- set the second derivative equal to zero (P’’(t) = 0) and rived by inspecting the time series of a state’s relative solve for t. This process generated the critical points power, as measured by the GPI.† F(t) is a polynomial: summarized in the report for the various examples.

* Doran, 1991, p. 264; Tessman and Chan, 2004, p. 135. † The maximum value, or upper asymptote, is equal to (k + d), and the minimum value, or lower asymptote, is equal to d.

23 Notes 9 Graham Allison’s work has catalyzed public discussion of the risk that the United States and China could fall into conflict; see 1 U.S. Department of Defense, Summary of the 2018 National Allison, 2017. For an academic survey of the lessons of history and Defense Strategy of the United States of America: Sharpening the the risk of a war between the United States and China, including American Military's Competitive Edge, Washington, D.C., 2018, two critiques of Allison’s “Thucydides Trap” argument, see Rich- p. 1. ard N. Rosecrance and Steven E. Miller, eds., The Next Great War? 2 Eliot A. Cohen, The Big Stick: The Limits of Soft Power and The Roots of World War I and the Risk of U.S.-China Conflict, the Necessity of Military Force, New York: Basic Books, 2016; Cambridge, Mass.: MIT Press, 2015. Ash Carter, “Remarks Previewing the FY 2017 Defense Budget,” 10 Another form of influence is economic influence, such as the February 2, 2016. infrastructure investments China is pursuing through its Belt and 3 There are numerous ways that scholars have defined power Road Initiative. in international relations. One taxonomy differentiates between 11 For example, in its annual Military Balance report, the Interna- hard power, which describes the power to coerce—in an extreme tional Institute of Strategic Studies (IISS) generally uses exchange case, through military invasion—and soft power, which describes rates to measure the military budgets of most countries, but it uses one’s ability to attract—for example, through persuasive media or PPP conversion rates for Russia and China owing to caveats con- appealing culture. Hard power has been at the center of competi- cerning the reliability of official economic statistics and to China tions between major powers. While balance of power can refer to being a transitional economy “whose productive capabilities are a very specific condition where no single state is strong enough to similar to those of developed economies, but where cost and price militarily dominate all others, we use the term here in a broader structure are often much lower than world levels.” However, IISS sense to mean “what is the distribution of power among the largest goes on to caution that “No specific PPP rate exists for the military states, and what are their trajectories?” For example, is the United sector, and its use for this purpose should be treated with caution. States still the most powerful state? Is its advantage eroding? For Furthermore, there is no definitive guide as to which elements of an overview of the role of power in international relations, see military spending should be calculated using the limited PPP rates David A. Baldwin, Power and International Relations: A Concep- available.” IISS, Military Balance, Vol. 118, No. 1, 2018, p. 511. tual Approach, Princeton, N.J.: Princeton University Press, 2016. 12 For example, China continues to use Russian-built engines for 4 Meredith Reid Sarkees and Frank Whelon Wayman, “Resort its J-20 advanced fighter because China’s indigenously produced to War: 1816–2007,” version 4.0, database, Washington, D.C.: CQ engine for the J-20 remains behind schedule. See Guy Norris, “Key Press, 2010. Milestones Loom for New Civil and Military Engines,” Avia- 5 See, for example, Graham Allison, Destined for War: Can tion Week and Space Technology, December 22, 2017, p. 84; and America and China Escape Thucydides’s Trap? Boston, Mass.: Andrew Tate, “Aerial Refuelling Capability of China’s J-20 Fighter Houghton Mifflin Harcourt, 2017. Confirmed,” Jane’s Defence Weekly, November 19, 2018. 13 6 For example, it is broadly agreed that economic power plays United Nations University International Human Dimensions Programme and United Nations Environment Programme, Inclu- an important role in understanding the rise and fall of states, sive Wealth Report 2014: Measuring Progress Toward Sustainability, although there are competing views on its role. Members of the Cambridge, United Kingdom: Cambridge University Press, 2014. realist school tend to focus on economic power as a latent military power that could be mobilized in the event of a major conflict, a 14 Economists have raised concerns that China has significantly prime example being the United States mobilizing before WWII. overstated its GDP for at least 20 years; see, for example, Thomas Liberal internationalists tend to devote more attention to how Rawski, “What Is Happening to China’s GDP Statistics?” China economic power influences international institutions and norms. Economic Review, Vol. 12, No. 4, 2001, pp. 347–354; and Alice Siqi The post-WWII era provides an example: American economic Han, “China’s Economic Black Box,” Foreign Policy, March 17, preeminence underwrote the Bretton Woods economic system. 2019. Some economists have even argued that there is a fundamental divergence between China’s reported GDP and the country’s 7 A prominent example of the declinism school is Paul Kenne- underlying economy; see Michael Pettis, “What Is GDP in China?” dy’s The Rise and Fall of the Great Powers: Economic Change China Financial Markets, blog, Carnegie Endowment for Interna- and Military Conflict from 1500 to 2000, New York: Vintage tional Peace, January 16, 2019. Books, 1987, while a more recent example is Fareed Zakaria’s The Post-American World, New York: W. W. Norton & Company, 2008. 15 David M. Lampton, The Three Faces of Chinese Power: Might, The most famous work of triumphalism is Francis Fukuyama’s Money, and Minds, Oakland, Calif.: University of California Press, The End of History and the Last Man, New York: Free Press, 1992. April 2008; Michael Pillsbury, China Debates the Future Security For a recent example of a counterdeclinist argument, see Bruce Environment, Washington, D.C.: National Defense University Jones, Still Ours to Lead: America, Rising Powers, and the Tension Press, December 2004. Between Rivalry and Restraint, Washington, D.C.: Brookings, 2014. 16 See, for example, Yan Xuetong, “The Rise of China and Its For more background on declinism see Eric Edelman, Under- Power Status,” Chinese Journal of International Politics, Vol. 1, standing America’s Contested Primacy, Washington, D.C.: Center 2006, pp. 5–33; and Huang Shuofeng, Zonghe Guoli Lun for Strategic and Budgetary Assessments, 2010. [On Comprehensive National Power], Beijing: Zhongguo Shehui 8 Hal Brands and Charles Edel, The Lessons of Tragedy: State- Kexue Chubanshe, 1992. craft and World Order, New Haven, Conn.: Yale University Press, 17 See J. David Singer, Stuart Bremer, and John Stuckey, “Capabil- 2019. ity Distribution, Uncertainty, and Major Power War, 1820–1965,” version 5, data set, Beverly Hills, Calif.: Sage, 1972.

24 18 Michael Beckley, “The Power of Nations: Measuring What uses to explain why wars occur. “First image” theories focus Matters,” International Security, Vol. 43, No. 2, 2018, p. 10. on human nature as the fundamental cause of conflict, “second image” theories focus on the nature of regimes, and “third 19 See, for example, Ashley J. Tellis, Janice Bially, Christopher image” theories focus on the nature of the international system Layne, and Melissa McPherson, Measuring National Power (Kenneth Waltz, Man, the State, and War, New York: Columbia in the Postindustrial Age, Santa Monica, Calif.: RAND Corp., University Press, 1959). An extensive RAND study synthesized a MR-1110-A, 2000; and Anthony Lanyi and Kimberly Brickell, variety of international relations theories to develop a statistical Institutional, Economic, and Organizational Basis of Military model of both interstate and intrastate wars (Stephen Watts, Capability, College Park, Md.: Center for Institutional Reform Bryan Frederick, Jennifer Kavanagh, Angela O’Mahony, and the Informal Sector, University of Maryland, 1999. Thomas S. Szayna, Matthew Lane, Alexander Stephenson, and 20 Human capital considerations include such things as the Colin P. Clark, A More Peaceful World? Regional Conflict Trends training and motivation of military personnel, as well as the and U.S. Defense Planning, Santa Monica, Calif.: RAND Corpo- effectiveness of military organizations. It is hard to quantify the ration, RR-1177-A, 2017). effectiveness of these sorts of factors, which may lead to their 28 A great deal of the literature on conflict forecasting concerns importance being underestimated. See Lanyi and Brickell, 1999. predicting intrastate conflicts, which is not directly relevant 21 See NIC Global Trends reports (National Intelligence Coun- to this report’s topic (major interstate wars). See, for example, cil, “Global Trends,” webpage, Office of the Director of National Frank D. W. Witmer, Andrew M. Linke, John O’Loughlin, Intelligence, undated). There have been industrial-age measures Andrew Gettelman, and Arlene Laing, “Subnational Violent of power, and there is a broad literature on how to update them Conflict Forecasts for Sub-Saharan Africa, 2015–65, Using Cli- for the 21st century. For an example, see Tellis et al., 2000. mate-Sensitive Models,” Journal of Peace Research, Vol. 54, No. 2, 2017. The U.S. government has also funded multiple attempts to 22 For more on how the IF model uses the GPI and for an exam- create crisis forecasting systems that were much more ambitious ple forecast using it, see Jonathan Moyer and Alanna Markle, than this report’s focus on the balance of power between the Relative National Power Codebook, Denver, Colo.: Frederick S. largest states and the implied risk of major war among them. Pardee Center for International Futures, Josef Korbel School of See Sean O’Brian, “Crisis Early Warning and Decision Support: International Studies, University of Denver, 2017. Contemporary Approaches and Thoughts on Future Research,” 23 The choice of denominator—here the total power of all G20 International Studies Review, Vol, 12, No. 1, 2010, pp. 87–104. countries—makes no mechanical difference, because what 29 Power cycle theory was developed by Charles Doran; see matters is the power of the countries displayed relative to each Doran, 2003, and Charles F. Doran, Systems in Crisis: New other. The eight countries we examine have a smaller share of Imperatives of High Politics at Century’s End, Cambridge, United total global power than total G20 power, but their power relative Kingdom: Cambridge University Press, 1991. For an example of to each other is unchanged—the United States would still have a critique of power cycle theory, see Sarah Hülser, “Power Cycle twice the global power of China in 2007. In other words, the Theory Reconsidered: Is China Going to Destabilize the Inter- choice of denominator may double or halve the y-axis measure national Order?” Berlin: NFG Research Group, Freie Universität of power, but the shape and relation of the power curves, and Berlin, Working Paper Series, No. 6, July 2013. the resulting implications of the power cycle theory we use to assess risk, are identical regardless of whether the denominator 30 Wars have tended to be longer and involved more casualties is all countries, the G20, or just the power of the eight countries when participants are passing through critical points. See, for included in our examples. example, Charles F. Doran and Wes Parsons, “War and the Cycle of Relative Power,” American Political Science Review, Vol. 74, 24 For example, the full GPI uses working-age population No. 4, 1980, pp. 947–996; and Jacob L. Heim, “Tapping the Power adjusted by years of schooling, while our modified GPI uses total of Structural Change: Power Cycle Theory as an Instrument working-age population, so China’s lower rate of education in its in the Toolbox of National Security Decision-Making,” SAIS workforce reduces the relative contribution of its population to Review, Vol. 29, No. 2, 2009, pp. 113–127. A case study of the start the full GPI. of WWI illustrates how the theory explains the cases of the war; 25 In other words, this is the smallest margin of advantage that see Doran, 1991, pp. 125–140. the majority of the U.S. population has ever experienced (because 31 For example, George Modelski’s theory of “long cycles” about half of the U.S. population was 37 years old or younger in (George Modelski, “The Long Cycle of Global Politics and 2017). “Annual Estimates of the Resident Population by Single the Nation State,” Comparative Studies in Society and History, Year of Age and Sex for the United States: April 1, 2010 to July Vol. 20, 1978, pp. 214–235). 1, 2017,” U.S. Census Bureau, Population Division, American FactFinder database, release date: June 2018. 32 Lee Kuan Yew assessed that “the Chinese have figured out that if they stay with ‘peaceful rise’ and just contest for first 26 Charles F. Doran, “Economics, Philosophy of History, and the position economically and technologically, they cannot lose” ‘Single Dynamic’ of Power Cycle Theory: Expectations, Competi- (Graham Allison and Robert D. Blackwill, “Will China Ever Be tion, and Statecraft,” International Political Science Review, No. 1?” YaleGlobal Online newsletter, February 20, 2013). Vol. 24, No. 1, 2003. 33 For example, “. . . we’ve been told for a long, long time that 27 There are several different taxonomies that one can use to Deng Xiaoping’s action was this: ‘Hide your strength, bide your organize the various theories of why wars occur. A classic frame- time, never take the lead.’ Xi Jinping in his last five years turned work developed by Kenneth Waltz divided theories into three that on its head, now we see consciously and deliberately a more broad types, according to the independent variable the theory overtly activist Chinese foreign policy and security policy and

25 international economic policy in the world at large” (Tobin Term Policy Analysis, Santa Monica, Calif.: RAND Corporation, Harshaw, “Emperor Xi’s China Is Done Biding Its Time: A Q&A MR-1626-RPC, 2003. Robust decision making has been applied with Former Australian Prime Minister Kevin Rudd, Who Is to a wide variety of policy questions; see, for example, Robert Going to School to Study the Chinese President,” Bloomberg J. Lempert, Drake Warren, Ryan Henry, Robert W. Button, News, March 3, 2018). Jonathan Klenk, and Kate Giglio, Defense Resource Planning Under Uncertainty: An Application of Robust Decision Making to 34 One scholar likens China attempting to maintain a low profile Munitions Mix Planning, Santa Monica, Calif.: RAND Corpora- in 2018 to an elephant trying to hide behind a small tree (Ruan tion, RR-1112-OSD, 2016. Zongze, “Chinese Scholar Ruan Zongze Comments on Whether China Has Stopped ‘Hide and Bide’ Strategy,” Huanqiu Shibao 43 A useful overview of the use of scenarios in defense analy- Online in Chinese, May 11, 2018). sis can be found in chapter 3 of Paul Davis, Analysis to Inform Defense Planning Despite Austerity, Santa Monica, Calif.: RAND 35 “Less Biding and Hiding,” The Economist, December 4, 2010; Corporation, RR-482-OSD, 2014. Shin Kawashima, “The Development of the Debate Over ‘Hiding One’s Talents and Biding One’s Time,’” Asia-Pacific Review, 44 Kees van der Heijden, Scenarios: The Art of Strategic Conver- Vol. 18, No. 2, 2011, pp. 14–36. sation, Hoboken, N.J.: Wiley, 2005. 36 China’s claims to the South China Sea date to the founding 45 As mentioned in other endnotes, analysts could also use a of the country, and it fought skirmishes over the decades to computer model, such as the IF model, to include feedback loops defend and expand its influence there. China’s island-building and systemic effects when constructing future scenarios. campaign, however, has dwarfed the scale and speed of those 46 For an example of an argument that the Chinese economy of other countries. In less than two years, it created approxi- faces a serious risk of an extended recession, or “lost decade,” see mately 3,000 acres of artificial islands in the South China Sea, Michael Pettis, Avoiding the Fall: China’s Economic Restructur- two orders of magnitude more area than any other claimant. ing, Washington, D.C.: Carnegie Endowment for International Vietnam, for example, has reclaimed the most land other than Peace, 2013. Our report does not comment on the likelihood of a China, and Vietnam has created only 80 acres. Matthew South- decline in Chinese economic growth; we focus on demonstrating erland, China’s Island Building in the South China Sea: Damage that the implications of such assumptions on the risk of global to the Marine Environment, Implications, and International Law, conflict can be compared with the implications of alternative Washington, D.C.: U.S.-China Economic and Security Review assumptions about the future. Commission Staff Report, April 12, 2016. 47 We calculated the GPI for all members of the G20, and the 37 A state’s role can manifest multiple ways, including by being y-axis of Figure 11 reflects share of GPI among the G20. As included in important organizations, by being excluded from shown in Figure 5, the top six countries account for almost international pressure to change domestic policies, and by resolv- 75 percent of the total global power, making them more likely to ing international crises. See Brock F. Tessman, “The Evolution of have meaningful changes in their share of global power under Chinese Foreign Policy: New Incentives with Slowing Growth,” different scenarios. Asian Security, Vol. 5, No. 3, 2009, pp. 296–318. 48 Doran and Parsons (1980) used a 16-year window, while 38 Doran and Parsons (1980) conducted an early test for a Hebron and James (1997) used a ten-year window. Brock F. relationship between critical points and major wars during the Tessman and Steve Chan, in “Power Cycles, Risk Propensity, and period 1815–1975. Later work by Heim (2009) tested the relation- Great-Power Deterrence,” Journal of Conflict Resolution, Vol. 48, ship over a longer historical period, using a different data set. No. 2, 2004, ran a sensitivity analysis using windows ranging Other tests of the theory have been conducted by Houweling and from 16 to five years and found that power cycle theory still had Siccama (1991) and Hebron and James (1997) (Henk Houweling strong explanatory power even with smaller windows. and Jan G. Siccama, “Power Transitions and Critical Points as Predictors of Great Power War: Toward a Synthesis,” Journal of 49 UN population forecasts are available at UN Population Divi- Conflict Resolution, Vol. 34, No. 4, 1991, pp. 642–658; Lui Hebron sion, “World Population Prospects 2019,” database, 2019. and Patrick James, “Great Powers, Cycles of Relative Capability, 50 and Crises in World Politics,” International Interactions, Vol. 23, Keywan Riahi, Detlef P. van Vuuren, Elmar Kriegler, Jae No. 2, 1997, pp. 145–173). Edmonds, Brian C. O’Neill, Shinichiro Fujimori, Nico Bauer, Katherine Calvin, Rob Dellink, Oliver Fricko, Wolfgang Lutz, 39 Doran, 2003. Alexander Popp, Jesus Crespo Cuaresma, Samir KC, Marian Leimbach, Leiwen Jiang, Tom Kram, Shilpa Rao, Johannes 40 For example, recent RAND research explored this lens; see Emmerling, Kristie Ebi, Tomoko Hasegawa, Petr Havlik, Florian Watts et al., 2017. Humpenöder, Lara Aleluia Da Silva, Steve Smith, Elke Steh- 41 One could further adjust the balance of power analysis to fest, Valentina Bosetti, Jiyong Eom, David Gernaat, Toshihiko include the aggregated power of alliances rather than individual Masui, Joeri Rogelj, Jessica Strefler, Laurent Drouet, Volker Krey, countries, which could shift when a power transition occurs. See, Gunnar Luderer, Mathijs Harmsen, Kiyoshi Takahashi, Lavinia again, Watts et al., 2017. Baumstark, Jonathan C. Doelman, Mikiko Kainuma, Zbig- niew Klimont, Giacomo Marangoni, Hermann Lotze-Campen, 42 For example, the robust decision making methodology Michael Obersteiner, Andrzej Tabeau, and Massimo Tavoni, examines a wide range of plausible futures to help create adaptive “The Shared Socioeconomic Pathways and Their Energy, Land and robust strategies. An early overview can be found in Robert Use, and Greenhouse Gas Emissions Implications: An Over- J. Lempert, Steven W. Popper, and Steven C. Bankes, Shaping the view,” Global Environmental Change, Vol. 42, 2017, pp. 153–168. Next One Hundred Years: New Methods for Quantitative, Long-

26 51 Zeke Hausfather, “Explainer: How ‘Shared Socioeconomic 53 Marshall Burke, Solomon M. Hsiang, and Edward Miguel, Pathways’ Explore Future Climate Change,” CarbonBrief, April 19, “Global Non-Linear Effect of Temperature on Economic Pro- 2018. duction,” Nature, Vol. 527, 2015, pp. 235–239. Burke, Hsiang, and Miguel (2015) take as inputs the population assumptions associ- 52 Brian C. O’Neill, Elmar Kriegler, Kristie L. Ebi, Eric ated with SSP3 and SSP5 and the temperature changes associated Kemp-Benedict, Keywan Riahi, Dale S. Rothman, Bas J. van with RCP8.5. To avoid a sudden jump in GDP when switching Ruijven, Detlef P. van Vuuren, Joern Birkmann, Kasper Kok, Marc between data sources, we do not use the GDP levels from the Levy, and William Solecki, “The Roads Ahead: Narratives for Burke, Hsiang, and Miguel methodology but rather replace the Shared Socioeconomic Pathways Describing World Futures in the baseline annual growth rates with growth rates that match those 21st Century,” Global Environmental Change, Vol. 42, 2017, pp. found when replicating the Burke, Hsiang, and Miguel methodol- 169–180. ogy.

References Edelman, Eric S., Understanding America’s Contested Primacy, Washington, D.C.: Center for Strategic and Budgetary Allison, Graham, Destined for War: Can America and China Assessments, 2010. Escape Thucydides’s Trap? Boston, Mass.: Houghton Mifflin Harcourt, 2017. Federation of American Scientists Nuclear Notebook, “Nuclear Arsenals of the World,” database, accessed on December 20, Allison, Graham, and Robert D. Blackwill, “Will China Ever Be 2019. As of December 20, 2019: No. 1?” YaleGlobal Online newsletter, February 20, 2013. As of https://thebulletin.org/nuclear-notebook-multimedia/ October 4, 2019: https://yaleglobal.yale.edu/content/will-china-ever-be-no-1 Fukuyama, Francis, The End of History and the Last Man, New York: Free Press, 1992. “Annual Estimates of the Resident Population by Single Year of Age and Sex for the United States: April 1, 2010 to July 1, 2017,” Han, Alice Siqi, “China’s Economic Black Box,” Foreign Policy, U.S. Census Bureau, Population Division, American FactFinder March 17, 2019. database, release date: June 2018. Harshaw, Tobin, “Emperor Xi’s China Is Done Biding Its Time: A Baldwin, David A., Power and International Relations: A Q&A with Former Australian Prime Minister Kevin Rudd, Who Conceptual Approach, Princeton, N.J.: Princeton University Is Going to School to Study the Chinese President,” Bloomberg Press, 2016. News, March 3, 2018. Beckley, Michael, “The Power of Nations: Measuring What Hausfather, Zeke, “Explainer: How ‘Shared Socioeconomic Matters,” International Security, Vol. 43, No. 2, 2018, pp. 7–44. Pathways’ Explore Future Climate Change,” CarbonBrief, April 19, 2018. Brands, Hal, and Charles Edel, The Lessons of Tragedy: Statecraft and World Order, New Haven, Conn.: Yale University Press, 2019. Hebron, Lui, and Patrick James, “Great Powers, Cycles of Relative Capability and Crises in World Politics,” International Burke, Marshall, Solomon M. Hsiang, and Edward Miguel, Interactions, Vol. 23, No. 2, 1997, pp. 145–173. “Global Non-Linear Effect of Temperature on Economic Production,” Nature, Vol. 527, 2015, pp. 235–239. Heim, Jacob L., “Tapping the Power of Structural Change: Power Cycle Theory as an Instrument in the Toolbox of National Carter, Ash, “Remarks Previewing the FY 2017 Defense Budget,” Security Decision-Making,” SAIS Review, Vol. 29, No. 2, 2009, February 2, 2016. As of October 4, 2019: pp. 113–127. https://dod.defense.gov/News/Speeches/Speech-View/ Article/648466/remarks-previewing-the-fy-2017-defense-budget/ Houweling, Henk W., and Jan G. Siccama, “Power Transitions and Critical Points as Predictors of Great Power War: Toward a Cohen, Eliot A., The Big Stick: The Limits of Soft Power and the Synthesis,” Journal of Conflict Resolution, Vol. 35, No. 4, 1991, Necessity of Military Force, New York: Basic Books, 2016. pp. 642–658. Davis, Paul K., Analysis to Inform Defense Planning Despite Huang Shuofeng, Zonghe Guoli Lun [On Comprehensive National Austerity, Santa Monica, Calif.: RAND Corporation, RR-482- Power], Beijing: Zhongguo Shehui Kexue Chubanshe, 1992. OSD, 2014. As of October 7, 2019: https://www.rand.org/pubs/research_reports/RR482.html Hülser, Sarah, Power Cycle Theory Reconsidered: Is China Going to Destabilize the International Order? Berlin: NFG Research Doran, Charles F., Systems in Crisis: New Imperatives of Group, Freie Universität Berlin, Working Paper Series, No. 6, July High Politics at Century’s End, Cambridge, United Kingdom: 2013. Cambridge University Press, 1991. IISS—See International Institute of Strategic Studies. ———, “E c onom ic s , Ph i los ophy of H i s tor y, a nd t he ‘ Si ng le Dynamic’ of Power Cycle Theory: Expectations, Competition, International Centre for Tax and Development, “Government and Statecraft,” International Political Science Review, Vol. 24, Revenue Dataset,” database, 2019. As of December 20, 2019: No. 1, 2003, pp. 13–49. https://www.ictd.ac/dataset/grd/ Doran, Charles F., and Wes Parsons, “War and the Cycle of International Institute of Strategic Studies, The Military Balance, Relative Power,” American Political Science Review, Vol. 74, No. 4, Vol. 118, 2018. 1980, pp. 947–965. International Monetary Fund, “World Economic Outlook Database,” database, April 2018.

27 Jones, Bruce, Still Ours to Lead: America, Rising Powers, and O’Neill, Brian C., Elmar Kriegler, Kristie L. Ebi, Eric Kemp- the Tension Between Rivalry and Restraint, Washington, D.C.: Benedict, Keywan Riahi, Dale S. Rothman, Bas J. van Ruijven, Brookings, 2014. Detlef P. van Vuuren, Joern Birkmann, Kasper Kok, Marc Levy, and William Solecki, “The Roads Ahead: Narratives for Shared Kawashima, Shin, “The Development of the Debate over ‘Hiding Socioeconomic Pathways Describing World Futures in the 21st One’s Talents and Biding One’s Time,’” Asia-Pacific Review, Century,” Global Environmental Change, Vol. 42, 2017, Vol. 18, No. 2, 2011, pp. 14–36. pp. 169–180. Kennedy, Paul, The Rise and Fall of the Great Powers: Economic Pettis, Michael, Avoiding the Fall: China’s Economic Change and Military Conflict from 1500 to 2000, New York: Restructuring, Washington, D.C.: Carnegie Endowment for Vintage Books, 1987. International Peace, 2013. Koninklijk Nederlands Meteorologisch Instituut, “KNMI ———, “W hat Is GDP i n C h i na? ” China Financial Markets, blog, Climate Explorer,” database, undated. As of October 7, 2019: Carnegie Endowment for International Peace, January 16, 2019. http://climexp.knmi.nl/plot_atlas_form.py Pillsbury, Michael, China Debates the Future Security Lampton, David M., The Three Faces of Chinese Power: Might, Environment, Washington, D.C.: National Defense University Money, and Minds, Berkeley and Los Angeles, Calif.: University Press, January 2000. of California Press, 2008. Rawski, Thomas G., “What Is Happening to China’s GDP Lanyi, Anthony, and Kimberly Brickell, eds., Institutional, Statistics?” China Economic Review, Vol. 12, No. 4, 2001, Economic, and Organizational Basis of Military Capability, pp. 347–354. College Park, Md.: University of Maryland Center for Institutional Reform and the Informal Sector, 1999. Riahi, Keywan, Detlef P. van Vuuren, Elmar Kriegler, Jae Edmonds, Brian C. O’Neill, Shinichiro Fujimori, Nico Bauer, Lempert, Robert J., Steven W. Popper, and Steven C. Bankes, Katherine Calvin, Rob Dellink, Oliver Fricko, Wolfgang Lutz, Shaping the Next One Hundred Years: New Methods for Alexander Popp, Jesus Crespo Cuaresma, Samir KC, Marian Quantitative, Long-Term Policy Analysis, Santa Monica, Calif.: Leimbach, Leiwen Jiang, Tom Kram, Shilpa Rao, Johannes RAND Corporation, MR-1626-RPC, 2003. As of October 7, 2019: Emmerling, Kristie Ebi, Tomoko Hasegawa, Petr Havlik, Florian https://www.rand.org/pubs/monograph_reports/MR1626.html Humpenöder, Lara Aleluia Da Silva, Steve Smith, Elke Stehfest, Lempert, Robert J., Drake Warren, Ryan Henry, Robert W. Valentina Bosetti, Jiyong Eom, David Gernaat, Toshihiko Button, Jonathan Klenk, and Kate Giglio, Defense Resource Masui, Joeri Rogelj, Jessica Strefler, Laurent Drouet, Volker Planning Under Uncertainty: An Application of Robust Decision Krey, Gunnar Luderer, Mathijs Harmsen, Kiyoshi Takahashi, Making to Munitions Mix Planning, Santa Monica, Calif.: RAND Lavinia Baumstark, Jonathan C. Doelman, Mikiko Kainuma, Corporation, RR-1112-OSD, 2016. As of October 7, 2019: Zbigniew Klimont, Giacomo Marangoni, Hermann Lotze- https://www.rand.org/pubs/research_reports/RR1112.html Campen, Michael Obersteiner, Andrzej Tabeau, and Massimo Tavoni, “The Shared Socioeconomic Pathways and Their Energy, “Less Biding and Hiding,” The Economist, December 4, 2010. Land Use, and Greenhouse Gas Emissions Implications: An Overview,” Global Environmental Change, Vol. 42, 2017, Maddison Project Database, version 2018. Jutta Bolt, Robert pp. 153–168. Inklaar, Herman de Jong, and Jan Luiten van Zanden, “Rebasing ‘Maddison’: New Income Comparisons and the Shape of Long- Rosecrance, Richard N., and Steven E. Miller, eds., The Next Run Economic Development,” Maddison Project Working Great War? The Roots of World War I and the Risk of U.S.-China Paper 10, 2018. As of December 27, 2019: Conflict, Cambridge, Mass.: MIT Press, 2014. https://www.rug.nl/ggdc/historicaldevelopment/maddison/ releases/maddison-project-database-2018 Ruan Zongze, “Chinese Scholar Ruan Zongze Comments on Whether China Has Stopped ‘Hide and Bide’ Strategy,” Huanqiu Modelski, George, “The Long Cycle of Global Politics and the Shibao Online in Chinese, May 11, 2018. Nation-State,” Comparative Studies in Society and History, Vol. 20, No. 2, 1978, pp. 214–235. Sarkees, Meredith Reid, and Frank Whelon Wayman, “Resort to War: 1816–2007,” version 4.0, database, Washington, D.C.: CQ Moyer, Jonathan, and Alanna Markle, Relative National Press, 2010. Power Codebook, Denver, Colo.: Frederick S. Pardee Center for International Futures, Josef Korbel School of International Singer, J. David, Stuart Bremer, and John Stuckey, “Capability Studies, University of Denver, 2017. Distribution, Uncertainty, and Major Power War, 1820–1965,” version 5, data set, in Bruce Russett, ed., Peace, War, and National Intelligence Council, “Global Trends,” webpage, Numbers, Beverly Hills, Calif.: Sage, 1972. As of October 4, 2019: Office of the Director of National Intelligence, undated. As of http://www.correlatesofwar.org/data-sets/national-material- October 4, 2019: capabilities https://www.dni.gov/index.php?option=com_ content&view=article&id=398&Itemid=776 Southerland, Matthew, China’s Island Building in the South China Sea: Damage to the Marine Environment, Implications, and NIC—See National Intelligence Council. International Law, Washington, D.C.: U.S.-China Economic and Security Review Commission, April 12, 2016. Norris, Guy, “Key Milestones Loom for New Civil and Military Engines,” Aviation Week and Space Technology, December 22, Stockholm International Peace Research Institute, “SIPRI 2017. Military Expenditure Database,” database, accessed on October 20, 2019. As of October 20, 2019: O’Brian, Sean P., “Crisis Early Warning and Decision Support: https://www.sipri.org/databases/milex Contemporary Approaches and Thoughts on Future Research,” International Studies Review, Vol. 12, No. 1, 2010, pp. 87–104. Tate, Andrew, “Aerial Refuelling Capability of China’s J-20 Fighter Confirmed,” Jane’s Defence Weekly, November 19, 2018.

28 Tellis, Ashley J., Janice Bially, Christopher Layne, and Melissa U.S. Department of Defense, Summary of the 2018 National McPherson, Measuring National Power in the Postindustrial Age, Defense Strategy of the United States of America: Sharpening the Santa Monica, Calif.: RAND Corporation, MR-1110-A, 2000. As of American Military's Competitive Edge, Washington, D.C., 2018. October 4, 2019: https://www.rand.org/pubs/monograph_reports/MR1110.html van der Heijden, Kees, Scenarios: The Art of Strategic Conversation, 2nd ed., Hoboken, N.J.: Wiley, 2005. Tessman, Brock F., “The Evolution of Chinese Foreign Policy: New Incentives with Slowing Growth,” Asian Security, Vol. 5, No. 3, Waltz, Kenneth, Man, the State, and War, New York: Columbia 2009, pp. 296–318. University Press, 1959. Tessman, Brock F., and Steve Chan, “Power Cycles, Risk Watts, Stephen, Bryan Frederick, Jennifer Kavanagh, Angela Propensity, and Great-Power Deterrence,” Journal of Conflict O’Mahony, Thomas S. Szayna, Matthew Lane, Alexander Resolution, Vol. 48, No. 2, 2004, pp. 131–153. Stephenson, and Colin P. Clark, A More Peaceful World? Regional Conflict Trends and U.S. Defense Planning, Santa Monica, Calif.: UN—See United Nations. RAND Corporation, RR-1177-A, 2017. As of October 4, 2019: https://www.rand.org/pubs/research_reports/RR1177.html United Nations Educational, Scientific and Cultural Organization, “UNESCO Institute for Statistics,” database, accessed on Witmer, Frank D. W., Andrew M. Linke, John O’Loughlin, December 20, 2019. As of December 20, 2019: Andrew Gettelman, and Arlene Laing, “Subnational Violent http://data.uis.unesco.org/ Conflict Forecasts for Sub-Saharan Africa, 2015–65, Using Climate-Sensitive Models,” Journal of Peace Research, Vol. 54, United Nations Population Division, “World Population Prospects No. 2, 2017, pp. 175–192. 2019,” database, 2019. World Trade Organization, “Statistics: Trade and Tariff Data,” United Nations University International Human Dimensions database, accessed December 2019. As of March 3, 2020: Programme and United Nations Environment Programme, https://www.wto.org/english/res_e/statis_e/statis_e.htm Inclusive Wealth Report 2014: Measuring Progress Toward Sustainability, Cambridge, United Kingdom: Cambridge Yan Xuetong, “The Rise of China and Its Power Status,” Chinese University Press, 2014. Journal of International Politics, Vol. 1, 2006, pp. 5–33. Zakaria, Fareed, The Post-American World, New York: W. W. Norton & Company, 2008.

About the Authors Acknowledgments Jacob L. Heim specializes in strategic assessment and The authors would like to thank Cedric Kenney, Sonni defense analysis with a background in international rela- Efron, Andrew Parasiliti, Robin Meili, Alan Vick, Rob tions and mathematics. In over ten years at RAND, he has Lempert, Mike Spirtas, and King Mallory for their assis- conducted research on a variety of topics, including the mil- tance, support, and insights throughout the development of itary balance in the Western Pacific and China’s evolving this report. The authors also thank Timothy J. Smith, Bryan nuclear deterrent. Heim served as a strategic analyst in the Frederick, and Greg Treverton for reviews that improved office of the Deputy Assistant Secretary of Defense (DASD) the quality of this work. for Strategy and Force Development, where he advised the DASD on U.S. defense strategy and the evolution of military capabilities. In 2017, the Secretary of Defense awarded him the Defense Medal for Exceptional Public Service. Before RAND, Heim was a senior operations research analyst with the MITRE Corporation. Benjamin M. Miller is an economist at the RAND Corporation and a professor at the Pardee RAND Graduate School. He conducts economic assessments that improve public systems, including infrastructure systems, regulatory systems, and financial systems. Other recent works examine approaches for deterring attacks against the electric power grid in the modern age and the impacts of natural disasters. Previously, Miller spent several summers as a statistician at the U.S. Census Bureau. Miller holds a Ph.D. in economics from the University of California, San Diego, and a B.S. in economics from Purdue University.

29 About This Report About the RAND Center for Global There is growing concern that U.S. power has been Risk and Security declining relative to the growing power of Russia and The Center for Global Risk and Security works across China. This concern renews long-standing questions about the RAND Corporation to develop multidisciplinary how we should measure national power, which nations research and policy analysis dealing with systemic risks to have the most power, which states are gaining and losing global security. The center draws on RAND’s expertise to power, and when such shifts in relative or perceived power complement and expand RAND research in many fields, might portend conflict. In this report, the authors explore including security, economics, health, and technology. A these questions, proposing a new tool for measuring the board of distinguished business leaders, philanthropists, interstate balance of power and identifying periods during and former policymakers advises and supports the center which shifts in the balance of power could potentially activities, which are increasingly focused on global security portend conflict. The research was conducted within the trends and the impact of disruptive technologies on risk and RAND Center for Global Risk and Security. security. For more information about the RAND Center for Global Risk and Security, visit www.rand.org/international/ Funding cgrs. Funding for this report was provided by gifts from RAND supporters and income from operations.

30 C O R P O R A T I O N

There is growing concern that U.S. power has been declining relative to the growing power of Russia and China. This concern renews longstanding questions about how we should measure national power, which nations have the most power, which states are gaining and losing power, and when such shifts in relative or perceived power might portend conflict. The authors explore these questions, illustrating a quantitative, scenario-based approach for policymakers who are interested in measuring the interstate balance of power, assessing the impact of shocks on the balance of power, and identifying periods during which shifts in the balance of power could potentially portend conflict between major powers. The methodology defines how different climate change scenarios, population projections, or economic growth forecasts lead to different balances of global power, then uses power cycle theory to map those changes in the distribution of global power to changes in the risk of conflict between major global powers. The authors demonstrate the potential of this methodology using three illustrative examples. They find that global power can be “sticky,” meaning it takes drastic assumptions about the future to create meaningful changes in the global balance of power. Further, because their model treats global power as a relative concept, the authors find that the types of shocks that affect the risk of conflict are the types that create relative “winners” and “losers.”

$20.00

ISBN-10 1-9774-0462-6 ISBN-13 978-1-9774-0462-6 52000

www.rand.org 9 781977 404626

RR-2989-RC