Beckley Power
Beckley Power
Beckley Power
Michael Beckley is Assistant Professor of Political Science at Tufts University and an associate in the Inter-
national Security Program at the Belfer Center for Science and International Affairs at the John F. Kennedy
School of Government at Harvard University. Parts of this article draw on chapter 2 of his book, Unrivaled:
Why America Will Remain the World’s Sole Superpower (Ithaca, N.Y.: Cornell University Press,
2018).
Previous versions of this article were presented at the Belfer Center for Science and International
Affairs at Harvard’s Kennedy School, the Dickey Center for International Understanding at
Dartmouth College, the ISSS-ISAC Joint Annual Conference, and the annual meetings of the
Southern Political Science Association and the American Political Science Association. The author
thanks Stephen Brooks, Jonathan Caverley, Tyson Chatagnier, Michael Desch, Jeffrey Friedman,
Bryan Greenhill, Jennifer Lind, Jonathan Markowitz, Daryl Press, Richard Rosecrance, Benjamin
Valentino, Stephen Walt, William Wohlforth, and the anonymous reviewers for excellent feedback.
1. David A. Baldwin, Power and International Relations: A Conceptual Approach (Princeton, N.J.:
Princeton University Press, 2016), p. 1; Martha Finnemore and Judith Goldstein, eds., Back to Ba-
sics: State Power in a Contemporary World (New York: Oxford University Press, 2013), pp. 4–5; Rob-
ert O. Keohane, “Big Questions in the Study of World Politics,” in Christian Reus-Smit and
Duncan Snidal, eds., The Oxford Handbook of International Relations (Oxford: Oxford University
Press, 2008), p. 709; and Harold D. Lasswell and Abraham Kaplan, Power and Society: A Framework
for Political Inquiry (New York: McGraw-Hill, 1950), p. 75.
2. Bertrand Russell, Power: A New Social Analysis (London: George Allen and Unwin, 1938),
pp. 10–12; and John J. Mearsheimer, The Tragedy of Great Power Politics (New York: W.W. Norton,
2014), p. 12.
3. See, for example, Alexandre Debs and Nuno P. Monteiro, “Known Unknowns: Power Shifts,
Uncertainty, and War,” International Organization, Vol. 68, No. 1 (January 2014), pp. 1–32, doi:
10.1017/S0020818313000192; A.F.K. Organski and Jacek Kugler, The War Ledger (Chicago: Univer-
sity of Chicago Press, 1980); Robert Powell, “War as a Commitment Problem,” International Organi-
zation, Vol. 60, No. 1 (January 2006), pp. 169–203, doi:10.1017/S0020818306060061; and William C.
Wohlforth, “The Stability of a Unipolar World,” International Security, Vol. 24, No. 1 (Summer
1999), pp. 5–41, doi:10.1162/016228899560031.
4. See Glenn H. Snyder, Alliance Politics (Ithaca, N.Y.: Cornell University Press, 2007); Stephen M.
Walt, The Origins of Alliances (Ithaca, N.Y.: Cornell University Press, 1987); and Kenneth N. Waltz,
Theory of International Politics (Reading, Mass.: Addison-Wesley, 1979).
5. See, for example, Kenneth W. Abbott and Duncan Snidal, “Why States Act through Formal In-
ternational Organizations,” Journal of Conºict Resolution, Vol. 42, No. 1 (February 1998), pp. 3–32,
doi:10.1177/0022002798042001001; James D. Fearon, “Bargaining, Enforcement, and International
Cooperation,” International Organization, Vol. 52, No. 2 (Spring 1998), pp. 269–305, doi:10.1162/
002081898753162820; and Lloyd Gruber, Ruling the World: Power Politics and the Rise of Supranational
Institutions (Princeton, N.J.: Princeton University Press, 2000).
International Security, Vol. 43, No. 2 (Fall 2018), pp. 7–44, https://doi.org/10.1162/ISEC_a_00328
© 2018 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology.
7
International Security 43:2 8
6. See, for example, Michael C. Desch, “War and Strong States, Peace and Weak States?” Interna-
tional Organization, Vol. 50, No. 2 (Spring 1996), pp. 237–268, doi:10.1017/S0020818300028551.
7. See, for example, Joanne Gowa and Edward D. Mansªeld, “Power Politics and International
Trade,” American Political Science Review, Vol. 87, No. 2 (June 1993), pp. 408–420, doi:10.2307/
2939050; and Albert O. Hirschman, National Power and the Structure of Foreign Trade (Berkeley: Uni-
versity of California Press, 1945).
8. See, for example, Nuno P. Monteiro and Alexandre Debs, “The Strategic Logic of Nuclear Prolif-
eration,” International Security, Vol. 39, No. 2 (Fall 2014), pp. 7–51, doi:10.1162/ISEC_a_00177.
9. See, for example, Seva Gunitsky, Aftershocks: Great Powers and Domestic Reforms in the Twentieth
Century (Princeton, N.J.: Princeton University Press, 2017).
10. Ashley J. Tellis et al., Measuring Power in the Postindustrial Age (Santa Monica, Calif.: RAND
Corporation, 2000), pp. 1–11.
11. Joseph S. Nye Jr., “The Changing Nature of World Power,” Political Science Quarterly, Vol. 105,
No. 2 (Summer 1990), p. 177, doi:10.2307/2151022.
12. Joseph S. Nye Jr., The Future of Power (New York: PublicAffairs, 2011), p. 3.
13. Nye, “The Changing Nature of World Power,” p. 178.
14. Mearsheimer, The Tragedy of Great Power Politics, p. 60; Nye, The Future of Power, p. 8; and Wil-
liam C. Wohlforth, The Elusive Balance: Power and Perceptions during the Cold War (Ithaca, N.Y.: Cor-
nell University Press, 1993), pp. 4, 10.
15. On the link between resources and power, see Paul M. Kennedy, The Rise and Fall of the Great
Powers: Economic Change and Military Conºict from 1500 to 2000 (New York: Random House, 1987);
Mearsheimer, The Tragedy of Great Power Politics, pp. 55–138; Nye, The Future of Power, pp. 25–81;
and Tellis et al., Measuring Power in the Postindustrial Age, pp. 1–33.
The Power of Nations 9
16. David J. Singer, Stuart Bremer, and John Stuckey, “Capability Distribution, Uncertainty, and
Major Power War, 1820–1965,” in Bruce M. Russett, ed., Peace, War, and Numbers (Beverly Hills,
Calif.: Sage, 1972), pp. 19–48.
17. See, for example, Klaus Knorr, The War Potential of Nations (Princeton, N.J.: Princeton Univer-
sity Press, 1956), p. 231; Mearsheimer, The Tragedy of Great Power Politics, pp. 61–63; and Michael
Beckley, “China’s Century? Why America’s Edge Will Endure,” International Security, Vol. 36, No. 3
(Winter 2011/12), pp. 58–59, doi:10.1162/ISEC_a_00066.
18. Mearsheimer, The Tragedy of Great Power Politics, pp. 55, 60–68; and Tellis et al., Measuring Power
in the Postindustrial Age, p. 26.
International Security 43:2 10
ing the United States as the world’s leading power.23 The U.S. National
Intelligence Council has issued multiple reports advising the president to pre-
pare the country for the rise of China and the reemergence of multipolarity by
2030.24 China’s president, Xi Jinping, has invoked the Thucydides Trap in pub-
lic speeches; and the author of Death by China has become President Donald
Trump’s top adviser on trade.
The hype about China’s rise, however, has been based largely on gross indi-
cators that ignore costs. When costs are accounted for, it becomes clear that the
United States’ economic and military lead over China is much larger than typi-
cally assumed—and the trends are mostly in America’s favor.
This article proceeds in eight sections. The ªrst section explains why schol-
ars typically measure power in terms of resources. The second explains why
scholars should measure resources in net rather than gross terms. The third
discusses speciªc indicators of gross and net resources. Sections four through
seven test the validity of these indicators with case studies, large-n statistics,
and replication analyses. The ªnal section discusses implications of these anal-
yses for scholarship and policy.
Power can be measured in two main ways.25 The most common approach, and
the one I focus on in this article, measures power by tallying the wealth
and military assets of each country. The logic of this “power as resources” ap-
proach is straightforward.26 Wealth enables a country to buy inºuence through
aid, loans, investment, and bribes and to cultivate soft power (the ability of a
country to attract and co-opt others) by, among other things, funding global
propaganda campaigns, building huge skyscrapers, and hosting international
expositions and sporting events.27 Military resources (e.g., troops and weap-
ons), on the other hand, enable a country to destroy enemies; attract allies; and
extract concessions and kickbacks from weaker countries by issuing threats of
violence and offers of protection.
23. Pew Research Center, “Global Indicators Database” (Washington, D.C.: Pew Research Center,
2017), http://www.pewglobal.org/database/.
24. National Intelligence Council (NIC), Global Trends 2030: Alternative Worlds (Washington, D.C.:
U.S. Government Printing Ofªce, 2012).
25. Mearsheimer, The Tragedy of Great Power Politics, pp. 57–60; Nye, The Future of Power, pp. 3–25;
and Tellis et al., Measuring Power in the Postindustrial Age, pp. 12–24.
26. Kennedy, The Rise and Fall of the Great Powers; Mearsheimer, The Tragedy of Great Power Politics,
pp. 55–138; Nye, The Future of Power, pp. 12–81; and Tellis et al., Measuring Power in the
Postindustrial Age, pp. 53–132.
27. On soft power, see Joseph S. Nye Jr., Soft Power: The Means to Success in World Politics (New
York: PublicAffairs, 2004).
International Security 43:2 12
28. For one of many examples, see David A. Baldwin, Paradoxes of Power (Cambridge, Mass.: Basil
Blackwell, 1989).
29. In Robert A. Dahl’s famous terms, power is A’s ability to get B to do something that B other-
wise would not. See Dahl, “The Concept of Power,” Behavioral Science, Vol. 2, No. 3 (July 1957),
pp. 201–215, doi:10.1002/bs.3830020303.
30. Peter Bachrach and Morton S. Baratz, “Two Faces of Power,” American Political Science Review,
Vol. 56, No. 4 (December 1962), pp. 947–952, doi:10.2307/1952796; and Robert O. Keohane and Jo-
seph S. Nye Jr., Power and Interdependence (New York.: HarperCollins, 1989), pp. 32–33.
31. Michael Barnett and Raymond Duvall, “Power in International Politics,” International Organi-
zation, Vol. 59, No. 1 (January 2005), pp. 39–75, doi:10.1017/S0020818305050010; Stefano Guzzini,
“Structural Power: The Limits of Neorealist Power Analysis,” International Organization, Vol. 47,
No. 3 (Summer 1993), pp. 443–478, doi:10.1017/S0020818300028022; and Steven Lukes, Power:
A Radical View (Basingstoke, U.K.: Palgrave Macmillan, 2005).
32. On the importance of specifying scope and domain when measuring power, see Baldwin,
Power and International Relations.
33. Andrew J.R. Mack, “Why Big Nations Lose Small Wars: The Politics of Asymmetric Conºict,”
World Politics, Vol. 27, No. 2 (January 1975), pp. 175–200, doi:10.2307/2009880; and Ivan Arreguín-
Toft, “How the Weak Win Wars: A Theory of Asymmetric Conºict,” International Security, Vol. 26,
No. 1 (Summer 2001), pp. 93–128, doi:10.1162/016228801753212868.
34. Peter Morris, Power: A Philosophical Analysis (Manchester, U.K.: Manchester University Press,
2002), pp. 25–28.
The Power of Nations 13
35. In a recent study, Robert J. Carroll and Brenton Kenkel try to get around this problem by using
machine learning techniques, CINC data, and militarized interstate dispute (MID) data to develop
a proxy for power called the Dispute Outcome Expectations (DOE) score, which is directly inter-
pretable as the probability of victory in a militarized interstate dispute. Unfortunately, however,
DOE scores cannot be used to study dispute and war outcomes, because the scores are based on
war and dispute outcome data and would thus be endogenous in statistical models. Moreover,
DOE scores are based on CINC, which I show to be a severely ºawed measure of power resources.
See Carroll and Kenkel, “Prediction, Proxies, and Power,” Florida State University and Vanderbilt
University, 2016, http://doe-scores.com/doe.pdf.
36. Mearsheimer, The Tragedy of Great Power Politics, p. 60; Nye, The Future of Power, p. 8; and
Wohlforth, The Elusive Balance, pp. 4, 10.
37. Mearsheimer, The Tragedy of Great Power Politics, p. 60.
International Security 43:2 14
generalizable, one that faithfully reºects the past but also can be applied to the
present and projected into the future.
38. Kingsley Davis, “The Demographic Foundations of National Power,” in Monroe Berger, Theo-
dore Abel, and Charles H. Page, eds., Freedom and Control in Modern Societies (New York: Van
Nostrand, 1954), pp. 206–242; and Katherine Organski and A.F.K. Organski, Population and World
Power (New York: Alfred A. Knopf, 1961).
39. Klaus Knorr calls these resources the “disposable surplus.” See Knorr, The War Potential of Na-
tions, p. 231.
40. Stephen Biddle, Military Power: Explaining Victory and Defeat in Modern Battle (Princeton, N.J.:
Princeton University Press, 2004).
The Power of Nations 15
Welfare costs are subsistence costs; they are the expenses a nation pays to
keep its people from dying in the streets and include outlays on basic items
such as food, health care, social security, and education.
Security costs are the price a government pays to police and protect its citi-
zens. The logic of deducting assets tied up in domestic law enforcement and
homeland security is simple: police and military units that are bogged down
chasing criminals, quelling rebellions, or defending borders against foreign in-
vasions cannot project power abroad or create wealth at home. Measuring se-
curity costs thus accounts for the fact that two nations with identical sets of
gross resources may, nevertheless, wield vastly different levels of power if one
country is surrounded by enemies and wracked by internal strife whereas the
other is stable and surrounded by allies.
Needless to say, production, welfare, and security costs add up. In fact,
for most of human history, they consumed nearly all of every country’s re-
sources.41 Even today, they tie down large amounts of the world’s economic
and military assets. To assess the balance of power, therefore, analysts must
deduct these costs by using net indicators.
Power Indicators
In this section, I show that the most commonly used indicators of economic
and military resources ignore production, welfare, and security costs. After
highlighting this problem, I discuss how scholars can address it by using
net indicators.
gross indicators
Most scholars and analysts measure power using gross indicators, including
various measures of economic input (e.g., on research and development
[R&D] spending, capital investment, and energy consumption); economic out-
put (e.g., GDP, manufacturing, and industrial output); trade and ªnancial
ºows; and “bean counts” of military spending, platforms, and personnel. Ac-
cording to a review of the literature, scholars and government analysts pro-
duced at least sixty-nine power measurement frameworks from 1936 to 2010,
and forty-two of these frameworks were composed solely of some combina-
tion of the gross indicators listed above.42
41. Ian Morris, Why the West Rules—For Now: The Patterns of History, and What They Reveal about the
Future (New York: Farrar, Straus and Giroux, 2010).
42. The remaining twenty-seven formulas combined gross material indicators with nonmaterial
factors (e.g., morale, prestige, and diplomatic skill), per capita material indicators, or both. See
Karl Hermann Höhn, “Geopolitics and the Measurement of National Power,” Ph.D. dissertation,
University of Hamburg, 2011.
International Security 43:2 16
The most popular indicator is GDP, which records the value of all goods and
services produced within a country over a ªxed period of time. GDP has been
described as “the leading indicator” and “the Zeus of the statistical pantheon,”
because governments, organizations, and scholars around the world use it to
gauge states’ raw capabilities.43 Although GDP is technically an economic in-
dicator, proponents argue that it captures both economic and military capacity,
because states can easily convert economic resources into military might. In
short, GDP is considered to be fungible; it can be turned into “any mix of mili-
tary, economic, and political” resources, just as a person can use cash to buy
many forms of inºuence.44
Despite the widespread use of GDP, however, few people know what it actu-
ally measures or recognize that it does not deduct costs.
To begin, GDP counts production costs (inputs and externalities) as output.
Spending money always increases GDP, even if the funds are wasted on boon-
doggles; in fact, the most common method of calculating GDP is called the
“expenditure method” and involves simply adding up all of the spending
done by the government, consumers, and businesses in a country in a given
time period.45 Thus, hiring workers always increases GDP, even if they spend
all day getting drunk in the break room. Boosting production always increases
GDP, even if the goods rot on the shelf and tons of toxic waste are released in
the process. In fact, a country can increase its GDP by dumping toxic waste
into the streets and then spending billions of dollars to clean it up.
GDP also does not deduct welfare costs. Money spent feeding people is
counted the same as proªts earned selling supercomputers on world markets.
Consequently, populous countries generate considerable economic activity
simply by existing. Even a nation caught in a Malthusian hell, in which all out-
put is immediately devoured, will post a large GDP if it has a big population.
Finally, GDP counts security spending as economic output. GDP does not
distinguish between guns and butter. It counts a $100 million gulag the same
as a $100 million innovation center. Hence, GDP fails to account fully for the
economic costs of domestic instability and international conºict. In fact, GDP
usually rises when a country mobilizes for war. To be sure, military invest-
ments can sometimes yield economic dividends. For example, the internet and
43. Zachary Karabell, The Leading Indicators: A Short History of the Numbers That Rule Our World
(New York: Simon and Schuster, 2014), p. 158.
44. Emilio Casetti, “Power Shifts and Economic Development: When Will China Overtake
the USA?” Journal of Peace Research, Vol. 40, No. 6 (November 2003), p. 663, doi:10.1177/
00223433030406003.
45. Diane Coyle, GDP: A Brief but Affectionate History (Princeton, N.J.: Princeton University Press,
2014).
The Power of Nations 17
the Global Positioning System began as U.S. military research projects. In gen-
eral, however, resources devoted to policing and protection drain wealth
rather than create it.46
Besides GDP, the other most commonly used indicator is “war potential,”
which combines measures of gross economic output and gross military re-
sources.47 The logic of this approach is that power ultimately depends on the
ability to win major wars, and doing that requires a big army backed by a
hefty military budget and substantial industrial might.
Governments are fond of this approach. For example, the U.S. National
Intelligence Council, a body that advises the president, gauges global power
trends with an index that combines military spending, GDP, population, and
R&D spending.48 Academics, too, typically measure the power of nations in
terms of war potential. As noted, more than 1,000 peer-reviewed studies have
used CINC, which combines data on military spending, troops, population,
urban population, iron and steel production, and energy consumption.49
Measures of war potential, however, suffer from the same problem as GDP:
they are gross measures that do not deduct production, welfare, or security
costs. They count military units the same, regardless of their level of skill or
technology, the welfare costs of supporting those units,50 or whether they are
projecting power abroad or imposing order at home. They also treat military
spending and other inputs, such as energy consumption or R&D spending,
as if they were outputs, so a country could substantially increase its CINC
score by making enemies and then raising a huge, oil-guzzling army to at-
tack them.
Ultimately, all gross indicators are one-dimensional; they measure only the
size of a country’s resources, not how efªciently a country uses them.
net indicators
How can scholars address the shortcomings of the standard indicators of na-
tional power discussed above? The ideal solution would be to deduct costs
and thereby measure net stocks of economic and military resources directly.
For example, if a country cuts down a forest to build a new ofªce park, then
the value of the forest would show up as a loss on the country’s balance sheet.
If a country spends $50 billion ªghting a war—or growing food to feed its peo-
ple or cleaning up toxic waste or hosting the Olympics—then $50 billion
would be deducted from its stock of assets. In short, there would be no
free lunch.
The obvious problem with such an approach, however, is that compiling
balance sheets for every country is a painstaking process that requires substan-
tial data and time. The World Bank and the United Nations, working with doz-
ens of economists from leading universities and research organizations, have
recently taken up the task and published rough estimates of countries’ net
stocks of resources.51 These databases, however, go back only to 1990 and are
therefore of limited use for studying long-term trends or general patterns in in-
ternational relations. To do that, scholars need a proxy for net resources that
has data covering many countries going back many decades. Does such an in-
dicator exist?
In an oft-cited statistical reference, the historian Paul Bairoch suggested that
the “strength of a nation could be found in a formula combining per capita and
total GDP.”52 Bairoch did not elaborate on this point, but subsequent research
supports his intuition: as noted, scholars already believe that GDP represents
the gross size of a state’s economic and military output, and there is a large lit-
erature showing that GDP per capita serves as a reliable proxy for economic
and military efªciency.
Economists, for example, use GDP per capita to measure economic develop-
ment, because rich countries are, almost by deªnition, more efªcient than poor
countries—the main exceptions to this rule are petro-states, such as Saudi
Arabia, that can grow rich simply by pumping oil. Military studies also show
that the higher a country’s GDP per capita, the more efªciently its military
ªghts in battle.53 The reason is that a vibrant civilian economy helps a country
produce advanced weapons, train skillful military personnel, and manage
complex military systems.
51. Glenn-Marie Lange and Kevin Carey, eds., The Changing Wealth of Nations 2018: Building a Sus-
tainable Future (Washington, D.C.: World Bank, 2018); and United Nations University International
Human Dimensions Programme on Global Environmental Change (UNU-IHDP) and United Na-
tions Environment Programme (UNEP), Inclusive Wealth Report 2014: Measuring Progress toward
Sustainability (New York: Cambridge University Press, 2014).
52. Paul Bairoch, “Europe’s Gross National Product: 1800–1975,” Journal of European Economic His-
tory, Vol. 5, No. 2 (Fall 1976), p. 282.
53. Michael Beckley, “Economic Development and Military Effectiveness,” Journal of Strategic
Studies, Vol. 33, No. 1 (February 2010), pp. 43–79, doi:10.1080/01402391003603581.
The Power of Nations 19
GDP per capita thus provides a rough but reliable measure of economic and
military efªciency. This ªnding is not surprising, because population size is
the main driver of production, welfare, and security costs. The bigger a coun-
try’s population, the more people the government must protect and provide
for. Therefore, dividing GDP by population controls for some of the costs that
make the difference between a state’s gross and net resources. Combining GDP
with GDP per capita thus yields an indicator that accounts for size and ef-
ªciency, the two main dimensions of net resources.
To create a rough proxy for net resources, I follow Bairoch’s advice by sim-
ply multiplying GDP by GDP per capita, creating an index that gives equal
weight to a nation’s gross output and its output per person. This two-variable
index obviously does not measure net resources directly, nor does it resolve all
of the shortcomings of GDP and CINC. By penalizing population, however, it
provides a better sense of a nation’s net resources than GDP, CINC, or other
gross indicators alone.
Future studies can experiment with ways to improve this measure by adjust-
ing the weights or, even better, by expanding the databases created by the
World Bank and the United Nations or developing new measures of net stocks
of resources. For now, however, multiplying GDP by GDP per capita yields
a primitive proxy that scholars can use to evaluate the importance of net
resources in international politics. The following sections conduct such
an evaluation.
Research Design
I use three methods to compare the importance of gross versus net resources in
international politics. First, I conduct case studies of extended great power ri-
valries in which one nation had a preponderance of gross resources while the
other had a preponderance of net resources. I focus on “extended” rivalries,
meaning geopolitical competitions that lasted for several decades or longer,
because they provide more information about each nation’s relative power
than does a single war or crisis. I focus on great powers, because minor
power competitions are often shaped by great power politics and thus may not
reveal much information about the relative power of the minor powers them-
selves. Additionally, I focus on cases in which one side had a preponderance of
gross resources while the other side had a preponderance of net resources, be-
cause these cases constitute head-to-head tests of the importance of gross
versus net resources in geopolitical competition.
Second, I use large datasets to assess how well some of the single-variable
indicators highlighted above (GDP, CINC, and GDP ⫻ GDP per capita) predict
International Security 43:2 20
the winners and losers of international disputes and wars. In essence, I use
GDP and CINC as representatives for the standard, gross approach to measur-
ing power; and I use GDP ⫻ GDP per capita as the representative for my alter-
native, net approach. I focus on dispute and war outcomes, because they are
especially revealing about nations’ relative power: in peacetime, countries may
be able to exaggerate their power; but in times of conºict, bluffs get called,
vulnerabilities get exposed, and stronger nations usually emerge victorious.
Obviously, no measure of power will predict all dispute and war outcomes—
resolve, strategy, luck, and selection effects also play a role, as I explain
below—but a valid measure of power should perform better than random
chance at predicting dispute and war outcomes, and, all else equal, scholars
should prefer the measure that predicts the most outcomes.
Third, I analyze how well each measure performs as a control variable when
plugged into existing models of international relations. Many studies use sta-
tistical models that control for power to isolate correlations among other vari-
ables. I replicate two dozen of these studies and substitute measures of gross
and net resources to see how each indicator affects the models’ in-sample
goodness-of-ªt. All else equal, scholars should prefer the measure of power
that maximizes the goodness-of-ªt in the most models, meaning the meas-
ure that explains the most variance in the data.
Each of these methods has strengths and weaknesses. My goal is to compen-
sate for the weaknesses of one with the strengths of others. The case study
method allows me to analyze cases in detail and incorporate a range of indica-
tors consistent with each measurement framework, rather than relying solely
on single-variable proxies. The large-n results, by contrast, sacriªce the detail
of the case studies, but help ensure that my ªndings apply broadly across
many cases. Finally, the replication analyses test the resilience of each meas-
ure to a variety of model speciªcations and across numerous areas of interna-
tional relations.
Case Studies
As noted, the ideal case to test the relative importance of gross versus net
resources in international politics would be an extended great power rivalry in
which one nation had a preponderance of gross resources while the other had
a preponderance of net resources. According to widely used datasets, there
have been fourteen great power rivalries since 1816 that lasted at least twenty-
ªve years.54 From this list, I select the rivalries with the largest gaps between
the balance of gross and net resources.
54. Correlates of War Project. 2017. “State System Membership List, v2016,” http://
The Power of Nations 21
To measure the balance of gross resources in a given rivalry, I take the aver-
age of one nation’s share of the sum of the two sides’ GDPs and of the two
sides’ CINC scores. For example, imagine that country A and country B are
rivals. Country A’s share of gross resources would be calculated as:
To measure the balance of net resources, I calculate that same nation’s share
of the sum of the two sides’ GDP ⫻ GDP per capita. In my hypothetical exam-
ple, this would be:
To calculate the gap between the balance of gross and net resources in a ri-
valry in a given year, I simply subtract country A’s share of gross resources in
that year from its share of net resources in that year and take the absolute
value of the difference:
For each of the fourteen great power rivalries mentioned above, I perform
this calculation for every year of the rivalry and then take the average. These
averages are displayed in descending order in table 1.
Given space constraints, I focus on the rivalries with the largest gaps, which
I arbitrarily deªne as those with at least a 20 percentage-point difference be-
tween the average balance of gross resources and the average balance of net re-
sources. Six cases meet this criterion; however, I ultimately exclude two of
them—France versus China (1860–1929) and Britain versus the Soviet Union
(1946–91)—because these rivalries were sideshows in larger geopolitical com-
petitions and thus do not constitute independent cases. In the nineteenth cen-
tury, France challenged China only after Britain had already brought China to
its knees in the Opium Wars.55 And during the Cold War, Britain’s rivalry with
correlatesofwar.org; and Scott Bennett, “Coding Notes for Interstate/Enduring Data,” Pennsylva-
nia State Department of Political Science, updated version, May 1, 2017.
55. For this reason, I add the Britain versus China (1839–1911) rivalry to the list even though it was
not included in the original rivalry datasets.
International Security 43:2 22
the Soviet Union was shaped by the larger U.S.-Soviet rivalry. In sum, I am left
with four cases for further study, which are highlighted in gray in table 1.
Before I analyze the cases, it is worth noting that the nine cases with the larg-
est gaps between the balance of gross and net resources, including the four
cases I study below, involve Russia or China. This is not surprising, because
China and Russia are the only countries in the past 200 years to have led the
world in gross resources while lagging behind other great powers in net re-
sources. Their experience in competitions with smaller but more developed
countries thus provides the most straightforward test of my contention that
power stems from net, rather than gross, resources.
56. Christopher J. Fariss, Jonathan Markowitz, and Therese Anders, “Over 500 Years of Latent
GDP and Population Estimates,” University of Michigan and University of Southern California,
2017; Stephen Broadberry, Hanhui Guan, and David Daokui Li, “China, Europe, and the Great Di-
vergence: A Study in Historical National Accounting, 980–1850,” Discussion Papers in Economic
and Social History, No. 155 (Oxford: University of Oxford, April 2017); and Chiu Yu Ko, Mark
Koyama, and Tuan-Hwee Sng, “Uniªed China and Divided Europe,” International Economic Re-
view, Vol. 58, No. 1 (January 2018), pp. 285–327, doi:10.1111/iere.12270.
The Power of Nations 23
signiªcant territory and most of its sovereign rights, ªghting at least a dozen
wars on its home soil—and losing every single one of them.
The most important of these conºicts were two “opium wars” with Britain.57
For centuries, European merchants had traveled to China to swap silver for tea
and silk. In the nineteenth century, however, British traders discovered that
they could obtain better terms of trade by growing opium in India and selling
it on the black market in China.58 Despite China’s long-standing ban on
opium, British dealers smuggled into China nearly twelve tons of the drug an-
nually, enough to keep 3 million addicts high year-round.59 This inºux of nar-
cotics eroded 20 percent of China’s wealth from 1828 to 1836.60
To stem the opium epidemic, the Chinese government declared a war on
drugs in 1839, and Chinese ofªcials began seizing opium from British mer-
chants and dumping it into the sea. Britain responded by sailing sixteen war-
ships into Chinese waters and sinking China’s navy. From 1839 to 1842, in
what is now called the First Opium War, British forces occupied most of
China’s major coastal cities and brought Beijing to the brink of famine by
blockading the Grand Canal, the lifeline linking the Chinese capital to China’s
rice ªelds in the south.
Overmatched, the Chinese government capitulated in 1842 and signed the
Treaty of Nanjing, which gave Britain $21 million in reparations, a perpetual
lease on Hong Kong, access to ªve port cities, unprecedentedly low Chinese
tariffs, and immunity from Chinese law for British citizens living in China.
Fifteen years later, Britain upped the ante by demanding full economic ac-
cess to all of China and the right to sell opium legally throughout the country.
When the Chinese government resisted, Britain again used military force,
sparking what is now known as the Second Opium War. In January 1858,
British forces occupied Guangzhou, the largest port in China; and in April
1858, British forces, joined by French troops and Russian and U.S. diplomats,
occupied Tianjin, the commercial hub of northern China only 100 miles
from Beijing.
China mustered little resistance, in part because its military was busy sup-
57. W. Travis Hanes III and Frank Sanello, The Opium Wars: The Addiction of One Empire and the Cor-
ruption of Another (Naperville, Ill.: Sourcebooks, 2004).
58. Frances V. Moulder, Japan, China, and the Modern World Economy: Toward a Reinterpretation of
East Asian Development ca. 1600 to ca. 1918 (Cambridge: Cambridge University Press, 1977),
pp. 100–102. See also Peter Ward Fay, The Opium War, 1840–1842 (Chapel Hill: University of North
Carolina Press, 1998). For a detailed study of Britain’s opium trade, see Carl Trocki, Opium, Empire,
and the Global Political Economy: A Study of the Asian Opium Trade, 1750–1950 (London: Routledge,
1999).
59. Morris, Why the West Rules—For Now, p. 515.
60. Julia Lovell, The Opium War: Drugs, Dreams, and the Making of Modern China (London: Picador,
2011), pp. 36–37.
International Security 43:2 24
61. Stephen R. Platt, Autumn in the Heavenly Kingdom: China, the West, and the Epic Story of the
Taiping Civil War (New York: Alfred A. Knopf, 2012).
62. Fariss, Markowitz, and Anders, “Over 500 Years of Latent GDP and Population Estimates”;
Peer Vries, “Public Finance in China and Britain in the Long Eighteenth Century,” Working Paper
No. 167/12 (London: London School of Economics and Political Science, August 2012), p. 19; and
Lovell, The Opium War, pp. 111–113.
63. Morris, Why the West Rules—For Now, p. 502; and Paul Bairoch, “International Industrialization
Levels from 1750 to 1980,” Journal of European Economic History, Vol. 11, No. 2 (Fall 1982), p. 281.
The Power of Nations 25
0% 0% 0%
Britain China Britain China Britain China
SOURCES: Christopher J. Fariss, Jonathan Markowitz, and Therese Anders, “Over 500 Years
of Latent GDP and Population Estimates,” University of Michigan and University of South-
ern California, 2017; and David J. Singer, Stuart Bremer, and John Stuckey, “Capability
Distribution, Uncertainty, and Major Power War, 1820–1965,” in Bruce Russett, ed., Peace,
War, and Numbers (Beverly Hills, Calif.: Sage, 1972).
NOTE: GDP stands for gross domestic product. CINC stands for Composite Indicator of
National Capability.
64. Kennedy, The Rise and Fall of the Great Powers, p. 145.
65. Robert C. Allen et al., “Wages, Prices, and Living Standards in China, 1738–1925: In Compari-
son with Europe, Japan, and India,” Economic History Review, Vol. 64, No. S1 (February 2011),
pp. 8–38, doi:10.1111/j.1468-0289.2010.00515.x.
International Security 43:2 26
so the central government had to keep taxes low to appease local rulers while
keeping military spending high to sustain large internal security forces.66
These competing demands plunged China into ªscal crisis. China’s tax rev-
enues in the nineteenth century were 50 percent lower than they were in the
seventeenth century and were ªve times smaller than Britain’s in aggregate
and one hundred times smaller on a per capita basis.67 Meanwhile, China’s
military spending consumed 50 to 70 percent of government revenues in
peacetime and 100 percent or more during wars.68
Production, welfare, and security costs also drained China’s seemingly vast
military resources. To begin, China’s military was unskilled and under-
equipped compared with Britain’s. As one study concludes, “In all areas
of equipment—weaponry, forts, and most critically ships—Chinese equipment
lagged behind that of the British . . . the British had long moved into the era of
ªrepower, while parts of the Chinese army hung on to bows, swords, spears,
and rattan shields.”69 The best Chinese ªrearm was the matchlock, a muzzle-
loading musket developed in the ªfteenth century that required soldiers to
light a match each time it was ªred; British regiments, by contrast, were
equipped with ºintlocks or breech-loading percussion locks.70 Chinese war-
ships carried 10 cannons each, whereas British ships had 120 or more, and
Chinese cannons lacked sights and swivels and thus could not target moving
objects, such as enemy ships and soldiers.71 Repeatedly during the Opium
Wars, therefore, Chinese armies of thousands were routed in minutes by a few
hundred, or even a few dozen, British troops.
Security costs also degraded China’s military power. China’s forces were
“scattered through the empire, far too busy with domestic peace-keeping du-
ties (killing bandits or rebels; carrying out disaster relief; guarding prisons; po-
licing smugglers) to be spared for the quarrel with the British.”72 At any given
time, 50,000 Chinese soldiers were in transit around the country suppressing
revolts. Consequently, Chinese garrisons often had only a quarter of their
troops on hand to counter British assaults.73
66. Tuan-Hwee Sng, “Size and Dynastic Decline: The Principal-Agent Problem in Late Imperial
China, 1700–1850,” Explorations in Economic History, Vol. 54 (October 2014), pp. 107–127,
doi:10.1016/j.eeh.2014.05.002.
67. Ibid., p. 120.
68. Vries, “Public Finance in China and Britain in the Long Eighteenth Century,” p. 19; and Albert
Feuerwerker, The Chinese Economy, 1870–1949 (Ann Arbor: University of Michigan Press, 1969),
pp. 80–83.
69. Lovell, The Opium War, p. 111.
70. Haijian Mao, The Qing Empire and the Opium War: The Collapse of the Heavenly Dynasty (Cam-
bridge: Cambridge University Press, 2016), pp. 27–30.
71. Ibid., pp. 30–38.
72. Lovell, The Opium War, p. 113.
73. Ibid.
The Power of Nations 27
74. Peter Duus, The Rise of Modern Japan (Boston: Houghton Mifºin, 1976), pp. 120–135; and Jona-
than D. Spence, The Search for Modern China (New York: W.W. Norton, 1991), pp. 216–244.
75. W.G. Beasley, Japanese Imperialism, 1894–1945 (Oxford: Clarendon, 1987), chap. 7.
76. Ibid., chap. 8.
77. These events are described in Peter Duus, Japan’s Quest for Autonomy: National Security and For-
eign Policy, 1930–1938 (Princeton, N.J.: Princeton University Press, 1966), chaps. 2–3; Michael A.
Barnhart, Japan Prepares for Total War (Ithaca, N.Y.: Cornell University Press, 1987), chap. 1.
78. For a comprehensive account of the Japanese invasion and occupation of China, see Rana
Mitter, China’s War with Japan, 1937–1945: The Struggle for Survival (London: Allan Lane, 2013).
International Security 43:2 28
0% 0% 0%
Japan China Japan China Japan China
SOURCES: Christopher J. Fariss, Jonathan Markowitz, and Therese Anders, “Over 500 Years
of Latent GDP and Population Estimates,” University of Michigan and University of South-
ern California, 2017; and David J. Singer, Stuart Bremer, and John Stuckey, “Capability
Distribution, Uncertainty, and Major Power War, 1820–1965,” in Bruce Russett, ed., Peace,
War, and Numbers (Beverly Hills, Calif.: Sage, 1972).
NOTE: GDP stands for gross domestic product. CINC stands for Composite Indicator of
National Capability.
than China, with lower production, welfare, and security costs, and was thus
able to garner a preponderance of net resources.
First, Japanese industry was more productive than China’s. In 1913, Japan’s
labor productivity was three times greater than China’s overall. By 1930, Japan
was producing 150 times as much iron and steel as China and controlled
80 percent of the global silk market, China’s top export industry.79
Second, welfare costs worked in Japan’s favor. Whereas Chinese economic
output hovered around subsistence levels in the early twentieth century,
Japan’s economy grew ªve times faster than its population, “making it possi-
ble to feed the increasing number of Japanese born every year with enough left
over to ªnance both the government’s modernization efforts and investment
in the modern sectors of the economy.”80 Whereas agriculture tied down
79. Bairoch, “International Industrialization Levels from 1750 to 1980,” p. 281; Allen et al., “Wages,
Prices, and Living Standards in China, 1738–1925,” p. 20; Singer, Bremer, and Stuckey, “Capability
Distribution, Uncertainty, and Major Power War, 1820–1965”; and Debin Ma, “Why Japan, Not
China, Was the First to Develop in East Asia: Lessons from Sericulture, 1850–1937,” Economic De-
velopment and Cultural Change, Vol. 52, No. 2 (January 2004), pp. 369, 373, doi:10.1086/380947.
80. Duus, The Rise of Modern Japan, p. 137.
The Power of Nations 29
81. Feuerwerker, The Chinese Economy, 1870–1949, p. 121; and David Flath, The Japanese Economy
(Oxford: Oxford University Press, 2014), p. 52.
82. Feuerwerker, The Chinese Economy, 1870–1949, p. 168; Moulder, Japan, China, and the Modern
World Economy, p. 191; and Spence, The Search for Modern China, p. 235.
83. E. Sydney Crawcour, “Industrialization and Technological Change, 1885–1920,” in Kozo
Yamamura, ed., The Economic Emergence of Modern Japan (Cambridge: Cambridge University Press,
1997), p. 52.
84. Hsi-sheng Chi, “The Military Dimension, 1942–1945,” in James C. Hsiung and Steven I. Le-
vine, eds., China’s Bitter Victory: War with Japan, 1937–45 (New York: Routledge, 1992).
85. Marvin Williamsen, “The Military Dimension, 1937–1941,” in Hsiung and Levine, China’s Bit-
ter Victory, p. 135.
86. Ibid., pp. 170–171.
87. Chi, “The Military Dimension, 1942–1945,” p. 173.
88. Williamsen, “The Military Dimension, 1937–1941,” p. 148.
89. Chi, “The Military Dimension, 1942–1945,” pp. 168–169.
International Security 43:2 30
The typical Chinese military unit spent the bulk of its time and energy simply
trying to preserve its existence. It expected to have to take care of its own
needs, including food, clothing, conscripts, weapons, and transportation.
Fighting consumed too much energy, so ªghting was done only when abso-
lutely necessary. When sufªciently desperate, soldiers would not hesitate to
pillage the very same people they were supposed to protect. This in turn pro-
voked numerous incidents of friction between the army and the civilian popu-
lation. Probably the worst case occurred in Honan during the early phase of
Operation Ichigo. When the Chinese troops retreated in defeat, more soldiers
were killed by the indignant local population than by the Japanese.90
Finally, Chinese forces suffered substantial security costs. China was inter-
nally divided prior to the Japanese invasion; indeed, historians call the period
from 1916 to 1928 the “Warlord Era,” because China was chopped up among
rival military cliques. The Nationalist Party, led by Chiang Kai-shek, took con-
trol of China in 1928, but its rule remained contested by warlords, communists,
and various ethnic separatist groups. During the war with Japan, therefore, the
Chinese government stationed troops throughout the country to prevent do-
mestic rivals from seizing power or seceding.91 With its forces dispersed, the
Chinese military often found itself outnumbered in battles with the Japanese
despite its four-to-one advantage in troops overall.92
0% 0% 0%
Germany Russia Germany Russia Germany Russia
SOURCES: Christopher J. Fariss, Jonathan Markowitz, and Therese Anders, “Over 500 Years
of Latent GDP and Population Estimates,” University of Michigan and University of South-
ern California, 2017; and David J. Singer, Stuart Bremer, and John Stuckey, “Capability
Distribution, Uncertainty, and Major Power War, 1820–1965,” in Bruce Russett, ed., Peace,
War, and Numbers (Beverly Hills, Calif.: Sage, 1972).
NOTE: GDP stands for gross domestic product. CINC stands for Composite Indicator of
National Capability.
tary and making plans to eliminate the Russian and French threat through a
preventive war.95
The result—World War I—was catastrophic for Russia: Germany annihilated
Russia’s military; exacted a large indemnity; and forced Russia to give up terri-
tory comprising parts of modern-day Estonia, Latvia, Lithuania, Poland,
Belarus, and Ukraine. Only Germany’s defeat by a coalition led by Britain,
France, and the United States saved Russia from bearing the full brunt of these
losses. Nevertheless, Russia was devastated by the war, and by 1920 it was en-
gulfed in a bloody civil war.
What explains Russia’s poor performance against Germany? Russia had
higher production, welfare, and security costs than Germany and thus had far
fewer net resources available for geopolitical competition (ªgure 3).
To begin, in the early twentieth century, Russia was literally “the least devel-
oped European power,” lagging behind its neighbors in terms of per capita in-
95. Jack Snyder and Keir A. Lieber, “Correspondence: Defensive Realism and the ‘New’ History of
World War I,” International Security, Vol. 33, No. 1 (Summer 2008), pp. 174–194, doi:10.1162/
isec.2008.33.1.174.
International Security 43:2 32
come, output per worker, and other measures of economic development such
as literacy and health.96 In 1910, Russia was only 40 percent as productive as
Germany overall and 20 percent as productive in heavy industries.97
In addition, most of Russia’s economic output was consumed by welfare
costs. Russia’s GDP grew steadily during the nineteenth century, but nearly all
of this growth stemmed from population growth.98 The demands of feeding
this growing population forced 90 percent of Russia’s labor force into agricul-
ture.99 With more and more mouths to feed, Russia failed to accumulate wealth
at the rate of other great powers: whereas the real per capita wealth of
Germany grew 3 percent annually from 1890 to 1917, Russia’s increased by
only 1 percent.100
Security costs also took a large toll on Russia’s economy. With a territory that
stretched across one-sixth of the Earth’s landmass, Russia had to maintain
large military forces just to police its own borders and prevent remote regions
from breaking away.101 From 1870 to 1913, peacetime defense spending con-
sumed 5 percent of Russia’s GDP and 80 percent of the Russian government’s
revenues annually. In Germany, by contrast, peacetime defense spending ac-
counted for 3 percent of GDP.102
Russia’s army was twice the size of Germany’s and had a bigger budget, but
Germany’s advantages in technology and skill enabled it to outªght Russia on
a soldier for soldier basis. Whereas German troops were well trained and
armed, a majority of Russian troops were untrained conscripts and sent into
battle without riºes, where they were expected to scavenge weapons from the
dead.103 Russia also lacked railroads in its western regions, which made it
difªcult for Russia to move its armies around the Russo-German border.104
Germany, on the other hand, had a well-developed railroad system, so it could
move its forces quickly to that same border.
Furthermore, throughout the nineteenth and early twentieth centuries, secu-
96. Alec Nove, An Economic History of the USSR, rev. ed. (New York: Penguin, 1989), p. 5.
97. Bairoch, “Europe’s Gross National Product.”
98. Raymond W. Goldsmith, “The Economic Growth of Tsarist Russia, 1860–1913,” Economic De-
velopment and Cultural Change, Vol. 9, No. 3 (April 1961), p. 441, doi:10.1086/449917.
99. Robert C. Allen, Farm to Factory: A Reinterpretation of the Soviet Industrial Revolution (Princeton,
N.J.: Princeton University Press, 2005), p. 25.
100. Goldsmith, “The Economic Growth of Tsarist Russia, 1860–1913,” pp. 472–475.
101. William C. Wohlforth, “The Russian-Soviet Empire: A Test of Neorealism,” Review of Interna-
tional Studies, Vol. 27, No. 5 (December 2001), pp. 213–235, doi:10.1017/S0260210501008099.
102. John M. Hobson, “The Military Extraction Gap and the Wary Titan: The Fiscal-Sociology of
British Defense Policy, 1870–1913,” Journal of European Economic History, Vol. 22, No. 3 (January
1993), pp. 478–479; and Kennedy, The Rise and Fall of the Great Powers, p. 174.
103. Jonathan R. Adelman, Revolution, Armies, and War: A Political History (Boulder, Colo.: Lynne
Rienner, 1985), pp. 88–92.
104. Mearsheimer, The Tragedy of Great Power Politics, p. 72.
The Power of Nations 33
rity costs sapped Russian military power. As the historian Paul Kennedy
notes, “The great part of the Russian army was always pinned down by inter-
nal garrison duties, by police actions in Poland and the Ukraine, and by other
activities, such as border patrol.”105 Consequently, “in every war waged by
Russia throughout the reign [of the Russian Empire], its generals were chroni-
cally embarrassed by a shortage of troops.”106 This shortage became acute in
the years prior to World War I: mass uprisings increased tenfold from 1909 to
1913 as the tsarist government’s hold on power deteriorated and the Russian
Revolution gathered pace.107
105. Kennedy, The Rise and Fall of the Great Powers, p. 172.
106. Fuller, Strategy and Power in Russia, 1600–1914, p. 253.
107. Kennedy, The Rise and Fall of the Great Powers, pp. 236–237.
108. Singer, Bremer, and Stuckey, “Capability Distribution, Uncertainty, and Major Power War,
1820–1965”; Robert S. Norris and Hans M. Kristensen, “Global Nuclear Weapons Inventories,
1945–2010,” Bulletin of the Atomic Scientists, Vol. 66, No. 4 (July 2010), pp. 77–83, doi:10.2968/
066004008; National Science Foundation (NSF), Science and Engineering Indicators, 1989 (Washing-
ton, D.C.: NSF, 1989), p. 313; NSF, Science and Engineering Indicators, 1987 (Washington, D.C.: NSF,
1987), p. 228.
109. Ray S. Cline, World Power Assessment 1977: A Calculus of Strategic Drift (Boulder, Colo.:
Westview, 1977).
110. Stephen G. Brooks and William C. Wohlforth, “Power, Globalization, and the End of the Cold
War: Reevaluating a Landmark Case for Ideas,” International Security, Vol. 25, No. 3 (Winter 2000/
01), pp. 5–53, doi:10.1162/016228800560516.
International Security 43:2 34
0% 0% 0%
United Soviet United Soviet United Soviet
States Union States Union States Union
SOURCES: Christopher J. Fariss, Jonathan Markowitz, and Therese Anders, “Over 500 Years
of Latent GDP and Population Estimates,” University of Michigan and University of South-
ern California, 2017; and David J. Singer, Stuart Bremer, and John Stuckey, “Capability
Distribution, Uncertainty, and Major Power War, 1820–1965,” in Bruce Russett, ed., Peace,
War, and Numbers (Beverly Hills, Calif.: Sage, 1972).
NOTE: GDP stands for gross domestic product. CINC stands for Composite Indicator of
National Capability.
111. Joshua R. Itzkowitz Shifrinson, Rising Titans, Falling Giants: How Great Powers Exploit Power
Shifts, Texas A&M University, chap. 3.
112. Hal Brands, Making the Unipolar Moment: U.S. Foreign Policy and the Rise of the Post–Cold War
Order (Ithaca, N.Y.: Cornell University Press, 2016).
The Power of Nations 35
113. William Easterly and Stanley Fischer, “The Soviet Economic Decline,” World Bank Economic
Review, Vol. 9, No. 3 (September 1995), pp. 341–371, at p. 341, doi:10.3386/w4735; International
Monetary Fund, A Study of the Soviet Economy (Washington, D.C.: International Monetary Fund,
1991); Brooks and Wohlforth, “Power, Globalization, and the End of the Cold War,” pp. 14–20; and
Allen, Farm to Factory, pp. 190–200.
114. Yegor Gaidar, Collapse of an Empire: Lessons for Modern Russia (Washington, D.C.: Brookings In-
stitution Press, 2007), p. 75.
115. John R. Oneal, “Measuring the Material Base of the Contemporary East-West Balance of
Power,” International Interactions, Vol. 15, No. 2 (January 1989), pp. 177–196, doi:10.1080/
03050627908434727.
116. NSF, Science and Engineering Indicators, 1989, p. 278; and NSF, Science and Engineering Indica-
tors, 1987, p. 228.
117. Gur Ofer, “Soviet Economic Growth: 1928–1985,” Journal of Economic Literature, Vol. 25, No. 4
(December 1987), pp. 1767–1833.
118. Vladimir Kontorovich, “The Long-Run Decline in Soviet R&D Productivity,” in Henry S.
Rowen and Charles Wolf Jr., eds., The Impoverished Superpower: Perestroika and the Soviet Military
Burden (San Francisco, Calif.: ICS Press, 1990).
119. U.S. Bureau of the Census, USA/USSR: Facts and Figures (Washington, D.C.: U.S. Government
Printing Ofªce, 1991), pp. 4–9.
120. International Monetary Fund, A Study of the Soviet Economy, p. 53, table II.2.3.
121. Susan B. Carter et al., eds., Historical Statistics of the United States: Earliest Times to the Present,
millennial ed. (Cambridge: Cambridge University Press, 2006).
122. Noel E. Firth and James H. Noren, Soviet Defense Spending: A History of CIA Estimates, 1950–
1990 (College Station: Texas A&M University Press, 1998), pp. 129–130.
International Security 43:2 36
4 percent of GDP annually propping up its allies with aid and arms.123 The
United States, by contrast, spent 0.3 percent of its GDP annually on all forms of
foreign assistance combined.124
The Soviet military, too, suffered from high production, welfare, and secu-
rity costs. The Soviet Union spent 2 to 3 percent of its GDP on military R&D,
but lagged a generation or more behind the United States in ªfteen of the
twenty most critical military technologies and was merely on par with
the United States in the remaining ªve categories.125 The Soviet military was
hobbled by a rigid command structure, and its ofªcers lacked initiative; its
troops lacked basic skills, such as map-reading; language barriers created seri-
ous communication problems among different divisions; and 25 percent
of Soviet forces were made up of fresh conscripts with little to no training.126
Soviet forces trained so little in part because Soviet weapons systems were so
fragile; for example, Soviet ªghters required overhauls at triple the rate of
many Western aircraft, and Soviet tank engines wore out after 500 hours or less
of use.127 Plagued by equipment failures, the government kept most weapons
systems “packed away like a family’s best china,” using them only for special
exercises once or twice a year.128 The resulting skill deªciencies, plus the tech-
nology issues discussed above, probably made a successful Soviet invasion of
Central Europe impossible.129
Finally, and perhaps most important, the Soviet military confronted an ex-
tremely hostile security environment.130 By the 1980s, the United States had six
times as many allies as the Soviet Union, and this American alliance network
had three times the population and gross resources of the Soviet Union and its
123. Charles Wolf et al., The Costs of the Soviet Empire (Santa Monica, Calif.: RAND Corporation,
1983).
124. Ibid.; and Carter et al., Historical Statistics of the United States.
125. Oneal, “Measuring the Material Base of the Contemporary East-West Balance of Power,”
p. 182.
126. John J. Mearsheimer, “Numbers, Strategy, and the European Balance,” International Security,
Vol. 12, No. 4 (Spring 1988), p. 183, doi:10.2307/2539001; William W. Kaufmann, “Who Is Conning
the Alliance?” Brookings Review, Vol. 5, No. 4 (Fall 1987), p. 13; and William E. Odom, The Collapse of
the Soviet Military (New Haven, Conn.: Yale University Press, 1998).
127. Richard K. Betts, Military Readiness: Concepts, Choices, Consequences (Washington, D.C.:
Brookings Institution Press, 1995), pp. 156–159.
128. Ibid., p. 156.
129. John J. Mearsheimer, “Why the Soviets Can’t Win Quickly in Central Europe,” International
Security, Vol. 7, No. 1 (Summer 1982), pp. 3–39, doi:10.2307/2538686; Mearsheimer, “Numbers,
Strategy, and the European Balance”; Barry R. Posen, “Measuring the European Conventional Bal-
ance: Coping with Complexity in Threat Assessment,” International Security, Vol. 9, No. 3 (Winter
1984/85), pp. 47–88, doi:10.2307/2538587; and Barry R. Posen and Stephen Van Evera, “Defense
Policy and the Reagan Administration: Departure from Containment,” International Security, Vol. 8,
No. 1 (Summer 1983), pp. 3–45, doi:10.2307/2538484.
130. Stephen M. Walt, “The Case for Finite Containment: Analyzing U.S. Grand Strategy,” Interna-
tional Security, Vol. 14, No. 1 (Summer 1989), pp. 5–49, doi:10.2307/2538764.
The Power of Nations 37
allies.131 Facing a robust U.S. containment barrier abroad and disaffected citi-
zens at home, the Soviet Union had to expend signiªcant resources just to de-
fend its borders and prevent restive regions and satellite countries from
breaking away. The United States, by contrast, had a secure home base in the
Western Hemisphere and dozens of rich allies around the world. It therefore
had more leeway to choose where and when to project military power and was
able to ofºoad part of the burden of defending the free world to others.
131. Stephen M. Walt, “Alliance Formation and the Balance of World Power,” International Secu-
rity, Vol. 9, No. 4 (Spring 1985), pp. 34–35, doi:10.2307/2538540.
132. Analyses of war and dispute outcomes may also suffer from selection effects, because weak
countries are unlikely to pick ªghts with stronger countries unless they have some advantage (e.g.,
more resolve, favorable terrain, or the element of surprise) that offsets the stronger side’s material
superiority. Such effects are probably present in my samples of wars and disputes, but I do not be-
lieve that selection effects or omitted variables bias my results, because I am comparing among in-
dicators of power, not between indicators of power and other factors. To guard against omitted
variable bias, however, I replicate several existing studies of war and dispute outcomes that con-
trol for many factors. The results, which are presented in the next section, show that the balance of
net resources is a better predictor of war and dispute outcomes than the balance of gross resources.
133. Dan Reiter, Allan C. Stam, and Michael C. Horowitz, “A Revised Look at Interstate Wars,
1816–2007,” Journal of Conºict Resolution, Vol. 60, No. 5 (August 2016), pp. 956–976, doi: 10.1177/
0022002714553107.
International Security 43:2 38
Wars MIDs
GDP 68% 64%
CINC 70% 64%
GDP ⫻ GDP per capita 78% 70%
Number of observations 54 276
NOTE: MIDs are militarized interstates disputes. GDP stands for gross domestic product.
CINC stands for Composite Indicator of National Capability.
134. Faten Ghosen, Glenn Palmer, and Stuart A. Bremmer, “The MID3 Data Set, 1993–2001: Proce-
dures, Coding Rules, and Description,” Conºict Management and Peace Science, Vol. 21, No. 2 (April
2004), pp. 133–154, doi:10.1080/07388940490463861.
135. See the online supplementary materials, doi:10.7910/DVN/58KDCM.
The Power of Nations 39
First, Russia, China, and Israel have been involved in a large percentage of all
militarized conºicts: Russia and China have been the ªrst and fourth most mil-
itarily active nations in the world over the past two centuries, accounting for
19 percent and 11 percent of all militarized interstate disputes since 1816; and
Israel has been one of the most militarily active states since its founding, hav-
ing been involved in 8 percent of all militarized interstate disputes since 1948.
Second, these nations consistently fought rivals that had a different balance
of gross versus net resources: as noted, Russia and China led the world in
gross resources, but lagged behind their great power rivals in terms of net re-
sources; and Israel has a smaller population and set of gross resources than
its Arab enemies, but is more developed and therefore has often had greater
net resources.
Replication Analyses
My proxy for net resources more accurately accounts for the outcomes of great
power rivalries and international wars and disputes, but scholars often want
to control for relative power in statistical models of various aspects of interna-
tional relations. As a ªnal test, therefore, I replicate a random sample of recent
international relations studies and substitute measures of gross and net re-
sources into their statistical models to see how each measure affects the mod-
els’ in-sample goodness-of-ªt.
My sample consists of all studies published in seven leading political science
journals from January 2012 to April 2017 that included a control variable for
relative power and for which replication data were publicly available.136 These
criteria left me with twenty-four studies, which are listed in the online appen-
dix.137 These studies analyze a variety of aspects of international relations, in-
cluding nuclear proliferation, terrorism, trade, immigration, international law,
alliance formation, and the onset and outcomes of wars and disputes. I pro-
vide more information about each study in a separate bibliography, which is
included in the online appendix.
I replicated the main model of each study using CINC, GDP, and GDP ⫻
GDP per capita, respectively, as the variable for relative power and calculated
the in-sample goodness-of-ªt using the Akaike Information Criterion (AIC),
136. The seven journals were American Journal of Political Science, American Political Science Review,
International Organization, International Studies Quarterly, Journal of Conºict Resolution, Journal of Poli-
tics, and Security Studies. I identiªed studies that control for power by searching for citations of
commonly used databases for CINC, GDP, and military spending data.
137. See the online supplementary materials.
International Security 43:2 40
The AIC is commonly used in model selection, with lower values repre-
senting better ªt.138 When no main model was apparent, I used the most fully
speciªed model.
Table 3 in the online appendix summarizes the results of the replication
analyses. In seventeen of twenty-four studies, the models using GDP ⫻ GDP
per capita achieved a better goodness-of-ªt than the models using CINC.
Among these seventeen cases, the average ⌬AIC (i.e., the difference in AIC be-
tween the models using CINC and those using GDP ⫻ GDP per capita) is 6,
which implies that the models using CINC are .05 times as likely as the models
using GDP ⫻ GDP per capita to minimize information loss on average. A com-
mon rule of thumb is that a ⌬AIC greater than 2 provides statistically
signiªcant evidence of one model being better than another.139 By this crite-
rion, eleven of the seventeen cases are signiªcant. In three other studies, the
models using GDP ⫻ GDP per capita and CINC achieved the same goodness-
of-ªt, leaving only four studies in which CINC performed better than GDP ⫻
GDP per capita.
In eleven of the twenty-four studies, the models using GDP ⫻ GDP per
capita achieved a better goodness-of-ªt than the models using GDP, and the
⌬AIC in eight of these eleven cases was 2 or greater. In seven other studies,
GDP ⫻ GDP per capita and GDP produced the same goodness-of-ªt, and in six
studies GDP performed better than GDP ⫻ GDP per capita.
In sum, replacing standard gross measures of power with my proxy for net
resources improves the model ªt in a plurality of recent international relations
studies. Given the small number of replicated studies, these results certainly
do not prove that power is a function of net rather than gross resources. They
are consistent, however, with such a claim, as well as with the narrower claim
that GDP ⫻ GDP per capita is a better single-variable indicator of relative
power than CINC or GDP.
Conclusion
I have argued that scholars should measure power in terms of net resources
rather than gross resources; developed a rough but ready indicator for doing
138. Hirotugu Akaike, “A New Look at the Statistical Model Identiªcation,” IEEE Transactions on
Automatic Control, Vol. 19, No. 6 (December 1974), pp. 716–723.
139. Kenneth P. Burnham and David R. Anderson, Model Selection and Multimodel Inference: A Prac-
tical Information-Theoretic Approach, 2nd ed. (New York: Springer, 2002), p. 70.
The Power of Nations 41
so; and demonstrated that this indicator does a better job than standard gross
indicators at tracking the rise and fall of great powers, predicting the outcomes
of international disputes and wars, and serving as a control variable in quanti-
tative studies of international relations. There are two main implications of
these results.
First, an enormous literature in international relations has been built on a
ºawed conception of power, so existing studies may need to be reevaluated
with new measures. As noted, more than 1,000 peer-reviewed studies have
used CINC to measure power. Yet, this indicator severely mischaracterizes the
balance of power in some of the most important geopolitical events of the past
200 years. CINC also suggests, nonsensically, that Israel is, and has always
been, one of the weakest countries in the Middle East; Singapore is one of the
weakest in Southeast Asia; Brazil dominates South America with roughly
ªve times the power resources of any other state; Russia dominated Europe
throughout the 1990s, with more power resources than Germany, France, and
the United Kingdom combined; and China has dominated the world since
1996 and currently has twice the power resources of the United States.
Given the size and scope of these errors, one has to wonder whether the re-
sults of some seminal studies would change if scholars replaced CINC with
more accurate measures of power. I can only conjecture which sets of studies
might be ripe for reevaluation, but several immediately come to mind.
One is the literature on war and militarized dispute outcomes. Many studies
argue that military power is shaped by nonmaterial factors, such as strategy,
culture, and domestic politics.140 These claims have been bolstered by promi-
nent statistical studies that ªnd little relationship between the balance of mate-
rial power, as measured by CINC and GDP, and the outcomes of wars and
militarized disputes.141 These nonªndings, however, may be artifacts of ºawed
indicators. When power is measured in net terms, I ªnd that the side with
greater resources has won 70 percent of disputes and nearly 80 percent of wars
over the past two centuries. Thus, material resources, when properly mea-
sured, may be a more signiªcant source of military power than a large litera-
ture in international relations suggests.
Another literature that might need to be reevaluated is power transition
theory. For decades, scholars have debated whether power parity increases or
decreases the likelihood of war between states.142 Nearly all of these studies,
140. For a review of this literature, see Nye, The Future of Power.
141. Zeev Maoz, “Resolve, Capabilities, and the Outcomes of Interstate Disputes, 1816–1976,”
Journal of Conºict Resolution, Vol. 27, No. 2 (June 1983), pp. 195–229, doi:10.1177/0022002783
027002001; and Biddle, Military Power.
142. For the most recent study on this issue, see Allison, Destined for War.
International Security 43:2 42
however, measure power in gross terms, with the most popular indicators be-
ing GDP and CINC.143 As I have shown, these indicators misrepresent the bal-
ance of power in many cases, so it is possible that the power transition
literature is littered with false positives and false negatives; in other words,
many of the cases identiªed as power transitions may not have involved an ac-
tual transition in power, and, conversely, many genuine power transitions may
not have not been identiªed as such.144 Given that power transition theory re-
lies on a precise measure of power, it is vital that scholars reevaluate it with
sound indicators.
The second implication of my results is that they challenge the conventional
wisdom about current trends in the balance of power. Since the 1990s, and
especially since the 2008 ªnancial crisis, hundreds of books and thousands of
articles and reports have asserted that the United States’ economic and mili-
tary edge over other nations is eroding and that the world will soon be-
come multipolar.
The main evidence typically cited for these trends is China’s rising GDP and
military spending and various statistics that are essentially subcomponents of
GDP—most notably, China’s massive manufacturing output; volume of ex-
ports; trade surplus with the United States; infrastructure spending; consumer
spending; and large government bureaucracy and scientiªc establishment.145
The problem, however, is that these are the same gross indicators that made
China look like a superpower during its century of humiliation: in the mid-
1800s, China had the world’s largest economy and military; led the world in
manufacturing output; ran a trade surplus with Britain; presided over a tribu-
tary system that extended Chinese trade and investment, infrastructure pro-
jects, and soft power across continental East Asia; and was celebrated in the
West for its consumer market potential and tradition of bureaucratic compe-
tence and scientiªc ingenuity.
Obviously China is not as weak today as it was in the nineteenth century,
but neither is it as powerful as its gross resources suggest. China may have the
world’s biggest economy and military, but it also leads the world in debt; re-
source consumption; pollution; useless infrastructure and wasted industrial
capacity; scientiªc fraud; internal security spending; border disputes; and pop-
143. Indra de Soysa, John R. Oneal, and Yong-Hee Park, “Testing Power-Transition Theory Using
Alternative Measures of National Capabilities,” Journal of Conºict Resolution, Vol. 41, No. 4 (August
1997), pp. 509–528, doi:10.1177/0022002797041004002.
144. For evidence of this problem, see Carsten Rauch, “Challenging the Power Consensus: GDP,
CINC, and Power Transition Theory,” Security Studies, Vol. 26, No. 4 (July 2017), pp. 642–664,
doi:10.1080/09636412.2017.1336389.
145. See, for example, Rachman, Easternization; Zakaria, The Post-American World; NIC, Global
Trends 2030; Allison, Destined for War; and Subramanian, Eclipse.
The Power of Nations 43
GDP CINC
20 25%
United States
China
trillions of dollars
15 20%
share of world
15%
10
10%
5 5%
0 0%
1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015
8
trillions of dollars
trillions of dollars
150 300
6
100 200
4
2 50 100
0 0 0
1990
1995
2000
2005
2010
2015
1990
1995
2000
2005
2010
1995
2000
2005
2010
2014
SOURCES: Christopher J. Fariss, Jonathan Markowitz, and Therese Anders, “Over 500 Years
of Latent GDP and Population Estimates,” University of Michigan and University of South-
ern California, 2017; David J. Singer, Stuart Bremer, and John Stuckey, “Capability Distri-
bution, Uncertainty, and Major Power War, 1820–1965,” in Bruce Russett, ed., Peace, War,
and Numbers (Beverly Hills, Calif.: Sage, 1972); Glenn-Marie Lange and Kevin Carey, eds.,
The Changing Wealth of Nations, 2018 (Washington, D.C.: World Bank, 2018); and United
Nations University International Human Dimensions Programme on Global Environmen-
tal Change (UNU-IHDP) and United Nations Environment Programme (UNEP), Inclusive
Wealth Report, 2014: Measuring Progress toward Sustainability (New York: Cambridge
University Press, 2014).
ulations of invalids, geriatrics, and pensioners. China also uses seven times the
input to generate a given level of economic output as the United States and is
surrounded by nineteen countries, most of which are hostile toward China,
politically unstable, or both.
Accounting for even a fraction of these production, welfare, and security
costs substantially reduces the signiªcance of China’s rise. As shown in ªg-
ure 5, if power is measured in terms of GDP or CINC, China already appears
to be the most powerful country in the world; by contrast, if power is mea-
International Security 43:2 44
sured with my proxy for net resources or the UN or World Bank’s measures of
net resources (or other measures of net stocks of economic and military re-
sources not shown here), then China lags far behind the United States and
looks set to do so for the foreseeable future.146
Clearly, a great deal hangs in the balance with regard to how scholars meas-
ure power. The most important point to be made, therefore, is that the
measurement of power needs to receive the same kind of sustained and rigor-
ous study that has been given to the effects of power. Power is the central vari-
able in the ªeld of international politics, yet scholars still lack a sound means
of measuring it. With so many policy decisions and academic theories relying
on accurate assessments of relative power, it is imperative that scholars get
those assessments right.
146. These points are detailed in Michael Beckley, Unrivaled: Why America Will Remain the World’s
Sole Superpower (Ithaca, N.Y.: Cornell University Press, 2018), pp. 33–97.
This article has been cited by:
1. Andrew Kydd. 2020. Switching sides: changing power, alliance choices and US–China–Russia
relations. International Politics 57:5, 855-884. [Crossref]
2. Peter J. Katzenstein. 2020. Protean power: a second look. International Theory 12, 1-19.
[Crossref]
3. Germán Baldi. 2020. Nature protection across countries: Do size and power matter?. Journal for
Nature Conservation 56, 125860. [Crossref]
4. Therese Anders, Christopher J Fariss, Jonathan N Markowitz. 2020. Bread Before Guns or
Butter: Introducing Surplus Domestic Product (SDP). International Studies Quarterly 64:2,
392-405. [Crossref]
5. Dani Belo. 2020. Conflict in the absence of war: a comparative analysis of China and Russia
engagement in gray zone conflicts. Canadian Foreign Policy Journal 26:1, 73-91. [Crossref]
6. Igor Davidzon. Three Models of Nuclear Weaponization Influence and Their Origins 33-45.
[Crossref]
7. Nicholas Khoo. 2019. Retooling great power nonproliferation theory: Explaining China's North
Korea nuclear weapons policy. The Pacific Review 43, 1-24. [Crossref]
8. R. Väyrynen. 2019. Models of a New World: towards a Synthesis. Outlines of global
transformations: politics, economics, law 12:3, 189-206. [Crossref]
9. Benjamin Schreer. 2019. Why US-Sino strategic competition is good for Australia. Australian
Journal of International Affairs 73:5, 431-448. [Crossref]
10. Adam Knight. 2019. Power Sharing and Power Relations after Civil War. Civil Wars 21:3,
434-436. [Crossref]
11. Caleb Pomeroy, Michael Beckley. 2019. Correspondence: Measuring Power in International
Relations. International Security 44:1, 197-200. [Citation] [Full Text] [PDF] [PDF Plus]