Declining competitiveness soothes war

Our primary finding is that great powers deal with decline by retrenching promptly and proportionately. Regardless of regime type, retrenchment is the most common response to decline, and preventive wars almost never happen during decline. Retrenchment helps great powers regain their former positions with some regularity, but states that fail to retrench never recover their rank. Although decline promotes retrenchment, the rate of decline is only a rough predictor of the extent of retrenchment. Great powers have a tendency to retrench less than might be expected given underlying power trends, but most errors are modest. Without question, our theory has anomalies, but rival views are only theories of anomalies.

Secondarily, we find that states that fall at similar rates behave similarly—there is a stepwise relationship between non-decliners, small decline-ers, medium decliners, and large decliners. Declining powers spend less on defense and slow the growth of military personnel more than other states. They are no more likely to form new alliances, but they are less likely to participate in disputes. We also find evidence that states refine retrenchment policies to prevailing structural conditions. While our structural modifiers have modest explanatory power on their own, collectively they perform impressively. Declining powers that are most favorably positioned have almost all retrenched.

To go about testing the arguments, the first section provides an overview of our research design. It bounds the scope of our analysis, proposes a measure of decline, describes our cases, and discusses the measurement of key variables. The second section presents our initial findings. It provides a coarse- grain overview of strategic adjustment across our sixteen cases and assesses the merits of rival theories in light of the evidence. The third section buttresses this analysis with a comparison of declining and non-declining powers across a series of quantitative indicators. All in all, the empirical record suggests that moments of decline may be fraught, but that prudence tends to prevail.

Research Design

To test theories of decline, we examine the grand strategies of sixteen declining great powers since 1870. Our main unit of analysis is the great powers. Drawing on the Correlates of War (COW) Project, we identify nine states that qualify as great powers during this period.1 Our main independent variable is the depth of decline. As defined in the introduction, acute relative decline (or simply decline from this point on) is when a great power loses its ordinal rank for at least five years. The depth of decline measures how far a state falls relative to other great powers during the transition time. In choosing to define decline in this way, we assume that what matters most is not short-term or random oscillations of power, but sustained trends that shake up the great power pecking order.

We measure relative power by examining a state's share of gross domestic product (GDP) among the great powers. Our data come from the pioneering work of Angus Maddison and date back to 1870, the first year for which we have reliable cross-national data.2 If a country's share of great power GDP drops a rank and remains there for at least five years, we count this as a moment of decline. To count as a great power, a state must be on the COW list, but to count as a case of decline, a great power must be above a 10 percent share of total great power GDP.

We measure the depth of decline by calculating the total decline in great power share of GDP for the five years following the shift in ordinal rank¬ings. We categorize declines that total less than a 2 percent drop in relative share of great power GDP as small; those totaling from 2 to 4 percent as medium; and those totaling more than 4 percent as large. To take a familiar example: in 1908 Germany accounted for 15.7 percent of great power GDP, surpassing Great Britain for the first time in its history. Over the next five years, Great Britain continued to lose ground to Germany, with its share of great power GDP reduced by a total of 1.4 percent. Thus, the 1908 British case represents a moment of small decline.

No measures are perfect, even when they are the best among flawed alternatives. Though parsimonious, cross-national GDP data must be viewed with skepticism, especially over long periods of time.3 GDP was invented as a concept relatively recently, and projecting it backwards in time is an arduous art. GDP is sensitive to the domestic division of labor, culturally specific behaviors, and regulatory environments. For some coun¬tries in the data set, we have had to estimate GDP using less reliable mea¬sures of output from specific economic sectors.4 What constitutes international power is intricate, evolving, and in part intangible. Our rebuttal is that using GDP is an elegant and conventional approach to analyzing power. Many studies operationalize power using some measure of eco¬nomic output—whether energy consumption, steel production, or GDP5 And GDP is the most frequently used for sterling reasons: it is arguably the most comprehensive, parsimonious, and dependable measure of economic performance. No other measure offers the historical reach, analytical objectivity, and scientific comparability of GDP.

Another concern is that GDP may not be the best measure of decline. States could care more about political, military, or cultural decline—or about some dynamic basket of these.6 Still, these indices have not been generalized across time and space, and face objections that they are arbitrary or indeterminate. We use economic data to identify decline not because we believe that policymakers are driven only by money, but because these moments are likely to correspond with crises. No other form of power is as fungible as economic power, and it can be converted into political, military, or cultural clout. Power and plenty have been closely correlated throughout history, and sustained periods of slow or negative economic growth can have profound political consequences.7 Other more nuanced forms of decline may also encourage retrenchment, but if there is a situation that is likely to elicit a retrenchment response, economic decline is it.

A different objection is that decision-makers may not know what the great power rankings are, much less that an ordinal transition is taking place. Many policymakers had faulty or imprecise GDP data for large stretches of time, while others had no GDP data at all.8 Yet we assume that decision makers acted as if they had access to GDP data, or something close to it. This is not unrealistic: political elites monitor a variety of economic indicators including agricultural production, industrial output, commodity prices, tax receipts, and import and export totals. Their measures may have been imperfect, but many correspond to what we now call GDP. To validate this point, we compared Maddison's measure of GDP with Arthur Banks's cross- national indicators of national performance. Although there are limitations to the Banks data, we found that GDP was highly correlated with a country’s revenues, expenditures, imports, and exports.9 These are precisely the types of indicators political elites tend to track, which gives us confidence that policymakers' perceptions of power will correlate with GDP.10

A final concern is that GDP is not the optimal measure for this study. GDP, like income, is a flow measure and gauges the market value of all finished goods and services produced within a country in a given year. If money were power—and it is only somewhat so—the best measure would be national wealth, a stock measure.11 Unfortunately, no one keeps such a measure for long periods of time. In the final analysis, however, the rank ordering depends on who makes the most, year in and year out. In any given year, a lackluster income can be compensated for by a large nest egg from prior years. Conversely, poor investment, profligate consumption, or high debt can mute the advantages of an enviable income. In this sense, GDP is a leading indicator, but one whose effect can be dampened or amplified by the national balance sheet. An advantage of focusing on ordinal transitions is that they reflect sustained trends and thus dampen the downsides of using an annual flow measure. Fundamentally, states cannot maintain their position if rivals have a persistently higher economic output.

Using GDP as a measure for relative power, we identify sixteen cases of decline since 1870, the point at which reliable cross-national figures begin for all the great powers. 82 Of the nine great powers in our sample, five underwent decline at some point in their tenure. Two great powers, Italy and Austria-Hungary, remained firmly planted in the bottom rungs of the great power ladder and never experienced ordinal transitions.13 Two other powers, the United States and China, have not experienced decline, although the former is on pace to do so sometime in the coming decade.

Great Britain was the most experienced declining power, accounting for six of the sixteen cases. In terms of rising powers, five different states managed to eclipse a rival great power. Germany dominated this list, accounting for eight of our sixteen cases.

Our sixteen cases of decline are evenly distributed in terms of depth.14 Six of our cases were small declines, five were medium declines, and five more were large declines. The smallest decliner, 1879 Russia, involved a cumulative drop of 0.4 percent share of great power GDP over five years. The largest decliner, 1987 Soviet Union, consisted of a cumulative loss of 9 percent share of great power GDP over five years. The 1987 Soviet Union case was a notable outlier, however: the median and modal depth of decline was a 3.2 percent share of great power GDP.

One interesting feature of our cases is that the depth of decline tends to increase in more recent cases. All six cases of small declines, for example, took place prior to the First World War. In contrast, three of the five large declines took place after the Second World War. Polarity explains much of why the depth of decline has increased over time. In 1870, for example, five great powers had more than a 10 percent share of great power GDP. In 1946, in contrast, only three great powers possessed more than a 10 percent share. Given the paucity of great powers and the lofty perches that they occupied, it is unsurprising that declines in this period involved sizable shifts.

In contrast to depth, our sixteen cases of decline are not evenly distributed across time. Eight of our cases occur in the pre-World War I period; five in the interwar period; two during the Cold War; and one in the post- Cold War period. In other words, decline itself appears to be on the decline. Once again, the most plausible explanation here seems to be the shift from a multipolar to a more bipolar balance of power in the international system. During the Cold War, only three great powers other than the United States or Soviet Union ever managed to achieve shares greater than 10 percent of great power GDP, and no more than two did so at the same time. Since there are fewer great powers able to challenge on the global stage, moments of decline occur less often.

There are two ways to critique the number of our cases: one is that we have too many. Only one of our cases, 1872 United Kingdom, involves a hegemon being eclipsed by a rising challenger. Because of the impressive economic performance of the United States since 1872, no other hegemonic transitions occurred. A critic could object that this makes most of our cases irrelevant to hegemonic stability theory in general and the Sino-American transition in particular. We think that would be a big mistake. For one, it would be dangerous to focus on the top dyad when great power dynamics are systemic. For another, no theory is as hegemon-centric as the critic's case implies. Although preventive war theorists focus on the hegemon, they are promiscuous in their case selection, not just focusing on the top two powers in the system. Robert Gilpin, for instance, applies his logic to the Goths, Venice, and Austria-Hungary.15 Lastly, there are solid reasons to suspect that hegemons are more alike than different from lesser great pow¬ers during decline, and lopping off non-hegemonic cases would obscure the patterns that studying all great power declines could teach us.

A benefit of our universe of cases is that it reveals the systemic dynamics of decline. Though we occasionally check dyadic relations to test rival views, our theory is systemic and our empirical strategies are positioned to observe the interplay between the great powers. Nearly all of our cases involve an ordinal shift from the second or third position in the great power rankings. Of the thirteen declines involving second or third ranked powers, eleven involve rival European great powers. So close together, these secondary power transitions will be highly salient to the participants, though they did not involve a hegemon and rising challenger, and may help illuminate future transitions.

Alternatively, it could be objected that we have too few cases. No selection criteria cast so fine a net that they catch all the cases of interest, and ours are no exception. A great power that experiences a sustained decrease in relative power, yet does not trade places with another power, would not appear in our cases. We accept this limitation, but choose to focus on ordinal transitions because they are most likely to prompt a response from a declining power. The emergence of a new geopolitical peer is hard to ignore and often creates security challenges that declining powers are compelled to address. It is also possible that policymakers may believe their state is in decline, even when our metrics indicate it is not. Yet that would mean leaders, selected for their political perceptiveness, were frequently out of touch with the most basic reading of relative power. The compensatory virtues of our method are that it does not rely on our own arbitrary assessments, and produces a list of plausible cases, many of which are thought archetypal.

Our primary dependent variable is the grand strategic response of declining great powers, which can fall into one of three categories. First, declining powers can expand the costs of their foreign policy. They can increase the pace of their military spending and accelerate the growth of their military. They can negotiate or strengthen offensive alliances aimed at particular targets. They can initiate fresh diplomatic crises and espouse policies that heighten the risk of war. Second, declining powers can maintain their existing foreign policy commitments in an attempt to preserve the status quo. They can continue present trends in terms of military spending and the size of their military. They can honor old alliance commitments, but not contract new ones. They can respond to potential crises in much the same manner as they have in the past. Third, declining powers can retrench and reduce their foreign policy burdens. They can slow the pace of their military spending and decelerate the expansion of their military. They can negotiate new defensive alliances to help protect vulnerable commitments. They can seek to minimize their participation in inter-national crises, and find peaceful ways to settle those that do break out.

In practice, measuring the extent to which a great power expands or contracts its foreign policy commitments is difficult. Great powers can follow ambiguous policies, whose intent is difficult to discern. They may adopt ambivalent policies, which are pacific in some regions, but belligerent in others. Their policies may vacillate as objectives shift, assessments change, or new options become available. Even policies that achieve their objectives in seemingly reasonable ways may appear more planned and coherent than they actually were. For all these reasons, we must evaluate any shift in a great power's grand strategy with care.

Our solutions to these problems are caution and crosschecking. We use a mixture of qualitative and quantitative measures to track grand strategic adjustments over time. In terms of timing, we focus on the five years before and five years after an ordinal transition. Given our description of policy-makers as rational, purposive pragmatists, this is a fair wager. Great powers helmed by far-sighted and savvy policymakers may be able to anticipate an ordinal transition and modify their policies earlier. Those with more myopic and stubborn leaders may change course only in the wake of disastrous events after rivals have overtaken them. We are agnostic as to whether great powers react at the outset, halfway through, or near the end of their decline. What matters more is whether policymakers modify their strategic approach with some proximity to the onset of decline. If strategic adjustment never takes place, or only takes place a decade or more after the fact, this would call our theory into question.

Qualitatively, we examine the foreign policy record of each of our sixteen cases for clear evidence of either the expansion or contraction of grand strategic ambitions. To keep the subject tractable, we zero in on the policy choices great powers make in the military and diplomatic spheres. For military policy, we track trends in defense spending and military manpower. How much are states spending on defense compared to previous years? We also investigate how states allocate military resources. Do they prioritize investments in certain types of capabilities? We record how states deploy their military assets. Do they position them abroad or closer to home? For diplomatic policy, we consider how a state manages its interests across regions important to its foreign policy. Which interests and which regions are stressed? We also assess the character of a great power's alliance commitments. Does it negotiate new agreements, with whom, and to what ends? We also study how a great power behaves in crises. Does it choose to confront or negotiate with rivals? What risks does it run in disputes and what settlements is it willing to accept?

We supplement these qualitative assessments with a variety of quantitative indicators. The first is the annual change in military expenditure and personnel in the declining power.16 Combined with a qualitative analysis of acquisition, training, doctrine, and deployment, this allows us to measure military responses to decline. The second is the number of new alliance agreements signed by a declining power.17 We couple our overview of alli¬ances with a qualitative assessment of the substance of the agreements and how new agreements fit in a state's broader alliance portfolio. This allows us to unpack the diplomatic responses to decline. Third, we survey declining powers' behavior in militarized interstate disputes.18 We examine how often declining powers participate and initiate disputes, as well as how far these disputes escalate. We combine this analysis with a qualitative assessment of the goals and strategies used during disputes and the outcomes of these disputes. This allows us to develop a general sense of the role of force during decline.

Our dependent variable is inescapably controversial. Great powers some-times move forward and backward simultaneously across various spheres; these problems are inherent in any study of grand strategy. Our proxy measures are also imperfect: states reduce their military forces and contract alliances for many reasons, only some of which are the result of shifts in power. We seek to minimize these issues through transparency. Reasonable people might disagree on how to interpret a state's policy responses. We welcome these disputes, code the debatable cases within a range, and report our findings within these parameters. We include detailed information about the coding of all sixteen of our cases in an online appendix.19

It is very important to note that our assessments are descriptive and comparative, not normative. Though previous scholarship graded foreign policies against the threats they faced—frequently scoring them as over- or under-reacting—we take a different tack.20 While evaluating the efficacy of foreign policy strategies is a useful exercise, for present purposes we prefer to judge foreign policies with comparable cases. The issue is not one of better or worse policies, but a relative increase or decrease in grand strategic ambition. This commitment to description brings diagnostic benefits: it sidesteps hindsight bias, groups like cases, and provides yardsticks to measure how far from the mean anomalies are.

Initial Findings

The pessimistic view of decline receives little support. A complete coding of the dependent variable for all sixteen cases of decline can be found in table 1 in chapter 1. Against arguments that retrenchment is rare, we find that declining powers retrenched in at least ten and at most thirteen of our sixteen cases, a range of 63-81 percent. On any accounting, the majority of declining powers began to retrench immediately before or shortly after their ordinal transi¬tion. We further find that great powers maintained policies of the status quo in at least two and at most five of our sixteen cases (a range of 13-31 percent). This finding suggests domestic interests can constrain retrenchment, but only in unusual circumstances. We also find that declining powers rarely take up policies of expansion. We find unambiguous evidence of expansion in only one of our sixteen cases: 1931 Germany. Aggressive responses to decline appear to be the exception, rather than the rule.