Power Shifts and War
Summary and Keywords
A large body of theoretical work posits that power shifts or expected power shifts cause war. Power transition theory, cyclic theories of war, preventive war arguments, and the bargaining model of war are discussed in this article. Indeed, shifting power is one of the most popular and venerable explanations for war. Its origins go at least as far back as Thucydides, who famously wrote, “What made war inevitable was the growth of Athenian power and the fear this caused in Sparta.” Two major points must be discussed. First, there is an impressive correlation between major power war and shifting power, a correlation consistent with the arguments of several systemic theories of war. Second, much of the empirical research examining power shifts and war suffers from endogeneity and model specification concerns. Regarding endogeneity, more effort should be placed on identifying valid instruments and conducting experiments. Regarding model specification, more attention needs to be paid to scope conditions. Shifting power is not expected to cause war in all contexts. Precisely defining the relevant contexts and modeling them empirically is necessary to evaluate the shifting power and war hypothesis.
Power shifts cause war. This is the view of a large and diverse body of theoretical work on the causes of war. Power transition theory, long cycle theory, other cyclic theories, preventive war theories, and the bargaining model of war each expect that significant changes or expected changes in power between countries cause war. Indeed, shifting power may be the most popular cause of war in the theoretical study of war. It is certainly one of the most common explanations, along with information problems and diversionary arguments. Depending on the theoretical perspective, shifting power arguments are considered as arguments about the rise and decline of either the major powers or countries in general, preventive war, or bargaining failure. Two major points are outlined in this article. First, there is an impressive correlation between major war and shifting power, a correlation consistent with the arguments of several systemic theories of war. Second, much of the empirical research examining power shifts and war suffers from endogeneity and model specification concerns.
First, several key concepts are defined to begin the discussion: specifically power, power shifts, and war. Each of these concepts is complex, in the sense that each is multidimensional. One implication of this conceptual complexity is that empirical measures are especially difficult to devise. In turn, there is a need to improve extant measures. Second, the reasons that power shifts cause war are explored. Most theoretical work on power shifts and war focuses on one of three causal mechanisms linking power shifts and war: fear, credible commitment problems, or a disjunction between a country’s relative power and the distribution of benefits it is receiving. In the process of explaining how power shifts cause war, several subsidiary questions must be addressed. Do all rising powers want to change the status quo? Why does the declining state not compromise so as to avoid war? And most importantly, how do power shifts occur? The ways that theories answer these questions is critical for identifying the relevant scope conditions for empirical testing and the diplomatic strategies that may promote peace. Third, key empirical literature on shifting power and war is presented. Surprisingly little scientific research (i.e., research that employs statistical tests of hypotheses) has examined the relationship between shifting power and war. At the same time, a large body of detailed case studies exists in this research area. On the one hand, the case-study research and large-N observational research lends credence to the claim that there is an association between shifting power and war. On the other hand, the extant empirical research can only suggest a correlation between shifting power and war because this research has not fully addressed endogeneity. Endogeneity concerns in this context include the omission of observable confounders and not addressing unobservable or difficult to measure confounders. The latter stems from the fact that the posited cause is not randomized across cases. Finally, suggestions for improving the credibility of the claim that shifting power causes war are discussed.
To understand the theoretical and empirical relationship between power shifts and war, a clear understanding of several key concepts is needed. The first relevant concept is power. Power is the ability to get an actor to do something that she would not otherwise do (Dahl, 1957, pp. 202–203). Somewhat more formally, actor A’s power is “the difference in the probability of an event, given certain action by A, and the probability of an event given no such action by A (Dahl, 1957, p. 214).” There are three key aspects of this conceptualization. First, in a twofold sense, power is relative. Actor A’s ability to get another actor to do something depends on how much power each has. Actor A may be more powerful than actor B but less powerful than actor C. In addition, actor A’s power relative to actor B can vary by the context. In a military domain, for example, actor A may be more powerful than actor B, but in a moral domain actor A may be less powerful than actor B. Second, there may be a difference between actual power and latent power (Mearsheimer, 2001). Actual power is just what it suggests. Latent power is an actor’s potential for developing or increasing power in short period of time. In contemporary world politics, Japan’s actual military power is not that great, relative to the most powerful countries in the world. However, Japan has great latent power. In Japan’s case, the difference between its actual and latent power stems primarily from a political choice to limit their military despite their large and modern economy. Third, power is multifaceted. The primary facets of power are economic power, moral power, political power, and military power. When we talk about power shifts causing war, the focus is on military power.
What is military power? The primary components of military power are weapons, personnel, training, armed forces organization, and strategy. Among these components, weapons or weapon systems are typically the most important element. Yet even with a focus on weapons, it is difficult to create a valid measure of military power. Weapons are multidimensional and there is no clear unit of measurement. Often, the focus is on some aggregation of military expenditures and military personnel. In turn, measures of shifts in power are typically some function of military expenditures and personnel.1
The second key concept to define is a shift in power. A shift in power occurs when the relative power between actors A and B is different at time T compared to time T − X, where X is some unit of time. Most research, theoretical and empirical, on shifting power and war focuses on either a decrease in the relative power difference between the actors or a power transition between the actors. Generally, this research is not concerned about situations in which country A has quite a bit more power than country B at time T and even more power at time T+X. Instead, the concept “a shift in power” is typically reserved for situations in which country A is stronger than country B but A’s strength relative to B decreases over time such that the two countries become near equals in military power. “Near equals” often means that country B has at least 75% of country A’s power or country A has at least 75% of country B’s power. The concept “shift in power” depends not only a clear definition and measure of military power but also some specification of the amount of time relevant for assessing a shift. Should shifts in power be studied by examining changes in power over 50 years, 25 years, 10 years, 1 year or some other amount of time? While some of the theories specify time intervals for examining a shift in power, in none of these theories is the shift precisely determined by theory. To summarize, research highlighting the concept “shift in power” tends to focus on decreasing relative power, which explains why preventive war occupies a central place in shifting power arguments.
The primary outcome of interest for this article is war. War is a political event characterized by violence between groups and the loss of many lives. There are two key aspects to this definition. First, war is political; the object of war is to change the institutions that organize the lives of the competing groups, change the distribution of resources between the groups, or both. Second, war involves violence and the loss of life. Like other political scientists, this essay adopts the 1,000-fatality threshold for identifying wars (Small & Singer, 1982). While most of the research discussed in this article aims to explain the onset of major power interstate wars or interstate wars in general, some research examines less severe military conflicts.
Why Do Power Shifts Cause War?
Why do power shifts cause war? Extant theories of power shifts and war focus on one of three causal mechanisms: fear, a disjunction between relative power and preference attainment, or a credible commitment problem.2 While the disjunction mechanism has received the most attention, the essay argues that it can be subsumed by the credible commitment mechanism.
One of the oldest explanations for war centers on shifting power and posits fear as the causal mechanism. In the second chapter of book one of his history of the Pelopennisian War, Thucydides (1954) writes: “What made war inevitable was the growth of Athenian power and the fear this caused in Sparta.” The historian Michael Howard (1984, p. 100) offers the same explanation for both World War I and World War II. Regarding World War I, Howard states:
It was the general perception of the growth of German power that was awakened by the naval challenge, and the fear that a German hegemony on the Continent would be the first step to a challenge to her own hegemony on the oceans, that led Britain to involve herself in the continental conflict in 1914 on the side of France and Russia.
Van Evera (1999) and Copeland (2000) offer similar preventive war perspectives. For Van Evera (1999), preventive war represents a window of opportunity, which he considers a primary cause of war. Windows cause war because they increase fear, which leads declining countries to take more risks (p. 79), reduce threat credibility (p. 80), and increase misperceptions (pp. 81–86). For Copeland (2000, p. 233) “fear of deep decline,” particularly in a bipolar system, is the primary explanation for major wars.3
Why does a shift in power create fear? From a realist perspective the answer is straightforward. The rising power will want to revise the status quo, and the declining power will want to resist such a revision. This explanation raises a number of additional questions. Do all rising powers want to change the status quo? Why does the declining state not compromise to avoid war? To be fair, these are questions suggested by more modern rationalist perspectives. Explanations anchored in fear, however, are explicitly nonrationalist. Nonrationalist means that a choice for war, which we might consider a choice mistake, is motivated by psychological biases and not by informational problems or rational calculations (see Jervis, 1988 for a broader discussion of misperception and war). One does not know whether the other side wants to revise the status quo, but one fears that it will. Moreover, this fear does not need a firm grounding in reality because the diversity of human beings and groups means that there will always be some difference between groups that might make one uncertain about how the other will behave when they have more power. In the case of the Peloponnesian War, there were significant political differences between the Spartans and Athenians. The latter was more open and more democratic and more of a trading state than the former. Nevertheless, there were also similarities between Sparta and Athens. Both were Greek city-states. Neither was a liberal or a full democracy in the modern sense of the terms. Only 50 years prior to the outbreak of their war, Athens and Sparta fought on the same side against the Persians, most famously in the Battle of Thermopylae. The similarities between Sparta and Athens suggest that a shift in power may not always activate fear. An important empirical implication follows from this. Empirical tests of shifting power and war need to pay careful attention to the relevant scope conditions for assessing the hypothesis.
Disjunction Between Relative Power and Preferences
Another argument connecting shifting power to war is disjunction, specifically a disjunction between the most powerful actor and the preferences that dominate in the relevant geographic region. Shifting power between a declining and rising power creates the disjunction. There are several disjunction theories, but three are worth particular attention: power transition theory, long cycle theory, and power-preference disparity theory.
Power transition theory has four key components (Organski, 1958; Organski & Kugler, 1981). First, the international system is hierarchical, whereas international politics are primarily anarchic. Hierarchy here means that one country is more powerful than others and that the dominant country has a disproportionate influence on the institutions that govern the international system. Lemke (2002) extends power transition theory to explain regional hierarchies and wars. Second, some countries are more satisfied with the international status quo than others. Viewed from the opposite perspective, some major powers are more dissatisfied with the status quo than others. Why some countries are more satisfied than others is not clear in the theory. Third, power is primarily conceived of in economic terms. The dominant country is the one with the largest gross domestic product. It is assumed that all major powers convert their resources into military power, that they convert similar percentages of their resources, and that they are similarly efficient. Thus, the state with the largest economy will naturally be the one with the most powerful military. Fourth, major power war occurs when there is a power transition between the dominant country and a dissatisfied major power contender. Which specific country is more likely to initiate war and the precise timing of a war are not discussed here. What matters is that war is most likely to occur when major powers are experiencing a power transition, which means they have roughly equal power, and the rising power is dissatisfied with the status quo.
Long cycle theory is primarily a theory of global leadership, economic growth, and technological innovation, but it has implications for major power war (Modelski & Thompson, 1989). There are five key components of long cycle theory. First, economic growth occurs in long waves that are usually about 50 years long. Second, long waves are a function of technological innovation in leading economic sectors. Third, after the development of a new leading technology, relative economic decline sets in. Fourth, for the last 500 or so years (many long cycle analyses begin in 1494) the country with the world’s leading technology has also had the world’s most powerful navy. Fifth, war is a method for choosing the system leader. War occurs because of relative decline or shifting power. On the one hand, the focus on the system leader and shifting power makes long cycle theory similar to power transition theory. On the other hand, long cycle theory does not posit a necessary relationship between shifting power and dissatisfaction with the system leader. Long cycle theory allows for nonmilitary avenues of choosing the system leader.
Power-preference disparity theory is a theory of war and not just war between major powers or war for system leadership. The core argument is straightforward: war is most likely to occur when relative power and the relative distribution of benefits is out of sync (Powell, 1999). Powell (1999) does not explain why such a disparity occurs, but we can stipulate that such a disparity is most likely to occur when there is a shift in power without a commensurate shift in benefits or vice versa.
There are several other disjunctive theories of war; most of these are typically described as rise-and-decline arguments. The most prominent works are Gilpin’s (1981) War and Change in World Politics, Kennedy’s (1987) The Rise and Fall of the Great Powers, Wallerstein’s (1983) Three Instances of Hegemony in the World Economy, and Goldstein’s (1988) Prosperity and War in the Modern Age. Pollins (1996) offers an excellent summary of much of this research.
The central theoretical limitation of the disjunction mechanism is that it does not explicitly explain why the actors are not able to reach a peaceful settlement. If war is costly for each, why do the declining and rising state not work out an agreement to avoid war? Similarly, why does the rising state not peacefully yield? Why are any promises from the rising state not credible? Is there anything the rising state could do to avoid a preventive attack? To answer these questions, the credible commitment mechanism, which subsumes the disjunction mechanism but clarifies key causal processes, is considered.
Credible Commitment Mechanism
Credible commitment problems are one of the primary explanations of war highlighted by the bargaining model of war. The canonical conceptual example of a credible commitment problem is preventive war between a declining and a rising state. In the bargaining model of war, however, credible commitment problems are not triggered by shifts in power but by expected shifts in power (Fearon, 1995; Powell, 2006).4 To be more precise, in the bargaining model, a credible commitment problem is only triggered by an expected, large, and rapid shift in relative power. The bargaining model answers the two questions raised previously: Do all rising powers want to change the status quo? Why is compromise not possible? Regarding the status quo, in the bargaining context the actors’ preferences are assumed to oppose each other. The focus on a large and rapid shift explains why compromise is not possible. Under a gradual shift in power, the declining state can give the rising state more benefits so that war is not immediately preferred to peace and allowing the actors time to work out a deal to forestall the relative change in power. Thus, a gradual shift in power allows peace under two conditions. First, the declining state has to allocate more benefits to the rising state. Second, the declining state has to increase its relative power to reverse the adverse shift. The second condition is the real challenge. Thinking about it from the perspective of the rising state may make it clearer. In a shifting power context war can be avoided if the rising state transfers the sources of its military power to the declining state. With a large and rapid shift in power, the rising state has no incentive to transfer the sources of its military power. It is one thing, for example, to transfer resources to stop a preventive attack. It is another thing, and an irrational choice, to transfer resources once one has become stronger than one’s adversary. Still, the bargaining model of war highlights an important condition that limits credible commitment problems and the outbreak of war. This condition is the cost of war. When war is expected to be too costly, even an expected, large, and rapid shift in power will not make war rational. As much as anything, bargaining breakdown and the onset of war is about war having an expected small cost.
How Power Shifts Occur
To better understand the causal relationship between power shifts and war, it is necessary to understand how power shifts occur. Power shifts occur in one of two ways. First, one country may acquire more or better weapons than its competitor. Second, one country may acquire more or more powerful allies than its competitor. It is also possible for a country to do both simultaneously, that is, acquire more weapons and more allies. How does this happen?
There are three general reasons for differential internal power growth. First, there may be differences in relative economic performance, an argument that is central to long cycle theory. Long cycle theory highlights the role of technological innovations in leading economic sectors (Thompson, 1990). For example advances in electronics and aerospace helped make the United States the world’s dominant economy in the 20th century. While technological innovations propel an economy forward, they also lead to relative decline over time, unless that actor develops a new lead technology. Relative economic decline sets in because of technological diffusion, increased competition from others trying to imitate and improve upon the dominant technology, excess supply, and diminishing marginal returns (Thompson, 1990). With economic growth happening in waves, relative economic decline is likely to occur, and with relative economic decline comes relative military decline, that is, a shift in power. Other shifting power arguments highlight different reasons for relative economic decline. For Gilpin (1981), the primary reason for shifting power is technological diffusion. For Kennedy (1987), shifting power occurs because of military overstretch and bureaucratic sclerosis.
Second, differential internal power growth may occur because of changes in domestic state capacity, which is the core perspective presented by power transition theory (Organski, 1958). Power transitions happen because countries industrialize at different times. Over the last 200 years, industrialization has been associated with changes in state capacity. One reason for a dramatic change in state capacity and a resulting change in industrialization and military power is revolution. Skocpol (1979), for example, contends that social revolutions lead to “a centralized mass incorporating state with enhanced great-power potential in the international arena” (p. 162). The dramatic differences in military power and effectiveness pre and post revolution in France, Russia, and China underscore how domestic changes influence military power (Carter et al., 2012). Revolution is only one reason why rapid industrialization occurs. To understand power shifts, it is important to remember that countries industrialize at different times and different rates.
Third, power shifts may occur because one country chooses not to invest as much in its military as others. Since the end of World War II, both Japan and Germany have chosen to limit their military power in a way that is not typical for countries with a large and advanced economy. For example, in 2015, Japan and Germany had the world’s fifth and sixth largest economies, but they ranked eighth and ninth in aggregate military expenditures and they had smaller defense burdens than any other country in the top 15 for military spending (SIPRI, 2015). These choices stem from several factors, especially the U.S. military alliance and the memory of World War II. Great Britain’s decision to limit its military spending in the 1930s is another example of shifting power resulting from political choices. Powell (1996) contends that Britain’s choice stemmed from uncertainty about Germany’s intentions. Whatever the reason, a shift in power occurred between Great Britain and Germany primarily because of political choices and not because of major changes in technology or state capacity.
Shifts in power may also occur because of alliance formation. One of the primary purposes of an alliance is to aggregate military capabilities. The Franco-Russian War of 1894 was meant to counter the military power of Germany and the Triple Alliance. Taylor (2001 ) suggests that the absence of an alliance between Great Britain and the Soviet Union contributed to the outbreak of World War II. Kim’s (1991) alliance transition theory incorporates alliances into power transition theory, arguing that war is most likely not when there is power parity associated with a power transition but power parity associated with an alliance transition and dissatisfaction with the status quo. By alliance transition, Kim means the difference in power between country A and its allies versus country B and its allies. In brief, the role of third parties is often central to shifting power explanations of war.
Several key points are important in this brief review of the primary causal mechanisms linking shifting power to war. First, shifting power explanations of war are more liberal than realist ones. Legro and Moravcsik (1999) posit that the best demarcation between liberal and realist theories is that the latter do not allow preferences to vary across units while the former does. Most of the shifting power explanations of war assume that preferences are independent of power and that they vary across countries. Second, two of the most popular explanations for war, deterrence failure and diversionary impulses, play little role in shifting power theories. To be fair, each may be viewed as affecting the cost of war, a critical parameter explicitly in credible commitment explanations and implicitly in all other explanations. Deterrence failure may also be relevant if a power shift occurs because one country consciously limits its military power despite a growth in an adversary’s power. Still, the dominant aspects of diplomatic discourse in the context of shifting power are preventive war, sanctions to forestall an adversary’s growth, and appeasement. Third, each theory of shifting power and war identifies scope conditions. For example, if war is too costly, then shifting power will not result in war or if there is no dissatisfaction, then a power transition is unlikely to lead to war. Fourth, it is important to understand how power shifts occur to understand the diplomatic tools and calculus in a particular context. For example, preventive war may seem more beneficial when a power shift stems from revolutionary domestic change than from the development of a new leading technology. Combining aspects of the third and fourth implications leads to a fifth: alliances and expectations of third parties are critical to decisions for war and peace. Alliances affect not only whether a shift in power occurs between coalitions but also a country’s expected cost of war.
Statistical Research on Power Shifts and War
Despite its theoretical prominence, there is little statistical research on power shifts and war. Central to statistical research is the testing of hypotheses to rule out chance relationships. While statistical research increases the confidence that there really is a relationship between a hypothesized cause and an effect, qualitative research also provides valuable insights. Notable case-study research regarding shifting power and war includes Kennedy (1987), Van Evera (1999), and Copeland (2000). Notwithstanding the value of the case study work, statistical research is the focus of this article. The statistical research that exists on this topic generally shows a correlation between power shifts and war, but there is much work to be done before a causal relationship can be concluded. Two areas for improvement are research designs that better incorporate scope conditions and addressing endogeneity concerns.
Much of the statistical empirical work on power shifts and war is part of either the power transition or long cycle theory research programs. Next, some prominent research in this area is outlined according to three questions. First, are the measures valid indicators of the key concepts? Absent valid measures, hypothesis tests cannot be conclusive. Furthermore, the wrong relationship may be tested or the measures used may introduce endogeneity into the testing. Second, does the research design address endogeneity concerns? There are two central endogeneity concerns: the omission of observable confounders and the omission of unobservable or difficult to observe confounders? The latter stems from the fact that the causal variable of interest, shifting power, is not randomized across cases. When the causal variable is not randomized, it is very difficult to ensure that the cases with the posited cause are the same as the cases without the posited cause, except for the posited cause. For example, none of the empirical research on shifting power and war fully deals with the following confounders: expectations about resolve, expectations about the cost of war, expectations about the role of third parties, variations across political institutions, and individual-level behavioral traits. Third, does the research design address scope conditions? Scope conditions are the conditions under which a hypothesis is expected to hold. For example, in the absence of sunlight, the addition of fertilizer is not expected to make plants grow. One way to incorporate scope conditions is to include interaction terms. A second way is to stratify the sample based on the relevant conditions.
Power Transition Theory
Organski and Kugler (1981) place more focus on power parity than on power transitions. They examine the relationship between power parity and war in a small number of great power dyads between 1870 and 1970. Parity is a dichotomous variable equal to one if the weaker state’s power is at least 80% of the stronger state’s power. A power transition occurs if one state’s power surpasses the other state’s within the last 20 years. They find that power parity is positively correlated with the onset of war. Notwithstanding its relationship to power transition theory, this research is noteworthy for showing a positive relationship between a balance of power/power parity and war. This concept is inconsistent with the popular view that a balance of power promotes peace. While this is an important piece of research, it does not fare well with regard to the five questions previously outlined. It does not include control variables or address the non-randomness of power transitions nor does it include a measure of dissatisfaction.
Other notable work on power transition theory includes Kim (1991, 1992), Kim and Morrow (1992), Lemke and Werner (1996), and Lemke (2002). Kim (1991) expands the empirical analysis to all great power dyads between 1816 and 1975, and more importantly incorporates a measure of satisfaction as well as measuring power in terms of alliance coalitions. The incorporation of the latter has led to this being labeled alliance transition theory (Kim & Gates, 2016). Kim (1991) differs from Organski and Kugler (1981) in other ways. Instead of using gross national product to measure power, as Organski and Kugler do, Kim (1991) uses the Correlates of War composite capabilities index (Singer, Bremer, & Stuckey, 1972). Kim measures power transitions in the same way as Organski and Kugler but also includes a measure of shifting power in the empirical model. Shifting power is measured as the difference in the rate of power growth between ten-year periods. This shifting power variable is not statistically significant in Kim’s analysis. The measure of satisfaction is alliance portfolio similarity. The idea is that countries that have an alliance portfolio that is similar to the system leader’s portfolio have more satisfaction than countries with different alliance structures. Kim (1992) extends the empirical analysis back to 1648! These are notable contributions, though the findings are only correlational because of the omission of other likely confounders such as regime differences and expectations of resolve.
Kim and Morrow (1992) add a new parameter to a country’s war calculus: risk attitude. They argue that power shifts and dissatisfaction are not the only factors that matter. A challenging country’s degree of risk acceptance also affects the choice for war. Their measure of risk acceptance is drawn from Morrow (1987), who builds on work by Bueno de Mesquita (1985). The basic idea motivating the measure is that countries that form alliances to increase their security are more risk averse. In other words, the most risk-acceptant country is one with no alliances and not that much power. While they address concerns about one omitted variable, others are not accounted for here.
The work of Lemke and Werner (1996) is noteworthy for employing a new measure of dissatisfaction: military buildups. The argument is that countries experiencing a military buildup are more likely to be dissatisfied with the status quo than countries not experiencing a significant buildup. This is a valuable contribution because it more directly measures the concept of interest. However, this research also suffers from many of the same endogeneity concerns mentioned previously.
The work of Lemke (2002) is the best piece of scientific research on power transition theory. First, he extends power transition theory to account for regional hierarchies. Thus, the modified theory can explain both major wars and minor regional ones. Second, and more importantly for this article, his empirical analysis includes multiple control variables. This greatly increases the confidence one can have in the findings. However, other potential omitted variables are not accounted for here, including economic interdependence, participation in intergovernmental organizations, territorial disputes, and a measure of risk orientation to name a few. Third, this research contains an excellent discussion of measurement of the key variables. What is missing is more sensitivity analysis with alternative measures. For example, what happens if the Kim (1991) measure of dissatisfaction is used? Are the results sensitive to how power and power parity are measured? Lemke includes some sensitivity analysis with different measures, but not all reasonable measures are examined. Notwithstanding its many merits, Lemke’s research, like the other research discussed here, does not fully account for the nonrandom assignment of the treatment. Some likely confounders include expectations about the cost of war, expectations about the role of third parties, variations across political institutions, and individual-level behavioral traits.
Long Cycle Theory
Long cycle theory is another major research program examining the relationship between shifting power and war. At its core, long cycle theory is a theory of global leadership. The world leadership cycle can be categorized in four phases: world power, delegitimation, deconcentration, and global war (Modelski & Thompson, 1987). As the phase labels suggest, the theory also aims to explain the occurrence of global war. The ordering of the phases is not meant to imply that the occurrence of war increases linearly throughout the cycle. Rather, long cycle theory expects that military conflicts are more common in the second than in the third phase. A significant corpus of empirical research finds evidence consistent with this expectation as well as a number of related hypotheses that flow from long cycle theory. Thompson and Rasler (1988), for example, examine nearly 500 years of data on systemic capability concentrations and warfare. Their analysis shows that long cycle theory best fits the periodic occurrence of global war (see also Modelski & Thompson, 1989, for a review of related research).
Like power transition theory, long cycle theory expects that war will be less frequent in unipolar systems than bipolar or multipolar ones. In contrast, Waltz’s neorealism expects bipolarity to be the most stable system. Thompson (1986) examines the relationship between polarity and war occurrence between 1484 and 1993 and finds that unipolarity is the most stable type of system. Central to this research is its measurement of polarity in terms of the concentration of blue water naval capability. This is a valid measure for long cycle theory, but as Thompson notes Waltz’s concept of power is not the same. In addition to questions of measurement, this research does not incorporate control variables, though it seems unlikely that the relationship between capability concentration and global war is likely to weaken significantly. After all, power transition theory finds a consistent correlation between power preponderance and peace.
Thompson (1992) augments long cycle theory and presents additional empirical evidence on the occurrence of war that is consistent with the theory. As previously noted long cycle theory has a global focus. Are there regional cycles? If so, how do they relate to the global cycle? Ludwig Dehio (1962) advanced a thesis about regional cycles. Thompson (1992) integrates Dehio’s arguments about the rise and decline of European regional powers into long cycle theory on the rise and decline of global powers. The essay labels this augmented long cycle theory. The novel expectation of the augmented theory is that regional and global power cycles are not in sync. Changes in the concentration of global power are measured in terms of naval power, and changes in the concentration of Western European regional power are measured in terms of changes in land power. Accordingly, this research presents a new measure of land power, namely the size of a country’s army in terms of personnel. Consistent with the augmented long cycle theory, Thompson finds that for nearly 500 years “the rhythms of continental and sea power are not in phase with one another” (p. 142).
Shifting power is the key process leading to major power war in long cycle theory as well as power transition theory. Augmented long cycle theory clarifies the difference between the two perspectives. In power transition theory, war occurs when there is a power transition and the rising power is dissatisfied with the international status quo. In augmented long cycle theory, war is most likely to occur when there is a power shift between a regional leader and the global leader. From the perspective of power transition theory, a regional leader is dissatisfied with the status quo. Augmented long cycle theory indicates why this is the case. Regional leaders focus on land power, while global leaders focus on sea power. “The most dangerous challenger tends to be a state that either has already ascended, or aspires to ascend, to a dominating regional position through the development of its land power (Thompson, 1996, p. 173). The different military emphases lead to different views on trade, conquest, and possibly regime type. Thompson (1996) shows that these regional-global power transitions are correlated with major power war.
Rasler and Thompson (2001) offer additional tests of augmented long cycle theory. They argue that the dissynchronization between regional and global power not only accounts for the occurrence of global war but also the war aims of major powers, which leads them to create a new measure for war aims. Noting the complexity of discerning war aims, they suggest that a binary measure, limited or unlimited, is a useful starting place. Unlimited war aims are rare, with only Hitler, Napoleon, Louis XIV, and Philip II having such grand aims. They find strong support for the expectation that structural contexts, specifically the dissynchronization between regional and global power concentration, facilitate war and unlimited war aims (Rasler & Thompson, 2001, p. 75).
Reuveny and Thompson (1999) also test additional implications of long cycle theory. They examine whether political and military factors lead to economic growth and innovation or if the reverse expectation has more support. They find that leading economic sector variables “Granger cause” military-political variables. Economic innovation, then, is especially central to global leadership.
As this brief discussion illustrates, long cycle theory is a fruitful and robust research program. The empirical relationship between regional-global power transitions and war, owing to a dissynchronization in the concentration of land and sea power, is consistent with long cycle theory. This research program has also found empirical support for a number of additional hypotheses. One concern with some parts of this research program is that the empirical analyses include few control variables. The work of Rasler and Thompson (2005) is an exception. They include systemic trade in their analysis. In addition to finding that war is associated with contestation for global leadership, they find that global leadership is associated with increased trade, just as long cycle theory expects.
Pollins (1996) also extends the long cycle theory, arguing that the long wave of economic change and global leadership cycles affect the conflict behavior of all countries. He finds that a model including cyclic phases better explains the amount of interstate conflict initiation over the last 200 years than a model that omits such phases. However, before a causal relationship is accepted, some research design concerns need to be better addressed.
First, dating of different phases varies across these theories, which raises reliability concerns. Second, it is not clear why the cycles need four phases instead of two or three. Indeed, Gilpin (1981) suggests that there are only two phases: hegemony and not hegemony. Organski and Kugler’s (1981) work also suggests that there are only two phases: preponderance and parity. Future research should work to validate the number of phases. Third, in some of this research, World War II is considered an extension of World War I. This seems ad hoc. Fourth, most of this research does not include control variables. Potential confounders include world trade (Rasler & Thompson, 2005, is an exception), the distribution of different types of political regimes, the number of states in the system, and measures of the relative distribution of benefits among the major powers. Fifth, while long cycle theory is a systemic theory, in the future it would be worthwhile to provide clearer microfoundations to justify the choices that actors make and to create multilevel expectations.
A third body of scientific research on shifting power and war is motivated by rational choice and preventive war theories.
Werner (1999) aims to test Powell’s (1999) formal model linking war to the disparity between power and preferences. To do so, she argues that a disparity between relative power and the relative distribution of benefits is more likely when relative power is changing. She assumes that the distribution of relative benefits changes more slowly, thereby justifying a focus on shifting power. This is the first study to analyze changes in relative power for all dyads for the period 1816–1992. Werner also employs a different measure of shifting power; she examines the absolute difference in the annual change in power growth rates, with power operationalized as the Correlates of War composite capabilities index. The dependent variable in this research is the onset of a militarized interstate dispute (MID), and Werner finds that shifting power is statistically and positively related to a MID. While the analysis includes a number of control variables, it remains plausible that expectations of conflict lead to a change in power and conflict itself.
Reed et al. (2008) also aim to test Powell’s argument. Unlike Werner, they include a measure of the distribution of benefits. In addition, where Werner measures changes in power, they focus on the disparity between the distribution of power and distribution of benefits. Conflict is most likely, they argue, when there is power parity but an imbalance in the distribution of benefits, and this is precisely what they find. Carroll and Kenkel (forthcoming) create a new measure for the expected outcome of a militarized dispute. They also find support for Powell’s theory that militarized conflict is less likely when the distribution of power and distribution of benefits is nearly equal, but they reach a different conclusion. However, unlike Reed et al. (2008), they find little support for the power parity expectation: power parity is not associated with an increase in conflict. This research further underscores the importance of creating valid measures of key concepts and testing key propositions with multiple plausible indicators. All of the research discussed in this article would benefit improved measures and additional tests.
Lemke (2003) is the first large-sample study examining the preventive war motivation. A country has a preventive motive when it is declining in power relative to another country. Specifically, the preventive motive variable is the regression coefficient of a state’s share of dyadic power regressed against time, the previous 20 years (Lemke, 2003, p. 278). The dyads in this analysis are all dyads that went to war between 1816 and 1992 and a random sample of politically relevant dyads that did not go to war. He does not find a statistically significant relationship between the preventive motivation for war (shifting power) and war onset.
Bell and Johnson (2015) note that credible commitment problems associated with the preventive war motivation arise because of expected power shifts, not shifts in power in the recent past. To test this hypothesis, they create a measure of the expected change in relative power. Instead of looking at a country’s recent military spending patterns and forecasting them forward, they regress military power for all countries through year t − 1 on a variety of country attributes and then forecast military power based off the regression coefficients and country attributes in year t. They also examine three- and five-year forecasts, and they find that the expected shift in power variable is positively correlated with war initiation. Like other researchers, they control for a host of relevant potential confounders, yet it seems likely that variables are still omitted. Furthermore, they do not include measures for territorial disputes or other factors related to the expectation of conflict.
Power Shifts and War: Causal or Correlational?
Do power shifts cause war? On the one hand, a large body of theoretical work says that power shifts or expected power shifts cause war. This work includes power transition theory, cyclic theories of war, preventive war arguments, and the bargaining model of war. On the other hand, extant research on power shifts and war has not fully addressed endogeneity concerns and relevant scope conditions. Perhaps the strongest empirical support for this expectation is found in research on cyclic theories of rise and decline (see Modelski & Thompson, 1989). These cyclic theories are at the system level of analysis. If the relevant actors have agency, however, then the causal processes in these theories are theoretically incomplete.
The central causal challenge stems from the fact that power shifts are not randomly assigned; as a result, it is very difficult to fully account for all confounders and identify a precise functional form for a regression model. Some potential confounders are very difficult to observe. For example, it is likely that hostile intent or expectations of war drive a change in power and the subsequent decision to enter a war. The best approach for increasing internal validity is to conduct an experiment. Given concerns about external validity, experiments in international relations have been rare. Instead of viewing experiments as a cure-all solution, they should be viewed as complementing observational research. Tingley (2011) offers a blueprint for experiments related to the shifting power. While his experiments focus on the shadow of the future, it would be possible to modify the experimental protocol to more fully evaluate shifting power concerns.
In observational research there are several strategies one can employ to improve the credibility of one’s inferences (see Angrist & Pischke, 2010), yet surprisingly, none of these have yet been implemented in the context of shifting power and war. First, a fixed-effects regression model can be created. Fixed effects account for all time-invariant omitted variables. Nevertheless, in most cases the likely omitted variables are not time invariant. Second, a regression discontinuity model can be evaluated. The challenge here is to identify a forcing variable, that is, an exogenous reason for the discontinuity in the treatment. Third, a difference-in-difference model can be explored. All of these have potential for this research area. The idea would be to compare the difference in changes in power between likely and unlikely belligerents. If hostile intent is the real cause, then we expect a significant difference in difference. But this idea assumes that we can correctly classify dyads as likely belligerents. Fourth, an instrumental variable model can be created. These are straightforward models to estimate, though identifying valid instruments is difficult. At the same time, researchers are increasingly identifying valid instruments in other contexts.
Future empirical research also has to give greater attention to relevant scope conditions, measurement, and variables related to the strategic context. From the perspective of bargaining theory, the effects of shifting power are conditional on the expected cost of war. Satisfaction with the status quo is likely a function of political similarity, yet aspect this has received little attention. Shifting power is only expected to increase the probability of war for certain pairs of states, but more work needs to be to identify the relevant sample. Does the presence of a territorial dispute indicate dissatisfaction or a disparity in benefits? Are the relevant dyads great powers, minor powers, politically relevant dyads, rivals, or some other set? Finally, most extant research does not consider expectations of third-party behavior.
Angrist, J. D., & Pischke, J.-S. (2010). The credibility revolution in empirical economics: How better research design is taking the con out of econometrics. Journal of Economic Perspectives, 24(2), 3–30.Find this resource:
Bell, S. R., & Johnson, J. C. (2015). Shifting power, commitment problems, and preventive war. International Studies Quarterly, 59(1), 124–132.Find this resource:
Bueno De Mesquita, B. (1985). The war trap revisited: A revised expected utility model” American Political Science Review, 79(1), 156–177.Find this resource:
Carroll, R., & Kenkel, B. (2017). Forthcoming. Prediction, proxies, and power. American Journal of Political Science.Find this resource:
Carter, J., Bernhard, M., & Palmer, G. (2012). Social revolution, the state, and war how revolutions affect war-making capacity and interstate war outcomes. Journal of Conflict Resolution, 56(3), 439–466.Find this resource:
Copeland, D. C. The origins of major war. Ithaca, NY: Cornell University Press, 2000.Find this resource:
Dahl, R. A. (1957). The concept of power. Behavioral Science, 2(3), 201–215.Find this resource:
Dehio, L. (1962). The Precarious Balance: Four Centuries of the European Power Struggle, translated Charles Fullman. New York: Vintage Books.Find this resource:
Fearon, J. D. (1995). Rationalist explanations for war. International Organization, 49(3), 379–414.Find this resource:
Gilpin, R. (1981). War and change in world politics. New York: Cambridge University Press.Find this resource:
Goldstein, J. S. (1988). Long cycles: Prosperity and war in the modern age. New Haven: Yale University Press.Find this resource:
Howard, M. (1984). The causes of wars. Wilson Quarterly (1976–), 8(3), 90–103.Find this resource:
Jervis, R. (1976). Perception and misperception in international politics. Princeton, NJ: Princeton University Press.Find this resource:
Jervis, R. (1988). War and misperception. Journal of Interdisciplinary History 18(4), 675–700.Find this resource:
Kennedy, P. (1987). The rise and fall of the great powers. New York: Vintage.Find this resource:
Kim, W. (1991). Alliance transitions and great power war. American Journal of Political Science, 35(4), 833–850.Find this resource:
Kim, W. (1992). Power transitions and great power war from Westphalia to Waterloo. World Politics, 45(1), 153–172.Find this resource:
Kim, W., & Morrow, J. D. (1992). When do power shifts lead to war? American Journal of Political Science, 36(4), 896–922.Find this resource:
Kim, W., & Gates, S. (2016). Power transition theory and the rise of China. International Area Studies Review, 18(3), 219-226.Find this resource:
Legro, J. W., & Moravcsik, A. (1999). Is anybody still a realist? International Security, 24(2), 5–55.Find this resource:
Lemke, D. (2002). Regions of war and peace. Cambridge: Cambridge University Press.Find this resource:
Lemke, D. (2003). Investigating the preventive motive for war. International Interactions, 29(4), 273–292.Find this resource:
Lemke, D., & Werner, S. (1996). Power parity, commitment to change, and war. International Studies Quarterly, 40(2), 235–260.Find this resource:
Levy, J. (1987). Declining power and the preventive motivation for war. World Politics, 40(1), 82–107.Find this resource:
Mearsheimer, J. J. (2001). The tragedy of great power politics. New York: W. W. Norton.Find this resource:
Modelski, G., & Thompson, W. R. (1987). Testing cobweb models of the long cycle. In G. Modelski (Ed.), Exploring long cycles. Boulder, CO: Lynn Rienner.Find this resource:
Modelski, G., & Thompson, W. R. (1988). Seapower in global politics, 1494–1993. Seattle: University of Washington Press.Find this resource:
Modelski, G., & Thompson, W. (1989). Long cycles and global war. In M. Midlarsky (Ed.), Handbook of war studies (pp. 23–54). Ann Arbor: University of Michigan Press.Find this resource:
Morrow, J. D. (1987). On the theoretical basis of a measure of national risk attitudes. International Studies Quarterly, 31(4), 423–438.Find this resource:
Organski, A. F., & Kugler, J. (1981). The war ledger. Chicago: University of Chicago Press.Find this resource:
Organski, A. F. K. (1958). World politics. New York: Knopf.Find this resource:
Pollins, B. M. (1996). Global political order, economic change, and armed conflict: Coevolving systems and the use of force. American Political Science Review, 90(1), 103–117.Find this resource:
Powell, R. (1996). Uncertainty, Shifting Power, and Appeasement. American Political Science Review, 90(4), 749–764.Find this resource:
Powell, R. (1999). In the shadow of power: States and strategies in international politics. Princeton, NJ: Princeton University Press.Find this resource:
Powell, R. (2006). War as a commitment problem. International Organization, 60, 169–203.Find this resource:
Rasler, K., & Thompson, W. (2001). Malign autocracies and major power warfare: Evil, tragedy and international relations theory. Security Studies, 10(3), 46–79.Find this resource:
Rasler, K., & Thompson, W. (2005). War, trade and the mediation of systemic leadership. Journal of Peace Research, 42(3), 251–269.Find this resource:
Reuveny, R., & Thompson, W. (1999). War, system leadership and economic innovation: The United States case. Journal of Conflict Resolution, 43(5), 570–595.Find this resource:
Reed, W., Clark, D., Nordstron, T., & Hwang, W. (2008). War, power, and bargaining. Journal of Politics, 70(4), 1203–1216.Find this resource:
Singer, J. D., Bremer, S., & Stuckey, J. (1972). Capability distribution, uncertainty, and major power war, 1820–1965. In B. Russett (Ed.), Peace, war, and numbers (pp. 19–48). Beverly Hills, CA: SAGE.Find this resource:
Skocpol, T. (1979). States and social revolutions: A comparative analysis of France, Russia and China. New York: Cambridge University Press.Find this resource:
Small, M., & Singer, J. D. (1982). Resort to arms: International and civil wars, 1816–1980. New York: SAGE.Find this resource:
Stockholm International Peace Research Institute. (2015). Trends in world military expenditures.
Taylor, A. J. P. (1954). The struggle for mastery in Europe, 1848–1918. New York: Oxford University Press.Find this resource:
Taylor, A. J. P. (2001). The origins of the second world war. London: Penguin.Find this resource:
Thompson, W. R. (1990). Long waves, technological innovation, and relative decline. International Organization, 44(2), 201–233.Find this resource:
Thompson, W. R. (1986). Polarity, the long cycle, and global power warfare. Journal of Conflict Resolution, 30(4), 587–615.Find this resource:
Thompson, W. R. (1992). Dehio, long cycles, and the geohistorical context of structural transitions. World Politics 45(October), 127–152.Find this resource:
Thompson, W. R. (1996). Balances of power, transitions, and long cycles. In J. Kugler & D. Lemke (Eds.), Parity and war: A critical reevaluation of the war ledger. Ann Arbor: University of Michigan Press.Find this resource:
Thompson, W. R., & Rasler, K. (1988). War and systemic capability reconcentration. Journal of Conflict Resolution, 32(2), 335–366.Find this resource:
Thucydides. (1954). History of the Peloponnesian War (Rex Warner, Trans.). New York: Penguin Books.Find this resource:
Tingley, D. H. (2011). The dark side of the future: An experimental test of commitment problems in bargaining. International Studies Quarterly, 55(2), 521–544.Find this resource:
Van Evera, S. (1999). Causes of war. Ithaca, NY: Cornell University Press.Find this resource:
Wallerstein, I. (1983). Three instances of hegemony in the history of the world economy. International Journal of Comparative Sociology, 24, 100–108.Find this resource:
Werner, S. (1999). Choosing demands strategically the distribution of power, the distribution of benefits, and the risk of conflict. Journal of Conflict Resolution, 43(6), 705–726.Find this resource:
(1.) Some research on shifting power and war examines weapons stocks, specifically naval arsenals. To compare navies, this research measures naval power in terms of tonnage and the number of capital ships countries have (Modelski & Thompson, 1988).
(2.) More broadly, research on war has identified five general causal mechanisms. In addition to the two listed above, there are informational, domestic-political, and risk mechanisms. Domestic-political is itself multifaceted and includes diversionary, institutional, economic, and cultural explanations. In many, if not most, cases it is likely that more than one mechanism is operating. Fearon (1995, p. 408), for example, notes that an information problem likely accompanies a credible commitment problem.
(4.) Fearon (1995) identifies three types of credible commitment problems: preemptive war and offensive advantages, preventive war, and contests over strategic territory. In addition to the three large and rapid changes in power that Fearon discusses, Powell (2006) identifies three other types of credible commitment problems: a large and rapid shift in political preferences, issue indivisibility, and maintaining the status quo is too costly.