Do Sunspots Cause War?
Find the working paper here.
Using a simple formal model, I argue for extrinsic uncertainty, archetypified as a sunspot, as a (quasi-)rationalist explanation for war. A sunspot is a random variable that has no effect on any of the fundamentals of the setting: not the distribution of resources, nor the countries' preferences, nor the technology of militarization, nor the technology of conflict. Nevertheless, war can transpire with sunspots and not transpire without sunspots, so that sunspots cause war in a strictly counterfactualist sense. The phenomenon works through the mechanism of self-fulfilling prophecy: some of the countries think that sunspots matter, so they wind up mattering. The result obtains even when the countries are armed with objective and correct probabilities about sunspots and rational expectations about what happens within each sunspot-realized world. An empirical test confirms the result.
Using a simple formal model, I argue for extrinsic uncertainty, archetypified as a sunspot, as a (quasi-)rationalist explanation for war. A sunspot is a random variable that has no effect on any of the fundamentals of the setting: not the distribution of resources, nor the countries' preferences, nor the technology of militarization, nor the technology of conflict. Nevertheless, war can transpire with sunspots and not transpire without sunspots, so that sunspots cause war in a strictly counterfactualist sense. The phenomenon works through the mechanism of self-fulfilling prophecy: some of the countries think that sunspots matter, so they wind up mattering. The result obtains even when the countries are armed with objective and correct probabilities about sunspots and rational expectations about what happens within each sunspot-realized world. An empirical test confirms the result.
Prediction, Proxies, and Power (with Brenton Kenkel)
Find the paper and data at the DOE project's website.
Many enduring questions in international relations theory focus on power relations, so it is important that scholars have a good measure of relative power. The standard measure of relative military power, the capability ratio, is barely better than random guessing at predicting militarized dispute outcomes. We use machine learning to build a superior proxy, the Dispute Outcome Expectations score, from the same underlying data. Our measure is an order of magnitude better than the capability ratio at predicting dispute outcomes. We replicate Reed et al. (2008) and find, contrary to the original conclusions, that the probability of conflict is always highest when the state with the least benefits has a preponderance of power. In replications of 18 other dyadic analyses that use power as a control, we find that replacing the standard measure with DOE scores usually improves both in-sample and out-of-sample goodness of fit.
Many enduring questions in international relations theory focus on power relations, so it is important that scholars have a good measure of relative power. The standard measure of relative military power, the capability ratio, is barely better than random guessing at predicting militarized dispute outcomes. We use machine learning to build a superior proxy, the Dispute Outcome Expectations score, from the same underlying data. Our measure is an order of magnitude better than the capability ratio at predicting dispute outcomes. We replicate Reed et al. (2008) and find, contrary to the original conclusions, that the probability of conflict is always highest when the state with the least benefits has a preponderance of power. In replications of 18 other dyadic analyses that use power as a control, we find that replacing the standard measure with DOE scores usually improves both in-sample and out-of-sample goodness of fit.
War and Peace in the Marketplace
Read a working version of the paper.
I argue that competition is the driving force in international relations and that it manifests through exchange and conflict. Exchange plays out in markets, while conflict takes the form of war. Because of their shared heritage in competition, neither ought to be assigned primacy. These concerns necessitate the construction of a general formal framework of international politics. All states compete over all commodities, where both kinds of competition are made available to the states. Posited behavior takes the form of the dual-competitive equilibrium: a set of prices that clear markets and a set of military deployments that constitute mutual best response. I prove its existence under weak conditions, study its welfare properties, and demonstrate its ramifications for the study of international relations.
I argue that competition is the driving force in international relations and that it manifests through exchange and conflict. Exchange plays out in markets, while conflict takes the form of war. Because of their shared heritage in competition, neither ought to be assigned primacy. These concerns necessitate the construction of a general formal framework of international politics. All states compete over all commodities, where both kinds of competition are made available to the states. Posited behavior takes the form of the dual-competitive equilibrium: a set of prices that clear markets and a set of military deployments that constitute mutual best response. I prove its existence under weak conditions, study its welfare properties, and demonstrate its ramifications for the study of international relations.
Making Peace on the Cheap
Read a working version of the paper.
I develop a general equilibrium model of intervention, trade, and war where a third party revises the disputants' economic and military resources. After the adjustment, the states make a trade and determine whether to fight for the right to consume all of both commodities. I study a very general version of the model from the differentiable point of view, which provides a deeper geometric understanding of the canonical bargaining model of war and its relationship with the neoclassical theory of exchange. While peace is infinitely improbable, it is supportable as a Walrasian equilibrium if the third party chooses the appropriate intervention. A simple parametric case provides intuitions, and a case develops proof of concept.
I develop a general equilibrium model of intervention, trade, and war where a third party revises the disputants' economic and military resources. After the adjustment, the states make a trade and determine whether to fight for the right to consume all of both commodities. I study a very general version of the model from the differentiable point of view, which provides a deeper geometric understanding of the canonical bargaining model of war and its relationship with the neoclassical theory of exchange. While peace is infinitely improbable, it is supportable as a Walrasian equilibrium if the third party chooses the appropriate intervention. A simple parametric case provides intuitions, and a case develops proof of concept.
Civil War in the Shadow of Intervention
Read a working version of the paper.
I argue that expectations of third-party intervention shape the decision to start a civil war. To assess the claim, I develop a theoretical model of onset and intervention and a subsequent statistical approach built on a predictive measure of intervention. Intervention expectations improve model fit and out-of-sample performance relative to a standard model and a similar model in the literature. Not all interveners are the same: major powers, rivals, neighbors, and coethnics play unique and quantifiable roles. More interestingly, intervention expectations did not play a uniformly important role throughout the second half of the 20th Century, and the end of the Cold War appears to have brought on a new intervention-onset process. Remarkably, intervention expectations seem to explain away the well-known effect of population on the onset of violence. These results suggest that the shadow of intervention looms large in powerful but nuanced ways.
I argue that expectations of third-party intervention shape the decision to start a civil war. To assess the claim, I develop a theoretical model of onset and intervention and a subsequent statistical approach built on a predictive measure of intervention. Intervention expectations improve model fit and out-of-sample performance relative to a standard model and a similar model in the literature. Not all interveners are the same: major powers, rivals, neighbors, and coethnics play unique and quantifiable roles. More interestingly, intervention expectations did not play a uniformly important role throughout the second half of the 20th Century, and the end of the Cold War appears to have brought on a new intervention-onset process. Remarkably, intervention expectations seem to explain away the well-known effect of population on the onset of violence. These results suggest that the shadow of intervention looms large in powerful but nuanced ways.
Policy Devolution and Cooperation Dilemmas
(with David Konisky and Christopher Reenock)
Read a working version of the paper.
How do public good spillovers influence agent effort in an interconnected environment? We develop a model for a general class of problems in which agents share the provision of a public good and ‘downstream’ agents experience positive spillovers. We characterize endogenous agent effort as the result of the boundary quality within which agents operate, the size of their relative jurisdiction and their reward structure. We examine our findings with an empirical exercise centered on the regulation of 6,000 major water pollution sources under the U.S. Clean Water Act. We construct a novel dataset of U.S. state water pollution regional offices and use geographic information on agency jurisdictions, watershed boundaries, and elevation-induced streamflow to characterize the propensity for shirking in an agent’s local environment. Our empirical results support our expectations.
How do public good spillovers influence agent effort in an interconnected environment? We develop a model for a general class of problems in which agents share the provision of a public good and ‘downstream’ agents experience positive spillovers. We characterize endogenous agent effort as the result of the boundary quality within which agents operate, the size of their relative jurisdiction and their reward structure. We examine our findings with an empirical exercise centered on the regulation of 6,000 major water pollution sources under the U.S. Clean Water Act. We construct a novel dataset of U.S. state water pollution regional offices and use geographic information on agency jurisdictions, watershed boundaries, and elevation-induced streamflow to characterize the propensity for shirking in an agent’s local environment. Our empirical results support our expectations.
Using Item Response Theory to Improve Measurement in Strategic Management Research: An Application to Corporate Social Responsibility
(with David M. Primo and Brian Kelleher Richter)
Read a working version of the paper.
We introduce item response theory (IRT) to management and strategy research. IRT explicitly models firms’ and individuals’ observable actions in order to measure unobserved, latent characteristics. IRT models have helped researchers in other social science disciplines measure traits like political ideology, and they can help strategic management researchers improve their measures. To demonstrate this potential, we show how the method improves upon the de facto best measure of corporate social responsibility (CSR), the KLD Index, by creating IRT Responsibility scores from the underlying data along with estimates of how accurate these measures are. We show, for instance, that firms like Apple may not be as socially responsible as previously thought, while firms like Walmart may be more responsible than typically believed.
We introduce item response theory (IRT) to management and strategy research. IRT explicitly models firms’ and individuals’ observable actions in order to measure unobserved, latent characteristics. IRT models have helped researchers in other social science disciplines measure traits like political ideology, and they can help strategic management researchers improve their measures. To demonstrate this potential, we show how the method improves upon the de facto best measure of corporate social responsibility (CSR), the KLD Index, by creating IRT Responsibility scores from the underlying data along with estimates of how accurate these measures are. We show, for instance, that firms like Apple may not be as socially responsible as previously thought, while firms like Walmart may be more responsible than typically believed.
Rubberstamping (with Peter Bils and Lawrence Rothenberg)
Read a working version of the paper.
Informal, "notice-and-comment," rulemaking is the prototypical mechanism employed by U.S. regulators. However, agencies frequently claim themselves exempt from notice-and-comment, and courts typically agree. Hence, agencies face a strategic choice between rulemaking processes. To assess the implications, we study a series of models with possible exemption. In a baseline model with one group and a policy-agnostic agency, there is rubberstamping: the agency always successfully applies for an exemption. However, the group captures the agency, as its preferred rule is promulgated. When first competing groups are added and, then, the agency is assumed to have policy preferences, not only does notice-and-comment sometimes occur but there are conditions when a strong form of rubberstamping takes place by which the agency is exempt and implements its ideal policy. Our results are roughly consistent with empirical observations, provide a variety of testable observations, and offer insights into how the rulemaking structure impacts societal welfare.
Informal, "notice-and-comment," rulemaking is the prototypical mechanism employed by U.S. regulators. However, agencies frequently claim themselves exempt from notice-and-comment, and courts typically agree. Hence, agencies face a strategic choice between rulemaking processes. To assess the implications, we study a series of models with possible exemption. In a baseline model with one group and a policy-agnostic agency, there is rubberstamping: the agency always successfully applies for an exemption. However, the group captures the agency, as its preferred rule is promulgated. When first competing groups are added and, then, the agency is assumed to have policy preferences, not only does notice-and-comment sometimes occur but there are conditions when a strong form of rubberstamping takes place by which the agency is exempt and implements its ideal policy. Our results are roughly consistent with empirical observations, provide a variety of testable observations, and offer insights into how the rulemaking structure impacts societal welfare.
Costly Signaling in Autocracy
Read a working version of the paper.
Those who would revolt against an autocrat often face a dilemma: they would like to revolt if the ruler would respond with democratization, but they would prefer to concede if the ruler would choose instead to repress. Consequently, the autocrat must decide how to best signal his resolve in the hopes of deterring revolt. Using a simple signaling model, we find that rulers cannot meaningfully convey their type by transferring wealth to the citizenry; however, they can convey their type through escalation, where the resolved autocrat has a competitive advantage in escalation. The results illustrate a more general result in a broad class of signaling models: information transmission is only possible when the cost of the signal is smaller for the type that wants to distinguish himself.
Those who would revolt against an autocrat often face a dilemma: they would like to revolt if the ruler would respond with democratization, but they would prefer to concede if the ruler would choose instead to repress. Consequently, the autocrat must decide how to best signal his resolve in the hopes of deterring revolt. Using a simple signaling model, we find that rulers cannot meaningfully convey their type by transferring wealth to the citizenry; however, they can convey their type through escalation, where the resolved autocrat has a competitive advantage in escalation. The results illustrate a more general result in a broad class of signaling models: information transmission is only possible when the cost of the signal is smaller for the type that wants to distinguish himself.
Smooth International Systems
Read a working version of the paper.
Smoothness plays an important, though oft-implicit, role in the study of international relations. Many theories turn on smoothness, but there has yet to be much formal traction on the subject. This is especially problematic given the important role smoothness plays in the study of institutional design and the nature of our empirical models. To help remedy this problem, I develop a theory of smooth international systems. All the world's states engage in simultaneous exchange and conflict. The world's prices and the states' military deployments act in constant mutual reinforcement. Under very weak conditions, equilibrium prices and deployments form a set with a very particular topological structure; in particular, they are no different from Euclidean space. What is more, the maps linking data and prediction are nearly always continuous, too. All told, these results provide a general possibility result as to when the analyst may safely assume nicely-behaved equilibria.
Smoothness plays an important, though oft-implicit, role in the study of international relations. Many theories turn on smoothness, but there has yet to be much formal traction on the subject. This is especially problematic given the important role smoothness plays in the study of institutional design and the nature of our empirical models. To help remedy this problem, I develop a theory of smooth international systems. All the world's states engage in simultaneous exchange and conflict. The world's prices and the states' military deployments act in constant mutual reinforcement. Under very weak conditions, equilibrium prices and deployments form a set with a very particular topological structure; in particular, they are no different from Euclidean space. What is more, the maps linking data and prediction are nearly always continuous, too. All told, these results provide a general possibility result as to when the analyst may safely assume nicely-behaved equilibria.