What you don’t know can hurt you: Information,external transparency,and interstate conflict,1982

Abstract

This paper examines the role that information plays in conflict behavior by focusing on the external component of governmental transparency. Building from the crisis bargaining and the diversionary strategic conflict avoidance literature, I argue that governments that are more externally transparent are less likely to initiate conflict and are less likely to have the opportunity to use force for diversionary purposes. Using original data on the access that foreign media has to states, a large-N statistical analysis is implemented. The empirical analysis suggests that states that are more externally transparent are less likely to initiate conflict and that these same states are less likely to respond to domestic challenges with diversionary uses of force. This research demonstrates the important role of governmental transparency and provides support for some of the conclusions of crisis bargaining models and the strategic conflict avoidance argument in the diversionary literature.

Keywords

Crisis bargaining interstate conflict diversionary conflict strategic avoidance transparency

Introduction

On 23 September 2007, “60 minutes” conducted an interview with the President of Iran, Mahmoud Ahmadinejad. Recent tensions between the USA and Iran served as the primary focus of the interview. Scott Pelley presented the following statement to Ahmadinejad: “President Bush has pledged that you will not be allowed to possess a nuclear weapon and will use military force if necessary”.¹ This statement contains both a clear demand (to cease any nuclear weapons program) and a clear threat (that military force will be used) to the leader of Iran. Ahmadinejad’s initial response was as follows:

I think Mr. Bush, if he wants his party to win the next election, there are cheaper ways and ways to go about this. I can very well give him a few ideas so that the people vote for him … help the people, the victims of Katrina. People will vote for them if they do these things. But if they insist on what they are saying right now, this will not help them.

There are two important pieces of information that Ahmadinejad points to in his response to the demand and threat that Pelley reports. First, he suggests that he believes that the US president is engaging in aggressive behavior for partisan electoral purposes. Second, he points to the fact that the Republican Party is experiencing electoral challenges and low popularity as a result of domestic political failures like Hurricane Katrina. There is a clear belief on the part of Ahmadinejad that President Bush was stepping up his rhetoric and making threats to divert from domestic political problems.²

This response from Ahmadinejad suggests that he had a clear picture of the domestic political environment that Bush was facing. The relatively transparent nature of American political institutions was crucial in allowing access to the information that Ahmadinejad identified as a potential incentive for threatening Iran. Later in the interview, Pelley asks Ahmadinejad about his military ambitions and he provides the following two responses. He states that “Iran is not walking toward war” and that “Iran does not need nuclear weapons”. This very well could have been cheap talk, but there is a clear attempt here toward the end of the interview to tone down the rhetoric and to make it as clear as possible that Iran does not want war and that they are not violating the demands of Bush (producing nuclear weapons). The above example demonstrates how information, and more specifically transparency, conditions the strategic environment between states. In responding to the threat from Bush, a leader from a relatively transparent state, Ahmadinejad clearly accounted for the internal political dynamics in formulating his response to the threat.

The role of information and misperception is a central focus of the study of why states fight (Fearon, 1995; Jervis, 1976; Morrow, 1989). A central dilemma associated with this literature results from the difficulties in empirically testing many of the theoretical models that point to information and uncertainty as important conditioning factors in the march to war. Some research has looked towards domestic regime type, more specifically, democracy, to proxy information-rich environments (Schultz, 1999). I argue here that we need to think about the state characteristics that allow for greater levels of information revelation, apart from democracy.

The state characteristic, transparency, is receiving greater attention in the interstate conflict literature (Finel and Lord, 2002; Lebovic, 2006). Here I focus on what I term external transparency and the role it plays in diminishing the likelihood of armed conflict between two states. The theoretical research that identifies the importance of information and uncertainty is primarily focused on the role that information revelation between states plays. It is this emphasis on information revelation between states that forces a focus on the state characteristics that make information revelation between states more likely. External transparency is defined here as the inability of government leaders to hide government processes, decision-making and events taking place within a state from external audiences. By thinking about external transparency as the mechanism through which information becomes available between states, it is possible to test some of the core theoretical conclusions from the crisis bargaining and the diversionary strategic conflict avoidance literature.

The remainder of this paper focuses on the role of external transparency in light of some of the core conclusions from the formal theoretical work examining interstate conflict in general and diversionary conflict. After establishing some of the theoretical foundations that guide the empirical inquiry, I discuss how the concept of external transparency is measured and the research design implemented to test the hypotheses presented below. I rely on a measure of transparency based on the access that the New York Times has to report within the borders of other states. Using this original data collection, I conduct an empirical test, with the results indicating that externally transparent states are less likely to initiate conflict. The results demonstrate a general relationship between external transparency and armed conflict and highlight how external governmental transparency can dampen the ability of leaders to engage in diversionary conflict.

Theoretical foundations

Information, external transparency and war

It is useful to start with two conclusions from the crisis bargaining literature. First, interstate conflict is the result of bargaining failure (Fearon, 1995). In other words, when assuming perfect and complete information in a bargaining setting, there is always a bargaining space that is preferred by both actors to war. The second, and more important, implication of the crisis bargaining literature is that finding acceptable negotiated outcomes is complicated by information asymmetries/uncertainty, commitment problems and issue indivisibilities (Fearon, 1995). Here I focus on how external transparency helps in overcoming information asymmetries and subsequent incentives to misrepresent.³

The crisis bargaining literature has advanced from making general claims about the role of information in a bargaining setting to more specific claims concerning the level of information about specific state characteristics, that is, power (Bearce et al., 2006; Reed, 2003), resolve (Fearon, 1994, 1997; Schultz, 1998), and domestic political conditions (Smith, 1996; Tarar, 2006). From the general claim that credible information availability makes it easier for states to overcome uncertainty and commitment problems, I expect that increases in external transparency should lead to a decrease in the probability of conflict initiation. To further specify this link it is useful to think about how information about specific state characteristics makes conflict less likely and in turn how external transparency provides the link between information and conflict.

An element of information that is frequently referenced in the crisis bargaining literature is resolve (Fearon, 1994, 1997; Morrow, 1989). Overestimating the resolve of another state can cause a state to give in to demands of another state, thus averting a conflict. However, uncertainty of resolve can also make conflict more likely. If one leader underestimates the resolve of another leader, she runs the risk of demanding too much or of not ceding enough ground in bargaining, thus making militarized disputes more likely. In addition, there are clear incentives to misrepresent how resolved you are in order to achieve a bargaining outcome closer to your ideal point (Fearon, 1995). This conclusion has produced what is considered war as a result of mutual optimism (Blainey, 1988). This is the idea that two states will fight because they both believe that they can gain benefits from fighting that exceed the costs of war. Under this condition we observe war between two rational actors (Slantchev and Tarar, 2011). External transparency provides a mechanism through which resolve can more easily be observed and possibly evaluated, limiting this possibility for mutual optimism.

Schultz (1998, 2001) points to one mechanism through which credible resolve can be signaled, domestic political opposition to the governing party’s foreign policy. He suggests that the ability of domestic opposition to effectively signal resolve has to come in the democratic context. Slantchev (2006) shows theoretically that a free domestic media provides the micro-foundations for audience costs in democratic regimes.

Here I argue that, regardless of regime type, if the coalition that a leader is responsible to (maybe the military in some autocracies and an opposition party in a democracy) is voicing opposition, and that state is externally transparent, adversaries can get a better gauge of how resolved their opponent is or the costs that a leader may incur by initiating conflict. High levels of external transparency increase the ability of adversaries to observe the behavior of opposition groups and other state characteristics, thus increasing the ability to gauge resolve. This increased ability to estimate resolve that comes through external transparency produces a decrease in the probability of interstate armed conflict initiation.

Hypothesis 1a: The more externally transparent a state is, the less likely it is to initiate a militarized interstate dispute (MID).

It is important to note that there are arguments suggesting that transparency might have little effect on conflict and in some cases make conflict more likely (Finel and Lord, 1999). Finel and Lord argue that increased transparency leads to an information overload, with little ability to determine what information is reliable.⁴ As a result, increases in transparency do not necessarily decrease the probability of conflict. Schultz (2002) suggests that on average transparency will decrease the probability of conflict between states, but that there are special cases where it can make conflict more likely. In particular, there is the possibility that the availability of information through transparency can reveal information about growing capabilities or a particularly hostile adversary. These types of information revelation can potentially open up the possibility of conflict spirals, commitment problems and an increased probability of war (Zagare and Kilgour, 1998). Whether or not this greater availability of information increases or decreases the likelihood of interstate conflict is an empirical question and one that the analysis here can start to provide further answers to.

Hypothesis 1b: The more externally transparent a state is, the more likely it is to initiate a MID.

Having discussed how information about resolve makes conflict less likely in general, I now turn to another specific state characteristic, domestic challenges, and evaluate how information about domestic political conditions, via external transparency, also decreases the probability of conflict. What this allows for is a more direct test of the mechanism argued to be at work here. Greater information, through transparency, may have an effect on conflict, but this relationship is more convincing if I can demonstrate that information about a specific state characteristic has an effect on conflict behavior. The diversionary strategic avoidance literature is a useful point of departure for this discussion. It is important to point out here that, while the foregoing discussion is reliant on models with unitary actors, the discussion to follow drops that assumption and builds from models with non-unitary actors.

Diversion and strategic conflict avoidance arguments

Just like the literature on information and general conflict behavior, there is a growing examination of the role that strategic interaction and information play in diversionary conflict in particular. Building from Downs and Rocke’s (1994) formalization of diversionary theory, Smith (1996) starts this research program by suggesting that democratic states are less able to engage in diversionary conflict because potential targets can observe when democratic states are most in need of a diversionary target and adjust their behavior to avoid serving as a scapegoat. This certainly does not mean that democratic states are never able to engage in diversion, but it does suggest that diversion is less likely for these types of states. A great deal of scholarly research followed Smith’s (1996) claim and this literature can be divided in to two groups. The first tests this argument in terms of US foreign policy and finds that, when modeled correctly, evidence suggests that the USA is less able to engage in diversionary conflict as a result of strategic conflict avoidance (Clark, 2003; Fordham, 2005). In particular, this research shows that, when US leaders are in trouble domestically, the USA is less likely to be targeted by other states. Once taking this into account, diversion is no longer likely by US leaders when they are facing domestic challenges. The other branch of literature tests the strategic avoidance argument in cross-national samples (Chiozza and Goemans, 2004; Leeds and Davis, 1997; Miller, 1999; Mitchell and Prins, 2004) and finds more generally that democratic regimes are less able to engage in diversion than autocratic regimes.

There is a common theme in much of the above literature. Both the original theoretical work (Smith, 1996) and the subsequent empirical work assume either implicitly or explicitly that democracy and transparency are one and the same. Much of this literature has used democracy as a proxy (Kisangani and Pickering, 2007; Leeds and Davis, 1997;⁵ Miller, 1999; Mitchell and Prins, 2004). This existing research suggests that there is not always a target available for a state seeking to engage in diversionary behavior and that this challenge for finding targets is most likely to be found in democratic regimes. Using democracy as a proxy for information or transparency does not appropriately test the theoretical mechanism at work in Smith’s theory. Democracies and autocracies can vary independently in their levels of external transparency.

It is also important to note that there is another body of research that suggests theoretically and finds empirically that democracies are more likely to engage in diversionary behavior than autocracies, the exact opposite of what the strategic avoidance literature suggests. For example, Gelpi (1997) argues and finds that democratic regimes are more likely than autocratic regimes to engage in diversionary actions because democratic leaders do not have the option to repress in response to domestic challenges while autocrats do. Others have shown that specific types of democratic regimes are predisposed towards engaging in diversionary conflict. Kisangani and Pickering (2007) show that mature democracies are predisposed toward diversionary behaviors and are more likely to benefit from diversion compared with other democracies. Brule and Williams (2009) compare differences in diversionary propensity across democratic regimes based on the coalition structure of the government and find that coalition democracies are less likely to engage in diversionary conflict than other democracies. There is certainly not a consensus on the role that regime type plays in mediating the likelihood that a leader engages in interstate conflict for domestic political purposes. The theoretical argument and analysis here conclude that it is not necessarily about democracy. Whether a state is externally transparent, something that can be a characteristic of both autocratic and democratic regimes, is a crucial state characteristic that conditions the ability of a leader to find diversionary targets.

Information, external transparency and domestic challenges

Why look at external transparency instead of democracy or general transparency when evaluating diversionary claims? External transparency, both the concept and its measurement, captures the component of transparency that allows other states to observe political events and other state characteristics. There are many characteristics about democracy that might make a leader more or less likely to engage in armed conflict for domestic political reasons. Using democracy as a proxy we run into an identification problem. Much of the literature in international relations has already identified democracy as an important variable for predicting the occurrence of conflict in general and diversionary conflict. The concept and measurement of external transparency is an attempt to focus on the causal mechanism that makes it less likely that some states will engage in diversionary behavior.

The key in reformulating the strategic avoidance argument is in identifying the core mechanism in the theory, information about domestic political conditions (or more specifically domestic challenges). It is helpful to go back to the diversionary argument to work through the logic of how external transparency influences the opportunity for states to engage in diversionary conflict. The standard diversionary claim is that, when a potential initiator experiences a domestic challenge, it has an increased incentive to use force to demonstrate competence (Downs and Rocke, 1994). Another way to think about this is that the leader of that state now has an increased payoff (or benefit) for engaging in conflict. Whether the leader of the potential initiator is experiencing a domestic challenge becomes an important piece of information to a potential target. This information is important to the target, because there are high costs to engaging in conflict and the target is now dealing with another state that might actually attain a domestic payoff from conflict (Tarar, 2006). External transparency plays an important role here as the mechanism through which the target observes whether a state is experiencing a domestic challenge (which generates the incentive to engage in diversionary behavior). If the target observes the domestic challenge occurring to the potential initiator, it is more likely to engage in placating behavior to avoid serving as a scapegoat. As a result, the potential initiator is less likely to find a diversionary target, leaving it with less opportunity for diversionary conflict.

If a leader can hide the fact that she is facing a domestic challenge, whether it is low popularity, economic failures or domestic uprisings, the opportunity for diversion is more available. On the contrary, if a state is externally transparent, a leader has a difficult time hiding these domestic challenges, allowing target states an opportunity to avoid diversionary uses of force. External transparency provides a window into the political and economic conditions of a state, allowing other states to adjust their behavior. If a state is not externally transparent, or in other words if a leader can hide the fact that she is facing a domestic challenge, she is in a position to externalize problems and use force for diversionary purposes. The more externally transparent a state is, the more difficult it becomes for the leader of that state to use force to divert. What this story implies is that it is not information through transparency on its own that decreases the probability of conflict, but information about specific state characteristics.

Hypothesis 2: Externally transparent states are less likely to initiate MIDs in response to a domestic challenge than externally opaque states.

It is important to point out the conditional nature of this hypothesis. It is not simply that being transparent will make a state less likely to initiate conflict (like hypothesis 1). Instead, transparent states are in less of a position to respond to a domestic challenge with conflict initiation, while opaque states have the opportunity to respond to domestic challenges by initiating conflict. The effect of a domestic challenge is conditional on the level of external transparency.

Research design

In order to test the above hypotheses I look toward a directed dyad level of analysis and implement data from the time period 1982–1999.⁶ Both hypotheses are about conflict initiation and require a directed dyad structure (Davies, 2002; Sobek, 2007). The dyadic setup is especially important here, because the theory about external transparency considers how a potential target state adjusts its behavior given a characteristic of the potential initiating state, external transparency. The directed dyad framework allows the effect of this characteristic of the potential initiator to be tested for, while also controlling for many of the variables that we understand are important in the choice to initiate and that cannot be controlled for in a monadic framework.⁷ This problem and solution are discussed in Sobek (2007). He states that “Traditionally, when testing diversionary theory, scholars have used monadic studies. This type of analysis contains no information on who was targeted by the diverting state, which leaves open the possibility that the characteristics of the targeted state caused the initiation” (p. 35). He continues on to say that “The directed dyad better controls for extraneous factors while more directly testing the diversionary hypothesis” (p. 35). I agree with this approach and similarly worry that testing at the monadic level leaves out important variables that are necessary to control for.

The following section begins with a discussion of the primary independent variables, external transparency and domestic political challenges. Finally I discuss how I specify the statistical models: specifically the measurement of the dependent variable, the unit of analysis, the estimator and the selection of control variables.

Measuring external transparency

In order to measure external transparency I look to the origin of newspaper articles from the New York (NY) Times. The NY Times has global coverage and frequently picks up articles from the Associated Press, Reuters, United Press International, Agence France Presse, etc. If the NY Times is not allowed into a country, these news organizations regularly have a presence. If foreign reporters from these papers file news stories about a country within the borders of that country, it suggests some level of external transparency. Whether the country has specific rules against allowing foreign journalists, the country is too dangerous for journalists and editorial choices are reflected by the number of articles originating from within a state, serving as an indication of the inability of leaders to hide the internal events and processes inside a state.

Using the archives of NY Times articles, I count the number of times a country is mentioned in the headline of a newspaper article. The search is restricted to news articles. The headline is used because this forces the requirement that an article is specifically about the country in question. I search for both the country name and the demonym. So, in the case of France, both France and French are searched for in the headline. This provides the denominator of the equation used to calculate the measure of external transparency.

The numerator is the number of articles returned in the first search that have either the country name or a city within the country in the dateline. The dateline indicates where the story was filed. This provides a count of all the times that an article about a specific country is filed within the borders of that country. In the case of France, anytime France, Paris, Lyon, Nice, etc. are mentioned in the dateline, it is counted in the numerator. I begin with the most populous city and continue to add cities in the country until three consecutive cities do not add to the total of datelines filed. In most cases when a story is filed from a smaller city, the country name is also included in the dateline, making it unlikely that datelines are missed. The data are collected on a yearly basis.

The ratio provides a cross-national, cross-temporal measure of external transparency. This serves as more than just a proxy of external transparency. The access that foreign newspapers have to a state plays a crucial role in determining the quality of information observed by outsiders. The presence of journalists in a state should increase the quality of information available to potential adversaries in the international system. It is important to note that the goal here is to measure access (or even potential access), not volume. Volume is more likely to be the result of editorial decisions and events going on within a state, something that this measure is not intended to capture.

For some countries the changes from year to year in this measure are relatively volatile. This is especially true of states that have fewer reports for each year. For example, Trinidad and Tobago in 1996 has three out of six datelines filed from within the country. In 1997 there are two out of nine datelines filed within the country. This is a drop from 50 to 22%. This is primarily a problem in smaller states that receive less coverage. By using a moving average, the consequences of these changes are muted. As a result I have also created a measure that is a five-year moving average.⁸ This moving average takes into account years t − 4, t − 3, t − 2, t − 1 and t. This helps to overcome the problem of volatility in the measure and helps to decrease the number of missing observations that result from there being zero mentions in the headline count for a country in a year (which makes it impossible to compute the measure). A five-year moving average of the dateline and headline counts are first generated. Percentages are then generated from those two measures. The number of country–years in the external transparency data increases by 361 from the regular calculation of the percentage of datelines to the moving average percentage of datelines. It is important to note that this is not 361 observations in the final dataset, this is 361 state–years for which dyads are added to the analysis. Also, the standard deviation of the measure decreases by 0.05 in the moving average, suggesting a decline in the volatility of the measure. Using the five-year moving average helps to fill in some of the missing observations that result from states not being mentioned in the NY Times through an entire year as well as decreasing the volatility in the measure that is unlikely to reflect real changes in the level of external transparency.

Why is this measure used as opposed to existing measures of domestic media freedom? Existing research examines the role of domestic media freedom, but in that research, both the theories tested and the data implemented focus on government control of domestic media (Van Belle, 1997; Choi and James, 2007). Existing theoretical work by Slantchev (2006) suggests that the presence of a free domestic media allows states to make credible threats. However, this does not consider the possibility that some states, while not having a domestic media, instead have foreign media that provides this outlet for information to outside observers. The measure here captures this possibility and isolates the presence of foreign media in a state, regardless of domestic media freedom. It is impossible from the Van Belle measure to know the extent to which domestic media sources are available outside the state and whether there are different rules for foreign and domestic media inside the state.⁹

A law reported on from Kenya on 28 January 1996 in the NY Times provides an example of the potential for divergence between domestic and international media access. The report in the NY Times indicates that a law was pushed through by the single-party government to stifle the press. It is stated in the report that “the code would also extend to foreign correspondents, who often are able to report what Kenyan publications cannot” (E12). This quote provides direct evidence from a reporter that within Kenya there are different rules and different degrees of access between the domestic and foreign media. A greater convergence between foreign and domestic media access resulted from the passage of this law in 1996, but the report makes the important point that, prior to the law, there were different rules for foreign and domestic media sources. An important take-away point from this is that the two can diverge from each other, not that they always will.

It is also worth noting how this measure correlates with a conventional measure of regime type, Polity IV. Given the argument that external transparency is distinct from democracy it is important that this measure is distinct from the polity data. The two measures are correlated at 0.303. Unsurprisingly there is a positive correlation between external transparency and democracy, but that relationship is not very strong. Much of this is the result of greater variability in the transparency measure among autocratic regimes. While autocratic regimes have an incentive to remain closed to their domestic population, those incentives are less clear for remaining closed to outside observers, particularly with growing international economic integration.

Figure 1(A and B) presents maps of the level of external transparency cross-nationally, breaking the data into quintiles for the pre- and post-Cold War time periods. The starkest change in these two maps is the growth in external transparency in Africa from the first to the second time period. This is unsurprising given the changes in technology and infrastructure that made moving into and through these countries easier. In addition, it is apparent that some of the most opaque states to outsiders are in the Middle East for both time periods. More than anything these maps illustrate the global distribution of external transparency, and begin to highlight how this state characteristic is distributed differently than democratic institutions.

Figure 1.

Global distribution of external transparency.

Figure 1(C and D) plots the relationship between democracy and transparency for the same two time periods. These plots demonstrate a couple of interesting points. First, it appears that for both time periods there is a wider variance or dispersion among the more autocratic regimes. This is unsurprising as democratic regimes are more likely to be uniformly transparent, while autocratic regimes are going to be more or less open depending on pressures from outside actors and the whims of those in authority positions. In addition, there are some interesting changes across the two time periods. Most interestingly, what this comparison shows is that there is a leveling out of transparency in the 1990s. This coincides with great increases in globalization and international economic integration. This is the time period where information revelation, particularly by less developed autocratic regimes, is more likely to be demanded by outside observers. As a result it is unsurprising that the relationship between regime type and transparency flattens out, with democratic states not much more likely to be open to outside observers compared with autocratic regimes. This leveling out also coincides with increases in external transparency on the African continent demonstrated in Figure 1(A and B).

In the models predicting initiation, I include the external transparency measure as the five-year moving average for state A of the dyad. This is because the hypotheses above suggest that the level of transparency of the potential initiator is the relevant variable for understanding initiation.¹⁰

Other independent variables

I account for the presence of a domestic challenge with two variables, indicators of mass unrest and elite unrest from the Banks Cross-Sectional Time-Series Data (Banks, 2008). These measures are also implemented by Kisangani and Pickering (2007) and provide an indicator of domestic challenges that are measured consistently across the time period and cross-sections in the sample for this paper. Mass unrest is a count of all general strikes against government, riots and anti-government protest. Elite unrest is a count of government crises and purges. This is to capture the conditional relationship suggested by the strategic avoidance argument and my hypothesis.

A number of control variables are included that are commonly found to be strong predictors of conflict. There is a long list of variables that can be included in a conflict model. The variables included here are common in the literature, but also do not lead to the loss of too many observations because of missing data. As implemented in Russett and Oneal (2001) a weak link operationalization of regime type is included in the models (Marshall and Jaggers, 2004). Existing research has shown that similarity in preferences is important to control for (Gartzke, 2000). This is especially true when testing theories linked to bargaining models. To capture preference similarity, I include a measure of UN voting affinity gathered by Gartzke and Jo (2002). Relative capabilities (Singer et al., 1972) is included in the model, operationalized as the capabilities of state A, the potential initiator in the dyad, over the sum of the capabilities in the dyad. This standardizes the measure from 0 to 1, with lower values indicating that state A is weaker than state B and higher values indicating increasing the strength of state A relative to state B.

I also included whether the two states in a dyad are contiguous. While contiguity is conventionally included as a control variable, its inclusion in this model also helps to account for the possibility that neighboring states are more likely to have media that can report on the events going on within a neighboring state’s borders. This is something that the NY Times-based measure cannot capture. Including the contiguity variable decreases the potential bias that results from this unaccounted source of external transparency. This variable is generated in Eugene (Bennett and Stam, 2000). Finally I include a control variable for state population of state A to insure against the possibility that the unrest variables are simply capturing population size, with larger states typically having more unrest. Including this variable allows for ruling out that higher counts of domestic challenges are simply an artifact of those states having larger populations. In addition, inclusion of this population variable also controls for the possibility that larger states simply receive more media attention. This is likely to be related to both the transparency measure and the dependent variable, making it important to account for. This variable comes from the World Development Indicators 2011 (World Bank).

Model and dependent variable

For the dependent variable I use MID initiation from the Correlates of War Militarized Interstate Dispute data (Jones et al., 1996). This codes whether there is any MID initiated by state A in the dyad. The initiator across the types of conflict is coded as the side in the conflict that took the first codeable action in the dispute (Jones et al., 1996). I estimate a series of probit models because the dependent variable is binary. Robust standard errors are used to control for spatial heterogeneity across dyads. Finally, I include a cubic polynomial of time to control for temporal dependence (Carter and Signorino, 2010).¹¹

Results and discussion

The results in Table 1 provide a test of both hypotheses presented above. For the coefficient on the external transparency variable in model 1, the expectation is that external transparency decreases the probability of MID initiation. The coefficient on external transparency is negative and statistically significant, providing support for the above logic. This model provides evidence that, as a state becomes more externally transparent, it is less likely to initiate a MID.

Table 1.

Probit model: transparency and MID initiation

	(1)	(2)
	Hypothesis 1	Hypothesis 2
Transparency	−0.871***	−0.760***
Transparency	(0.101)	(0.102)
Elite unrest	0.096**	0.246**
Elite unrest	(0.030)	(0.083)
Transparencyelite*		−0.296*
Transparencyelite*		(0.168)
Population	1.80 × 10⁻¹⁰ *	2.46 × 10⁻¹⁰ **
Population	(8.57 × 10⁻¹¹)	(8.08 × 10⁻¹¹)
Regime low	−0.002	−0.001
Regime low	(0.003)	(0.003)
UN affinity	−0.528***	−0.539***
UN affinity	(0.065)	(0.066)
Contiguity	1.630***	1.630***
Contiguity	(0.038)	(0.038)
Relative capabilities	0.423***	0.439***
Relative capabilities	(0.060)	(0.060)
Mass unrest	0.016***
Mass unrest	(0.003)
Constant	−3.149***	−3.180***
Constant	(0.153)	(0.156)
Observations	316300	316572
Psuedo-R	0.28	0.28

Standard errors in parentheses; all variables lagged one year.

p < 0.05, ** p < 0.01, *** p < 0.001.

This is in line with much of the theoretical literature claiming that a more information-rich environment produces a bargaining setting conducive to peaceful settlements (Fearon, 1995; Morgan, 1984; Morrow, 1989). In particular, this shows that even one single state in the dyad being externally transparent makes the initiation of a MID less likely.¹² This finding is also in contradiction with the argument put forward by Finel and Lord (1999) that transparency will actually lead to an overload of information, making conflict more likely or having no effect at all. It is also important to note that this result holds when controlling for regime type, suggesting that external transparency has its own separate effect on conflict initiation.

The coefficient on relative capabilities is positive and statistically significant in this model. This suggests that, as the stronger state, A is the more likely to initiate a MID against the potential target. The coefficient for the UN affinity variable is negative and statistically significant, indicating unsurprisingly that, as two states converge over their preferences, they are less likely to experience a MID initiation (Gartzke, 1998). The coefficient on contiguity is positive and statistically significant. As expected, contiguity in a dyad increases the probability of onset (Bremer, 1992). Finally, the coefficient on the weak link regime type variable is not statistically significant. The coefficient on the population variable is positive and statistically significant. Finally, the coefficients on the mass and elite unrest variables are positive and statistically significant. We get a clearer picture of the role of the unrest variables in the models with interaction terms to follow.

Model 2 presents the results with an interaction term between elite unrest and external transparency.¹³ The results here provide support for hypothesis 2. The interaction term in these models is best examined by plotting out the effects graphically (Brambor et al., 2005). In particular, the expectation here is that, at low levels of external transparency, elite unrest leads to an increased probability of MID initiation, while at higher levels of transparency, elite unrest will no longer be related to initiation.

Figure 2 shows that domestic challenges increase the probability of initiation by opaque states, but not for externally transparent states. This figure demonstrates the marginal effect of elite unrest at varying levels of external transparency, producing two important conclusions. First, the effect of elite unrest on MID initiation declines with increasing levels of external transparency. Second, at higher levels of external transparency (above approximately 0.6), elite unrest no longer has an effect on initiation, with the confidence interval of the marginal effect crossing over zero. Again, leaders in opaque states appear to be more likely to engage in diversionary behavior than their more transparent counterparts.

Figure 2.

Marginal effect of elite unrest as external transparency changes.

While hypothesis 2 receives partial support in implementing the elite unrest variable of domestic challenges, it does not when the mass unrest variable is implemented. Neither the mass unrest variable nor the interaction between mass unrest and external transparency is statistically significant at conventional levels. This may be the result of the measurement of the external transparency data. Most of the reporters filing from within countries file and operate from within the capital. Being in the capital of a country is likely to leave a reporter predisposed to reporting about elite political machinations as opposed to popular unrest, which could be occurring elsewhere in a country. If a government crisis is occurring, this is very likely to be reported on by foreign media, especially the NY Times. This is of course not the case with large protests in the capital cities, but a small protest in a city outside of the capital may not get as much attention from foreign news media. The literature about protest movements in the USA finds that there is a bias in media coverage against covering protests movements. In addition, in the unlikely event that a protest is covered, the issue of the protest is rarely covered (Smith et al., 2001). If the foreign media in a country does not cover protests, the information that a leader is challenged domestically will not be provided to outside observers. As a consequence, the strategic environment between potential diverting and target states is unaffected.

This partial support for hypothesis 2 is consistent with the strategic avoidance argument (Smith, 1996), but it still tells a somewhat different story. Rather than strategic avoidance being a phenomenon that occurs for only democratic states, it is specifically a phenomenon related to external transparency. The more externally transparent a state is, the less likely it is to divert in response to elite challenges.

Conclusion

The theoretical implications of bargaining models and the diversionary strategic avoidance arguments suggest that we should see specific strategic behaviors in the international system under limited information structures. This research finds that we need to consider the independent role that external transparency plays in conflict behavior. External transparency provides the micro-foundations through which information affects conflict behavior. This article demonstrates empirically that external transparency plays a role in whether states initiate MIDs. In addition, the tests of hypothesis 2 provide a direct examination of how information about a specific state characteristic, domestic challenges, can alter the strategic environment between states.

The theoretical implications of Smith (1996) and the subsequent empirical literature suggest that a large number of the states (democracies) in the international system are left with little opportunity to engage in diversionary conflict. These findings have both theoretical and empirical implications for the strategic conflict avoidance thesis. It is not democracy alone that makes it more difficult for leaders to engage in diversionary conflict. Instead it is external transparency that makes states less able to use diversionary conflict. Many democracies can be characterized as externally transparent, but it is not a characteristic unique to democratic states. This allows us to conclude that the class of states that have a diminished opportunity to engage in diversionary conflict might be much larger.

Footnotes

Acknowledgements

The author wishes to thank Michael Bell, David Cingranelli, David Clark, K. Chad Clay, Benjamin Fordham, Richard Frank, Jesse Johnson, Andrew Long, Carla Martinez Machain, Amanda Murdie, Jeff Pickering, Patrick Regan, Brandon Zicha, Lucie Zicha, the Editor of Conflict Management and Peace Science and the anonymous reviewers for their helpful comments and suggestions. This research was previously presented at the Jan Tinbergen European Peace Science Conference.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Notes

References

Banks

(2008) Cross-National Time Series, 1815–2008 [Computer file]. ICPSR edn. Ann Arbor, MI: Inter-university Consortium for Political and Social Research [producer and distributor]; doi:10.3886/ICPSR07412.

Bearce

Flanagan

Floros

(2006) Alliances, internal information, and military conflict among member-states. International Organization 60(3): 595–625.

Bennett

Stam

(2000) EUGene: A conceptual manual. International Interactions 26: 179–204. Available at: http://eugenesoftware.org

Blainey

(1988) The Causes of War. New York: The Free Press

Brambor

Clark

Golder

(2005) Understanding interaction models: Improving empirical analysis. Political Analysis 13: 1–20.

Bremer

(1992) Dangerous dyads: Conditions affecting the likelihood of interstate war. Journal of Conflict Resolution 36(2): 309–341.

Brule

Williams

(2009) Democracy and diversion: Government arrangements, the economy, and dispute initiation. Journal of Peace Research 46(6): 777–798.

Carter

Signorino

(2010) Back to the future: Modeling temporal dependence in binary data. Political Analysis 18(3): 271–292.

Chiozza

Goemans

(2004) Avoiding diversionary targets. Journal of Peace Research 41(4): 423–443.

10.

Choi

S-W

James

(2007) Media openness, democracy, and militarized interstate disputes. British Journal of Political Science 37(1): 23–46.

11.

Clark

(2003) Can strategic interaction divert diversionary behavior? A model of U.S. conflict propensity. Journal of Politics, 65(4): 1013–1039.

12.

Davies

(2002) Domestic strife and the initiation of international conflicts: A directed dyad analysis, 1950–1982. The Journal of Conflict Resolution 46(5): 672–692.

13.

Downs

Rocke

(1994) Conflict, agency, and gambling for resurrection: The principal-agent problem goes to war. American Journal of Political Science 38(2): 362–380.

14.

Fearon

(1994) Domestic political audiences and the escalation of international disputes. American Political Science Review 88: 577–592.

15.

Fearon

(1995) Rationalist explanations for war. International Organization 49: 379–414.

16.

Fearon

(1997) Signaling foreign policy interests: Tying hands versus sinking costs. Journal of Conflict Resolution 41(1): 68–90.

17.

Finel

Lord

(1999) The surprising logic of transparency. International Studies Quarterly 43(2): 315–339.

18.

Fordham

(2005) Strategic conflict avoidance and the diversionary use of force. Journal of Politics 67(1): 132–153.

19.

Gartzke

(1998) Kant we all get along?: Opportunity, willingness, and the origins of the democratic peace. American Journal of Political Science 42(1):1–27.

20.

Gartzke

(2000) Preferences and the democratic peace. International Studies Quarterly 44(2): 191–210.

21.

Gartzke

D-J

(2002) United Nations General Assembly voting, 1946–1996. Version 3.0. Available at: http://www.columbia.edu/~eg589/datasets

22.

Gelpi

(1997) Democratic diversions: Governmental structure and the externalization of conflict. Journal of Conflict Resolution 41(2): 255–82.

23.

Jervis

(1976) Perception and Misperception in World Politics. Princeton, NJ: Princeton University Press.

24.

Jones

Bremer

Singer

(1996) Militarized interstate disputes, 1816–1992: Rationale, coding rules, and empirical patterns. Conflict Management and Peace Science 15(2): 163–213.

25.

Kisangani

Pickering

(2007) Diverting with benevolent military force: Reducing risks and rising above strategic behavior. International Studies Quarterly 51(2): 277–299.

26.

Lebovic

(2006) Democracies and transparency: Country reports to the UN Register of Conventional Arms, 1992–2001. Journal of Peace Research 43(5): 543–562.

27.

Leeds

Davis

(1997) Domestic political vulnerability and international disputes. Journal of Conflict Resolution 41: 814–835.

28.

Levy

(1989) The diversionary theory of war: A critique. In: Midlarsky

(ed.) The Handbook of War Studies. Boston, MA: Unwin Hyman, pp. 259–288.

29.

Marshall

Jaggers

(2004) The Polity IV Project. College Park, MD: Center for International Development and Conflict Management, University of Maryland. Available at: http:/www.bsos.umd.edu/cidcm/polity

30.

Miller

(1999) Regime type, strategic interaction, and the diversionary use of force. Journal of Conflict Resolution 43(3): 388–402.

31.

Mitchell

Prins

(2004) Rivalry and diversionary uses of force. Journal of Conflict Resolution 48(6): 937–961.

32.

Morgan

(1984) A spatial model of crisis bargaining. International Studies Quarterly December: 407–426.

33.

Morrow

(1989) Capabilities, uncertainty, and resolve: A limited information model of crisis bargaining. American Journal of Political Science 33(4): 941–972.

34.

Reed

(2003) Information, power, and war. The American Political Science Review 97(4): 633–641.

35.

Russett

Oneal

(2001) Triangulating Peace. New York: Norton.

36.

Schultz

(1998) Domestic opposition and signaling in international crises. The American Political Science Review 92(4): 829–844.

37.

Schultz

(1999) Do democratic institutions constrain or inform?: Contrasting two institutional perspectives on democracy and war. International Organization 52: 233–266.

38.

Schultz

(2001) Democracy and Coercive Diplomacy. Cambridge: Cambridge University Press.

39.

Schultz

(2002) Domestic political competition and transparency in international crises: The good, the bad, and the ugly. In: Finel

Lord

(eds) Power and Conflict in the Age of Transparency. New York: Palgrave Macmillan, pp. 57–82.

40.

Singer

Bremer

Stuckey

(1972) Capability distribution, uncertainty, and major power war, 1820–1965. In: Russett

(ed.) Peace, War, and Numbers. Beverly Hills, CA: Sage, pp. 19–48.

41.

Slantchev

(2006) Politicians, the media, and domestic audience costs. International Studies Quarterly 50(2): 445–477.

42.

Slantchev

Tarar

(2011) Mutual optimism as a rationalist explanation of war. American Journal of Political Science 55(1): 135–148.

43.

Smith

(1996) Diversionary foreign policy in democratic systems. International Studies Quarterly 40(1): 133–153.

44.

Smith

McCarthy

McPhail

Augustyn

(2001) From protest to agenda setting: Description bias in media coverage of protest events in Washington D.C. Social Forces 79(4): 1397–1423.

45.

Sobek

(2007) Rallying around the Podesta: Testing diversionary theory across time. Journal of Peace Research 44(1): 29–45.

46.

Tarar

(2006) Diversionary incentives and the bargaining approach to war. International Studies Quarterly 50(1):169–188.

47.

Van Belle

(1997) Press freedom and the democratic peace. Journal of Peace Research 40: 405–414.

48.

Zagare

Kilgour

(1998) Deterrence theory and the spiral model revisited. Journal of Theoretical Politics 10(1): 59–87.

What you don’t know can hurt you: Information,external transparency,and interstate conflict,1982–1999

Abstract

Keywords

Introduction

Theoretical foundations

Information, external transparency and war

Diversion and strategic conflict avoidance arguments

Information, external transparency and domestic challenges

Research design

Measuring external transparency

Other independent variables

Model and dependent variable

Results and discussion

Conclusion

Footnotes

Acknowledgements

Funding

Notes

References