Hypothesis 1a: Federal regimes are less prone to engage in war with each other than non-federal types of regimes.
Hypothesis 1b Federal regimes are just as prone to engage in war as other types of regimes.
Hypothesis 2: Federal regimes are as prone to engage in war/military conflict with each other as democratic regimes are prone to engage in war/military conflict with democratic regimes.
Hypothesis 3: Democratic federal regimes are less prone to engage in War with each other than all other regime types are prone to engage in war with other regime types (federal-autocratic, democratic non-federal, and autocratic non-federal) regimes.
Hypotheses 1a and 1b are related or overlapping; it is hoped that such a design of clustering hypotheses was expected to provide salience for interpreting the results of the proposed regressions and surveys of data. The initial survey of the research was intended to find out whether these assumptions held at face value before running logistic regressions on the model above.
INITIAL FEDERAL PEACE PAPER
In prior research on federalism and war, “Federalist Peace Theory, 1817-1992” (2001), I hypothesized that regardless as to whether these wars are with non-federal, unitary-democracies, and/or non-democratic nations, that of those wars in which federations take on or find themselves involved in, these same federations generally are successful in "winning the war". Finally, by combining several different data sets on international conflict, regime, and polity types of indicators, I found that (1) regardless of whether it is a relatively autocratic-federal regime or a democratic-federal one, federal structure-"in and of itself--has a robust and consistent negative effect on the likelihood of conflict or escalation, (2) both normative and structural models are supported by data, and (3) support for this federal normative-structural model is approximately as statistically robust as democratic peace models . This paper attempts in a more conceptual way to retest some of these hypothesis using a different data set.
BACKGROUND
As aforementioned, rationale behind this research derives from review of articles on (1) democratic peace and from (2a) studies of federal competition from the rational choice perspective and (b) from a review of analysis based on the research in the structural schools of economy and politics. Importantly, as noted above, according to "democratic peace" research and theory: Democracies, in and of themselves, are less likely to fight wars with each other--even as they are more likely to defeat an adversarial autocratic state (Lake, 1992). Democratic Peace is based on the assumption that both democratic and autocratic states are equally conflict-prone and war-oriented. (Maoz & Russett, 1993). Weart (1998), however, warns that, prior to the 20th century, other definitions of democracy dominated the landscape and “only very relaxed definitions of democracy” uphold democratic peace theory in earlier times . Just as importantly, Maoz and Russett (1993) have called into question the use of the simplified Polity I and II data sets, type espoused by Gurr (1974), Lake (1992), and others, used by many democratic peace theorists.
Some of these problems have been corrected in more recent Polity III (1993) and Polity IV (2001) updates. Nonetheless, the early inconsistencies in the Polity data sets had encouraged Maoz and Russett to develop their own 100 point scale for democracy basing it both on how democratic Side H scored on the Polity II scale plus how Side L scored. Then, in their subsequent formalization , Maoz and Russett, divided everything by how democratic Side H scored on the Polity scale while subtracting how Side L was scored. Maoz and Russett called this measure JOINREG .
However, since few have retested Maoz and Russetts dyadic data against the updated Polity IV scores, it is not clear whether the reformed versions of Polity might not be just as adequate. Therefore, in this paper I return to Polity IV for the overall Polity score. Nonetheless, in order to maintain some comparability with Maoz and Russett--as well as Lake--, this Polity score was cross checked with a dummy variable for democratic state of 1= (Polity>5) in regression analysis.
Eventually, federal peace theory intends to replicate much of Zinnes and Merrit (1993) and Lake (1992, as well as Maoz and Russett (1993), and their research on democratic peace. However, instead of focusing on the degree democracy plays in war making and foreign policy, another regime type, namely the federal regime, is being studied. As noted in aforementioned reviews, democratic-federal states are predicted to be less war-prone than non-democratic ones. This was also previously empirically tested (Stoda, 2001) so there were also a dummy variable for this dimension of variable interaction of these two variables. This was retested in this paper. The other very important test was to see if federal states as a whole really are, indeed, equally prone to war in general as other non-federal regime since this is an important assumption for comparison of regime types and has been a problematic assumption criticized by Gartzke (1998) and others.
DATA SETS AND ASSUMPTIONS BEHIND THE HYPOTHESES
The Polity IV data set includes data on regimes or states dating back to 1800. It uses a 21-point scale, which helps in coding regime type on a democratic-autocratic scale. Therefore, both federal and non-federal states can be rated according to the 21-point Polity scale, so as to classify them as non-democratic federal or as democratic-federal regimes. Frustratingly, the Polity IV data set no longer provides information on a three point scale for centralization, which had been a surrogate for federalism mentioned by Maoz and Russett. Therefore, I elected to build a new data set of federations or federal state actors from 1800 to 1994 relying primarily on Lemco (1991) and Elazar (1994b). This data set was largely based on both Jonathan Lemco’s (1991), Political Stability in Federal Governments, and Elazar’s (1994b), Federal Systems of the World: A Handbook of Federalism, Confederal and Autonomy Arrangements.
Lemco, in his research on the origin and stability of federations, found 44 historical federal regimes dating back to 1787. This is the basic list of states used in this research paper; however, several other federal regimes neglected by or unknown by Lemco, such as the United Arab Emirates and modern day Russia, were added. Finally, Elazar (1994b) provided another exceptional list of federations, which corroborated and expanded Lemco's list. Lemco's coding method required that a federal regime, at its origin and throughout his existence, include or exhibit at least one of the following: (1) a territorial divisions of power at two levels of governance, (2) a federal constitution, or (3) a mixed bicameral government. I have accordingly coded the addition of several federal states and have added them to Lemco's and Elazar’s lists, based on the existing presence of at least one of these three federal conditions of the regime. (See Appendix for complete list of federations by region.)
Regardless as to whether a federation is a democratic or non-democratic state, due to its matrix-like structure of governance, it was posited apriori to the investigation that: a federal regime is likely to avoid war with other federal states. War, in this paper, is defined as a measure of interstate dispute, using Militarized International Dispute data, that has moved to the highest level of conflict which involves 1000 or military deaths each year. Further, the wars considered relevant for study in this paper had to have occurred between 1946 and 1992 and are in the MID 2a and 2b data sets on international disputes . The MID 2a data set includes data on nearly 2100 conflict events involving actors in the international system. Meanwhile MID 2b consists of nearly 4800 observations of state actors in the international system involved in interstate conflicts on a fatality scale of 1 through 6 (with level 1 being no casualties and level 6 being a full-scale war involving 1000 casualties a year). For this paper, I also used a modified a MID year list available from R. Tucker (1997) provided for creating dyadic year data involving interstate system actors under the Correlates of War (COW) project.
Hypothesis 1b Federal regimes are just as prone to engage in war as other types of regimes.
Hypothesis 2: Federal regimes are as prone to engage in war/military conflict with each other as democratic regimes are prone to engage in war/military conflict with democratic regimes.
Hypothesis 3: Democratic federal regimes are less prone to engage in War with each other than all other regime types are prone to engage in war with other regime types (federal-autocratic, democratic non-federal, and autocratic non-federal) regimes.
INITIAL FEDERAL PEACE PAPER
In prior research on federalism and war, “Federalist Peace Theory, 1817-1992” (2001), I hypothesized that regardless as to whether these wars are with non-federal, unitary-democracies, and/or non-democratic nations, that of those wars in which federations take on or find themselves involved in, these same federations generally are successful in "winning the war". Finally, by combining several different data sets on international conflict, regime, and polity types of indicators, I found that (1) regardless of whether it is a relatively autocratic-federal regime or a democratic-federal one, federal structure-"in and of itself--has a robust and consistent negative effect on the likelihood of conflict or escalation, (2) both normative and structural models are supported by data, and (3) support for this federal normative-structural model is approximately as statistically robust as democratic peace models . This paper attempts in a more conceptual way to retest some of these hypothesis using a different data set.
BACKGROUND
As aforementioned, rationale behind this research derives from review of articles on (1) democratic peace and from (2a) studies of federal competition from the rational choice perspective and (b) from a review of analysis based on the research in the structural schools of economy and politics. Importantly, as noted above, according to "democratic peace" research and theory: Democracies, in and of themselves, are less likely to fight wars with each other--even as they are more likely to defeat an adversarial autocratic state (Lake, 1992). Democratic Peace is based on the assumption that both democratic and autocratic states are equally conflict-prone and war-oriented. (Maoz & Russett, 1993). Weart (1998), however, warns that, prior to the 20th century, other definitions of democracy dominated the landscape and “only very relaxed definitions of democracy” uphold democratic peace theory in earlier times . Just as importantly, Maoz and Russett (1993) have called into question the use of the simplified Polity I and II data sets, type espoused by Gurr (1974), Lake (1992), and others, used by many democratic peace theorists.
Some of these problems have been corrected in more recent Polity III (1993) and Polity IV (2001) updates. Nonetheless, the early inconsistencies in the Polity data sets had encouraged Maoz and Russett to develop their own 100 point scale for democracy basing it both on how democratic Side H scored on the Polity II scale plus how Side L scored. Then, in their subsequent formalization , Maoz and Russett, divided everything by how democratic Side H scored on the Polity scale while subtracting how Side L was scored. Maoz and Russett called this measure JOINREG .
However, since few have retested Maoz and Russetts dyadic data against the updated Polity IV scores, it is not clear whether the reformed versions of Polity might not be just as adequate. Therefore, in this paper I return to Polity IV for the overall Polity score. Nonetheless, in order to maintain some comparability with Maoz and Russett--as well as Lake--, this Polity score was cross checked with a dummy variable for democratic state of 1= (Polity>5) in regression analysis.
Eventually, federal peace theory intends to replicate much of Zinnes and Merrit (1993) and Lake (1992, as well as Maoz and Russett (1993), and their research on democratic peace. However, instead of focusing on the degree democracy plays in war making and foreign policy, another regime type, namely the federal regime, is being studied. As noted in aforementioned reviews, democratic-federal states are predicted to be less war-prone than non-democratic ones. This was also previously empirically tested (Stoda, 2001) so there were also a dummy variable for this dimension of variable interaction of these two variables. This was retested in this paper. The other very important test was to see if federal states as a whole really are, indeed, equally prone to war in general as other non-federal regime since this is an important assumption for comparison of regime types and has been a problematic assumption criticized by Gartzke (1998) and others.
DATA SETS AND ASSUMPTIONS BEHIND THE HYPOTHESES
The Polity IV data set includes data on regimes or states dating back to 1800. It uses a 21-point scale, which helps in coding regime type on a democratic-autocratic scale. Therefore, both federal and non-federal states can be rated according to the 21-point Polity scale, so as to classify them as non-democratic federal or as democratic-federal regimes. Frustratingly, the Polity IV data set no longer provides information on a three point scale for centralization, which had been a surrogate for federalism mentioned by Maoz and Russett. Therefore, I elected to build a new data set of federations or federal state actors from 1800 to 1994 relying primarily on Lemco (1991) and Elazar (1994b). This data set was largely based on both Jonathan Lemco’s (1991), Political Stability in Federal Governments, and Elazar’s (1994b), Federal Systems of the World: A Handbook of Federalism, Confederal and Autonomy Arrangements.
Lemco, in his research on the origin and stability of federations, found 44 historical federal regimes dating back to 1787. This is the basic list of states used in this research paper; however, several other federal regimes neglected by or unknown by Lemco, such as the United Arab Emirates and modern day Russia, were added. Finally, Elazar (1994b) provided another exceptional list of federations, which corroborated and expanded Lemco's list. Lemco's coding method required that a federal regime, at its origin and throughout his existence, include or exhibit at least one of the following: (1) a territorial divisions of power at two levels of governance, (2) a federal constitution, or (3) a mixed bicameral government. I have accordingly coded the addition of several federal states and have added them to Lemco's and Elazar’s lists, based on the existing presence of at least one of these three federal conditions of the regime. (See Appendix for complete list of federations by region.)
Regardless as to whether a federation is a democratic or non-democratic state, due to its matrix-like structure of governance, it was posited apriori to the investigation that: a federal regime is likely to avoid war with other federal states. War, in this paper, is defined as a measure of interstate dispute, using Militarized International Dispute data, that has moved to the highest level of conflict which involves 1000 or military deaths each year. Further, the wars considered relevant for study in this paper had to have occurred between 1946 and 1992 and are in the MID 2a and 2b data sets on international disputes . The MID 2a data set includes data on nearly 2100 conflict events involving actors in the international system. Meanwhile MID 2b consists of nearly 4800 observations of state actors in the international system involved in interstate conflicts on a fatality scale of 1 through 6 (with level 1 being no casualties and level 6 being a full-scale war involving 1000 casualties a year). For this paper, I also used a modified a MID year list available from R. Tucker (1997) provided for creating dyadic year data involving interstate system actors under the Correlates of War (COW) project.
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).
