Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire The System Made Me Stop Doing It The Indirect Origins of Commercial Peace Erik Gartzke UCSD egartzke@ucsd.edu May 7, 2016
Introduction There is a large literature on trade and conflict. Considerable theoretical consensus about the relationship Trade should inhibit interstate conflict... but with whom? Evidence decidedly mixed (c.f. Beck, et al. 1998) Thesis: we may be looking in the wrong place. Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Basic Argument Trade story, comparative advantage, is bilateral (dyadic) Liberal peace argument also posed/tested dyadically No reason to assume that this is correct Dyadic: Incentives of states are conflicted Trading states want to trade But may also have incentives to compete/fight PD/Chicken characteristics of the situation Extra-Dyadic: No tension between incentives Trading states clearly do not want other states to fight. Main implication: to the degree that international conflict inhibits trade, trading states should seek to deter third-party disputes. Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire Literature Three possible relationships connecting trade and conflict... Trade reduces conflict It seems natural to associate trade with peace. Commerce requires cooperation, if not exactly amity. Trading states should be reluctant to fight if war interrupts the flow of profits. (Montesquieu 1989, Kant 1957, Cobden 1903, Angell 1933, Axelrod 1984, Rosecrance 1985, Brooks 2005). Trading states could produce interests who prevail on leaders to refrain from fighting (Bentham 1781, McDonald 2009).
Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire Literature II It also seems natural to associate trade with war. The rise of Athens as a commercial empire, and the alarm which it inspired in Lacedaemon is said to have lead to the Peloponnesian war Thucydides 1960 (page 46). Realist theories of war rely on economic development as the primary mechanism for increasing state power (Morgenthau 1948, Waltz 1959, Claude 1962, Mearsheimer 2001). Socialists were equally persuaded that the search for foreign markets propelled class warfare (Hobson 1938, Lenin 1970). It is also possible that trade has no net effect on conflict. Effects of trade could cancel each other out (Morrow 1999). Trade is seldom valuable enough to interfere with the path to war (Gartzke, et al. 2001, Gartzke and Li 2003)
Literature III Other issues: The commercial enterprise can continue with third parties, even when conflicts interrupt direct bilateral ties. Recognition of this fact has led to attempts to estimate the elasticities of substitution present in sourcing goods and services elsewhere (Polachek, et al. 1999, Crescenzi 2002). Others have tried to model the effects of trade networks on whether states fight (Dorussen and Ward 2008, Hafner-Burton and Montgomery 2008, Lupu and Traag 2011) Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Theory 2 state world: States compete in a divide the dollar (like) game. Trade is a side payment for cooperating. Endogenous bargaining means that at least some of the deterrent effect of trade is absorbed in competing offers If the value of trade is low relative to value of dispute Fighting can still occur Trade must be large relative to cost of military contest in the game in order to have significant deterrent effect Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Theory II 3 state world: 2 states compete in a divide the dollar (like) game. 3rd state decides whether to exert influence/deter contest. 3 sets of trade relationships. 3rd state can intervene, prevent a dispute if: The 3rd state is large relative to the other two states It s value for trade is high relative to the cost of deterrence The implications of trade theory for conflict are clearer/less contingent for indirect trade relationships than for direct trade. Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Operationalizing Trade 3 types of trade relationships: Bilateral Monadic (state) Systemic Theoretical/empirical focus has been on bilateral trade Could use either/both bilateral/state variables for test Claims made in the literature apply equally to monadic trade If war is costly to trade (any trade), it should constrain/inform/transform trading states as predicted Trade with third parties is also impacted by conflict Conflicting incentives occur at both bilateral/state level Reverse causality problem strongest in bilateral data The clearest distinctions occur at the system level Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Figure 1: World Exports/Skewness of Exports Exports and Export Skewness 0 20 40 60 80 1950 1960 1970 1980 1990 2000 World Exports and Export Skewness (by dyad year)
Dependent Variable(s): Data MID Onset: standard COW MID Dataset, coding from Maoz. Independent Variables: Trade: state/dyad/system, Gleditsch (2002), Oneal % Russett (1999). Capabilities: state CINC scores, plus interaction term. Contiguity: six categories of decreasing physical proximity Distance: nat. log. of great circle distance between capitals. Alliance Status: COW dyadic defense pact/neutrality/entente. Major Power Status: COW criteria, dummy for status. Democracy: Polity IV data, monadic/dyadic/systemic values. Development: COW CINC, energy consumption. Temporal Dependence: Beck et al. (1998) Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire
Table 1: The Effect of Trade on Systemic Conflict (Annual Counts, Negative Binomial Regression) DV: Count of Basic Model Interpolated Development Temporal Trends MID Onset 1.1 1.2 1.3 1.4 1.5 1.6 Variable All MIDs Fatal MIDs All MIDs Fatal MIDs All MIDs Fatal MIDs Trade -0.2251-0.3187-0.1088-0.3651-0.3267-0.5414 (0.0936) (0.1387) (0.0293) (0.0955) (0.1017) (0.1509) Democracy 1.2139 2.9952-0.3471 0.0005 1.6684 1.1484 (0.8428) (1.4623) (1.3009) (0.0691) (0.9462) (1.1532) Development -1.1353 6.0005-13.9534 (0.3334) (3.2492) (4.4760) # of States -0.0038 0.0086 0.0018 0.0163-0.0110 0.0199 (0.0092) (0.0111) (0.0037) (0.0053) (0.0054) (0.0079) Year 0.0408 0.0062 0.0175 0.0257 0.0387 0.0043 (0.0302) (0.0337) (0.0044) (0.0065) (0.0124) (0.0239) Year 2-0.0005 0.0000 (0.0002) (0.0003) Year 3 0.0000 0.0000 (0.0000) (0.0000) Intercept -76.8246-12.3384-31.3643-48.7745 0.8693-1.0744 (58.2347) (65.0940) (8.1754) (12.1120) (0.2206) (0.4982) ln(α) -3.0523-15.6756-2.0615-14.01-2.1469-15.3017 (0.4734) (4.9907) (0.2478) (8.230) (0.2737) (3.8990) N 53 53 186 186 186 186 Log-likelihood -184.145-118.209-522.369-272.283-517.588-272.729 χ 2 (4,4,4,5,7,7) 22.43 7.71 924.59 245.65 291.55 283.42 Significance levels : : 5% : 1% : 0.1%. All reported significance levels are two-tailed.
Table 2: Effects of Systemic and Dyadic Trade on Dyadic Conflict (Annual Dummy, Logit) DV: MID Onset Basic Systemic Difference Interact Develop. Bilateral 2.1 2.2 2.3 2.4 2.5 2.6 Variable All MIDs All MIDs All MIDs All MIDs Fatal MIDs All MIDs State Trade (low) 1.942 2.823-1.558-23.29 107.5 0.322 (2.024) (2.026) (2.394) (18.35) (87.49) (37.57) State Trade (high) 2.547 20.78 14.86 17.72 (0.406) (3.286) (3.574) (3.248) System Trade -2.643-4.822-4.479 9.422-4.429 (1.340) (1.362) (1.385) (2.308) (1.372) State (low) Sys. 174.6-1056.1 34.11 (139.2) (775.4) (277.2) State (high) Sys. -142.3-88.11-120.7 (24.74) (29.16) (24.29) Bilateral (low) -0.0009 (0.0007) Bilateral (high) 0.0005 (0.0002) Bilat. (low) Sys. 0.0064 (0.0051) Bilat. (high) Sys. -0.0036 (0.0015) Democracy (low) -0.0851-0.0808-0.0872-0.0907-0.106-0.146 (0.0194) (0.0196) (0.0196) (0.0200) (0.0467) (0.0200) Democracy (high) 0.109 (0.0165) Distance -0.133-0.135-0.143-0.147-0.488-0.0983 (0.0315) (0.0317) (0.0306) (0.0306) (0.0290) (0.0626) Contiguity 2.253 2.244 2.246 2.239-0.329-0.547 (0.253) (0.256) (0.249) (0.250) (0.109) (0.104) Alliance 0.0182 0.0160-0.000155-0.0113-0.394 0.175 (0.139) (0.139) (0.135) (0.134) (0.219) (0.133) CINC (ratio) 2.090 2.041 2.123 2.175 2.154 2.079 (0.382) (0.382) (0.371) (0.361) (0.650) (0.352) Maj. Power 1.953 1.936 1.736 1.683 1.104 1.612 (0.175) (0.174) (0.192) (0.191) (0.281) (0.188) Econ. Dev. -0.0878 (0.0662) Intercept -3.204-2.945-2.729-2.765-0.0382-0.472 (0.294) (0.319) (0.324) (0.327) (0.717) (0.256) N 507283 501952 501952 501952 502105 501769 Significance levels : : 5% : 1% : 0.1%. All reported significance levels are two-tailed. The temporal Splines and peaceyear variables have been suppressed to save space. 16
Table 2: Blowup Effects of Systemic of and Key Dyadic Relationships Trade Dyadic Conflict in(annual Table Dummy, 2 Logit) DV: MID Onset Basic Systemic Difference Interact Develop. Bilateral 2.1 2.2 2.3 2.4 2.5 2.6 Variable All MIDs All MIDs All MIDs All MIDs Fatal MIDs All MIDs State Trade (low) 1.942 2.823-1.558-23.29 107.5 0.322 (2.024) (2.026) (2.394) (18.35) (87.49) (37.57) State Trade (high) 2.547 20.78 14.86 17.72 (0.406) (3.286) (3.574) (3.248) System Trade -2.643-4.822-4.479 9.422-4.429 (1.340) (1.362) (1.385) (2.308) (1.372) State (low) Sys. 174.6-1056.1 34.11 (139.2) (775.4) (277.2) State (high) Sys. -142.3-88.11-120.7 (24.74) (29.16) (24.29) Bilateral (low) -0.0009 (0.0007) Bilateral (high) 0.0005 (0.0002) Bilat. (low) Sys. 0.0064 (0.0051) Bilat. (high) Sys. -0.0036 (0.0015) Democracy (low) -0.0851-0.0808-0.0872-0.0907-0.106-0.146 (0.0194) (0.0196) (0.0196) (0.0200) (0.0467) (0.0200) Democracy (high) 0.109 (0.0165) Distance -0.133-0.135-0.143-0.147-0.488-0.0983
Relationship Between Dyadic Trade & MID Onset New MID onsets (within Dyad).00025.0003.00035.0004 0 20 40 60 80 100 Dyadic Trade (low): percentiles in sample from lowest to highest
Rel. Between Systemic Trade and Dyadic MID Onset New MID onsets (within Dyad) 0.0001.0002.0003 0 20 40 60 80 100 Systemic Trade: percentiles in sample from lowest to highest
Systemic & State Trade (high) & MID Onset 1 Pr MID 0.5 0 1.2 0.06 0.9 0.08 0.1 System Trade 0.6 0.12 0.3 0.14 0 State Trade high
Effect of Systemic Trade on MID Onset for One Weak and One Powerful State New MID onsets (within Dyad).0001.0002.0003.0004.0005 0 20 40 60 80 100 Systemic Trade: percentiles in sample from lowest to highest
Effect of Systemic Trade on MID Onset for Powerful States New MID onsets (within Dyad) 0.001.002.003 0 20 40 60 80 100 Systemic Trade: percentiles in sample from lowest to highest
Table 3: Effects of Systemic and Dyadic Trade on Dyadic Conflict (Annual Dummy, Logit) DV: MID Onset Basic Systemic Fatal 3.1 3.2 3.3 Variable All MIDs All MIDs All MIDs All MIDs Fatal MIDs Hegemon Capability 2.668 (0.987) 8.435 (1.309) 12.00 (1.946) Hegemon Trade -0.235 (0.0911) -0.382 (0.117) -1.167 (0.210) State Trade (low) -2.615 (2.620) -4.260 (17.93) 169.9 (86.28) State Trade (high) 2.357 (0.408) 23.83 (3.224) 16.77 (3.018) System Trade 12.27 (2.345) 21.30 (3.559) State System (low) 30.21 (137.2) -1623.3 (765.7) State System (high) -164.3 (24.44) -96.40 (25.09) Democracy (low) -0.0934 (0.0207) -0.105 (0.0451) Distance (log) -0.0444 (0.0613) -0.147 (0.0307) -0.488 (0.0280) Contiguity (dummy) -0.611 (0.103) 2.233 (0.250) -0.299 (0.105) Alliance (dummy) -0.0295 (0.137) -0.0105 (0.135) -0.496 (0.221) Capability Ratio 1.987 (0.370) 2.113 (0.362) 2.133 (0.634) Maj. Power (either) 1.700 (0.187) 1.610 (0.189) 0.986 (0.268) Development (energy) -0.00916 (0.0167) Intercept -0.807 (0.294) -5.638 (0.491) -4.855 (1.018) N 523663 496774 502105 Significance levels : : 5% : 1% : 0.1%. All reported significance levels are two-tailed. The temporal Splines and peaceyear variables have been suppressed to save space.
Erik Gartzke UCSD egartzke@ucsd.edu The System Made Me Stop Doing It The Indire Tentative Conclusions Trade affects conflict, but not (quite) in the way we thought: Trading states are not significantly less dispute prone Trade-intensive systems are significantly less disputatious The biggest effect of trade in diminishing conflict is on the most disputatious dyads (asymmetric trading relationships) The pacific effects of trade transfer to non-trading states. It seems reasonable to suppose that the chief beneficiaries of trade are responsible for shifting the effect of liberal peace Need to demonstrate the causal mechanism, etc.