Permalink
Switch branches/tags
Nothing to show
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
417 lines (297 sloc) 14.8 KB
---
title: "A Unified Explanation of Territorial Conflict"
subtitle: POSC 3610 -- International Conflict
author: Steven V. Miller
institute: Department of Political Science
titlegraphic: /Dropbox/teaching/clemson-academic.png
date:
fontsize: 10pt
output:
beamer_presentation:
template: ~/Dropbox/miscelanea/svm-r-markdown-templates/svm-latex-beamer.tex
latex_engine: xelatex
dev: cairo_pdf
fig_caption: false
slide_level: 3
make149: true
mainfont: "Open Sans"
titlefont: "Titillium Web"
---
```{r setup, include=FALSE, cache=F, message=F, warning=F, results="hide"}
knitr::opts_chunk$set(cache=TRUE)
knitr::opts_chunk$set(fig.path='figs/')
knitr::opts_chunk$set(cache.path='cache/')
knitr::opts_chunk$set(
fig.process = function(x) {
x2 = sub('-\\d+([.][a-z]+)$', '\\1', x)
if (file.rename(x, x2)) x2 else x
}
)
```
```{r loadstuff, include=FALSE}
knitr::opts_chunk$set(cache=TRUE)
options(knitr.kable.NA = '')
library(tidyverse)
library(stevemisc)
library(countrycode)
library(knitr)
library(stargazer)
library(dotwhisker)
```
# Introduction
### Goal for Today
*Highlight why disputes, once initiated, escalate to war and why selection matters.*
### Territorial Disputes and War
Vasquez (1993, 1995) argues territorial issues are a root cause of war.
- He draws from evolutionary psychology, primitive anthropology, and sociobiology to make that argument.
- However, data limitations hindered an issue-politics paradigm in IR.
The release of version 2.0 of CoW-MID changed this.
- From there, a slew of scholarship argued territorial MIDs are more likely to lead to war.
### Non-Random Sampling and Selection Bias
However, MIDs are not a "random data-generating process" (DGP).
- i.e. Second Kashmir War (MID\#1312) was not a roll of the dice.
- The India-Pakistan dyad is not randomly drawn from an urn.
### Non-Random Sampling and Selection Bias
However, *all* our statistical models assume random DGP.
- This is the inference in inferential statistics.
- i.e. we infer about the population with a sample of the population.
MIDs are not randomly sampled. They're event data.
### Non-Random Sampling and Selection Bias
Important for our purposes. MIDs are events that, by definition, carry a higher probability of war.
- We don't know yet if territorial issues are truly war-prone.
- We just know territorial MIDs are more likely to lead to war and most wars are fought over territory.
### Possible Relationships Between Territory and War
![Table 1 in Senese and Vasquez (2003)](tab1.png)
### Table 1 in Senese and Vasquez (2003)
Our intuition suggests the top-left quadrant.
- i.e. disputed territory leads to MIDs.
- MIDs over distribution of territory lead to war.
### Table 1 in Senese and Vasquez (2003)
Other empirical scenarios are plausible and consistent with the data.
- States may have a lot of territorial MIDs but those rarely escalate (i.e. top-right)
- Territorial MIDs lead to war but states are reticent to initiate MIDs (i.e. bottom-left)
- We're all wrong about the bellicosity of territorial issues (i.e. bottom-right)
# Senese and Vasquez (2003)
### Senese and Vasquez (2003)
Senese and Vasquez (2003) will use a selection model to test the following:
- Dyads with territorial claims are more likely to culminate in MIDs.
- Territorial MIDs are more likely to escalate toward war.
### Senese and Vasquez's (2003) Design
*DVs*:
- onset of a MID, war as highest fatality-level
This is functionally similar what Reed (2000) did in his model.
### Senese and Vasquez's (2003) Design
*Primary IVs*:
- Selection model: territorial claim (Huth, 1996)
- War/escalation model: Policy/Regime/Other MIDs to baseline of territorial MIDs.
If the steps-to-war intuition is correct:
- Dyad-years with territorial claims are more likely to lead to MIDs. And:
- MIDs over other issues should be less war-prone than territorial MIDs.
### Senese and Vasquez's (2003) Design
*Other IVs (i.e. controls)*
- Contiguity
- Minimum democracy (i.e. weak-link specification)
- Joint alliance
- Economic development
- Great power status (i.e. major-major, minor-minor)
- Baseline: a major-minor (e.g. USA-Canada) dyad-year
### Senese and Vasquez's (2003) Design
Important methodological notes:
- Model includes peace years/splines for temporal auto-correlation.
- *Unit of analysis*: non-directed dyad-years
- *Temporal domain*: 1919-1992
- *Statistical method*: Heckman selection model
- This will model selection into MID and then escalation to war simultaneously.
###
```{r sv2003-mod3-tab2, eval=F, echo=F, fig.width=14, fig.height=8.5, warning=F, message=F}
tribble(
~term, ~estimate, ~std.error,
"Policy MID", -1.107, .097,
"Regime MID", -1.081, .175,
"Other MID", -.913, .509,
"Minimum Democracy", -.023, .011,
"Joint Alliance", -.273, .120,
"Contiguity", -.193, .100,
"Economic Development", -.391, .105,
"Major-Major Status", .618, .175,
"Minor-Minor Status", -.316, .104
) %>%
mutate(statistic = estimate/std.error,
p.value = 1.96*pnorm(-abs(statistic)),
model = "War Onset") %>%
dwplot(.,dot_args = list(aes(colour = model, shape = model), size = 1.1)) + theme_steve_web() +
xlab("Coefficient Estimate") +
geom_vline(xintercept = 0, colour = "grey60", linetype = 2) +
theme(legend.position = "bottom") +
labs(title = "What Explains War Onset? (Model 3 in Senese and Vasquez' [2003] Table 2)",
subtitle = "MIDs over regime type and general policy concerns are much less likely to escalate to war than MIDs over territory.",
caption = "Reproduction of Model 3 in Table 2 of Senese and Vasquez (2003).") +
scale_colour_discrete(name = "Model") + scale_shape_discrete(name = "Model")
```
![](tab2.png)
### What Do We Know About War Onset?
Senese and Vasquez (2003) have the following findings about war onset:
- Regime and Policy MIDs are much less likely to escalate to war than territorial MIDs.
- Other MIDs drifts negative but estimated effect is diffuse.
- Joint democracy "weak-link" has a small but discernible negative effect on escalation.
- Allies are unlikely to escalate MIDs to war.
- Economic developed dyads are unlikely to escalate MIDs to war.
- Major-Majors (e.g. France-Germany) are more war-prone than Major-Minors (e.g. USA-Canada).
- Also: Minor-Minors (e.g. Canada-Mexico) are less war-prone than Major-Minors.
###
```{r sv2003-mod4-tab2, eval=F, echo=F, fig.width=14, fig.height=8.5, warning=F, message=F}
tribble(
~term, ~estimate, ~std.error, ~model,
"Territorial Claim", .716, .036, "MID Onset",
"Minimum Democracy", -.034, .002, "MID Onset",
"Joint Alliance", .027, .033, "MID Onset",
"Contiguity", .941, .032, "MID Onset",
"Economic Development", .222, .021, "MID Onset",
"Major-Major Status", .457, .069, "MID Onset",
"Minor-Minor Status", -.615, .026, "MID Onset",
"Policy MID", -1.107, .097, "War Onset",
"Regime MID", -1.081, .175, "War Onset",
"Other MID", -.913, .509, "War Onset",
"Minimum Democracy", -.023, .011, "War Onset",
"Joint Alliance", -.273, .120, "War Onset",
"Contiguity", -.193, .100, "War Onset",
"Economic Development", -.391, .105, "War Onset",
"Major-Major Status", .618, .175, "War Onset",
"Minor-Minor Status", -.316, .104, "War Onset"
) %>%
mutate(statistic = estimate/std.error,
p.value = 1.96*pnorm(-abs(statistic))) %>%
dwplot(.,dot_args = list(aes(colour = model, shape = model), size = 1.1)) + theme_steve_web() +
xlab("Coefficient Estimate") +
geom_vline(xintercept = 0, colour = "grey60", linetype = 2) +
theme(legend.position = "bottom") +
labs(title = "A Unified Model of MID Onset and War Onset (Model 4 in Senese and Vasquez's [2003] Table 2)",
subtitle = "Consistent with the steps-to-war argument, territorial claims to MIDs and MIDs over territory are the most war-prone type of MID.",
caption = "Reproduction of Model 4 in Senese and Vasquez (2003).") +
scale_colour_discrete(name = "Model") + scale_shape_discrete(name = "Model")
```
![](tab2.png)
### What Do We Know About Conflict?
Implications of Senese and Vasquez's (2003) unified model:
- *Importantly:* territorial claims lead to MIDs and MIDs over territory are the most war-prone.
- This is ultimately what they're focused on testing.
### What Do We Know About Conflict?
Other findings:
- Joint democracy is pacifying at both phases of conflict.
- Economically developed states have more MIDs, but are unlikely to escalate.
- Both are inconsistent with Reed (2000) whose findings may be sensitive to his temporal domain.
- Contiguity leads to MIDs but most wars involve non-contiguous states.
- This is still the case, btw, in the GML MID data.
###
```{r sv-tab3-initiation, eval=T, echo=F, fig.width=14, fig.height=8.5, warning=F, message=F}
tribble(
~Condition, ~fit, ~lwr, ~upr,
"Presence of a Territorial Claim", .0088, .0073, .0105,
"Absence of a Territorial Claim", .0012, .0011, .0013,
) %>%
ggplot(.,aes(colour = Condition)) +
# geom_hline(yintercept = 0.5, colour = gray(1/2), lty = 2) +
geom_linerange(aes(x = Condition, ymin = lwr,
ymax = upr),
lwd = 1, position = position_dodge(width = 1/2)) +
geom_pointrange(aes(x = Condition, y = fit, ymin = lwr,
ymax = upr),
lwd = 1/2, position = position_dodge(width = 1/2),
fill = "WHITE") +
coord_flip() + theme_steve_web() +
ylab("Predicted Probability (with Confidence Intervals)") +
xlab("") +
labs(title = "The Presence of a Territorial Claim Increases the Likelihood of MID Onset by Over 600%",
subtitle = "MIDs are still (fortunately) rare events but the presence of a territorial claim has a large effect on conflict initiation.",
caption = "Reproduction of first two rows in Table 3 of Senese and Vasquez (2003)")
```
###
```{r sv-tab3-escalation, eval=T, echo=F, fig.width=14, fig.height=8.5, warning=F, message=F}
tribble(
~Condition, ~fit, ~lwr, ~upr,
"Presence of a Territorial MID", .312, .277, .348,
"Presence of a Policy MID", .055, .043, .070,
"Presence of a Regime MID", .060, .033, .096
) %>%
ggplot(.,aes(colour = Condition)) +
# geom_hline(yintercept = 0.5, colour = gray(1/2), lty = 2) +
geom_linerange(aes(x = Condition, ymin = lwr,
ymax = upr),
lwd = 1, position = position_dodge(width = 1/2)) +
geom_pointrange(aes(x = Condition, y = fit, ymin = lwr,
ymax = upr),
lwd = 1/2, position = position_dodge(width = 1/2),
fill = "WHITE") +
coord_flip() + theme_steve_web() +
ylab("Predicted Probability (with Confidence Intervals)") +
xlab("") +
labs(title = "Territorial MIDs are More than Four Times as Likely to Culminate in War Relative to MIDs Over Other Issues",
subtitle = "The predicted probability of war onset is still around .3 but the difference in bellicosity is clear.",
caption = "Reproduction of first three rows in Table 3 of Senese and Vasquez (2003)")
```
# Conclusion
### Conclusion
Senese and Vasquez (2003) present a unified explanation of territorial conflict.
- Territorial claims between states are likely to become militarized.
- Territorial MIDs are likely to lead to war.
It's important to separate correlates of dispute onset from dispute escalation.
- Factors that promote MID onset need not promote war onset.
<!-- ### How This Relates to International Conflict
Consider the familiar argument about territorial disputes and war.
- We know wars are fought over territory more than any other issue.
- We have arguments (e.g. Vasquez, 1993/1995) to explain this.
So, we know territorial disputes are likely to lead to war.
- We *don't* know that territory is truly war-prone.
How does this make sense?
### Possible Relationships between Territory and War
![Table 1 in Senese and Vasquez (2003)](tab1.png)
### Table 1 in Senese and Vasquez (2003)
Our intuition suggests top-left quadrant.
- i.e. territorial issues lead to disputes. Territorial disputes escalate to war.
Other empirical scenarios are plausible.
- Top-right quadrant: states have a lot of territorial disputes, but those rarely escalate to war.
- Bottom-left quadrant: states are reticent to initiate territorial disputes because those that are initiated likely become wars.
- Bottom-right quadrant: we're all wrong about the bellicosity of territorial issues.
Notice the conceivable scenarios still consistent with the claim territorial disputes lead to war?
###
![Table 1 in Reed (2000)](reed2000-tab1.png)
### Interpreting Reed's (2000) selection model
Implications of Reed's (2000) model:
- Power parity leads to MIDs, but MIDs between equals are less likely to lead to war.
- *Joint democracy leads to fewer MIDs, not necessarily fewer wars.*
- Would be consistent with Senese's (1997) finding.
- Likewise: satisfied states have fewer MIDs, not necessarily fewer wars.
- Allies still have disputes, just unlikely to escalate to war.
###
![Table 2 in Senese and Vasquez (2003)](tab2.png)
### Table 2 in Senese and Vasquez (2003)
Table 2 supports the Senese and Vasquez' hypothesis about the danger of disputed territory.
- The war onset models tell us what we already know.
- Other issue-type MIDs are not as war-prone as territorial MIDs.
- The joint models illustrate territorial claims in a dyad lead to MIDs more than dyads without territorial claims.
- Other issue-type MIDs are not as war-prone as territorial MIDs in a sample selection model.
###
![Table 3 in Senese and Vasquez (2003)](tab3.png)
### Table 3 in Senese and Vasquez (2003)
How to read Table 3:
- Predicted probability of MID onset with territorial claim: .0088
- Predicted probability of MID onset without territorial claim: .0012
- These are fortunately "rare events" but notice the lower-bound of the territorial claim probability does *not* overlap the upper-bound "no-territorial claim" probability.
- Likewise: predicted probability of territorial MID escalating to war: .312
- Predicted probability of a policy or regime MID escalating to war: no higher than .06
## Strategic Factors
### Influence Strategies
- Bullying (escalation for any response short of total acquiescence)
- Reciprocating (tit-for-tat responses)
- Appeasing (opposite of bullying: use of positive inducements instead of negative ones)
- Trial-and-error (i.e. "going with what works")
### Results of Influence Strategies
- Bullying is most frequently used strategy; also most likely to escalate.
- Reciprocating is most successful
- Appeasing has bad name post-Munich Crisis; also war-prone
- Trial-and-error leads to mixed results.
# Conclusion
### Conclusion
It's important to separate correlates of dispute onset from dispute escalation.
- Factors that promote MID onset need not promote war onset.
- This can have important implications for scholarship like democratic peace. -->