War and statebuilding in Latin America

PSCI 2227: War and State Development

Prof. Brenton Kenkel
brenton.kenkel@gmail.com

Vanderbilt University

February 4, 2026

Recap

Warmaking and statemaking

  • Tilly’s theory
    • War makes states — develop or die out
    • Institutional developments persist into peace — and increase the incentive for further war
    • Commercial/capitalist states advantaged in both ends of process
  • Abramson’s skeptical take
    • States didn’t grow when war became more capital-intensive
    • Smaller states were more likely to survive
    • Urbanization drove the formation of — mostly small — states

Today’s agenda

  1. Latin America as a “hard case” for the theory
  2. Thies’ alternative: rivalry (threat of war) as driver of statebuilding
  3. Reanalyzing the data on war, rivalry, and tax capacity

Theory

Generality of the war-and-statebuilding logic

Reminder of the basic logic

  • Success in military competition requires troops and equipment
  • Troops and equipment cost money — a lot of money
  • Large territorial states are best equipped to raise that money
  • … and can use these tools to consolidate power internally too

Theory came out of studying European history

…but when you boil it down to these parts, none of it is Europe-specific

Problems for the Tilly story outside Europe

Why the theory is plausible for Europe

  • Europeans fought lots of wars against each other
  • Sovereign, territorial states emerged relatively early in Europe

But it’s hard to see “war made the state” elsewhere

  • Interstate war has been much less common outside Europe
  • And yet sovereign territorial states came to dominate worldwide
  • Maybe European colonialism can explain emergence of territorial states, but not as much persistence in post-colonial era

Latin America as test case

Latin America is a “hard case” for the non-European application of the theory

Basic facts to contend with:

  1. Vast majority of region gained independence 1800s–1830s

  2. Major interstate wars have rarely occurred in the region

War Years Belligerents Deaths
War of the Triple Alliance 1864–1870 Paraguay vs Argentina, Brazil, Uruguay ~300,000–400,000
War of the Pacific 1879–1884 Chile vs Bolivia, Peru ~15,000–20,000
Chaco War 1932–1935 Bolivia vs Paraguay ~90,000–100,000
  1. Borders have barely changed since independence movements

Stability of Latin American borders

1830

present

Thies’ theory: Rivalry made the state

Why is war important in Tilly’s model?

  1. Impetus to raise revenue and become more efficient at coercion
  2. Prunes out political organizations that fall behind

The first of these should still operate under just the threat of war

Key concept for Thies: interstate rivalry as longstanding war threat

  • “enduring” rivalry: 6+ low-level conflicts in 20-year period
  • “strategic” rivalry: policymakers perceive rival as serious military threat

Unlike war, rivalry varies a lot over space + time in Latin America

Not many wars in Latin America, 1900–2000

But many strategic rivalries, 1900–2000

(enduring rivalries are in between)

Thies’ expectations

  1. Sustained external rivalry spurs statebuilding
    • Provides impetus to raise revenue and become efficient
    • Expect gradual increase in state capacity rather than sudden bursts
  1. Internal conflict hurts statebuilding
    • Undermines legitimacy of state
    • Saps resources that would be used to build capacity
  1. Interstate war doesn’t help or hurt
    • Key difference with medieval Europe: common access to debt finance
    • Less need to raise revenue domestically
    • (more on this in a couple weeks when we read Queralt)

Data

Measuring statebuilding

Thies collects annual data on South America, 1900–2000

  • didn’t do Central America because some sociologist didn’t either 🤷🏻‍♂️

Dependent variable: tax ratio

  • Annual tax revenue as a fraction of gross domestic product
    • e.g.: tax_ratio for Argentina 1975 = 0.112
    • means tax revenues were 11.2% of Argentina’s economy
  • Proxy for the state’s extractive capacity (success as a stationary bandit)

Implicit assumption: how much the state actually raises is proportional to how much it can raise

Redoing the data analysis here

Sorry to say, this article is basically a disasterclass in how to not do statistics

  • No presentation of descriptive statistics, just complicated regressions
  • “Bad controls”: variables possibly affected by the treatment
    • Treatments = interstate war history, civil war history, rivalries
    • Bad controls = democracy, debt, GDP, inflation, trade, agriculture (i.e., everything but the time trend)
  • Technical: poor measures of statistical error (no clustering by country)
  • Makes lots of causal claims using exclusively correlational evidence

So I pulled the independent variables from the tables, reconstructed the tax ratio from MOxLAD, and redid the analysis how I like it

Interstate war and tax ratio

Civil war and tax ratio

Strategic rivalries and tax ratio

Did rivalry make the state in Argentina and Chile?

What about the rest of Latin America?

Tax ratio compared to regional average

Tax ratio compared to regional average

Regression results (pooled)

(1) (2) (3) (4)
* p < 0.05
p-values from wild bootstrap clustered by country (999 replications). All models include cubic time trend.
Interstate wars (last 10 years) -0.017*
SE = 0.007
p = 0.020
Civil wars (last 10 years) -0.005*
SE = 0.002
p = 0.005
Strategic rivalries -0.002
SE = 0.012
p = 0.894
Enduring rivalries -0.011
SE = 0.012
p = 0.382
Number of country-years 768 768 768 768
R2 0.279 0.289 0.270 0.272

Regression results (country fixed effects)

(1) (2) (3) (4)
* p < 0.05
p-values from wild bootstrap clustered by country (999 replications). All models include cubic time trend and country fixed effects.
Interstate wars (last 10 years) -0.004
SE = 0.004
p = 0.333
Civil wars (last 10 years) -0.001
SE = 0.001
p = 0.441
Strategic rivalries -0.010*
SE = 0.005
p = 0.042
Enduring rivalries 0.009
SE = 0.006
p = 0.182
Number of country-years 768 768 768 768
R2 0.661 0.661 0.667 0.661

Conclusions

Both types of war negatively correlated with tax ratios

Rivalry also negatively correlated with tax ratios

No correlation robustly distinguishable from statistical noise

“Eye test” doesn’t suggest any particularly strong patterns either

My conclusions:

  • Whatever variation exists in Latin American tax ratios, it’s hard to trace to actual war or preparations for wars that didn’t happen
  • Tilly is a much better guide to Europe (even if flawed) than to Latin America

Wrapping up

What we did today

  • Examined whether Tilly’s bellicist theory travels outside Europe
  • Latin America: few interstate wars, stable borders, yet sovereign states persisted
  • Thies’ alternative: interstate rivalry (threat of war) spurs statebuilding
  • Looked at the data: wars, rivalries, and tax ratios across South America 1900–2000
  • Found little robust evidence that war or rivalry drove state capacity in the region

Next time

War and statebuilding in China

  • Dincecco and Wang 2018, “Violent Conflict and Political Development over the Long Run: China Versus Europe”
  • Reading guide to be posted by Friday

Have a good weekend!