The Economic Consequences of U.S. Military Interventions

A data-driven investigation into how U.S. military interventions shape the U.S. economy by measuring their causal impact on stock performance across 15 sectors.

Scroll to explore

Introduction

$7 trillion. That's roughly how much the U.S. has spent on military operations since 2001 alone. But beyond the direct costs, every intervention sends ripples through financial markets, creating winners, losers and a trail of economic consequences that most investors never see coming.

In this project, we decode those ripples. By aligning 29 carefully selected U.S. military interventions spanning 58 years with daily stock data from 5,053 companies across 15 sectors, we reveal how geopolitical shocks translate into market volatility, sector rotations and measurable abnormal returns.

Our focus isn't on judging the interventions themselves... It's on understanding the economic footprint they leave behind. Which sectors suffer? Which ones thrive? And perhaps most importantly: Do the empirical insights align with public opinion?

Our Analysis at a Glance

0 Military Interventions
0 Economic Sectors
0 Years of Data
0 Stocks Analyzed

What We Expect to Find

Before diving into the data, we started with common assumptions about how markets react to military interventions. Here are the hypotheses we set out to test.

H1 Energy Dominance

"The Energy sector will be hit hardest by military interventions because wars disrupt oil supply chains."

Wars in oil-rich regions like the Middle East historically trigger oil price spikes and supply fears.

H2 Immediate Panic

"Markets will crash immediately when wars start, with panic selling across all sectors."

Uncertainty and fear should trigger immediate risk-off behavior from investors.

H3 Middle East Focus

"Interventions in the Middle East will have the strongest market effects due to oil dependency."

The region controls major oil reserves and strategic chokepoints like the Strait of Hormuz.

H4 Objectives Matter

"The objective of an intervention shapes market reaction, regime changes will rattle markets more than limited protective missions."

Aggressive regime changes signal prolonged instability, while protective missions are seen as limited in scope.

H5 Intensity-Impact

"High-intensity wars will cause bigger market crashes. More violence means more economic damage."

Intuition suggests that deadlier, longer conflicts should produce proportionally larger market reactions.

H6 Duration Matters

"Longer wars will cause more cumulative damage to markets than short, decisive operations."

Extended uncertainty and resource drain should compound negative effects over time.

Now let's evaluate those assumptions against 58 years of market data.

Finance 101

New to finance? Here's everything you need to understand our analysis.

CAPM, Capital Asset Pricing Model

Before we dive into the metrics, let's understand the foundation of our analysis. CAPM is a model that predicts what a stock should return based on how the overall market is doing.

The Weather Forecast Analogy

Think of CAPM like a weather forecast for stocks. Just as meteorologists predict tomorrow's temperature based on historical patterns and current conditions, CAPM predicts a stock's "expected return" based on how it has historically moved with the market. If the forecast says 70°F but it's actually 85°F, that 15° difference is "abnormal", something unusual happened. Same idea with stock returns.

How It Works

Expected Return = Risk-Free Rate + β × (Market Return - Risk-Free Rate)

The key ingredient is β (beta), a number that measures how sensitive a stock is to market movements. A β of 1.5 means the stock typically moves 1.5× as much as the market. We calculate β using "peacetime" (non-intervention) days, so we can isolate what's truly abnormal during wars.

Basic Stock Metrics

Adjusted Close

The stock's closing price after accounting for corporate actions like stock splits and dividends. We use this instead of raw prices because it makes returns comparable over long time periods.

When a company pays a dividend, its stock price drops because cash left the company. Without Adjusted Close, it would look like you lost money, even though you just received that cash in your pocket.
Volume

The number of shares traded in a day. High volume means lots of buying and selling activity, often indicating strong investor interest or reaction to news.

Like foot traffic in a store, more visitors usually means a big sale or an event is happening. Low volume is like a quiet Tuesday morning.
Return

The percentage change in price from one day to the next. A +2% return means the stock gained 2% in value, a -3% return means it lost 3%.

Formula: (Today's Price - Yesterday's Price) / Yesterday's Price × 100
Volatility

How much a stock's price swings up and down. High volatility = big price swings (risky). Low volatility = stable prices (safer). We measure this using a 20-day rolling window.

The Roller Coaster Analogy
Imagine two roller coasters: one with gentle hills and smooth turns, another with steep drops and sharp twists. Both might end at the same height, but the ride is very different. Volatility measures how wild the ride is, not where you end up, but how bumpy the journey. During wars, we expect more "wild rides" as investors react to uncertainty.

Event Study Metrics

To measure the impact of military interventions, we compare what actually happened to what we expected to happen based on normal market behavior. The following metrics allow us to isolate and analyze these anomalies.

AR, Abnormal Return

The difference between a stock's actual return and the return it was expected to achieve based on its normal behavior relative to the market on a normal day (no intervention). It isolates the specific impact of an event (like a war) by removing general market trends.

The Student Grade Analogy
Imagine a student who usually scores 80% on tests. If they score 90% after studying with a new method, the "abnormal" improvement is +10%, that's the effect of the new method. Similarly, if a stock usually returns 1% when the market returns 1%, but during a war it returns 3%, the abnormal return is +2%. That extra 2% is what we attribute to the war's impact.
Formula: AR = Actual Return - Expected Return (from CAPM)
Positive AR = Stock did better than expected
Negative AR = Stock did worse than expected
CAR, Cumulative Abnormal Return

The sum of all abnormal returns over a period (like the 20 days after an intervention starts). This shows the total impact of an event, not just the day-to-day changes.

The Savings Account Analogy
Think of CAR like tracking your savings over a month. Each day you might save a little more or less than planned (daily AR). At month's end, you add up all those differences to see your total "abnormal" savings. A CAR of +5% means the sector gained 5% more than expected over the entire event window, like finding an extra $50 in your account after a month of small daily surprises.
Formula: CAR = AR(day 1) + AR(day 2) + ... + AR(day n)
CAR = +5% = Sector gained 5% more than expected over the event window
CAR = -5% = Sector lost 5% compared to expectations (underperformed)
AV, Abnormal Volatility

Similar to abnormal return, but for volatility. It measures whether a stock became more or less volatile than expected during an event. Higher AV means more uncertainty and risk.

The Heart Rate Analogy
Your resting heart rate might be 70 bpm. During a scary movie, it jumps to 100 bpm, that extra 30 bpm is "abnormal" heart rate caused by the movie. Similarly, if a stock normally swings 2% daily but during a war it swings 5%, the abnormal volatility is +3%. It tells us how much more "nervous" the market became because of the event.
Formula: AV = Actual Volatility - Expected Volatility
Positive AV = More uncertainty/risk than expected (market is nervous)
Negative AV = Less uncertainty than expected (market stayed calm)
CAV, Cumulative Abnormal Volatility

The sum of abnormal volatility over a period. A high CAV indicates sustained uncertainty throughout an event, while a low CAV suggests markets remained calm.

The Stress Test Analogy
Imagine tracking your stress levels during a month-long project. Some days are more stressful than expected, others less. CAV is like adding up all those "extra stress" days. A high CAV means the market was consistently more jittery than normal throughout the war, like a month where every day felt more stressful than usual. A low or negative CAV means the market handled the uncertainty surprisingly well.
Formula: CAV = AV(day 1) + AV(day 2) + ... + AV(day n)
High CAV = Sustained period of high uncertainty and market stress
Low/Negative CAV = Market remained calm or stable throughout the event

Understanding the 15 Sectors

Our analysis groups stocks into 15 sectors. Here's what each one contains and why it might react differently to military interventions.

Energy

Oil, gas, coal and renewable energy companies.

Examples: ExxonMobil, Chevron, ConocoPhillips
Technology

Software, hardware, semiconductors and IT services.

Examples: Apple, Microsoft, NVIDIA
Health Care

Pharmaceuticals, biotech, hospitals and medical devices.

Examples: Johnson & Johnson, Pfizer, UnitedHealth
Financial Services

Banks, insurance, investment firms and credit services.

Examples: JPMorgan, Berkshire Hathaway, Visa
Consumer Cyclical

Retail, autos, hotels, restaurants, things people buy when times are good.

Examples: Amazon, Tesla, Home Depot
Consumer Defensive

Essential goods people buy regardless of economy, food, beverages, household products.

Examples: Procter & Gamble, Coca-Cola, Walmart
Consumer Staples

Food, beverages, tobacco and household essentials, stable demand sectors.

Examples: Costco, PepsiCo, Colgate-Palmolive
Industrials

Aerospace, defense, machinery, construction and transportation.

Examples: Boeing, Lockheed Martin, Caterpillar
Basic Materials

Mining, chemicals, metals and paper, raw materials for everything.

Examples: Dow, Freeport-McMoRan, Nucor
Communication Services

Media, entertainment, social platforms and interactive services.

Examples: Google (Alphabet), Meta, Disney
Utilities

Electric, gas and water utilities, essential services with stable demand.

Examples: Duke Energy, Southern Company, NextEra
Real Estate

REITs, property management and real estate development.

Examples: American Tower, Prologis, Simon Property
Telecommunications

Telecom carriers, wireless services and network infrastructure.

Examples: AT&T, Verizon, T-Mobile
Consumer Discretionary

Non-essential goods and services, luxury items, entertainment, travel.

Examples: Nike, Starbucks, Booking Holdings
Miscellaneous

Diversified companies and conglomerates that span multiple industries.

Examples: Various holding companies and diversified firms
Why Sectors Matter for War Analysis

Different sectors react differently to military interventions. Energy might seem most vulnerable to Middle East conflicts, but our data reveals surprises. Industrials (including defense contractors) often benefit from increased military spending. Consumer Defensive sectors are typically stable during crises, but showed unexpected volatility. Understanding these 15 sectors helps explain why our hypotheses were sometimes confirmed and sometimes challenged.

Data & Methods

Datasets

  • Military Intervention Project (MIP)
    376 U.S. military interventions from 1776 to 2017 (Fletcher Center for Strategic Studies), including start/end dates, objectives, target countries, hostility indicators and fatalities.
  • NASDAQ Stock Market Data (Kaggle)
    Daily OHLCV prices (Open, High, Low, Close, Adjusted Close, Volume) for 5,324 stocks, covering 1962-2020.
  • US Stock Metrics & Company Info (Kaggle)
    Sector and industry classifications merged from multiple sources (NASDAQ screener, US stock metrics dataset) to build a comprehensive stock universe.
  • Symbols Metadata
    Security names and exchange information for ticker symbol mapping and validation.

Methodology

  • Data Loading & Integration
    Load and merge stock market data with sector/industry classifications
  • Data Cleaning & Outlier Handling
    Remove ETFs, handle missing values, standardize formats and filter suspicious returns
  • Volume-Weighted Sector Aggregation
    Aggregate stock-level returns and volatility to sector level using volume weights
  • MIP Data Processing
    Clean intervention records, standardize dates and map objectives to canonical categories
  • PCA Intensity Index
    Combine duration, hostility and fatalities into a single intensity score (0-1)
  • Event Window Construction
    Define pre/during/post windows around each intervention
  • Overlap Resolution
    Keep non-overlapping interventions by prioritizing the most intense
  • Event Panel Merge
    Link daily sector data with intervention phases for causal analysis
  • CAPM-Based Abnormal Returns
    Estimate expected returns from normal days to isolate war-driven effects

Methodology Deep Dive

AGGREGATION
Volume-Weighted Sectors

Aggregate stock-level metrics to sector level using trading volume as weights.

Return
rs,t = Σ wi·ri
Volatility
vs,t = Σ wi·vi
Weight
wi = vi/Σvj
Why sector-level? Stock-level analysis is too noisy; industry labels are unreliable (>75% "Unknown"). Volume proxies market cap when unavailable, capturing liquidity & economic activity.
INTENSITY
PCA Intensity Index

Single score (0-1) combining multiple intervention characteristics via Principle Component Analysis.

Duration
log(1+days)
Escalation
hostility avg
Lethality
log(1+deaths)
Why PCA? Instead of arbitrary weights, PCA finds the linear combination capturing maximum variance across interventions. PC1 explains 55% of variance, sign is fixed so higher lethality = higher intensity.
TEMPORAL
Event Windows
Pre-Event
-w to -1 days

Market anticipation & geopolitical tension signals

During Event
Day 0 ? End

Active intervention with real-time market reactions

Post-Event
+1 to +w days

Recovery phase & persistent effect measurement

w = min(duration, 20 days), longer wars generate earlier rumors and media coverage, but short interventions shouldn't get oversized context windows.
MARKET MODEL
Abnormal Returns Calculation
Rs,t = αs + βs · Rm,t-1 + εs,t
Rs,t
Sector return
as
Sector intercept
βs
Market sensitivity
Rm,t
Market return
es,t
Residual (AR)
ARs,t = Rs,t - (α̂s + β̂s · Rm,t-1)
Key insight: We estimate α̂ and β̂ using only days with no active intervention. This gives us a baseline of how each sector typically moves with the market in peacetime. During war periods, any deviation from this baseline (the residual ε) is the abnormal return, the part not explained by general market movements.

Explore Our Data

Interactive visualizations of our stock market and military intervention datasets

Stock Market Universe

Our dataset includes 5,053 U.S. stocks spanning 15 economic sectors and 278 industries. Here's how they're distributed:

Sector → Industry Hierarchy

Click on a sector to explore its industries. Size represents number of stocks.

Hypothesis: Size ≠ Sensitivity

The treemap reveals a striking asymmetry between sector size and geopolitical sensitivity. Financial Services dominates our universe at 25.6% (1,293 stocks), followed by Health Care (16.6%) and Industrials (11.2%). Yet our central hypothesis posits that Energy, representing only 5.2% of stocks, may exhibit the most pronounced market reactions to military interventions.

This counterintuitive relationship stems from the fundamental economics of military operations: Interventions frequently target oil-producing regions (Middle East, North Africa, Venezuela), disrupt global supply chains and trigger commodity price shocks. The 1973 Yom Kippur War and subsequent OPEC embargo, the 1990 Gulf War's impact on Kuwaiti oil fields and the 2003 Iraq invasion all demonstrate how geopolitical events in energy-rich regions propagate through petroleum markets to affect sector returns.

Explore RQ1 to see how this hypothesis holds empirically across 29 interventions and 58 years of data.
Sector Return Correlations

Pearson correlation of daily returns between sectors. Blue = positive correlation, Red = negative.

Correlation Structure & Identification Strategy

The correlation matrix reveals a sparse, low correlation structure across most sector pairs. This is precisely what we need for causal identification, if sectors moved in lockstep, we couldn't distinguish war-specific effects from general market noise.

Strongest Correlations
  • Health Care ↔ Communication Services: ρ = 0.30
  • Real Estate ↔ Industrials: ρ = 0.31
  • Consumer Discretionary ↔ Communication Services: ρ = 0.28
  • Technology ↔ Real Estate: ρ = 0.27

Even the strongest correlations are moderate, indicating good sector independence.

Most Independent Sectors
  • Basic Materials: Near-zero with all sectors (ρ ≈ 0.02-0.04)
  • Consumer Defensive: Low correlation across the board (ρ ≈ 0.08-0.19)
  • Energy: Largely independent (ρ ≈ 0.09-0.28)
  • Miscellaneous: Minimal co-movement with most sectors

These sectors act as natural hedges during geopolitical shocks.

Why this matters for our analysis: The low inter-sector correlations (median ρ ≈ 0.15-0.20) mean that when we observe abnormal returns in Energy during a Middle East intervention, we can be confident this reflects sector-specific exposure to oil supply risks, not a spurious correlation with broader market movements.

Methodological Implication: Our CAPM-based abnormal return framework estimates each sector's β (market sensitivity) using only "peacetime" days. The low residual correlations confirm that after removing the common market factor, sector returns are approximately independent, satisfying a key assumption for valid event study inference.

Military Interventions Dataset

29 non-overlapping U.S. military interventions used in our causal analysis (filtered from 376 in the MIP dataset):

Intervention Objectives
Intensity Distribution
Strategic Objectives Analysis

The distribution reveals that "Maintain/Build Regime" dominates with 16 interventions (55%), reflecting Cold War-era containment policies and post-9/11 nation-building efforts. "Protect Military/Diplomatic Interests" accounts for 8 interventions (28%), typically shorter operations to secure U.S. assets or personnel.

Hypothesis: We expect "Remove Regime" interventions (3 cases) to produce the largest market impacts, since these operations involve maximum military commitment, prolonged uncertainty and fundamental geopolitical restructuring (e.g., Iraq 2003, Panama 1989).

Intensity Distribution Analysis

The histogram exhibits a pronounced right-skew (median = 0.289), indicating that the majority of U.S. military interventions were low-intensity operations. Over 75% of interventions fall below intensity 0.4, representing limited engagements such as shows of force, evacuations and peacekeeping missions.

The long right tail captures high-intensity conflicts: Vietnam (0.91), Afghanistan (1.0) and the Lebanese Civil War (0.73). These outliers, though rare, are hypothesized to drive the most significant and persistent market effects.

Geographic Distribution of Interventions

Countries targeted by U.S. military interventions. Red indicates high intervention count, blue indicates low. Click on a country to see intervention details.

Geographic Hotspots & the Oil Connection

The globe reveals clear geographic clustering: Asia (Vietnam, Korea, Afghanistan), the Middle East (Iraq, Syria, Lebanon, Iran), Latin America (Nicaragua, Panama, Haiti) and Africa (Libya, Somalia, Egypt). Many of these regions sit atop critical oil reserves or control strategic chokepoints (Strait of Hormuz, Suez Canal).

The Energy Hypothesis: We expect interventions in oil-producing regions to generate outsized Energy sector reactions. Historical precedent supports this: The 1973 Yom Kippur War triggered the OPEC embargo and a 300% oil price surge, Operation Desert Storm (1990) saw crude spike 130% on Kuwait invasion fears.

Explore the Globe: Click on any country to see which interventions occurred there, their intensity scores and strategic objectives.

Key Findings Preview

Overall Market Impact

On average, aggregate stock prices fall by approximately 7% following military interventions, indicating a broad risk-off reaction in the market. The Vietnam War alone produced a cumulative market impact of -91% over its 8-year duration.

Energy Sector: Small but Mighty

Despite representing only 5.2% of stocks, the Energy sector exhibits the most extreme and variable responses. During the Gulf of Sidra Incident, Energy showed +8.6% abnormal returns; during the Contra Affair, it dropped -69%. This volatility reflects direct exposure to oil supply disruptions in intervention regions.

Sector Heterogeneity

Responses vary dramatically by sector. Consumer Cyclical consistently underperforms (mean AR: -7.2%), while Utilities shows defensive characteristics with near-zero average abnormal returns. Financial Services, the largest sector, exhibits moderate sensitivity with high variance across interventions.

Market Anticipation

Trading volume is elevated before intervention dates, suggesting markets anticipate military actions through geopolitical tensions. Our event study shows abnormal volume already elevated at t=-10, with no sharp spike at t=0, indicating interventions are largely priced in.

Intensity Matters

Our PCA-based intensity index reveals a clear relationship: higher-intensity interventions produce larger absolute market impacts. The Vietnam War (intensity: 0.91) and War in Afghanistan (intensity: 1.0) show the most pronounced effects, while low-intensity operations like border skirmishes produce minimal market disruption.

Persistent Effects

Abnormal trading volume remains above baseline after interventions end, reflecting sustained portfolio adjustments. Effects persist for 20+ trading days post-intervention, with gradual mean reversion rather than immediate normalization.