The Verdict: What We Got Right and Wrong

We started with six common assumptions about how markets react to military interventions. Here's how they held up against 58 years of data.

H1

Energy Dominance

Refuted

"Energy would be hit hardest because wars disrupt oil supply chains."

Reality: Energy shows a modest -0.15% average abnormal return, far from the largest swings. Consumer Cyclical (-0.20%) and Utilities show more consistent losses.

Modern markets price in oil risk before interventions begin. The 1973 Yom Kippur War was the last time Energy truly dominated.

See RQ1 Analysis
H2

Immediate Panic

Refuted

"Markets would crash immediately with panic selling across all sectors."

Reality: All 15 sectors show median abnormal returns near zero. Markets don't systematically crash; they differentiate. Some sectors gain while others lose.

Modern markets are sophisticated. Investors rotate between sectors rather than flee entirely, seeking relative safety in defensive stocks.

See RQ1 Analysis
H3

Middle East Focus

Partial

"Middle East interventions would have the strongest market effects due to oil dependency."

Reality: Middle East interventions actually show +0.11% average returns. The Caribbean (-0.16%) shows more negative impact. Libya (4 interventions) is most targeted.

Geographic location matters less than intervention objective. Markets have learned to price in Middle East risk.

See RQ2 Analysis
H4

Objectives Matter

Confirmed

"Regime changes would rattle markets more than limited protective missions."

Reality: Regime Change operations produce extreme swings: Health Care +3.03% vs Consumer Cyclical -3.38%. Protect Military/Diplomatic shows near-zero average impact.

The "why" of intervention matters more than the "where." Aggressive objectives signal prolonged instability.

See RQ2 Analysis
H5

Intensity-Impact

Partial

"High-intensity wars would cause bigger market crashes proportionally."

Reality: Pearson correlation r=-0.3578 (p=0.0567), a moderate negative relationship, but not statistically significant at α=0.05. Duration confounds the relationship.

Intensity shows a moderate negative correlation with returns, but duration and overlapping events confound the relationship.

See RQ3 Analysis
H6

Duration Matters

Partial

"Longer wars would cause more cumulative damage than short, decisive operations."

Reality: Afghanistan shows -28% CAR vs Desert Storm's +7.5%. However, 20-year conflicts overlap with other major events (2008 crisis, Iraq War), making it impossible to isolate duration's true effect.

Correlation exists, but causation is unclear. Long wars inevitably coincide with other economic shocks, confounding the analysis.

See RQ3 Analysis
1 Confirmed
3 Partial
2 Refuted

Surprising Discoveries

Beyond testing our hypotheses, the data revealed several unexpected patterns that challenge conventional wisdom about war and markets.

Consumer Staples Volatility

The "boring" defensive sector showed surprising volatility during the Gulf of Sidra Incident, with a -3.0% abnormal return, one of the most extreme single-event reactions in our dataset.

Why it matters: Even "safe haven" sectors aren't immune. Supply chain disruptions can hit everyday goods harder than expected during geopolitical crises.
The Energy Paradox

Despite wars often occurring in oil-rich regions, Energy sector returns averaged only -0.15%, less volatile than Consumer Cyclical, Utilities and several other sectors.

Why it matters: Markets have evolved. The 1973 oil shock taught investors to hedge energy exposure. Today's markets price in geopolitical risk before conflicts escalate.
Modern Market Sophistication

Near-zero median returns across all sectors suggest markets don't panic; they differentiate. Technology and Real Estate showed consistent resilience across most intervention types.

Why it matters: The "flight to safety" narrative is oversimplified. Sophisticated investors rotate between sectors rather than flee markets entirely.
Regime Change Extremes

Regime Change operations produced the widest sector spread: Health Care gained +3.03% while Consumer Cyclical lost -3.38%, a 6.4 percentage point gap within the same intervention type.

Why it matters: Aggressive interventions create winners and losers simultaneously. Sector selection matters more than market timing during regime changes.
The Duration Correlation

Afghanistan (7,632 days) shows -28% CAR, while Desert Storm (348 days) achieved +7.5% CAR. However, longer conflicts overlap with other major events (2008 crisis, Iraq War), making causation difficult to isolate.

Caveat: This correlation doesn't prove causation. Multi-decade conflicts inevitably coincide with other economic shocks, confounding the relationship between duration and market impact.
Universal Market Response

The Libya-Egypt border tensions showed statistically significant abnormal returns across all 15 sectors, a rare systematic response that affected the entire market uniformly.

Why it matters: Some events transcend sector boundaries. When all sectors move together, it signals a fundamental shift in market risk perception.

Key Findings

Intensity Drives Negative Returns

Higher intensity interventions produce more negative market returns (correlation: -0.38). The Vietnam War (intensity: 0.91) caused -123.71% CAR, while low-intensity operations average +2.77% CAR.

Sector Heterogeneity

Responses vary dramatically by sector. Technology and Real Estate show resilience across most objectives, while Consumer Cyclical, Energy and Utilities consistently underperform.

Objectives Shape Outcomes

Regime Change operations produce the most extreme swings: Health Care +3.03% vs Consumer Cyclical -3.38%. Protect Military/Diplomatic operations show near-zero average impact, while defensive operations average +10.43% CAR.

Median Returns Near Zero

All sectors show median abnormal returns close to zero, indicating interventions don't systematically shift central tendency. However, Miscellaneous (IQR: 2.79%) and Energy (IQR: 1.80%) show the highest variance during conflicts.

Answering Our Research Questions

RQ1
How do U.S. military interventions influence stock returns and volatility across economic sectors?

Military interventions produce statistically significant abnormal returns across multiple sectors, but the direction and magnitude vary considerably:

  • Strongest Positive: Miscellaneous (+1.24% during Red Sea Minesweeping), Technology (+0.96% during Vargas Tragedy)
  • Strongest Negative: Vietnam War hit Basic Materials hardest (-0.74% daily AR), Financial Services suffered during Yom Kippur War (-0.74%)
  • Statistical Significance: Libya-Egypt border tensions showed universal significance across all 15 sectors, which is a rare systematic market response
  • Defensive Stability: Consumer Defensive (IQR: 0.78%) and Utilities (IQR: 0.90%) show the tightest distributions
Explore RQ1 Analysis
RQ2
How do objectives and geographic targets influence stock market reactions?

Both the stated objective and geographic target significantly influence market response patterns:

  • Dominant Objective: Maintain/Build Regime (55%) shows mixed impacts, Technology gains +0.34% while Consumer Cyclical loses -0.52%
  • Extreme Volatility: Regime Change operations produce the widest swings, Health Care +3.03% vs Consumer Cyclical -3.38%
  • Geographic Patterns: Middle East interventions yield +0.11% average returns, Caribbean shows -0.16%
  • Most Targeted: Libya (4), North Korea (3), Russia (3) in our filtered dataset
Explore RQ2 Analysis
RQ3
How does intervention intensity influence market reactions?

Our PCA-based intensity index (Lethality: 0.65, Escalation: 0.58, Duration: 0.49) reveals a negative correlation (-0.38) between intervention intensity and market returns:

  • High Intensity: Vietnam War (0.91) caused -123.71% CAR, War in Afghanistan (0.86) resulted in -20.92% CAR
  • Low Intensity: Operations below median intensity (0.29) average +2.77% CAR vs -1.13% for high-intensity
  • Best Performer: Reunification Talks skirmishes (intensity: 0.29) achieved +46.22% CAR
  • Variance Explained: First principal component explains 55% of total variance across the three dimensions
Explore RQ3 Analysis

Limitations

  • Data ends in 2020: Our stock market data coverage stops at 2020, meaning we cannot analyze recent conflicts where the U.S. provides support (e.g., Ukraine-Russia, Israel-Palestine)
  • Confounding economic events: Major crises occurred during our study period (the 2008 subprime crisis, the dot-com bubble in 2000, Black Monday in 1987 and the COVID-19 pandemic), making it difficult to isolate war effects from broader market shocks
  • Stock data coverage begins in 1962, limiting analysis of earlier interventions
  • CAPM model assumptions may not fully capture market dynamics during crises
  • Overlapping intervention windows can confound individual event effects
  • Survivorship bias in stock data may underestimate negative impacts
  • MIPs lasting fewer than 5 days were excluded (~25% of dataset)

Future Work

  • Extend data to present: Incorporate post-2020 market data to analyze U.S.-supported conflicts like Ukraine and Gaza
  • Extend analysis to international markets and cross-border spillovers
  • Incorporate alternative risk models (Fama-French, momentum factors)
  • Analyze high-frequency data for intraday response patterns
  • Study the role of media coverage in amplifying market reactions
  • Investigate long-run effects beyond the 20-day post-event window

Final Thoughts

"Markets are not merely reactive instruments. They are sophisticated processors of geopolitical risk, capable of distinguishing between temporary shocks and structural threats."

Our analysis of 29 U.S. military interventions spanning 58 years reveals a financial landscape far more nuanced than conventional wisdom suggests. The data tells a story of market evolution: from the oil-shock panic of 1973 to the measured sector rotations of modern conflicts, investors have learned to price geopolitical risk with increasing sophistication.

The most striking finding is not what moves markets, but what doesn't. Energy, long assumed to be the primary casualty of military conflict, shows remarkably muted responses. While longer conflicts like Afghanistan correlate with larger negative returns, we must be cautious: 20-year wars inevitably overlap with other major economic events (the 2008 financial crisis, concurrent conflicts in Iraq), making it impossible to isolate duration's true causal effect.

Perhaps most importantly, our research demonstrates that intervention objectives matter more than geography. Regime change operations create extreme winners and losers simultaneously, while protective missions barely register on market radar. This suggests that investors have developed sophisticated frameworks for assessing the long-term economic implications of different intervention types.

Correlation ≠ Causation

Long wars correlate with negative returns, but confounding events make causal claims difficult to support.

Sector Rotation, Not Flight

Modern investors rotate between sectors rather than flee markets entirely during geopolitical crises.

Objectives Trump Geography

The "why" of intervention predicts market response better than the "where."

The Bottom Line: Military interventions don't crash markets; they reshape them. Understanding these patterns isn't just academic; it's essential for anyone seeking to navigate the intersection of geopolitics and finance. As global tensions continue to evolve, the lessons from 58 years of data offer a roadmap for anticipating how markets will respond to the conflicts of tomorrow.

Resources

Created for CS-401 Applied Data Analysis at EPFL, Fall 2025