Introduction
Political earthquakes like Brexit don’t just rattle the nation that voted—they send seismic waves through the global financial system. A 2023 University of Surrey study, published in the International Journal of Finance & Economics, revealed that the UK’s 2016 decision to leave the EU triggered a cascade of volatility across European markets. The research exposed how tightly interwoven financial systems had become, with risk traveling across borders faster than ever before. This step-by-step guide walks you through the method used by researchers to trace that contagion—so you can apply the same lens to future political upheavals.
What You Need
- Historical market data: Daily closing prices for at least 10 European stock indices (e.g., FTSE 100, DAX, CAC 40) and volatility indices (e.g., VIX, VSTOXX) for the 12 months before and after the Brexit referendum.
- Political event timeline: Precise dates of key events (e.g., June 23, 2016 referendum, triggering of Article 50) to mark event windows.
- Statistical software: A package capable of running GARCH models and spillover indices (e.g., EViews, R with rugarch and vars packages, or Python with statsmodels).
- Risk measurement tool: Access to a function that calculates conditional correlations, such as DCC-GARCH or a rolling correlation matrix.
- Data cleaning skills: Ability to handle missing values and align time series across different time zones.
Step-by-Step Guide
Step 1: Identify the Trigger Event and Event Window
Pinpoint the exact moment the political shock occurred. For Brexit, that is June 23, 2016—the referendum date. But the contagion didn’t begin or end there. The Surrey study examined a window stretching from one year before the vote (to capture pre-referendum anxiety) to one year after (to see how risk transmitted during negotiations). You need to define two periods: a pre-event baseline and a post-event analysis window. This contrast is critical for isolating the shock’s impact.
Step 2: Collect and Align Market Data
Gather daily closing prices for a representative set of European stock indices (the study used 10 countries). Convert prices into daily logarithmic returns: return = ln(price_t / price_t-1). Also obtain volatility indexes—like the VSTOXX for Eurozone—to gauge implied fear. Ensure all series are aligned to the same calendar days; remove holidays where markets were closed. Check for any anomalies (e.g., flash crashes) that might distort results.
Step 3: Calculate Baseline Volatility and Correlations
Using the pre-event data (e.g., June 2015 to June 2016), run a GARCH(1,1) model on each index’s returns. This gives you the conditional variance—how volatile each market was on any given day before Brexit. Then compute the conditional correlations between every pair of indices using a DCC-GARCH model. This baseline correlation matrix tells you how closely markets already moved together before the shock. For example, the study found that UK and German equities had a pre-Brexit correlation of roughly 0.6. Record this as your benchmark.
Step 4: Estimate Spillover Indexes
Now apply the methodology proposed by Diebold and Yilmaz (2009) to measure directional spillovers. In practice, this means estimating a vector autoregression (VAR) model on the returns of all indices simultaneously. From the VAR, compute the forecast error variance decomposition to see how much of a shock to one market explains variations in another. Aggregate these into a total spillover index—a single number that captures the percentage of total forecast error variance coming from cross-market shocks. The Surrey research showed that Brexit increased the total spillover index across European markets by 12% in the year after the vote.
Step 5: Compare Pre- and Post-Event Spillover Dynamics
Repeat Step 4 using only the post-event data (June 2016 to June 2017). Compare the total spillover index and the directional spillovers (e.g., from UK to Germany vs. Germany to UK). The key finding from the study: post-Brexit, the UK became a net transmitter of volatility to the rest of Europe, whereas before it had been more of a receiver. You should also calculate rolling spillover indexes over 60-day windows to see how contagion ebbed and flowed during negotiations.
Step 6: Test for Statistical Significance
To ensure the observed changes aren’t just noise, perform a bootstrap test. Randomly shuffle the timeline of the pre-event data 10,000 times to create a null distribution of spillover index changes. Compare your actual shift (the 12% increase) against this distribution. The Surrey study reported a p-value below 0.01, confirming the spike was statistically significant. If your p-value exceeds 0.05, consider widening the event window or checking for other confounding events.
Step 7: Visualize the Contagion Network
Create a network graph where nodes are countries and directed edges represent the magnitude of volatility spillovers (thicker arrows = bigger transmission). Use the directional spillover values from Step 5. This visualization, similar to Figure 3 in the original paper, immediately highlights which markets became “super-spreaders.” In the Brexit case, the UK, France, and Germany emerged as the three biggest transmitters, while smaller markets like Greece and Portugal mainly received shocks.
Step 8: Interpret the Results in Context
Finally, tie the quantitative findings back to real-world events. The study noted that Brexit did not just shake Britain—it forced a repricing of risk across all European assets because investors recognized the political uncertainty could trigger a domino effect. Check if your data shows similar patterns: a spike in cross-border correlations during key negotiation periods (e.g., triggering Article 50 on March 29, 2017). Document any anomalies: for instance, the UK’s domestic volatility actually decreased relative to Europe after the vote, but its exported volatility grew.
Tips for Success
- Start with a smaller set of indices (e.g., UK, Germany, France) to debug your models before scaling up to 20 countries.
- Use the same GARCH order across all series to keep your models comparable. GARCH(1,1) is the standard, but check for residuals’ autocorrelation.
- Account for currency effects: If mixing indices from different currency zones (e.g., GBP vs. EUR), consider converting all returns to a common currency (e.g., USD) to avoid FX noise overwhelming the equity signal.
- Robustness checks matter: Rerun the analysis with different event windows (e.g., 6 months or 18 months) to ensure the 12% increase isn’t just a one-off artifact.
- Combine with qualitative news: The numbers alone don’t tell the story. Match spillover spikes to political headlines (e.g., Theresa May’s Lancaster House speech in January 2017) to build a richer narrative.
- Publish your code to allow replication. The Surrey researchers made their data and R scripts available, which accelerated peer verification.
- Watch for reverse causality: A spike in spillovers might be driven by a common third factor (e.g., a global financial crisis). Include a control event—like the 2016 US election—to isolate the Brexit effect.
By following this guide, you can transform any political shock—whether it’s a referendum, election, or trade war—into a quantifiable map of financial contagion. The same toolkit that revealed Brexit’s European shockwaves can now illuminate the hidden connections that make modern markets both resilient and fragile.