Brexit Without the Apocalypse

Brexit has acquired the unfortunate habit of explaining everything. Weak British investment? Brexit. High inflation? Brexit. German stagnation? Apparently also Brexit, if one is feeling adventurous. A useful empirical exercise begins by refusing that convenience.

The right question is not whether the British economy experienced disappointments after the June 2016 referendum. It plainly did. The question is whether those disappointments form a distinctive British break when compared with economies exposed to many of the same global shocks but not to Brexit.

Using RainbowStats, I compare the United Kingdom with Germany and with an equal-weight control made from Germany, France, and the Netherlands. I then add the United States. This is descriptive evidence rather than a complete causal design, but it immediately disciplines the argument.

1. The identification problem

Britain voted to leave the European Union on June 23, 2016. The new UK–EU Trade and Cooperation Agreement took provisional effect on January 1, 2021 and formally entered into force on May 1. Those are natural dates to mark, but they are not laboratory walls. Between them sit political uncertainty, a global pandemic, supply disruptions, a European energy crisis, and the largest inflation shock in a generation.

If Brexit caused a large UK-specific aggregate collapse, one should expect British outcomes to deteriorate sharply relative to reasonable European comparators near the referendum or the new trade regime. If Britain and its neighbors weaken together, the shared movement cannot simply be assigned to Brexit. Brexit could have generated spillovers, but this comparison alone cannot identify them.

The charts help separate the first term from the rest. They cannot perfectly divide the final three. That modesty is a feature.

2. GDP: the cliff is not obvious

Before the referendum, UK growth averaged 1.99% versus 1.48% in the Northwest European control. From 2016Q3 through 2019Q4, the corresponding averages were 2.00% and 2.07%. That is not the profile of an economy immediately separating from its neighbors in free fall.

The post-2022 result is even more awkward for a one-cause narrative. UK year-over-year growth averaged 1.98%; the control averaged 1.30%; Germany averaged only 0.38%. Britain did not look healthy so much as Germany looked unusually weak.

Across 2022–2025, the United States averaged 2.59% year-over-year growth. Britain, Germany, and the three-country control were all slower. The strongest visual result is therefore not “Brexit caused Europe to stagnate.” Brexit cannot plausibly be the UK-specific treatment and the general explanation for the untreated controls at the same time. The result is that Europe’s growth problem is larger than Brexit.

3. Relative gaps: there is no synchronized British collapse

GDP, fixed investment, and exports do not all turn down together at either Brexit marker. Investment is the most plausible source of lasting damage because uncertainty can delay projects long before customs rules change. Yet the automated RainbowStats split for investment occurs before the referendum, warning us that Britain entered the Brexit episode with pre-existing structural problems.

The EUKLEMS evidence reinforces that warning. British R&D investment underperformance is striking, but it begins too early to be attributed wholly to Brexit. Communications-technology investment tells the opposite story.

4. Exports: the most relevant warning, but not a cliff

Trade was one of the central economic concerns. Even a zero-tariff, zero-quota agreement adds rules of origin, declarations, regulatory checks, and fixed costs. Those burdens are especially important for smaller firms and for goods crossing borders repeatedly.

The binary split is revealing because the data select 2022Q2—not June 2016 or January 2021. In the regression of European-control export growth on UK export growth, the slope falls from 0.807 to 0.188. The associated R² values are 0.616 and 0.624. The correlation remains meaningful, but the amplitude changes sharply.

This is consistent with trade relationships becoming less stable after the new regime, but it is not uniquely diagnostic of Brexit. The selected date is also the start of the energy and Ukraine shock. Moreover, growth rates around the pandemic contain huge base effects. A stronger causal trade study would use export levels, bilateral destination shares, goods versus services, firm size, and a synthetic control.

The LP regression as a robustness check

The `LP_REGRESSION` alias runs a median, or least-absolute-deviation, regression. That is useful here because it does not let the most spectacular COVID observations dominate the fitted line. On the full 2010Q2–2026Q1 sample, it estimates a slope of 0.453 and an intercept of 2.089, with mean absolute error 2.628 and R² 0.415. Conventional OLS gives a slope of 0.521 and R² 0.423. The broad relationship survives the robust fit; neither method turns the chart into causal proof.

5. Labor held up; the price premium is harder to dismiss

The feared labor-market implosion does not appear. UK unemployment averaged 4.20% in 2016Q3–2019Q4, compared with 5.90% in the control. It remained lower in 2022–2025. That does not settle questions about participation, real wages, or migration composition, but it rules out the simplest jobs-disaster account.

Prices are less comforting. The UK inflation rate exceeded the control by an average 0.62 percentage points in the referendum-to-pandemic window and 1.08 points in 2022–2025. Sterling depreciation, energy exposure, and new trade frictions can all contribute. The chart identifies a British premium; it cannot allocate that premium among the mechanisms.

6. EUKLEMS: old weaknesses and real strengths

The productivity proxy—real value added per worker—reaches 130.2 in the UK and 128.3 in EU11 by 2021. That is not evidence of a post-referendum British productivity collapse relative to the aggregate. R&D investment, however, reaches only 113.6 in Britain versus 185.7 in EU11. The timing matters: this is a long-running British weakness, not something born in 2016.

Communications-technology investment reaches 539.6 in the UK, compared with 324.4 in EU11. Any serious account must be able to hold those facts together: weak research investment, strong communications investment, and broadly comparable aggregate productivity. “Brexit did everything” cannot.

7. What the evidence does—and does not—say

The Office for Budget Responsibility’s Brexit assumption is a counterfactual level claim: it expects UK trade intensity eventually to be 15% lower and long-run productivity 4% lower than they otherwise would have been. An economy can suffer that sort of loss and still grow at roughly the same rate as slow-growing neighbors. A missing cliff in year-over-year growth therefore contradicts the most theatrical forecasts, but it does not logically refute a gradual level loss.

Brexit may be a tax on British performance. It is not a sufficient theory of European stagnation.

That is the interpretation I find most consistent with the evidence. The UK did not visibly fall off an aggregate economic cliff relative to Germany, France, and the Netherlands. Trade relationships became noisier and less proportional after 2022; investment and inflation provide plausible channels of cost. But the larger comparison is transatlantic. The United States grew faster while multiple European economies struggled together.

The policy implication is less satisfying than a slogan and more useful than one. Britain can reduce avoidable trade frictions with Europe. Europe—including Britain—also needs to confront the shared problems of productivity, energy costs, capital formation, diffusion of technology, and scale. Re-litigating the referendum will not substitute for either job.

Methods, data, and caveats

1. European control. Equal-weight average of Germany, France, and the Netherlands. It is transparent and economically nearby, but it is not a formal synthetic control and should not be labeled the entire EU.
2. Frequency. Quarterly year-over-year growth rates from FRED/OECD. The UK series currently ends in 2026Q1, which defines the common GDP sample.
3. Periods. Pre-referendum: 2010Q2–2016Q2; uncertainty: 2016Q3–2019Q4; post-shocks: 2022Q1–2025Q4. The 2020–2021 interval is graphed but omitted from period averages because base effects dominate.
4. Split regression. RainbowStats chooses the split that minimizes regression error. Searching over dates means the split is descriptive and should not be treated like a pre-registered hypothesis test.
5. EUKLEMS. Annual data cover 1995–2021. The productivity proxy is real value added divided by employment. Endpoint values are indexed to 1995 = 100.
6. Causality. Shared shocks, sector mix, migration, fiscal policy, exchange rates, and measurement revisions remain confounders. The article tests whether a dramatic break is visible; it does not recover the full remain-in counterfactual.

Sources

• FRED, Federal Reserve Bank of St. Louis, carrying OECD quarterly national accounts, labor, and CPI series used by RainbowStats.
• EUKLEMS & INTANProd documentation, including coverage for EU member states, the UK, U.S., and Japan across output, labor, productivity, and tangible and intangible investment.
• Office for Budget Responsibility: initial impact of Brexit on UK–EU trade.
• UK Government: Trade and Cooperation Agreement implementation report.