The cost of EU Member States’ proximity to the war

The EU’s geographical proximity to Russia’s war of aggression against Ukraine has weighed on its economy since the full-scale invasion in February 2022, constraining the recovery from the pandemic and casting uncertainty over future prospects. This Special Topic explores the effect of that proximity on the economic performance of EU Member States, and the potential transmission channels of its impact over the past three years. The analysis confirms significant proximity-related differences in GDP growth and broader macroeconomic conditions. Over 2022-23, the loss in annual growth is estimated at around 2 pps. for every 1 000 kilometre reduction in distance. The output growth loss is particularly high around 1.4-1.8 pps. for Member States bordering the countries at war. When the analysis is extended to include 2024, the estimated effects appear to be milder. In line with previous literature, the results confirm that energy and financing costs, trade relations, employment growth, and government expenditure have played a role in transmitting the economic impact of proximity.

Member States that are closest to the region at war have underperformed in terms of real GDP growth since 2019. This becomes particularly apparent after accounting for their higher average growth rates over the preceding period 2013-19 (see Graph II.1.1 and Map II.1.1). The analysis in this Special Topic explores the extent to which this underperformance owes to Member States’ geographical proximity to the war, and the role played by the potential transmission channels of this impact over the past three years. 

Throughout the analysis, proximity is measured by 1) the average of the two distances between the capital of each Member State and Kyiv and Moscow, respectively and 2) the presence of a border. The group of ‘closer’ countries includes the thirteen Member States with lower distance than the median (higher proximity) and includes the Baltic States (Estonia, Latvia, Lithuania), two of the EU Nordics (Finland, Sweden), the Visegrad group (Czechia, Hungary, Poland, Slovakia), Austria, Bulgaria, Germany and Romania. Also assessed is the relative economic performance of the eight countries that border either Ukraine or Russia, namely Estonia, Finland, Hungary, Latvia, Lithuania, Poland, Romania, and Slovakia. Visual inspection of the growth paths for these proximity-based groups confirms that the negative impact of the war is more pronounced for those located nearer to the war. Those Member States that are closer to, or bordering, the war zone experienced above-average growth prior to the war, which made their post-war contraction appear even more pronounced (see Graph II.1.2). ⁽[11]⁾ 

Comparing economic outcomes to pre-war expectations

In 2022-23, the EU economy grew less than had been forecast in autumn 2021. The first step of the analysis is a comparison of the Autumn 2021 Forecast (AF21) – the last one prior to Russia’s full-scale invasion of Ukraine in 2022 – with the actual outcomes. By assuming that the observed discrepancies are largely due to the unforeseen consequences of the war, these deviations offer an approximate assessment of the war's economic impact. ⁽[12]⁾ The analysis focuses on the performance of the median ⁽[13]⁾ country within the group of the 13 EU Member States that are geographically closer to the war. The actual annual average GDP growth in the first two years of the war (2022–23) was 1.9 pps. lower than the pre-war forecast for the median EU Member State (see Graph II.1.3). 

Bordering countries appear most vulnerable to the negative economic spillovers from the war. In terms of potential growth, the median bordering country lost more than 1 pp. per year relative to the EU average. Similar proximity effects are observed in the growth of most expenditure components, namely private consumption, investment, exports, and imports, all of which grew more slowly than anticipated. The only exception is government consumption, for which no proximity effect is detected. It is also noteworthy that for all groups of countries considered, the negative forecast errors for imports are larger in magnitude than for exports, implying a positive impact on net exports.

Graph II.1.3: AF21 forecast errors for GDP and expenditure components

Graph II.1.4: AF21 forecast error for HICP, energy and unprocessed food inflation (average error in 2022-23)

The graph shows median values across proximity related country groups (‘closer’, ‘border’) defined previously, see Graph II.1.2.

Proximity to the war is also linked to higher inflationary shocks. Inflation forecast errors are positive for the EU as a whole, confirming the large inflationary fallout of the war. HICP inflation turned out to be 5.9 pps. per year above the pre-war forecast. ⁽[14]⁾ Proximity effects are evident: the median HICP inflation rate in Member States closer to the war was about 2.2 pps. higher than the EU average, and 3.7 pps. higher for those bordering the war. Similar proximity effects are also observed for food inflation (see Graph II.1.4).

The drivers of the economic impact of proximity to the war

There are several channels through which geographical proximity to the war can amplify the exposure to its economic consequences. First, countries closer to Russia and Ukraine often have stronger trade links with them, making these economies more vulnerable to disruptions from sanctions, supply bottlenecks, and changes in demand. Second, these economies tend to be more energy-intensive and depend more on Russia for gas and oil, which heightens their exposure to energy-supply disruptions and resulting price spikes. Third, proximity may also translate into higher financial risk premiums, particularly in non-euro-area economies, reflecting both heightened geopolitical uncertainty and tighter financing conditions. Fourth, neighbouring countries received significant numbers of displaced persons from Ukraine, which can boost GDP through higher consumption and labour supply, while also generating short-term fiscal costs. Finally, heightened security concerns may have influenced both public and private investment decisions, leading governments to increase defence spending and causing some firms to postpone or relocate planned investments. ⁽[15]⁾

Empirical studies confirm that geographical proximity exacerbated vulnerability to the impact of the war through these channels. Federle et al. (2024a) analysed the economic impact of wars in a historical sample spanning 150 years and 60 countries, and found that the economic costs of war spill over to countries that are geographically close to the war site through the contraction of trade. ⁽[16]⁾ The Spring 2022 Economic Forecast included a vulnerability heatmap highlighting high exposure to the fallout of the war in Member States with stronger energy, trade, and financial links to Russia. ⁽[17]⁾ Subsequent analyses by the European Parliament highlighted the war’s negative economic impact during the first two years, due to higher energy prices and supply chain disruptions, noting also the EU policy initiatives designed to boost defence spending and reduce strategic vulnerabilities in energy and food security. ⁽[18]⁾ Cobion et al. (2025) found that the war sharply reduced consumer confidence, especially in neighbouring countries, while Federle et al. (2024b) reported that stock markets in countries close to Ukraine fell sharply at the start of the invasion, whereas those farther away remained comparatively stable. ⁽[19]⁾ Using forecast-error analysis from the Winter 2022 Economic Forecast, Capoani (2024) shows that countries closer to the war experienced larger-than-expected GDP losses and stronger inflationary pressures in 2022. ⁽[20]⁾ Additional studies highlight specific impacts: Beckmann et al. (2023) document a surge in demand for euro cash, particularly in countries bordering Ukraine and Gluszak et al. (2025) showed that the arrival of persons fleeing Ukraine raised rents in Poland’s largest cities by around 0.7%, illustrating how migration pressures transmitted to housing markets. ⁽[21]⁾ Finally, Federle (2024a) found that geographically distant countries that are less exposed to adverse trade effects may even benefit from increased military expenditure. 

Testing for war proximity effects on growth and macroeconomic conditions

The second step of the analysis uses a panel data regression model to estimate and formally test whether war-proximity effects influence the macroeconomic conditions of EU Member States. The model is estimated over the period 2002–24 and includes fixed effects in both the country and time dimensions. This specification controls for unobserved country-specific, time-invariant characteristics, such as participation in the euro area, differences in historical growth paths or the relative size of economies, provided these differences remain constant over time. By also controlling for time-fixed effects, the model accounts for common shocks affecting all countries in a given year, thereby capturing common pre- and post-war shocks to growth and broader macroeconomic conditions.

The impact of proximity to the war is captured through an interaction term between a war-period dummy variable—set to one for the years 2022–24—and a measure of geographical proximity or distance. Four proximity measures are used in this analysis: (a) the distance variable described above (i.e. average distance in kilometres from each country’s capital to Kyiv and Moscow), entered with a negative sign so that higher values imply higher proximity; (b) the logarithm of this distance to capture possible nonlinear effects, again with a negative sign; (c) a dummy variable for the thirteen countries classified as “closer” to the war; and (d) a dummy variable for the countries at the “border”.

Proximity effects were found for the major channels of transmission suggested by the literature: trade, inflation and energy, uncertainty, population and labour, and government spending. ⁽[22]⁾ The results confirm that, since the start of the war, countries geographically closer to the war have reduced their overall trade openness as well as the share of their trade with Ukraine and Russia ⁽[23]⁾, while their current account balance and their terms of trade have deteriorated. Countries in proximity to the war also faced higher inflationary pressures, particularly in energy prices, and especially in the first two years of the war. This pattern is consistent with the negative impact of proximity on the energy intensity of GDP, reflecting an adjustment to higher energy prices and import costs. In line with earlier analyses, rising country risk premiums and deteriorating fiscal balances coincided with higher long-term government bond yields in countries geographically closer to the war. The data also show a rise in household saving rates in these countries, possibly reflecting precautionary behaviour and larger financial wealth losses due to inflation. On the labour market, despite an increase in displaced persons from Ukraine fleeing to neighbouring countries, employment growth remained weaker in countries near the conflict. On the fiscal side, these countries recorded higher government expenditure—particularly on defence and investment as a percentage of GDP—accompanied by a worsening of budget balances. 

Quantifying the economic cost of proximity 

GDP growth exhibits strong war-proximity effects according to the above test. For the first two years of the war, the test suggests a “cost of proximity” of around 2 pps. of GDP for every 1 000-kilometre decrease in distance, or roughly 0.039 pps. for each 1% reduction in distance. ⁽[24]⁾ Compared with the EU average distance, these estimates correspond to about 1.1–1.3 pps. lower growth for Member States closer to the conflict—depending on whether proximity is measured linearly or logarithmically—and around 1.4–1.8 pps. for those bordering the war. These results are broadly consistent with the analysis above based on AF21 forecast errors (see Graph II.1.5). The latter also shows that the effects for median countries within each group are smaller than for the average, indicating substantial variations among Member States—at least during the first two years of the war—with some experiencing notably higher proximity-related costs. 

Graph II.1.5: Impact of proximity on GDP growth

The graph shows deviations from the EU average cost of the war observed for countries closer or bordering the region at war. The AF21 forecast-error results, presented in terms of both medians and averages, are shown in blue bars. The proximity related country groups (‘closer’, ‘border’) are defined previously, see Graph II.1.2.

Including data for 2024 slightly reduces the estimated impact of proximity on GDP growth compared to the EU average, by around 0.1-0.2 pps. annually, depending on whether proximity is measured linearly or logarithmically. Graph II.1.6 visualises the difference in estimated costs across countries based on the two metrics. The log-distance model assigns a higher cost of proximity to the five countries closest to the war. Overall, the difference in annual GDP growth between the most distant and the closest country amounts to around 6 percentage points. The costs are also displayed in Map II.1.2.

Graph II.1.6: Impact of proximity on GDP growth, Member States, 2022-24

The graph shows deviations from the EU average in terms of average annual GDP growth over 2022-24 across countries, ordered along the horizontal axis by their geographical proximity to the conflict—from most distant to closest.

Map II.1.2: Economic cost of the war in Ukraine

Linear distance-based econometric estimates.

Controlling for growth drivers and distilling the proximity channels

The results remain unchanged when accounting for differences in income levels and capital accumulation. The robustness of the estimated war–proximity coefficient was assessed by including additional control variables beyond the unobserved country and year fixed effects. ⁽[25]⁾ The coefficient in specification (1) shown in Table II.1.1 remains broadly unchanged when GDP per capita (in purchasing power standards) and investment-to-GDP are included as additional growth drivers (specification 2). These variables, for which the proximity detection test (performed in the previous section) found weak or no evidence of war–proximity effects, have the expected signs: the coefficient on GDP per capita is negative, reflecting long term catching-up of less developed economies, while investment positively contributes to growth.

The main transmission channels suggested by the literature appear sufficient to explain the impact of proximity on GDP growth. In Table II.1.1 the coefficient on the proximity-to-war variable shrinks from 3.24 in specification (2) to roughly half of its absolute value and becomes statistically insignificant once energy intensity and energy prices are included (specification 3). This suggests that energy costs and inflation, which exhibit a post-war upward shift correlated with proximity, can effectively account for the associated negative impact on growth. By contrast, when government expenditures are added (specification 4), the coefficient on proximity becomes even larger in absolute value and more negative, reflecting an offsetting effect of increased public spending on growth. In specification (5), the coefficient on the proximity factor remains statistically significant but substantially smaller once employment, sentiment, and long-term interest rates (all with the expected signs) are included, indicating that these channels are indeed relevant but do not fully explain the proximity-related effects on GDP growth. Taken together (specification 6), the coefficient on the proximity factor continues to shrink and eventually becomes statistically insignificant, highlighting the effectiveness of the included channels in capturing the negative growth effects associated with proximity to the war.

Table II.1.1:  Selected robustness results

Panel least squares regressions with country and year fixed effects. The specifications include: (1) only unobserved fixed effects, (2) selected control variables, and (3–5) alternative sets of channels related to geographical proximity. Results for the linear distance case are qualitatively similar.