VoxEU Column Europe's nations and regions

Cultural proximity and trade: Evidence from Eurovision

This column introduces the use of Eurovision song contest scores as a measure of cultural proximity. Unlike other measures, such as common language or religion, Eurovision scores are asymmetric and time-varying, allowing estimates to distinguish between two potential channels through which cultural proximity might affect trade: trade costs and consumer preferences.

Each year in May, the European Broadcasting Union organises the Eurovision Song Contest. This is a huge pan-European televised show, where every participating country sends an artist to perform a song. The other countries grade those songs, either by televoting, or (in earlier times) by popular juries. In late May, Dima Bilan, a Russian singer, won the contest thanks to massive support from the votes of former USSR satellites.

Eurovision scores as indicators of cultural proximity

The Eurovision contest has long reflected countries’ commonalities or bilateral affinity (even sympathy) (Ginsburgh and Noury 2005). For example, Cyprus awards to Greece an average of 7.41 points more than Greece receives on average; Greece reciprocates by awarding an excess of 6.26 points beyond the Cypriot mean score. Over-generous and reciprocal relationships can be found for many Eastern European country pairs and, to a lesser extent, between Scandinavian countries. But, the scores also reveal the lack of affinity or even rivalry between some countries. This relationship need not be reciprocal. France grades Great Britain 0.86 points below average, while the Brits treat France almost neutrally. Not surprisingly, Cyprus and Turkey stand out as two countries that systematically award each other grades below average. Germany (and Austria) over-grade Turkey without being compensated; this likely reflects immigrants’ ethnic ties.

There is also plenty of anecdotal evidence that Eurovision scores reflect changes in attitudes and moods over time. For example, the Economist described the victories of Serbia in 2003 and Turkey in 2007 as signs that other European countries perceive them as less culturally remote (May 29, 2003; May 17, 2007). The successes of Estonia and Latvia in 2001 and 2002, respectively, were similarly commented (May 30, 2002). The success of Finland in 2006 was associated to the widespread perception of a successful “Finnish model” (July 6, 2006). Long-time BBC commentator Terry Wogan attributed Britain’s failure to earn a single point in 2003 to its decision to back the United States in its attack on Iraq.

While those examples suggest short-term effects, the data also reveal long-run trends. For example, the popularity of countries seems subject to cycles that are difficult to reconcile with underlying movements in musical quality. For example, the Economist relates the failure of France to win any contest since 1977 to the loss of French popularity and cultural clout in Europe (May 12, 2005). The example of Italy offers a similar, but less spectacular, picture.

Controlling for the quality of the song (using song-specific fixed effects), the Eurovision scores reveal cultural proximity between pairs of countries. In economics, cultural proximity is related to common language, religious proximity, or to common legal environment. In sociology, cultural proximity has a broader definition that includes history, clothing style, living pattern, and culture in several senses. Cultural proximity is not necessarily reciprocal and can vary across time. A country can display respect and sympathy for the cultural, societal, and technological achievements of another country without this feeling necessarily being reciprocal and everlasting.
The characteristics of the Eurovision scores satisfy this broader definition, once we control for the quality of the song. The scores vary over time and are potentially asymmetric and provide a good summary indicator of cultural proximity. Table 1 shows pair-wise correlation coefficients between quality-adjusted Eurovision Song Contest (ESC) scores and conventional measures.

Table 1 Coefficient of correlation between different measures of cultural proximity

  ESC scores Adjusted ESC scores
Adjusted ESR scores 79.34  
Common language 15.55 13.67
Common legal origin 11.45 20.33
Religious proximity 13.34 16.45
Ethnic ties 20.18 20.63
Genetic proximity 16.33 12.05

Source: Felbermayr and Toubal (2007).

The correlation between the quality-adjusted Eurovision score and the raw score data is 79.34. Hence, about a fifth of the variance in raw scores is accounted for by quality. The correlation coefficient between both Eurovision score measures and the conventional proxies of cultural proximity all have the right sign and are all statistically different from zero. The adjusted score displays larger correlation coefficients (with the exceptions of common language and genetic distance), signalling that quality adjustment improves our index of cultural proximity.

Quantifying the impact of cultural proximity on bilateral trade

Almost all empirical trade gravity models include variables such as common language, religion, or ethnicity to control for cultural proximity between countries. There seems to be a consensus on the overall quantitative and qualitative importance of these “standard” cultural proximity determinants for bilateral trade flows. Given the correlation between Eurovision scores and conventional measures of cultural proximity, it is not surprising that the scores are positively related to bilateral trade volumes. However, if the scores reflect only components of cultural proximity which are symmetric between country pairs and time-invariant, with their time and within-pair variation essentially noise, then we should not find any robust effect of Eurovision scores on trade. However, we show that both the time and the within-pair variation of Eurovision scores still contribute to explaining the variation in bilateral trade.

The “ad valorem tariff equivalent” of cultural proximity, which is the reduction in ad valorem trade costs that would be equivalent to the trade creation achieved by moving the respective measures of cultural proximity from their lowest to their highest sample realisations, ranges from 4% to 8%. The impact of cultural proximity is larger on trade in differentiated goods, for which there are no formal markets. Cultural proximity has no effect on trade of goods that are exchanged on organised markets.

How does cultural proximity affect bilateral trade?

So far the literature has had little to say on the mechanisms through which cultural proximity affects bilateral trade volumes. There are two channels through which cultural proximity matters for international trade. Cultural proximity is related to lower communication and transaction costs. This translates into higher bilateral trade volumes due to lower trade costs. On the other hand, cultural proximity may be related to preferences. Then, culturally close countries trade more because they have strong tastes for each other’s products. We attempt to disentangle the trade-cost and the preference channels of cultural proximity. This distinction is important, because (multilateral) cultural distance is related to a country's per capita income and welfare only if it affects trade volumes via the trade-cost channel.

If the Eurovision scores reflect the broad sociological notion of cultural proximity, one can exploit their time and within-pair variance to econometrically separate the preference channel from the trade-cost channel. For this purpose, we use two econometric models, which both help to identify the preference effect of cultural proximity separately. The first strategy assumes that the bilateral trade-cost channel is not affected by swings in bilateral attitudes or in the sheer popularity of a country: trade costs depend on the deep time-invariant components of cultural proximity (linguistic, religious or ethnic ties). In contrast, the preference channel depends on more short-lived fads and fashions, as well as on the time-invariant factors.

The second strategy assumes that the trade-cost channel depends only on the symmetric component of cultural proximity. This follows widespread practice in the empirical and theoretical trade literature, where trade costs are typically considered symmetric (the tunnel between Britain and France is the same in both directions). The preference channel, however, need not be symmetric (the French may dislike the British more than the British dislike the French).

Comfortingly, both empirical strategies yield similar results. Our estimates imply that one-third of the total effect of cultural proximity on bilateral trade is due to the preference effect. Hence, only about two-thirds of the total effect of cultural distance can be interpreted in welfare-theoretic terms.

We should not expect a preference effect when focusing on homogeneous goods such as steel or paper. Using the classification proposed by Rauch (1999), this prediction bears out nicely: there is a measurable preference effect only for differentiated goods.


Felbermayr, G., and F. Toubal (2007). “Cultural Proximity and Trade”, (former version, Tübinger Diskussionsbeitrag No 305, 2006)
Ginsburgh, V. A. and A. Noury (2005). "Cultural Voting: The Eurovision Song Contest". CORE Discussion Paper No. 2005/6
Rauch, J.E. (1999). “Networks Versus Markets in International Trade”. Journal of International Economics 48(1): 7-35.


2,520 Reads