Southern Europe is suspicious: the evolution of trust in the EU

Analysis of the eurozone crisis and its consequences has started to consider its social, as well as political and economic, dimensions. Bruegel analysis, for example, has considered distributional challenges, effects on poverty levels, social security systems, and changes in confidence towards political institutions. Academic economic research into the significance of cultural values, such as trust, suggests a viable new route to consider some of these deep-rooted social consequences of Europe’s economic woes, and we seek here to sketch some stylised empirics. If we think that such social values are both economically important and that it is very hard to change them, then the findings are worrying. Economic crisis has been associated with a striking degradation in the levels of interpersonal trust in southern Europe.

In an often cited quote from 1972, Kenneth Arrow wrote that “Virtually every commercial transaction has within itself an element of trust … It can be plausibly argued that much of the economic backwardness in the world can be explained by the lack of mutual confidence”. The recent literature investigating the economic ‘payoff’ of trust in society stretches back most notably to Knack & Keefer (1997) and La Porta et al. (1997), who used measures of generalised trust to proxy for levels of social capital and test the connection to economic performance. However, since definitions of social capital vary significantly between disciplines, here we stick to the narrow importance of trust while acknowledging the link to social capital. Much of the economic literature empirically tests the connection between trust and economic performance using data from the World Values Survey (WVS), which asks “Generally speaking, would you say that most people can be trusted or that you can’t be too careful in dealing with people?”

Fehr et al. (2004)’s experiments suggest that this question does indeed measure general levels of social trust, Bloom, Sadun & Van Reenen (2012) use this data to show that higher local levels of trust increase the aggregate productivity of firms, Knack & Keefer (1997) use it cross-sectionally to illustrate that trust has large impacts on aggregate economic activity as well as to argue that increasing levels of trust is extremely difficult, and La Porta et al. (1997) show it can explain substantial variation in political and economic institutions. While there is debate on the channels through which trust (and social capital) economically operates (e.g. Tabellini, 2008, Guiso et al. (2006)), the relevant lessons are threefold:

1. Generalised trust is significantly associated with economic performance.
2. It can be empirically proxied using WVS data.
3. It is slow and difficult to change.

The recent release of new waves from both the European Social Survey (ESS) in late 2013 and World Values Survey in April 2013 provides an opportunity to assess deep social consequences of the crisis. Just as trust can affect economic performance, so too can economic performance affect general levels of trust in a society and it is this reverse causality that the new data allows us to consider. Here we use here data from ESS, since it covers a wider range of European countries and asks all respondents exactly the same question as the WVS, and use WVS data as a robustness check. The only difference is that responses to the ESS question range from 0-10 whereas WVS responses are binary. We take the mean of responses across a given country, using the relevant weights, such that higher numbers indicate higher levels of generalised trust. The data used relates to waves from 2006/7, 2008/9, 2010/11 and 2012/13.

At the continental level, changes have been slightly negative: across the 10 countries out of EU-15 in the sample that have complete data – Belgium, Denmark, Finland, Germany, Ireland, Netherlands, Portugal, Spain, Sweden and the UK – trust has fallen by 0.8% between 2006/7 and 2012/13. Across the 16 countries from EU-28 with complete data – add Cyprus, Bulgaria, Estonia, Poland, Slovenia and Slovakia to the previous list – it has fallen by 1%.

However, if we disaggregate then the picture markedly changes. In particular, we create a ‘northern’ group (Belgium, Germany, Finland, Netherlands, UK), a ‘southern’ group (Cyprus, Spain, Ireland and Portugal), and an ‘eastern’ group (Bulgaria, Estonia, Poland, Slovenia and Slovakia). Some countries which would naturally fall into this group are excluded due to missing survey waves – for example, France was not included in the most recent wave (it sees a moderate fall until then), Greece lacks two of the four waves and Italy is included in none of the waves.

Below, we plot the levels of generalised trust across the three groups on the left axis using bars, and the percentage change on the right axis compared to 2006/7 using the lines.

Source: European Social Survey, Bruegel calculations

The graph is interesting for five reasons:

1. The southern group has seen levels of trust fall by over 7% since 2006/7 – Spain has seen small losses while Portugal, Cyprus and Ireland have all seen substantial decreases, by as much as nearly 15% in Cyprus.
2. The northern group sees an upswing in 2012/13 with very little change in the preceding two waves – Finland and the UK have very similar paths, with dips after 2006/7 before recovering in 2012/13 while Belgium, Germany and Netherlands see more consistent increases.
3. The eastern group sees moderate gains of just over 1%, with only Slovakia seeing a fall over time.
4. Averaging across waves, the level of trust in the northern group is 21% higher than the southern group and 29% higher than the eastern group.
5. However, the difference in levels between the southern and eastern group has almost entirely eroded, and the eastern group could be expected to overtake if trends continue.

Cross-checking with the latest data from the World Values Survey agrees with these results. Only two countries per group are included in both the 2005-2008 and the 2010-2014 waves – Germany and Netherlands from the northern group, Spain and Cyprus from the southern group, and Slovenia and Poland from the eastern group – but even so the northern group sees an increase, the eastern group sees a weaker increase and the southern group sees a substantial decrease.

It is worth noting that all these figures are highly stylized and aggregated. Values like trust vary significantly within countries, and further work could usefully seek to explore the links between regional changes in economic circumstances and changes in social attitudes.

Nonetheless, given the high persistence of social attitudes over time as well as their economic significance, the substantial and heterogeneous changes observed across Europe in the last eight years are striking. It is also interesting how closely the trajectories of southern and eastern Europe match the patterns in convergence observed on the tenth anniversary of EU enlargement. Economic crisis has dramatically affected not only livelihoods but – at least in a group of southern eurozone countries – deep and long-lasting social values.