Trust is a fundamental element of social capital – a key contributor to sustaining well-being outcomes, including economic development. In this entry we discuss available data on trust, as measured by attitudinal survey questions; that is, estimates from surveys asking about trusting attitudes.
Global comparisons of trust attitudes around the world today suggest very large time-persistent cross-country heterogeneity. In one extreme, in countries such as Norway, Sweden and Finland, more than 60% of respondents in the World Value Survey think that people can be trusted. And in the other extreme, in countries such as Colombia, Brazil, Ecuador and Peru, less than 10% think that this is the case.
Data from European countries shows that average trust in the police tends to be higher than trust in the political and the legal systems. And trust in the political system is particularly low – in fact much lower than interpersonal trust for all countries except Switzerland. On the other hand, trust in the police is notably high, and in the majority of European countries people trust the police more than they trust each other.
Long-run data from the US, where the General Social Survey (GSS) has been gathering information about trust attitudes since 1972, suggests that people trust each other less today than 40 years ago. This decline in interpersonal trust in the US has been coupled with a long-run reduction in public trust in government – according to estimates compiled by the Pew Research Center since 1958, today trust in the government in the US is at historically low levels.
Interpersonal trust attitudes correlate strongly with religious affiliation and upbringing. Some studies have shown that this strong positive relationship remains after controlling for several survey-respondent characteristics.1
This, in turn, has led researchers to use religion as a proxy for trust, in order to estimate the extent to which economic outcomes depend on trust attitudes. Estimates from these and other studies using an instrumental-variable approach, suggest that trust has a causal impact on economic outcomes.2 This suggests that the remarkable cross-country heterogeneity in trust that we observe today, can explain a significant part of the historical differences in cross-country income levels.
Measures of trust from attitudinal survey questions remain the most common source of data on trust. Yet academic studies have shown that these measures of trust are generally weak predictors of actual trusting behaviour. Interestingly, however, questions about trusting attitudes do seem to predict trustworthiness. In other words, people who say they trust other people tend to be trustworthy themselves.3
The World Value Survey allows cross-country comparisons of self-reported trust attitudes. This visualization shows estimates of the share of survey respondents agreeing with the statement “most people can be trusted”.4 Countries can be added by clicking on the option at the bottom of the chart.
As it can be seen, there are very large differences in levels, and trends tend to be fairly stable. For example, in China the share of people reporting to trust others is much higher than in Brazil – and this have been true across several surveys over the last couple of decades.
Using the same data discussed above, this map shows a global picture of cross-country differences in trust levels. Here, estimates correspond to the latest available data from the World Value Survey. Once again, heterogeneity stands out. In one extreme, in countries such as Norway, Sweden and Finland, more than 60% of respondents think that people can be trusted. And in the other extreme, in countries such as Colombia, Brazil, Ecuador and Peru, less than 10% think that this is the case. Notice that even in some relatively homogeneous regions, such as Western Europe, there are some marked differences: there is a twofold difference between France and neighbouring Germany.
The results from the World Value Survey discussed above show that there are very large, time-persistent cross-country differences in the share of people who report trusting others, even within European countries. But are these cross-country differences similarly large if we look at average ratings of trust in a scale that allows for differences in intensity? The following visualization shows the average rating of trust in others across European countries, using data from Eurostat. In this case, respondents answer to the question “would you say that most people can be trusted?” using an 11-point scale, ranging from 0 to 10. As it can be seen, results are consistent – countries with high interpersonal trust in the World Value Survey also have high trust ratings in the Eurostat survey. Yet heterogeneity on the more granular scale used by Eurostat is somewhat smaller.
The OECD, using Eurostat data, provides estimates of trust in public institutions that are comparable to estimates of ‘trust in others’ (i.e. interpersonal trust). The following visualization, taken from the OECD report How’s life? (2015) , shows average ratings of trust in (i) the political system, (ii) the police, and (iii) the legal system. These figures can be directly compared to those on interpersonal trust discussed in the visualization above. In both cases figures rely on Eurostat data from the same survey, so respondents here also rate trust in institutions using an 11-point scale, ranging from 0 to 10. As it can be seen, average trust in the police tends to be higher than trust in the political and the legal systems. And trust in the political system is particularly low – in fact much lower than interpersonal trust for all countries except Switzerland
The data suggests a broad correlation between trust in others, and trust in the different public institutions. Specifically, northern Europe (and Switzerland) report higher levels of trust, while southern and eastern Europe (and France) report lower levels across the board.
We mentioned above that the police is trusted more than other public institutions in most European countries. But do people in these countries trust the police more than they trust each other? This question is relevant, because trust in the police can become, in certain situations, an important substitute for interpersonal trust. This visualization plots Eurostat figures for trust in the police (y-axis) and trust in others (x-axis). The size of each dot represents national income (PPP-adjusted GDP per capita). We can see that there is a clear positive correlation; and in the majority of countries people report the same or higher trust in the police than trust in others. The clear exceptions are Greece, Czech Republic, Slovakia, Poland, and Slovenia – these are the countries that lie significantly below a hypothetical line with slope one (i.e. the line below which trust in others is higher than trust in the police). Denmark and the Netherlands, both with higher levels of trust than the mentioned countries, are also somewhat below this hypothetical line.
From an inter-temporal perspective, the data from the Gallup World Poll suggests that trust in public institutions has been going down recently in OECD countries. This visualization shows the OECD-average estimate of trust in governments over the period 2006-2014, using Gallup’s data. As it can be seen, the percentage of the population reporting confidence in the national government went down every year in the period 2009-2013.
The Pew Research Center recently constructed a series of long-run estimates of trust in the government for the US, staring 1958. This visualization uses their data, to plot the share of people who say they can trust the government in Washington always or most of the time.6 As it can be seen there are some clear patterns associated with political cycles, but in the long-run there is a negative trend. Today, trust in the government in the US is at historically low levels. The Pew Research Center has a dedicated website, with many interesting visualizations – including disaggregated trends by ethnicity and political affiliation. Further details and analysis available in the report Beyond Distrust: How Americans View their Government.
In the US, the General Social Survey (GSS) has been gathering information about trust attitudes since 1972. To our knowledge, this is the longest available time-series on interpersonal trust estimates in the world. This visualization uses this source to show the evolution of trust in the US. Specifically, this plot shows the share of respondents agreeing with the statement “most people can be trusted” in the surveys 1972-2014.7 As we can see, there are short-term fluctuations, but people in the US seem to trust each other less today than 40 years ago.
Trust is a key element of social capital – but it is not the only one. Data from the UK suggests that different aspects of social capital change in time at different rates. This chart, from the Centre for Social Investigation at Nuffield College, Oxford, shows that in the UK trust in other people fluctuates year by year, but there is no trend over the last couple of decades. This is consistent with the figures from the World Value Survey, where the UK shows little variation between the 1998 and 2009 surveys. Interestingly, however, associations with voluntary organisations declined significantly over the same period – the chart shows that the percentage of the UK population that is active with one or more organization fell from 52% in 1993 to 43% in 2012.
Trends in four measures of social capital, UK, 1991-2013 – Figure 2 in Centre for Social Investigation (2015)8
The data from Eurostat and the World Value Survey shows that Sweden is one of the countries with the highest levels of trust globally. This visualization from the SOM institute – an independent survey research organisation at the University of Gothenburg in Sweden –, shows that interpersonal trust in Sweden is not only high, but also very stable across time.
In this visualisation we see the degree of trust in others, raging from 0 to 10. We can see that estimates are very persistent – the share of individuals rating trust as low (0-3), medium (4-6) and high (7-10) has not changed significantly over the last two decades.
The data from the SOM in Sweden also allows an inter-temporal analysis of other measures of trust. This visualization shows estimates of general trust in politicians. In this case stability is reflected in the fact that political cycles are not associated with accentuated fluctuations. This seems to contrast with the data from the US, where public trust in government seems to be highly cyclical.
A number of academic studies have explored the link between religious beliefs and self-reported trust attitudes. This figure, from Guiso et al. (2006),11 summarizes the results from one such study using data from the World Values Survey. Specifically, the bars in this figure represent the effect of religious affiliation on trust, in percent of the sample mean of trust relative to “no religious affiliation”. The reported effects correspond to estimated coefficients in a regression where the dependent variable is trust in others (i.e. a variable equal to 1 if participants report that most people can be trusted), and there are controls for demographic characteristics (health, gender, age, education, social class, income). Since the data is available for the same country over several years, the authors also control for country-specific time-invariant characteristics (the so-called country fixed effects).
Taking these results at face value, the reported effects suggest that being raised religiously raises the level of trust by 2.6 percent; and regularly attending religious services (the author’s definition of being “religious” for the purpose of the figure), raises the level of trust by another 20 percent. Similarly, these results suggest that the effect of religion differs across denominations: self-reported Catholic and Protestant religious affiliation has a positive effect on trust; while Muslim, Hindu and Buddhist affiliation does not.
As usual, these results have to be interpreted with caution, since reported figures do not control for unobservable time-varying factors that may simultaneously affect attitudes towards religion and trust; in other words, it is likely that there are unaccounted sources of bias that undermine the causal interpretation of the coefficients. Indeed, other studies using attitudinal survey questions on trust have found different results. For instance, Alesina and La Ferrara (2000)12 use data from the General Social Survey in the US, and find that religious affiliation is not statistically related to trust after controlling for further characteristics, such as whether survey respondents had a history of traumatic experiences.
Effect of religious affiliation on trust relative to no religious affiliation – Guiso et al. (2006)13
In a much cited article, Arrow (1972)14 says that “Virtually every commercial transaction has within itself an element of trust, certainly any transaction conducted over a period of time.”
The extent to which trust is linked to economic development has been the subject of many academic papers in the economics literature on growth (see Guiso et al. 2006,15 Algan and Cahuc 2010,16 and the references therein). A common way to get a first-order approximation of this relationship is to estimate the correlations between trust and GDP per capita. This visualization provides evidence of this correlation, by plotting trust estimates from the World Value Survey against GDP per capita. Each dot on this scatter-plot corresponds to a different country. You can learn more about measures of national income in our entry on GDP data.
As it can be seen, there is a very strong positive relationship. Most academic studies find that this relationship remains after controlling for further characteristics. And similar results can also be obtained by looking at other measures of economic outcomes. Looking at outcomes across individuals, Guiso et al. (2006), for instance, report that trust has a positive and statistically significant correlation with the probability of becoming an entrepreneur, even after controlling for education, age and individual income. Their results also hold if religious affiliation of the respondents’ ancestors is used as a proxy for trust – they thus argue that, since ancestors’ religion correlates with respondents’ trust attitudes, this instrumental variable approach can be taken as evidence that the estimated relationship goes in the suggested direction (i.e. that trust leads to entrepreneurship, rather than the other way around).
Other studies using instrumental variables have also found similarly large effects. Algan and Cahuc (2010)17 predict that, according to their estimates, African countries would have a five-fold increase in GDP per capita if they had the same level of inherited social attitudes as Sweden, after controlling for lagged GDP per capita, contemporaneous political environment and time-invariant country characteristics.
Algan and Cahuc (2010) show that inherited trust of descendants of US immigrants is significantly influenced by the country of origin and the timing of arrival of their forebears. This is their instrumental variable: the inherited trust of descendants of US immigrants is used as a time-varying measure of inherited trust in the country of origin. This approach allows the authors to control for country fixed effects and interpret the effect of trust on growth causally. You can read a summary of their findings and approach in a voxeu.org article written by the researchers.
Nunn and Wantchekon (2011)18 provide evidence to explain mistrust in Africa: they show that current differences in trust levels within Africa can be traced back to the transatlantic and Indian Ocean slave trades. More specifically, they show that individuals whose ancestors were heavily raided during the slave trade are less trusting today – and using a variety of different econometric strategies, they claim that this relationship is causal.
Trust vs. GDP per capita, 2014 (or latest available data) 19
Cross-country data, as well as within-country data, suggest that economic inequality is negatively related to trust. This visualization provides evidence of this relationship: it shows a scatter plot of trust estimates from the World Value Survey against income inequality measured by the Gini index. Each dot on this scatter-plot corresponds to a different country, with colors representing different world regions and dot sizes representing population. A Gini index of 0 reflects perfect equality, so the observed negative correlation in this graph implies that higher inequality is associated with lower trust. In other words, we can see that countries with higher income inequality also tend to report lower levels of trust. You can read more about income inequality and the Gini index in our entry on income inequality.
This negative relationship can be explained through various mechanisms: social ties may imply that people are more willing to trust those who are similar to themselves, or higher inequality may lead to conflicts over resources. The empirical work from Alesina and La Ferrara (2000)20 provides evidence in support of the former mechanism. Jordahl, H. (2007)21 provides a discussion of these and other possible mechanisms.
One of the reasons to justify government intervention in the market for education, is that education generates positive externalities.22 This essentially means that investing in education yields both private and social returns. Private returns to education include higher wages and better employment prospects (as we discuss in our entry on Skill Premium). Social return include pro-social behaviour (e.g. volunteering, political participation) and interpersonal trust.
This chart uses OECD results from the Survey of Adult Skills to show how self-reported trust in others correlates with educational attainment. More precisely, this chart plots the percentage-point difference in the likelihood of reporting to trust others, by education level of respondents. Those individuals with upper secondary or post-secondary non-tertiary education are taken as the reference group, so the percentage point difference is expressed in relation to this group. As we can see, in all countries those individuals with tertiary education were by far the group most likely to report trusting others.
And in almost every country, those with post-secondary non-tertiary education were more likely to trust others than those with primary or lower secondary education. The OECD’s report Education at a Glance (2015) provides similar descriptive evidence for other social outcomes. The conclusion is that adults with higher qualifications are more likely to report desirable social outcomes, including good or excellent health, participation in volunteer activities, interpersonal trust, and political efficacy. And these results hold after controlling for literacy, gender, age and monthly earnings.
Social cohesion is often defined as the capacity of a country to support peaceful collective decision making. This pair of plots, from the World Development Report (2013),24 show the correlation between the index of peaceful collective decision making, and two key measures of social cohesion at the micro level: trust and civic engagement. The index of peaceful collective decision making is a quantitative indicator that, for each country, aggregates data on political stability, the absence of violence, and voice and accountability.
The figure shows a strong positive relationship: countries where people are more likely to report trusting others, are also countries where there is less violence and more political stability and accountability.
Share reporting trust in people and index of civic engagement vs index of peaceful collective decision making – World Development Report (2013)25
Attitudinal survey questions provide the main source of data to estimate interpersonal trust attitudes. Available evidence for countries with multiple such estimates, suggest that results are robust to the specific surveying methodologies. This scatter plot bears this out, by comparing cross-country estimates from different surveys. Specifically, this figure plots the estimated interpersonal trust levels as measured by the World Values Survey, against interpersonal trust levels as measured by the European Social Survey and the Afrobarometer Survey. The resulting correlation is positive and very high.
Interpersonal trust levels as measured by the World Values Survey and European Values Study, and the European Social Survey and Afrobarometer Survey – Inglehart & Welzel (2010)26
In an academic paper, Glaeser et al. (2000)27 examine the predictive power of two types of survey questions: questions about trusting attitudes and questions about past trusting behavior. The authors examine the predictive power of these questions by comparing survey answers with actual trusting behaviour in an incentivised experimental setting with monetary rewards. They show that, while measures of past trusting behavior are better than the abstract attitudinal questions in predicting subjects’ experimental choices, in general terms they are both weak predictors of trust. Interestingly, however, questions about trusting attitudes do seem to predict trustworthiness. In other words, people who say they trust other people tend to be trustworthy themselves.
OECD – Eurostat
- Data: OECD data on trust is published in the Society at a Glance – OECD Social Indicators. This source relies on the estimates from Eurostat, specifically the EU statistics on income and living conditions (EU-SILC)
- Geographical coverage: OECD member states
- Time span: Recent years
- Available at: online here.
The World Value Survey (WVS)
- Data: Data on trust and many other social and cultural characteristics from cross-national and time-series surveys
- Geographical coverage: The WVS covers almost 100 societies (nearly 90% of the world’s population). But not all countries have observations in each survey wave.
- Time span: Several waves of surveys from 1981 onwards
- Available at: http://www.worldvaluessurvey.org/WVSDocumentationWVL.jsp
US General Social Survey
- Data: Time-series data on trust and many other social and cultural characteristics
- Geographical coverage: US
- Time span: Yearly surveys since 1972
- Available at: https://gssdataexplorer.norc.org