In his paper on ethnic diversity and interpersonal trust, Andrew Leigh remarks that:

‘At the very least, trust appears to be a useful proxy variable for a variety of outcomes that are important to economists.’

And not just economists. But the note of caution in this statement is warranted. We can, I think, say fairly confidently that high levels of stated interpersonal trust are likely to be a good thing, but that lower levels are ambiguous – they might be a sign of trouble, but there are other possible interpretations.

One reason is the wording of the question. In the survey Andrew used, the 1997-98 Australian Community Survey, he coded as ‘trusting’ people who disagreed with the statement ‘generally speaking, you can’t be too careful in dealing with most Australians.’ That’s a common wording in the surveys on trust, but it seems to force people into a trust/don’t trust choice, precluding more nuanced responses. Trust/distrust is more like a continuum from naivety to paranoia about others than a simple choice between one or the other.

We can see the effects of more nuanced answer options in the 2005 Australian Survey of Social Attitudes (available here, though it is a user-unfriendly website). It has the same wording as the ACS, except that ‘Australians’ is replaced with ‘people’. Only small proportions are naive or paranoid – 2% saying that people can almost always be trusted, and 5% saying ‘almost always’ you can’t be too careful. 52% chose ‘people can usually be trusted’, and 39% ‘usually can’t be too careful’. The year before, in the Australian Election Survey, a simple two-choice answer option (though with a slightly different question ‘most people’, rather than ‘people’) saw 53% go for the negative response, compared to 44% in the social attitudes survey. Possibly offering a more qualified ‘trusting’ option improves apparent trust levels.

The broader problem with this question is that in practice trust is contextual, and so we are usually dealing with a narrower set of circumstances and persons than ‘Australians’ or ‘people’ in general. Other polling shows that we are more trusting when we are asked about a specific person or group of persons than asked very general questions. For example, we tend to rate specific politicans more highly for trustworthiness than politicians in general; we claim not to have confidence in banks in general, but all the individual banks have mostly satisfied customers – and who thinks that the banks will steal their money?

Because trust is contextual, I’m not at all sure how I would answer the more general question about trust. Would I leave my apartment door unlocked or post my PIN on my blog? No, I wouldn’t. Would I invest my savings in a get-rich-quick scheme? Again, no. So I do not believe that people can always be trusted. But I trust shops not to rip me off and not to serve me contaminated food, even if I have never been to them before and will never go to them again. I buy things on the Internet from foreign countries. I’m almost never concerned about my personal safety. So while I know there are untrustworthy people around, I am also confident that I can avoid them in my daily business. This enables me to act as if people are trustworthy.

Saying that people can usually be trusted in answer to these poll questions means, I think, that those respondents trust the people they need to deal with. But responding negatively does not necessarily mean a respondent is completely untrusting. They could be trusting in some contexts, but not others. So if ethnic diversity causes fewer people to express trust in general this may not be much of an issue, if they still have trust where that is useful in reducing monitoring and enforcement costs. The intriuiging aspect of Putnam’s latest research, as Andrew Leigh noted, is that he seems to be suggesting that in diverse cities people are less trusting of their own ethnic group as well as other people. It will be interesting to see more evidence on that.