Trust and Diversity

One of the thorniest issues I’ve delved into as a researcher is the relationship between trust and heterogeneity. In a piece in the AFR today, I discuss some of the findings from my research (warning: PDF with regression tables).

The full text of the AFR oped is over the fold. In my view, this is a hard issue, but it’s important that those who support immigration don’t shy away from the possibility that ethnic heterogeneity may have negative impacts. For pro-immigration social democrats, similar issues arise with regard to diversity and welfare.

Lies and Statistics, Australian Financial Review, 23 July 2005 

Trust makes society work better, says Andrew Leigh.

In places where people trust one another, institutions, markets and societies seem to work better. Trusting societies have more effective bureaucracies, schools that function more efficiently, less corruption, and faster growth. Trust acts as a kind of "social glue" that enables business and communities to operate more effectively. For these reasons, social capital, once solely the domain of sociologists, has increasingly attracted attention from economists.

What determines why some people are more trusting than others? To answer this question, I used data from the Australian Community Survey (carried out by Edith Cowan University and NCLS Research), which asked over 6000 respondents whether they agreed with the statement, "Generally speaking, you can’t be too careful in dealing with most Australians". Responses were used to class people as trusting or distrusting.

Using multiple regression analysis, it is possible to test the effect of various demographic factors, holding other factors constant. At an individual level, gender makes little difference, but better educated people and those who work full-time are more likely to be trusting. By contrast, longer commuting times are associated with lower levels of trust, which may be due to the fact that commuting time crowds out social activities.

It is also possible to observe the effect of neighbourhoods. We see how trust changes when we look at similar individuals who live in different communities. At a neighbourhood level, trust is higher in rural Australia than in cities, and higher in richer areas than in poor ones.

Neighbourhood-level analysis also throws up a startling finding: trust is lower in ethnically diverse neighbourhoods. Residents of multi-racial neighbourhoods are more likely to agree that "you can’t be too careful in dealing with most Australians". In particular, neighbourhoods where many languages are spoken tend to have lower levels of trust, suggesting the main issue may be whether people can communicate effectively with those living nearby.

The negative relationship between trust and ethnic diversity in Australia is similar in size to results found in the US by economists Alberto Alesina and Eliana La Ferrara. However, while they conclude the US effect is mostly driven by white Americans’ aversion to ethnic diversity, I find only limited evidence that those born in Australia are hostile to diversity. Instead, most of the negative relationship between diversity and trust appears to operate through those who are born overseas. Trust is particularly low among immigrants living in racially diverse neighbourhoods.

Since Australia and the US are both "settler societies" that experienced high levels of immigration in the postwar era, it is perhaps not surprising to find a negative relationship between ethnic heterogeneity and trust in both countries. But this should be weighed against the substantial benefits of immigration. Although reducing diversity might raise levels of trust, it would likely be detrimental to society on balance. Instead, policymakers should focus attention on the problem itself: building local trust in immigrant communities.

Andrew Leigh is an economist at the Australian National University. The full research paper is available at

This entry was posted in Australian issues. Bookmark the permalink.

14 Responses to Trust and Diversity

  1. That’s an interesting paper, Andrew.

    One thing that comes through in a lot of the literature I’ve looked at is fear of crime as a mediating variable. It would be interesting to build this into the analysis, but you’d need to look at a different data set, I suspect.

  2. SInclair Davidson says:

    In my Masters subject I’ve run a similar regression when talking about the impact of trust on firm size. Using ethno-inguistic fractionalisation as a proxy for culture i show (a) it appears ‘bad’ when looking at regular corporate governance variables, but (b) when you include a proxy for poor legal environment (French law dummy) the result is insignificant. So ethnic diversity is a ‘problem’ in poor legal environments (can’t trust the other and so on), but not in ‘good’ legal environments.

    Extending to your AFR piece we see a similar result. Trust plays a positive role in wealthy (law abiding areas governed by inter-personal norms and law etc) and a negative role is less wealthy areas (less law abiding, governed by personality etc).

  3. Though I am not sure to what extent these questions get at what we need to be concerned about. Would I leave my front door unlocked? I don’t trust the people in my area enough for that. But do I feel frightened in the street, believe local officials are corrupt, or that I will be ripped off in the shops? No, I don’t.

    It’s a while since I read your full paper – could another factor be that some immigrants come from low-trust societies, and it while take a while, perhaps a generation, before they get over their old suspicions?

  4. Andrew’s questions go to the difficulty of defining trust that is definitely there in much of the social capital literature.

    For instance, where I live (New Farm in Brisbane), I most certainly wouldn’t leave my door unlocked. I did once, and someone came in while I was in my study and knicked my backpack from the hall. Conversely, another junkie tried to break into my garage and was chased away by neighbours. On the traditional indices, the fact that most of us neighbours chat to each other, sometimes socialise, and watch out for each other would place us very high on the trust scale. But there are people on the streets (sometimes mentally ill homeless people) whom I’d cross the road to avoid. Then, the suburb has a nice community feel and I know and chat to local shopkeepers, waitresses etc.

    I’m inclined to think that it’s worth looking at the criminological literature on these questions as well, because these finer distinctions must necessarily be made when you’re interested in correlations between these factors, fear of crime, and crime prevention. Because I’ve read a fair bit of this, I’m a tad sceptical about coming up with a decent proxy for trust per se in this context. You can do so for things like business transactions, but I think community and everyday relations need other approaches.

    It might also be worth looking at the Australian and British literature on everyday low-level incivility.

    Hope this is helpful.

  5. Andrew, on the survey question, this just occurred to me. It’s also somewhat ambiguous – in that an immigrant might assume that “Australians” refers to Anglos, or people who’ve been here longer. The literature shows that people often take some years after immigration before they identify as “Australian”. Some Irish folks I know, who’ve lived here for over a decade, don’t think of themselves as Australian and talk about the way that “Australians” (ie not them) behave. It seems to me that if you’re measuring trust with regard to ethnic diversity, that’s potentially a big problem with this proxy.

  6. Andrew Leigh says:

    Mark, you make a good point. But the effect appears to be stronger on the question relating to “people in my local area” than “most Australians”. And yes, you’re right that it would be nice to get more data on the details. But even I baulk at running a regression of crime rates on ethno-linguistic fractionalization. Or am I wrong to do so?

    Sinc, your notion of looking at the interactions is very cool.

  7. Yes, I think you could do it, Andrew – but fear of crime would be a better measure than crime rates.

  8. Sacha Blumen says:

    It’s good to look at these questions. As researchers (in whatever field we work!), we have to be prepared to keep out minds open to possibilities that we might find difficult to otherwise consider.

    I’ve only read these posts – but I was just wondering if, in the survey, respondents were asked a question something like “people in my neighbourhood hold similar values to me” – ie, is it possible to attempt to discern a connection between trust and people feeling that others in their neighbourhood hold similar values/views to them. I’d guess that there would be a relation there – a sort of people feeling like they’re part of the tribe or less part of it.

    These are just some thoughts…

  9. Brenton Caffin says:


    Richard Layard from the LSE at his book launch on Happiness (sadly) admitted that ethnic diversity does appear to have an impact on trust and thus on levels of social capital and well-being.

    A slightly different aspect which he suggested may be worth exploring is transience – that percentage of the population which passes through a location vs the percentage of long-term residents. It’s the combination of high diversity and high transience which can seriously erode trust – one reason why some London Boroughs such as Southwark have begun to target retention as a key performance indicator in its regeneration programmes.

  10. Andrew Leigh says:

    Brenton, I find the same thing on mobility. But there are benefits to a mobile population too – especially as it reduces pockets of high unemployment. I’m not sure the Southwark strategy is the right one.

  11. Brenton Caffin says:


    I’m not sure I agree – encouraging greater mobility can entrench rather than reduce pockets of high unemployment (it’s ok, I haven’t worked for Southwark so I’m not getting defensive).

    If the only solution to reduce unemployment is to move people away to get better jobs, this reduces demand for housing, depressing prices, which in turn attracts the most recent wave of immigrants, who have the least well formed connections with the local community and employment prospects. This in turn increases ethnic diversity and further reduces trust. Your solution – while does indeed present benefits to the individual who leaves – creates a viscious circle at the neighbourhood level.

    While you can’t stop people moving to find work (nor should you), it is incumbent on councils like Southwark to find ways to reduce the need to move in order to break that cycle (e.g. attracting investment and developing local employment opportunities). I feel this is the best means of increasing the well-being of its residents.

  12. Andrew Leigh says:

    Brenton, I’m not sure your logic works. If people move in response to unemployment, you’ll get net outflows, not merely churn. Sure, this will drive house prices down, but it will also drive wages up (at least for employers that draw only on the local area, like corner stores).

    In such an environment, it’s hard to see why the ethnic composition of an area will necessarily change.

  13. Sacha Blumen says:

    Just as an aside, maybe it’s mixed up, maybe it’s not just predominantly A or B.

Comments are closed.