Diversity, Trust and Redistribution

As a perk for receiving the early career award from the Academy of the Social Sciences in Australia, I was asked to write a paper for their journal, Dialogue. I thought I’d pick something a bit provocative, so chose to write up some research I’ve been doing on the negative impacts of ethnic and linguistic diversity.

On balance, I think that immigration and diversity are beneficial for Australia. Immigration supplements our labour market with much-needed skills, it weaves new threads into our cultural tapestry, it gives us a better sense of the world around us, and it brings us new restaurants, and new products, that we would not have tried otherwise. But the evidence suggests to me that diversity is also likely to lead to lower levels of interpersonal trust, and may undermine support for the traditional Australian welfare state. And I think that social democrats ignore these factors at their peril.

The whole piece is only a couple of thousand words long, and has a few pretty pictures to make the point, so download it here if you’re interested.

This entry was posted in Australian Politics, Economics Generally. Bookmark the permalink.

18 Responses to Diversity, Trust and Redistribution

  1. Oz says:

    I’m just wondering Andrew were the people in those multi-ethnic communities mainly first-generation migrants as opposed to people who were born or who have grown up in Australia.

    I just would expect there would be a greater tendency for new migrants to be more likely to hunker down rather than those who are culturally diverse but who were born or who grew up here.

  2. Patrick says:

    Two points struck me:
    – The correlation from linguistic diversity to low trust on the neighborhood level seems very glibly made, especially just three sentences after the line:
    [trust is] higher in richer neighbourhoods than in poor ones.
    and not long before the line:
    Holding constant other factors such as income, I find little relationship between diversity and beliefs about redistribution.
    Doesn’t that suggest that income might be worth looking at more than linguistic diversity?

    – the second is that you seem to take it for granted that we actually want a more redistributive society. But this meets two objections, I believe
    (a) isn’t there a strong argument that our capacity to absorb immigration is partly determined by the extent to which we restrict redistribution (eg generally speaking the US absorbs immigrants well, Europe, except Britain, doesn’t), and
    (b) don’t we already redistribute about as much as anyone to people who actually need redistribution? Ie, ignoring attitudes, isn’t the effect of our restricted redistribution system that the actual money going to those in actual need is as much as pretty much anywhere else?

  3. Nathan Lambert says:


    Liked the paper. I think it’s important that advocates of increased immigration think long and hard about the potential drawbacks.

    With regard to the wording of the survey (it asked respondents to agree or disagree with the statement “Generally speaking, you can’t be too careful in dealing with most Australians”), are we sure ethnic respondents didn’t interpret the word “Australians” to mean “Skips” — ie, Australian-born citizens with broadly Anglo-Celtic heritage? Because that would measure something more specific than “trust”.

    And on that topic, is there a way to measure trust that forces us to be clearer about the conceptualisation? To me, trust is whether or not you watch your customers to check they’re not shoplifting. It’s how long your debtors take to pay their invoices — if they pay their invoices. It’s whether or not a cable technician admits they made a mistake, and whether or not they try to bill you for it regardless. But then again, when I think of “trust”, I also think of the ability to go around to a few of the other major suppliers in your line of work and get a casual, gentleman’s agreement that nobody’s going to undercut anyone too badly — ie, a good, old-fashioned cartel.

  4. Andrew Leigh says:

    Oz/Nathan, the trust finding is about the attitudes of native-born Australians. And yes, it would be nice to have more than one question measuring it, though fortunately there is a significant literature on what the ‘trust question’ actually measures.

    Patrick, income is certainly important. I spend more time on it in the academic version of the paper. But I think that’s a finding most people expect. And yes, I agree with you that we might not want more redistribution. The political message of the paper was really aimed at those who think that there’s no tension between diversity and a generous welfare state.

  5. Andrew,

    I should have been more specific about my concern.

    The trust question is a big question. It would be preferable to have a measure of revealed behaviour rather than stated behaviour. Eg, ‘Are opposing parties in the Small Claims Tribunal more likely to be ethnically heterogeneous?’.

    Also, is the Australian Community Survey a survey of church congregations?

  6. Pingback: Club Troppo » Friday’s Missing Link

  7. Andrew Leigh says:

    Nathan, I don’t quite follow why it would be interesting to look at the SCT. On the revealed preference issue, here’s the relevant extract from my Economic Record paper:

    What is the ‘Generally speaking, you can’t be too careful in dealing with . . . ’ question really measuring? Glaeser et al. (2000) surveyed a group of Harvard undergraduates, and then asked them to play a trust game. They found that the question used in the General Social Survey (‘Generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people?’) actually measures how trustworthy the respondent is, rather than how trusting of others they are. Similarly, Karlan (2005) finds that Peruvian respondents who agree that most people can be trusted are less likely to default on their microcredit loans. It is likewise possible that the question used in this paper is actually measuring how trustworthy rather than how trusting the respondent is.

    The ACS is a random sample of the population, though it was administered by people who were mostly interested in religious behaviour.

  8. Peter Whiteford says:


    I read your paper and found it very interesting, but I’m not convinced of the argument that the superficial similarity between social spending levels in the USA and Australia means that we really have similar attitudes to redistribution.

    Before expanding on this, I agree that there is a lot of evidence of a relationship between immigration, diversity and access to social welfare.

    In the first half of the 20th century in Australia this actually worked in the opposite way – the White Australia policy was intended to make sure that we had the “right sort” of immigrants.

    It tends to be forgotten now, but there were also racial (and moral) provisions in the Social Security system. From 1908, to be eligible for an age or disability pension apart from having a “good character” you also had not to be African or an “Asiatic”, as they were called then. This lead to some interesting bureaucratic conundrums – are Turks Europeans or Asiatic? Did it change according to which side of the Bosporus you were born on? Some Circassians came from below the Caucasus but they looked European, but confusingly the Circassians that came from the North Caucasus looked Turkish. Are African-Americans African or should they be treated as Americans?

    These are real examples. To find out more look at Alan Jordan, Of Good Character and Deserving of a Pension: Moral and Racial Provisions in Australian Social Security, March 1989, RP77 which used to be available from the Social Policy Research Centre at UNSW (but is not downloadable, and I don’t know if it is out of print.)

    In addition, of course not all indigenous Australians were covered by the social security system until 1966. I’m not 100% sure about this, but I think indigenous people in urban areas were, but those on reserves or missions were not – this was also to avoid cost-shifting from State governments as well as restricting welfare to people like “us”.

    From Alan Jordan’s paper I think that the racial provisions in access to social security were pretty much a dead letter by the 1950s, but as noted they lasted longer for some indigenous people.

    During this period, social welfare spending was low but this was mainly because of a young population age structure, low unemployment in the 50s and 60s and low divorce rates.

    However, I would argue that now the Australian social security system is actually particularly generous to immigrants. To be eligible for an age pension, apart from reaching 65 you simply have to have lived in Australia for 10 years (and be in Australia at the time you claim). I think in most of the Nordic countries to get minimum pensions you have to have been there for 20 years, and I think this is also the case in the Netherlands.

    To get full eligibility for a social insurance pension in most European countries you will typically have had to work and contribute for more than 30 years and sometimes close to 40 years, although there are of course lots of ways of retiring early.

    In addition, in Australia eligibility is simply based on a test of residence and you don’t ever have had to actually pay any tax to get a pension. Most migrants do actually work of course, but if you arrived in Australia in your 30s or even your 40s you would have the same entitlement to age, disability, widow and other pensions as someone who was born here and started work when they were 16.

    This has changed a bit over the past 15 years. The Superannuation system obviously rewards longer working careers in Australia. Since the early 1990s migrants have to wait up to two years before accessing unemployment payments and many other working age payments. But there is no waiting period for family payments and migrants also have immediate access to health care under Medicare. Refugees and people with humanitarian visas get immediate access to benefits. (Asylum seekers are in a different boat – so to speak.)

    Most importantly, the superficially similar levels of social spending in Australia and the US mask radically different welfare systems. As you know, I have argued that the Australian system is the most targeted in the OECD (and therefore probably the world). If you look at the share of transfer spending received by the poorest 20% of the population, then the US is slightly above the OECD average, while Australia is about twice the OECD average (and the highest in the OECD).

    If you take account of tax paid by the poor, then net transfers to the poorest 20% are about 6% of household net income in Australia but only about 1.4% of household net income in the US. On this measure, Australia is the most generous to the poor in the OECD and the US is about the fifth least generous. Interestingly, one of the countries with even lower generosity to the poor is Japan, a country that is more ethnically homogeneous than most European countries.

    If you look at social assistance welfare entitlements for people of working age relative to median equivalent incomes you get a similar picture. As a percentage of the poverty line Australian entitlements are about twice the American level and we pay these to a higher share of the population. On this measure of generosity, Australia is in the top 2 or 3 OECD countries, and the US is in the lowest five. (On the basis of entitlement levels, Japan is also very generous , but they pay their benefits to an even lower proportion of the population than the US).

    Now virtually all of the indigenous population are likely to be in the poorest 20% of the population and a good share of some immigrant groups, particularly those who have a relatively high refugee component. (Contrary to popular belief in some circles, New Zealanders make the greatest positive contribution to the Australian exchequer.) There appears to be a high level of reliance on welfare payments among the indigenous population, although I have been told that their use of the health care system is actually below the Australian average despite their greater health problems, due to problems of access.

    But overall, it appears to me that whatever attitudes the majority of Australians have to different ethnic groups, the way we have structured the welfare system is actually particularly generous to the people we supposedly disapprove of.

    I think part of the reason for differences between the US and Australia is that the African American population is a much higher share of the population and for historical and other reasons the relationship is more fraught. In addition, I would think that Australia has a relatively much lower share of illegal migration. I have seen figures of around 11 million illegal migrants in the USA, which if scaled down to Australian levels would be equivalent to roughly 750,000 people. One estimate of overstaying in Australia that I have seen is around 50,000 people.

  9. Andrew…

    – You say linguistically heterogeneous neighbourhoods have lower trust.
    – Kaplan says lower trust means more loan defaults.

    In syllogistic fashion we can conclude:

    – Linguistically heterogeneous neighbourhoods have more loan defaults.

    My point was that it would be nice to test that proposition directly.

    I perhaps don’t have the same faith as you in the transitive robustness of the results.

  10. The SCT suggestion was just an example of the same thing.

    – If linguistically heterogeneous neighbourhoods have lower trust…
    – And lower trust makes it more difficult to enforce contracts…


    – Linguistically heterogeneous neighbourhoods have more difficulty enforcing contracts.

    And I was trying to think of a metric for that. I don’t think the SCT would actually work because you’ve got no data on ‘successfully enforced’ contracts with which to compare it — obviously successful contracts don’t end up in court.

  11. Pingback: Tim Worstall

  12. Patrick says:

    Interestingly, a writer formerly known as the Economist (or a part thereof) has written a book on labour mobility in which he touches on this with (apparently) studies demonstrating that support for the welfare state is not dependent on immigration.

    He is more successful at rebutting the argument that taxpayers give willingly only to those with whom they feel some kinship and that immigration, therefore, jeopardises support for the welfare state. A willingness to pay taxes to support the poor is independent of levels of immigration, he shows.

    But telling, the (presumably sympathetic) reviewer concludes:
    The harsh truth is that voters find it easier to accept immigrants who look and behave as they do than those who are different. That, as a basis for policy, still leaves most of mankind outside the gates.

  13. Patrick says:

    Oops, forgot the link to the author’s webpage – little info but positive reviews from both Martin Wolf and the Guardian 🙂

  14. Pingback: Andrew Leigh » Blog Archive » Media diversity

  15. Pingback: Linguistic Diversity and Trust « Anggarrgoon

  16. Pingback: John Barrdear » Blog Archive » Libertarians, inequality and cultural homogeneity

  17. Pingback: Andrew Leigh

Comments are closed.