How far does the apple fall from the tree?

Australian sociologists have a long history of work on intergenerational mobility, but economists have come late to the party. Nonetheless, we have our own particular toolkit, which considers the questions somewhat differently from sociologists. I’ve just had a paper published in the Berkeley Electronic Press economics journals which, so far as I’m aware, is the first paper to estimate what economists call the ‘intergenerational earnings elasticity’ for fathers and sons. Here’s the gist of it:

Intergenerational Mobility in Australia, Berkeley Electronic Press: Contributions in Economic Analysis and Policy
Combining four surveys conducted over a forty year period, I calculate intergenerational earnings elasticities for Australia, using predicted earnings in parents’ occupations as a proxy for actual parental earnings. In the most recent survey, the elasticity of sons’ wages with respect to fathers’ wages is around 0.2. Comparing this estimate with earlier surveys, I find little evidence that intergenerational mobility in Australia has significantly risen or fallen over time. Applying the same methodology to United States data, I find that Australian society exhibits more intergenerational mobility than the United States. My method appears to slightly overstate the degree of intergenerational mobility; if the true intergenerational earnings elasticity in the United States is 0.4–0.6 (as recent studies have suggested), then the intergenerational earnings elasticity in Australia is probably around 0.2–0.3.

Note that the finding that Australia is more socially mobile than the United States (and probably less socially mobile than Scandinavia) accords with an earlier paper of mine with Dan Andrews, which found that more equal countries tend to also be more socially mobile. In other words, it’s easier to move from rags to riches in a more equal country.

More broadly, the advantage of looking at the intergenerational elasticity is that it’s a summary measure that facilitates comparisons across countries. But without the work of great ANU sociologists Leonard Broom and Frank Jones (now at UQ), who ran some innovative surveys in the 1960s and 1970s, it would not have been possible to look at how intergenerational mobility has changed across time in Australia. So I’m greatly in their debt.

As you’ll see if you read the paper, getting good measures of intergenerational mobility is pretty data-intensive, requiring long panels of earnings data for fathers and sons. As Australia’s HILDA survey runs on over the next decade or two, we’ll hopefully get more precise estimates of how hard it is to move from rags to riches in Australia. And while my study can only look at blokes (because of lower labour force participation rates for women), it will eventually be possible to get good estimates for mothers and daughters too.

Update: Andrew Norton’s blog has a characteristically thoughtful post, which has provoked an interesting comment thread.

About these ads
This entry was posted in Inequality. Bookmark the permalink.

9 Responses to How far does the apple fall from the tree?

  1. backroom girl says:

    “it will eventually be possible to get good estimates for mothers and daughters too”

    Is it really necessary to compare sons with fathers and daughters with mothers? Surely it should be possible to compare the occupational SES of the current generation with that of whichever parent had the higher SES job. You would get a closer match between me and my Dad, for example, than between me and my Mum, because he went to university and she didn’t. On that logic, of course, you would expect to find significant upward mobility among women in my generation, relative to their mothers, but such a finding would be somewhat spurious wouldn’t it?

    Sorry if that is all a bit off the point, but I’m only guessing what ‘intergenerational earnings elasticity’ means.

  2. Fred Argy says:

    It is a fascinating topic – to me anyway – and I am glad that your paper is getting wide recognition and discussion.

    In your comment, you remind us of the evidence that “more equal countries tend to be also more socially mobile”. In my 2006 discussion paper on Equality of Opportunity in Australia, I spelt out the positive economic spin-offs from “active” policies – those which seek to level out opportunities. I argued that, beyond a point and in the wrong form (passive transfers), redistribution policies are bad for economic growth and even counter-productive for the poor BUT, in moderation and with the right methods (active social investment), redistribution can be positive for growth. And one of the causation channels is precisely your point – that with more equal countries (and here I mean more equal opportunity countries), there is more income mobility and “it is easier to move from rags to riches”. You therefore get a more entrepreneurial and incentive-driven society and resources are more likely to be directed to people with the greatest potential to contribute to the economy.

    Incidentally, this may be one reason why empirical studies can find no clear correlation between inequality and economic growth (indeed the correlation is if anything negative).

  3. Fred Argy says:

    The full text of my 2006 discussion paper is on

  4. Andrew Leigh says:

    Backroom gal, the US studies that include mothers and daughters tend to focus on family income rather than earnings, since women’s earnings tend to be a poorer proxy for their access to resources than in the case of blokes. That means there’s no alternative to long-run panel datasets. Your approach of highest parental earnings is certainly novel, but would have the problem that it would conflate earnings within an occupation with whether both parents work (since a child of 2 working parents would on average have higher SES).

    Fred, I think looking at the causes and consequences of intergenerational mobility would be terrific research agenda. It’s just a pity it’s so darn data-intensive. As to inequality and growth, I actually think the evidence points somewhat strongly to a positive relationship; but that’s a debate for another day…

  5. Fred Argy says:

    Andrew, your comment about a strong positive relationship between inequality and growth (meaning more inequality leads to stronger growth) surprises me. A recent review of the literature by Quintin and Saving challenges the “old orthodoxy” and points to evidence why inequality might hinder growth and offers a rationale for it. See:

  6. Andrew Leigh says:

    Fred, I usually like these summary papers from the US regional Feds, but I didn’t find that one especially compelling. Eight years ago, Kristin Forbes’ AER paper showed that with panel data econometrics, the relationship was strongly positive. There’s some nice work since then by Sarah Voichovsky which qualifies Forbes, but I’m not aware of anyone who’s successfullly rebutted her. Oddly, QS don’t mention her.

  7. Fred Argy says:

    Thanks Andrew. Yes, I was aware of Forbes’ important 2000 article and noted his findings in my earlier mentioned discussion paper. But other (earlier and more recent) articles continue to be predominantly (although not overwhelmingly) in the other camp.

    As the empirical evidence and theory are both inconclusive and conflicting, the line I run with my students when I conduct policy workshops at the ANU is as follows:

    1. Liberal reform which promises higher economic growth can often increase income inequality (although inequality is also driven by many other factors unrelated to reform).

    2. The appropriate policy response is not to obstruct economic liberal reform (although some compromises may be needed) nor is to forget about redistribution (although some compromises are needed there too). One should instead try to get closer to a win/win outcome where reform gains are widely shared by selecting the “optimal” methods of redistribution (both on the spending and revenue side).

    3. The higher the proportion of social spending going into well-targeted “active” (opportunity-levelling, mobility-enhancing) programs – such as education, health, low-cost rental housing, public transport, removing obstacles to financial development, work incentives and wage subsidies – and the smaller the proportion going into passive unconditional transfers, the more favourable will be the impact on the economy.

    4. The economic balance sheet on redistribution is more likely to be positive if the revenue-raising measures are broad-based and non-distorting.

    5. If the optimal methods of redistribution are used, social spending is likely to have a fairly neutral incremental impact on the economy (as the efficiency costs of higher taxation will tend to be outweighed by the efficiency benefits on the spending side). The gains from the reform will be fully preserved.

    6. That said, the scale of social spending is crucial. Redistribution must reach a point where it becomes a drag on the economy and hence counter-productive. What that point is – and whether we have reached it in Australia – is in the end a judgmental issue.

  8. Kevin Cox says:

    It is interesting that FA argues for the first 5 points that it is not the inequality that matters but how and on what the money is spent that matters and then point 6 says that a more equal society will become a drag on the economy when it gets too equal. I suspect that he says this because he is thinking of methods of redistribution that might appear to destroy incentives but that is likely to be a problem in methods of redistribution not of inequality.

  9. Patrick says:

    It is interesting that he doesn’t actually say anything in point 6 about equality, only about the degree of redistribution. Redistribution is a factor in inequality but both high and low redistribution are compatible with both high and low inequality at any given point in time.

Comments are closed.