Margaret Wente and Inequality
I highly recommend a new StatsCan (Andrew Heisz) study of income inequality and redistribution – with one significant quibble.
Heisz confirms a great deal of what we know – family after tax income inequality has been growing apace in the 1990s, driven above all by increased inequality of market income. http://www.statcan.ca/english/research/11F0019MIE/11F0019MIE2007298.pdf
This is a methodologically sophisticated study. It confirms the main story line of Armine Yalinizyan’s recent update on the growing gap for the CCPA inequality project (itself based on more or less the same data base) but adds to it by standardizing the family income data for family size (giving us an important piece of methodological defence against the rampant inequality deniers) and also giving us several useful statistical measures of inequality and polarization.
One of Heisz’s key findings is important – it is the growth of market income inequality which explains growing after tax family income inequality, not a major reduction in the offsetting role of the tax/transfer system.
My quibble is this. Heisz tosses out the (unsubstantiated) thought (p9) that growing market income inequality is not explained by rising individual earnings inequality, but rather by changed family earnings patterns. He has picked up on the “homogamy” thesis – that there are many more high paid highly-educated women out there, and that they are mainly living with high-paid men, compounding the inequality effect of the individual earnings distribution. This was the theme picked up by Margaret Wente in today’s Globe. Conservatives are fine with this argument since it puts progressives on the defensive – after all we don’t want to worry about high paid women or consider taxing family rather than individual incomes (which we don’t want to do on gender equity grounds.) Homogamy has the implication that inequality is at least partly beyond our control.
My take is that “homogamy” does indeed play a role as suggested by earlier StatsCan work. However, the tax data for INDIVIDUALS clearly show a surge in top tail driven income inequality as the top 1% or so have disproportionately benefited from economic growth. This has been closely documented for Canada by Saez and Veall to 2000 (see their working paper at http://papers.nber.org/papers/W9607
My quick take on the most recent tax data is that the trend continues very strongly in that direction- the income share of those making more than $100,000 and especially more than $250,000 – the top one half of one percent or so – is rising fast. (Preliminary 2004 tax data show that the top one half of one percent making more than $250k now “earn” about 10% of all income, even more if you adjust for the untaxed portion of capital gains.)
As a further quibble, many statistical measures tend to under-state top tail driven inequality. The Gini mainly captures what is happening at the middle rather than at the extremes, and the often-used p90/p10 ratio – the minimum distance between the top and bottom 10% – doesn’t capture what is happening in the top 10%.
But count on Peggy to grab an aside in a relatively obscure StatsCan study to put us off on the wrong trail!
It seems that those few, speculative sentences in Heiszâ€™s paper garnered a great deal of attention. In todayâ€™s Ottawa Citizen, William Watson picked up on the same theme as Margaret Wente:
An ungated version of the Saez-Veall paper.
A fascinating piece of work. And I agree; it looks as though the gains are concentrated above the 99th percentile: much of the gains of the top decile are in the top 99.5% or even the top 99.9%.
I might have added a reference to Brahim Boudarbat, Thomas Lemieux and Craig Riddell
“Recent Trends in Wage Inequality and the Wage Structure in Canada” in David Green and Jon Kesselman (Eds) Dimensions of Inequality in Canada. UBC Press. 2006. This examines the changing individual wage structure and concludes as follows: “Our main finding, therefore, is that the wage inequality between more educated and less educated workers increased substantially over the past two decades, with particularly steep growth for men between 1995 and 2000.” (p.302) Somewhat ironically, the authors thank Andrew Heisz for providing special census data tabulations.
The Vancouver Sun has re-printed Watson’s column today:
Another piece of the puzzle is that Statscan, by its own admission, is undercounting the poorest households in is surveys. Marc Frenette, David Green and Garnet Picot find that, based on comparisons among the standard survey and more comprehensive census and tax data, incomes at the bottom are way smaller than officially stated. I blogged about it, with links to the paper, here: