Saturday, November 5, 2011

Income inequality in Canada's provinces

Kate Pickett and Richard Wilkinson's The Spirit Level compares countries by their Gini coefficient, a measure of income inequality, and also breaks the United States down into its states to compare those.

So I broke Canada down by its provinces too, using Gini data from Harvard's Center for European Studies, from 2005. The data is listed for the ten provinces, though unfortunately not for the three territories. (NOTE: On rechecking this source I realise it is not as accurate as I had hoped, with the Gini coefficients calculated with data from 1980 to 2003. I hope this will continue to have relevance into the 2000s but I realise this could create a lot of noise. Unfortunately I thought to double-check this only after the rest of the blog post was written - keep reading, but take with a pinch of salt.)

The Harvard data is also split into 'pregini' - income inequality at market level before government taxes and transfers are included - and 'postgini' - income inequality after transfers. Below I will look at both measures, and compare them in scatterplots with other social phenomena.

Remember what we are expecting to see here. If Pickett and Wilkinson are right, we should see correlations between income inequality and a host of social problems. I am reproducing here the same kind of graph they use repeatedly in the book. As far as possible I will aim for 2003 statistics. These graphs will include a regression line calculated by Excel, a statistical device Pickett and Wilkinson also use to show 'the line which best fits the trend through the data points.'

First, obesity, from Statistics Canada, the official statistics office, 2004.

Pregini and Obesity
Postgini and Obesity
This is a curious result. Before wealth transfers like social welfare are taken into account, there seems to be a strong correlation between income inequality and obesity. After transfers, this correlation is much weaker. Why?

Something makes one Canadian province have higher income inequality than another, before the state intervenes to create greater balance. Perhaps that unknown factor is what causes obesity, rather than actual income inequality.

I have suggested before that Pickett and Wilkinson use income and status interchangeably, and wondered if they should. I argued that there have been people with low status and high income, or high status and low income.

If Pickett and Wilkinson are right, that income inequality increases the rate of obesity, maybe it is the low status associated with poorer socioeconomic groups that is the problem, and not the actual amount of money they have after transfers. It seems plausible that one might be living comfortably enough on a council estate, but still be viewed as low-status by those living in private accommodation.

If this is the case then a large welfare state that addresses income inequality without somehow reducing pre-transfer inequality won't make much difference.

This is not an argument of Pickett and Wilkinson, who instead say that a welfare state is one successful way to reduce inequality:
Sweden does it through redistributive taxes and benefits and a large welfare state. As a proportion of national income, public social expenditure in Japan is, in contrast to Sweden, among the lowest of the major developed countries. Japan gets its high degree of equality not so much from redistribution as from a greater equality of market incomes, of earnings before taxes and benefits. Yet despite the differences, both countries do well...
Let's look at a few more results. Next infant mortality, 2005.

Pregini and Infant Mortality
Postgini and Infant MortalityWe see the opposite trend. Before transfers the highly unequal provinces do slightly better, while after transfers they seem to do worse, which challenges my earlier idea. Let's see some more, homicide rate, 2006, next.

Pregini and Homicide
Postgini and Homicide
Another puzzling one! Before transfers, the unequal provinces are safer. After transfers the unequal provinces are more dangerous! I'm really not sure what to make of that. Is there something inherent to those provinces which create initial market inequality that keeps homicides low (but makes people obese)? Next life satisfaction (percentage satisfied or very satisfied) from 2007.

Pregini and Life Satisfaction
Postgini and Life Satisfaction
Again, before transfers the unequal provinces are happier, after transfers the equal provinces are happier. The negative correlation between actual income inequality and life satisfaction is what Pickett and Wilkinson would have predicted. The other correlation - well I'm not sure what that means, if anything. (Any ideas?) Next perceived life stress ('quite a lot') in 2007.

Pregini and Perceived Life Stress
Postgini and Perceived Life Stress
And finally sense of belonging to local community (somewhat strong or very strong) in 2007:

Pregini Sense of Belonging to Community
Postgini Sense of Belonging to Community
Just to toy with us, this time inequality is slightly correlated with a sense of community!

Well at this stage I am sorely confused. For some factors we see the opposite of what The Spirit Level would predict. For most we do see post-transfer income inequality correlating with negative social traits.

Of course there could be unseen causes for all of these phenomena, causes which sometimes correlate with inequality and sometimes do not. Interesting though. Should I be eating a little humble pie since The Spirit Level's predictions hold true for a few of these indicators? Or gloating that they don't for all?

Dissatisfied with this result, I am going for two final graphs. The first compares annual charitable donations per person in 2004 in Canadian dollars with postgini:Different provinces have different levels of wealth and I'm not controlling for that so the result (which shows that people in more unequal provinces give more) may be skewed. Hence this, showing hours volunteered in 2004:

People in more unequal provinces volunteer more? Again independent verification of The Spirit Level's predictions eludes me. If income inequality harms social cohesion, why are people in less equal regions giving more time and money to help others? I remain sceptical.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.