A Crucial Factor in any Analysis of Inequality

Thesis: Properly implemented cost-of-living adjustments are often missing from analyses of inequality data, which often leads to overstated or outright incorrect results.

As the income gap between the wealthy and everybody else grows wider and wider in the United States, the search for the underlying causes has brought more and more economists, think tanks, and academics alike to analyzing data on inequality.  But, as with any statistical analysis, researchers must be careful with their underlying assumptions and variables before coming to conclusions.  One key variable that must be controlled for when comparing income levels across the country is cost of living: in different parts of the country, income levels will vary as employers must compensate labor differently based on the market they’re in, to compete with employers in lower cost of living markets.  For example, the mean annual wage of a nurse in California is $90,860 versus $67,140 in Pennsylvania (source: Bureau of Labor Statistics).  The map below, created by the Tax Foundation, provides a visualization of how cost of living varies across the country.  

$100bystateDespite the large variation in cost of living across the country, many studies seem to neglect controlling for the variable and the effect it has on nominal income.  For example, a paper written by David Autor and cited in the Wall Street Journal argues that the biggest societal factor for explaining inequality is that of education, pointing to the fact that the earnings gap between non-college educated households and those with a college education has grown at a faster pace than the gap between the 1% and the rest of the country.  But Autor’s study is missing a key piece: it doesn’t control at all for cost of living across earners.  It simply compares earnings by education level, with no regard for location – even though there is definitely a correlation between education level and cost of living.  More educated households are more likely to be located in areas like California and New England, where not only is the cost of living higher, but so are income levels as employers compensate for that cost of living.  The result is that Autor’s claims are overstated: while there is still certainly truth to his study, it is also the case that we would expect there to be a correlation between education and nominal income that is actually due simply to variation in cost of living.

On the flip side, there are cases where arguments about income inequality have been made that seem to overstate the cost of living effect.  An example that should be well known to economics students on this campus by now is the highly controversial Michigan Daily article Relative Wealth, which argues that one can be considered to be a middle class American even with an annual household income of $250,000, due to of cost of living variation.  While I won’t argue for or against the premise of this article, I will point out that Klein overestimates the degree to which higher cost of living areas dilute one’s income, and that weakens her argument.  While it is certainly true that higher major expenses (mainly rent/home prices) should correspond to a higher income in certain areas of the country, many discretionary expenses become relatively cheaper when one lives in a high cost of living area.  Goods which do not vary in price geographically (e.g. practically anything you can buy on the internet) cost a smaller portion of one’s income if they live in a higher cost of living area, giving them more purchasing power for certain goods.  This somewhat offsets the nominal wage differential of different cost of living areas.  And so those who wish to study income inequality must tread carefully: there are many confounding variables at play.  I don’t know what the best way to control for cost of living is, but those who ignore it will come to conclusions that may be over or understated.

Leave a Reply