Richard Green wonders if there are measures of the cost of living that vary according to income group:
Different Market Baskets for Different Income Levels, by Richard Green: I keep reading stories about food pantries being under particular pressure this year. The stories would be more helpful if they could explain explicitly why the food stamp program (perhaps the most successful anti-poverty program in the US) isn't sufficient to prevent this. We do know that not everyone eligible for food stamps uses them, but this has always been true, and so wouldn't explain a change in demand at the pantries; we need to look elsewhere for an explanation.
My suspicion is that an accurate measure of CPI would vary by income group. The most obvious example is that low income people spend a higher fraction of income on heat, electricity and transportation than higher income people. Even if the entire CPI is flat, if the energy and transportation sectors see large rises in prices, it will likely have a particularly large impact on the bottom quintile of the income distribution, and hence cause real incomes within this group to fall. Of course, gas and heating oil prices gave risen a lot over the past couple of years.
When I look at the BLS web site, I don't find anything about different market baskets for different income classes--I do wonder if there is something out there.
Anyone? The best I can do is this 1998 working paper from the IMF ("Is the United States CPI Biased Across Income and Age Groups?," by S. Nuri Erbas and Chera L. Sayers) showing that the CPI understates the true cost of living for older and/or poorer households, and overstates the rate for younger and/or richer households. See, in particular, Table 3, Tables I2 and I3 in the appendix, and Chart 1.
I can think of public policy reasons to avoid having more than one measure, e.g. the potential for perverse incentives when earning additional income can change the CPI used to adjust income for changes in the cost of living, the cost of calculating more than one measure, and the difficult theoretical, statistical, and political problem of defining official income classes (how many income classes, where to draw the line between classes, what income measure to use, what exclusions to allow, etc.). But even if problems such as these prevent us from actually implementing cost of living measures that differ by income group, the extent to which the market basket and the associated cost of living varies across income (and other) groups is an interesting and important question.