A survey of the remunerations of graduate students in math PhD programs in public universities initiated and conducted by Hans Christianson, the Director of Graduate Studies at UNC at Chapel Hill is in, and the results are not hugely surprising. With a few exceptions, the (adjusted for cost of living) yearly stipends are pretty comparable, as are the terms of employment.
The TL;DR of the survey is here. An extract from the spreadsheet showing the stipends and cost of living indices (using the data from salary.com) is shown below. Some of the numbers should be taken cum granum salis (Stony Brook, really? did they count Jim Simons?) but by and large they conform with the intuition.