10This depiction is perhaps a bit too bleak; Parker and Vissing-Jorgensen (2009) have done some impressive work that makes some progress in determining how the spending of different groups varies over the business cycle, arguing in particular that high-income and low-income households seem to bear more of the fluctuations than do middle-income households. But the amount of effort required to extract results of this kind from such a highly imperfect dataset is a formidable barrier to entry for other scholars, and skeptics can argue that other factors (like variation in survey participation over the cycle) could drive the results.