¹⁸See Aguiar and Bils (2011) for recent evidence of important systematic biases in the CE data, and Attanasio, Battistin, and Ichimura (2004) for some earlier evidence; see Meyer and Sullivan (2009) for comparisons of poverty as measured by income and consumption sources. Meyer and Sullivan (2009) have argued in a number of papers, however, that at the bottom of the distribution, consumption is likely better measured than income, which may come from informal or irregular sources that are not well captured by the usual survey methods such as the use of income tax returns.