Working Paper: NBER ID: w13413
Authors: Bruce D. Meyer; James X. Sullivan
Abstract: In the U.S., analyses of poverty rates and the effects of anti-poverty programs rely almost exclusively on income data. In earlier work (Meyer and Sullivan, 2003) we emphasized that conceptual arguments generally favor using consumption data to measure the well-being of the poor, and, on balance, data quality issues favor consumption in the case of single mothers. Our earlier work did not show that income and consumption differ in practice. Here we further examine data quality issues and show that important conclusions about recent trends depend on whether one uses consumption or income. Changes in the distribution of resources for single mothers differ sharply in recent years depending on whether measured by income or consumption. Measures of overall and sub-group poverty also sharply differ. In addition to examining broader populations and a longer time period, we also consider new dimensions of data quality such as survey and item nonresponse, imputation, and precision. Finally, we demonstrate the flaws in a recent paper that compares income and consumption data.
Keywords: Poverty; Wellbeing; Income; Consumption
JEL Codes: D31; I31; I32; I38
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| consumption (E21) | wellbeing (I31) |
| income (E25) | wellbeing (I31) |
| low consumption (E21) | adverse outcomes (I12) |
| low income (I32) | adverse outcomes (I12) |
| consumption (E21) | asset accumulation (G51) |
| income measures (I32) | asset accumulation (G51) |
| income trends (D31) | consumption trends (D12) |
| consumption poverty (I32) | income poverty (I32) |