Wal‐Mart and County‐Wide Poverty*

Objectives. This study seeks to identify the independent effect of Wal‐Mart stores on changes in U.S. family‐poverty rates at the county level. We draw on the contributions of a number of disciplines to enhance our understanding of the broader forces that influence poverty. Methods. A key innovation is that we estimate a two‐stage regression model, in which an instrument is created for new Wal‐Mart stores from a location equation; this reduces any potential endogeneity bias in the poverty‐change equation. In addition, we use spatial econometric methods to correct for spatial dependence bias. Results. After controlling for other factors determining changes in the poverty rate over time, we find that counties with more initial (1987) Wal‐Mart stores and counties with more additions of stores between 1987 and 1998 experienced greater increases (or smaller decreases) in family‐poverty rates during the 1990s economic boom period. Conclusions. Wal‐Mart creates both benefits and costs to communities in which the chain locates. These benefits and costs need to be weighed carefully by community decisionmakers in deciding whether to provide public subsidies to the chain.