This paper deals with the determinants of poverty in the U.S. at the census tract level. By proceeding in this way, potential aggregation errors that may occur when using larger areas as the unit of observation are avoided. Our empirical analysis exploits a geographically consistent data set covering the period 1970–2010 and all metropolitan statistical areas. The estimations have been carried out using a system generalized method of moments framework for dynamic panels. Our results show that poverty is a persistent phenomenon, especially in tracts with higher levels. Those variables with a more robust relationship with poverty reflect labor market conditions. We also obtain evidence of a higher vulnerability of women to poverty. These findings have been framed within the debate about the optimal design of local development strategies. The main conclusion drawn is that place‐based and person‐centered policies can be considered as complementary rather than as substitutes.

中文翻译：

---

1970-2010 年美国人口普查区贫困决定因素的评估

本文在人口普查区层面讨论美国贫困的决定因素。通过以这种方式进行，可以避免使用较大区域作为观察单位时可能出现的潜在聚合错误。我们的实证分析利用了地理上一致的数据集，涵盖了 1970-2010 年和所有大都市统计区域。已经使用动态面板的矩量框架的系统广义方法进行了估计。我们的研究结果表明，贫困是一个持续存在的现象，尤其是在水平较高的地区。那些与贫困关系更密切的变量反映了劳动力市场状况。我们还获得了妇女更容易陷入贫困的证据。这些发现是在关于地方发展战略的最佳设计的辩论中形成的。