@article{BradyGiesselmannKohleretal.2018,
  author    = {Brady, David and Giesselmann, Marco and Kohler, Ulrich and Radenacker, Anke},
  title     = {How to measure and proxy permanent income},
  series = {The Journal of Economic Inequality},
  volume    = {16},
  journal   = {The Journal of Economic Inequality},
  number    = {3},
  publisher = {Springer},
  address   = {Dordrecht},
  issn      = {1569-1721},
  doi       = {10.1007/s10888-017-9363-9},
  pages     = {321 -- 345},
  year      = {2018},
  abstract  = {Permanent income (PI) is an enduring concept in the social sciences and is highly relevant to the study of inequality. Nevertheless, there has been insufficient progress in measuring PI. We calculate a novel measure of PI with the German Socio-Economic Panel (SOEP) and U.S. Panel Study of Income Dynamics (PSID). Advancing beyond prior approaches, we define PI as the logged average of 20+ years of post-tax and post-transfer ("post-fisc") real equivalized household income. We then assess how well various household- and individual-based measures of economic resources proxy PI. In both datasets, post-fisc household income is the best proxy. One random year of post-fisc household income explains about half of the variation in PI, and 2-5 years explain the vast majority of the variation. One year of post-fisc HH income even predicts PI better than 20+ years of individual labor market earnings or long-term net worth. By contrast, earnings, wealth, occupation, and class are weaker and less cross-nationally reliable proxies for PI. We also present strategies for proxying PI when HH post-fisc income data are unavailable, and show how post-fisc HH income proxies PI over the life cycle. In sum, we develop a novel approach to PI, systematically assess proxies for PI, and inform the measurement of economic resources more generally.},
  language  = {en}
}
@article{KohlerKreuterStuart2018,
  author    = {Kohler, Ulrich and Kreuter, Frauke and Stuart, Elizabeth A.},
  title     = {Nonprobability Sampling and Causal Analysis},
  series = {Annual review of statistics and its application},
  volume    = {6},
  journal   = {Annual review of statistics and its application},
  publisher = {Annual Reviews},
  address   = {Palo Alto},
  issn      = {2326-8298},
  doi       = {10.1146/annurev-statistics-030718-104951},
  pages     = {149 -- 172},
  year      = {2018},
  abstract  = {The long-standing approach of using probability samples in social science research has come under pressure through eroding survey response rates, advanced methodology, and easier access to large amounts of data. These factors, along with an increased awareness of the pitfalls of the nonequivalent comparison group design for the estimation of causal effects, have moved the attention of applied researchers away from issues of sampling and toward issues of identification. This article discusses the usability of samples with unknown selection probabilities for various research questions. In doing so, we review assumptions necessary for descriptive and causal inference and discuss research strategies developed to overcome sampling limitations.},
  language  = {en}
}