Econ Papers Daily / 期刊论文
收入变化与家庭收入单一问题测量中的测量误差
Income changes and measurement error in single-question measures of household income*
摘要
I provide insights into income changes measured in surveys using single questions, by extending to longitudinal settings a measurement error model previously developed in the literature. In this framework, single-question income data are validated against a benchmark provided by detailed source-by-source questions, which are considered the best practice for measuring income in surveys. I outline the assumptions required to infer benchmark income changes between two time periods (e.g., two subsequent survey waves), both at the macro- and micro-levels, when income is collected using single questions. Recovering meaningful income changes from such measures requires assuming that respondents’ misreporting behaviour is time-invariant. The use of single-question income changes as dependent and independent variables in linear models, and potential heterogeneity in respondents’ misreporting behaviour are also discussed. I apply the methodology to estimate income changes in Italy between 2022 and 2023, using data from a new web-survey conducted by the Bank of Italy.
中文摘要
我通过将文献中先前开发的测量误差模型扩展到纵向情境，提供了对使用单一问题在调查中测量的收入变化的洞见。在该框架中，单问题收入数据以详细逐项来源问题所提供的基准进行验证，后者被视为调查中测量收入的最佳实践。我概述了当收入使用单一问题收集时，在宏观和微观层面推断两个时期（例如两个后续调查波次）之间基准收入变化所需的假设。从此类测量中恢复有意义的收入变化需要假设受访者的误报行为是时间不变的。还讨论了将单问题收入变化作为线性模型中的因变量和自变量使用，以及受访者误报行为中的潜在异质性。我将该方法应用于估计意大利2022年至2023年间的收入变化，使用的是意大利银行开展的一项新网络调查的数据。