Sources of the Union Wage Gap: Results from High-Dimensional Fixed Effects Regression Models

An interaction in a fixed effects (FE) regression is usually specified by demeaning the product term. However, algebraic transformations reveal that this strategy does not yield a within-unit estimator. Instead, the standard FE interaction estimator reflects unit-level differences of the interacted variables. This property allows interactions of a time-constant variable and a time-varying variable in FE to be estimated but may yield unwanted results if both variables vary within units. In such cases, Monte Carlo experiments confirm that the standard FE estimator of x ⋅ z is biased if x is correlated with an unobserved unit-specific moderator of z (or vice versa). A within estimator of an interaction can be obtained by first demeaning each variable and then demeaning their product. This “double-demeaned” estimator is not subject to bias caused by unobserved effect heterogeneity. It is, however, less efficient than standard FE and only works with T > 2.

Download Full-text

Language Variation Suite: A theoretical and methodological contribution for linguistic data analysis

Proceedings of the Linguistic Society of America ◽

10.3765/plsa.v1i0.3734 ◽

2016 ◽

Vol 1 ◽

pp. 29 ◽

Cited By ~ 2

Author(s):

Olga Scrivner ◽

Manuel Díaz-Campos

Keyword(s):

Fixed Effects ◽

Regression Models ◽

Quantitative Methods ◽

Language Variation ◽

Mixed Effects ◽

Logistic Regression Models ◽

Linguistic Data ◽

Diachronic Study ◽

Phonetic Context ◽

Fixed Effects Regression

In recent years there has been growing interest in quantitative methods for analyzing linguistic data. Advanced multifactorial statistical analyses, such as inferential trees and mixed-effects logistic regression models, have become more accessible for linguistic research as a result of the availability of an open source programming environment provided by the statistical software R. In the present paper, we introduce a novel toolkit, Language Variation Suite, a software program that offers a friendly environment for conducting quantitative analyses. We demonstrate how theory built on traditional monofactorial analysis can be extended to macro and micro multifactorial approaches allowing for a deeper understanding of language variation. The focus of the analysis is based on intervocalic /d/ deletion in Spanish from the Diachronic Study of the Speech of Caracas 1987 and 2004-2010. In contrast to traditional methodological approaches we have treated intervocalic /d/ as a continuous dependent variable according to the intensity ratio measurements obtained. Furthermore, we have integrated various syntactic, phonetic and sociolinguistic factors. Non-parametric and fixed-effects regression models revealed that overall age (younger speakers), sex (male speakers), phonetic context (low vowels), token frequency and morphosyntactic category (past participles) have a significant effect on the lenition of intervocalic /d/. In contrast, the mixed-effects model selected only phonetic context, frequency and category, showing that individual speaker variation is higher than group variation.

Download Full-text

On the Use of Two-Way Fixed Effects Regression Models for Causal Inference with Panel Data

Political Analysis ◽

10.1017/pan.2020.33 ◽

2020 ◽

pp. 1-11 ◽

Cited By ~ 2

Author(s):

Kosuke Imai ◽

In Song Kim

Keyword(s):

Panel Data ◽

Causal Inference ◽

Fixed Effects ◽

Regression Models ◽

Applied Research ◽

Causal Effects ◽

Difference In Differences ◽

The Difference ◽

Fixed Effects Regression ◽

Time Specific

Abstract The two-way linear fixed effects regression (2FE) has become a default method for estimating causal effects from panel data. Many applied researchers use the 2FE estimator to adjust for unobserved unit-specific and time-specific confounders at the same time. Unfortunately, we demonstrate that the ability of the 2FE model to simultaneously adjust for these two types of unobserved confounders critically relies upon the assumption of linear additive effects. Another common justification for the use of the 2FE estimator is based on its equivalence to the difference-in-differences estimator under the simplest setting with two groups and two time periods. We show that this equivalence does not hold under more general settings commonly encountered in applied research. Instead, we prove that the multi-period difference-in-differences estimator is equivalent to the weighted 2FE estimator with some observations having negative weights. These analytical results imply that in contrast to the popular belief, the 2FE estimator does not represent a design-based, nonparametric estimation strategy for causal inference. Instead, its validity fundamentally rests on the modeling assumptions.

Download Full-text

Flexible and fast estimation of quantile treatment effects: The rqr and rqrplot commands

10.31235/osf.io/4vquh ◽

2021 ◽

Author(s):

Nicolai T. Borgen ◽

Andreas Haupt ◽

Øyvind N. Wiborg

Keyword(s):

Quantile Regression ◽

Fixed Effects ◽

Regression Models ◽

Treatment Effects ◽

The Other ◽

High Dimensional ◽

Fast Estimation ◽

Quantile Treatment Effects

Using quantile regression models to estimate quantile treatment effects is becoming increasingly popular. This paper introduces the rqr command that can be used to estimate residualized quantile regression (RQR) coefficients and the rqrplot postestimation command that can be used to effortless plot the coefficients. The main advantages of the rqr command compared to other Stata commands that estimate (unconditional) quantile treatment effects are that it can include high-dimensional fixed effects and that it is considerably faster than the other commands.

Download Full-text

Gendered Wealth Losses after Dissolution of Cohabitation but not Marriage in Germany

10.31235/osf.io/rcuvd ◽

2019 ◽

Author(s):

Diederik Boertien ◽

Philipp M. Lersch

Keyword(s):

Gender Inequality ◽

Fixed Effects ◽

Regression Models ◽

Legal Protection ◽

Economic Consequences ◽

Union Dissolution ◽

Economic Wealth ◽

Marital Property ◽

Fixed Effects Regression ◽

Property Regimes

Objective: To document gender differences in how economic wealth changes following the dissolution of marriage and cohabitation in Germany.Background: Wealth can be an important resource to deal with the adverse economic consequences of union dissolution. Marital property regimes usually ensure that both partners receive a share of the couples’ wealth following a divorce. The dissolution of cohabiting unions is not governed by such rules in most countries, including Germany, which may lead to a more unequal division of wealth following the dissolution of cohabitation as compared to marriage.Method: The analysis consists of multivariable fixed-effects regression models based on longitudinal data from the German Socio-Economic Panel (N = 6,388 individuals) for the years 2002 to 2017.Results: Changes in wealth are relatively similar for men and women after the dissolution of marriage. The dissolution of cohabiting unions is related to losses in wealth for women, but not for men. Controlling for post-dissolution processes, gender inequality increases after the dissolution of cohabitations.Conclusion: Union dissolution is associated with wealth losses. The legal protection of the economically weaker spouse in marriage prevents gender inequality in these wealth losses. Lacking such legal protection, cohabitation is associated with gender inequality in the consequences of dissolution.

Download Full-text

Lockdown benefit varies among countries and sub-national units: a reanalysis of the data by Bendavid et al. (2021)

10.1101/2021.02.17.21251898 ◽

2021 ◽

Author(s):

Lukas Boesch

Keyword(s):

Growth Rates ◽

Fixed Effects ◽

Regression Models ◽

Empirical Support ◽

Mixed Effects ◽

Point Of View ◽

Pairwise Comparisons ◽

Fixed Effects Regression ◽

Pharmaceutical Interventions ◽

At Home

AbstractAre the lockdown measures limiting the propagation of COVID-19? Recent analyses on the effectiveness of non-pharmaceutical interventions in reducing COVID-19 growth rates delivered conflicting conclusions. While Haug et al. (2020) did find strong empirical support for reductions in COVID-19 growth rates, Bendavid et al. (2021) did not. Here, I present the results of a reanalysis of the data by Bendavid et al. (2021). Instead of relying on pairwise comparisons between 10 countries with fixed-effects regression models to isolate the effect of lockdown measures, I modelled the development of the pandemic with and without lockdown measures for the entire period and all countries included in the data with one mixed-effects regression model. My results reconciled the conflicting conclusions of Haug et al. (2020) and Bendavid et al. (2021): while mandatory business closure orders did not affect COVID-19 growth rates, a general decrease in COVID-19 growth rates was attributable to the implementation of mandatory stay-at-home orders. However, the effect of mandatory stay-at-home orders varied, being weaker, even zero, in some countries and sub-national units and stronger in others, where COVID-19 growth rates only decreased due to the implementation of mandatory stay-at-home orders. The heterogeneity in the effect of mandatory stay-at-home orders on the spread of COVID-19 is challenging from a scientific and political point of view.

Download Full-text