Propensity score matching (PSM) has become a popular approach to estimate causal treatment effects. It is widely applied when evaluating labour market policies, but empirical examples can be found in very diverse fields of study. Once the researcher has decided to use PSM, he is confronted with a lot of questions regarding its implementation. To begin with, a first decision has to be made concerning the estimation of the propensity score. Following that one has to decide which matching algorithm to choose and determine the region of common support. Subsequently, the matching quality has to be assessed and treatment effects and their standard errors have to be estimated. Furthermore, questions like 'what to do if there is choice-based sampling?' or 'when to measure effects?' can be important in empirical studies. Finally, one might also want to test the sensitivity of estimated treatment effects with respect to unobserved heterogeneity or failure of the common support condition. Each implementation step involves a lot of decisions and different approaches can be thought of. The aim of this paper is to discuss these implementation issues and give some guidance to researchers who want to use PSM for evaluation purposes.

/pdf/some-practical-guidance-for-the-implementation-of-propensity-2qsp4y0xke.pdf

Some practical guidance for the implementation of propensity score matching

We investigate conditions sufficient for identification of average treatment effects using instrumental variables. First we show that the existence of valid instruments is not sufficient to identify any meaningful average treatment effect. We then establish that the combination of an instrument and a condition on the relation between the instrument and the participation status is sufficient for identification of a local average treatment effect for those who can be induced to change their participation status by changing the value of the instrument. Finally we derive the probability limit of the standard IV estimator under these conditions. It is seen to be a weighted average of local average treatment effects.

Evolution and Rationality Some Recent Game-Theoretic Results. Identification and Estimation of Local Average Treatment Effects

Recently there has been a surge in econometric work focusing on estimating average treatment effects under various sets of assumptions. One strand of this literature has developed methods for estimating average treatment effects for a binary treatment under assumptions variously described as exogeneity, unconfoundedness, or selection on observables. The implication of these assumptions is that systematic (for example, average or distributional) differences in outcomes between treated and control units with the same values for the covariates are attributable to the treatment. Recent analysis has considered estimation and inference for average treatment effects under weaker assumptions than typical of the earlier literature by avoiding distributional and functional-form assump- tions. Various methods of semiparametric estimation have been proposed, including estimating the unknown regression functions, matching, meth- ods using the propensity score such as weighting and blocking, and combinations of these approaches. In this paper I review the state of this literature and discuss some of its unanswered questions, focusing in particular on the practical implementation of these methods, the plausi- bility of this exogeneity assumption in economic applications, the relative performance of the various semiparametric estimators when the key assumptions (unconfoundedness and overlap) are satise ed, alternative estimands such as quantile treatment effects, and alternate methods such as Bayesian inference.

/pdf/nonparametric-estimation-of-average-treatment-effects-under-35vd3gs0i5.pdf

Nonparametric Estimation of Average Treatment Effects under Exogeneity: A Review

Part I. Causality and Empirical Research in the Social Sciences: 1. Introduction Part II. Counterfactuals, Potential Outcomes, and Causal Graphs: 2. Counterfactuals and the potential-outcome model 3. Causal graphs Part III. Estimating Causal Effects by Conditioning on Observed Variables to Block Backdoor Paths: 4. Models of causal exposure and identification criteria for conditioning estimators 5. Matching estimators of causal effects 6. Regression estimators of causal effects 7. Weighted regression estimators of causal effects Part IV. Estimating Causal Effects When Backdoor Conditioning Is Ineffective: 8. Self-selection, heterogeneity, and causal graphs 9. Instrumental-variable estimators of causal effects 10. Mechanisms and causal explanation 11. Repeated observations and the estimation of causal effects Part V. Estimation When Causal Effects Are Not Point Identified by Observables: 12. Distributional assumptions, set identification, and sensitivity analysis Part VI. Conclusions: 13. Counterfactuals and the future of empirical research in observational social science.

Counterfactuals and Causal Inference: Methods and Principles for Social Research

Instrumental variable (IV) methods are commonly used in accounting research (e.g., earnings management, corporate governance, executive compensation, and disclosure research) when the regressor variables are endogenous. While IV estimation is the standard textbook solution to mitigating endogeneity problems, the appropriateness of IV methods in typical accounting research settings is not obvious. Drawing on recent advances in statistics and econometrics, we identify conditions under which IV methods are preferred to OLS estimates and propose a series of tests for research studies employing IV methods. We illustrate these ideas by examining the relation between corporate disclosure and the cost of capital.

https://papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID1505074_code113360.pdf?abstractid=694824&rulid=9740146&mirid=4

On the Use of Instrumental Variables in Accounting Research

Propensity score matching provides an estimate of the effect of a “treatment” variable on an outcome variable that is largely free of bias arising from an association between treatment status and observable variables. However, matching methods are not robust against “hidden bias” arising from unobserved variables that simultaneously affect assignment to treatment and the outcome variable. One strategy for addressing this problem is the Rosenbaum bounds approach, which allows the analyst to determine how strongly an unmeasured confounding variable must affect selection into treatment in order to undermine the conclusions about causal effects from a matching analysis. Instrumental variables (IV) estimation provides an alternative strategy for the estimation of causal effects, but the method typically reduces the precision of the estimate and has an additional source of uncertainty that derives from the untestable nature of the assumptions of the IV approach. A method of assessing this additional uncertainty...

/pdf/assessing-bias-in-the-estimation-of-causal-effects-rosenbaum-1ec6hndzbe.pdf

Assessing bias in the estimation of causal effects: Rosenbaum bounds on matching estimators and instrumental variables estimation with imperfect instruments

Using harmonized longitudinal data from the British Household Panel Survey (BHPS), the German Socio-Economic Panel (GSOEP), and the National Longitudinal Survey of Youth (NLSY), we trace career prospects after motherhood for five cohorts of American, British, and West German women around the 1960s. We establish wage penalties for motherhood between 9% and 18% per child, with wage losses among American and British mothers being lower than those experienced by mothers in Germany. Labor market mechanisms generating the observed wage penalty for motherhood differ markedly across countries, however. For British and American women, work interruptions and subsequent mobility into mother-friendly jobs fully account for mothers’ wage losses. In contrast, respective penalties are considerably smaller in Germany, yet we observe a substantial residual wage penalty that is unaccounted for by mothers’ observable labor market behavior. We interpret this finding as indicating a comparatively more pronounced role for statistical discrimination against mothers in the German labor market.

Motherhood, labor force behavior, and women’s careers: An empirical assessment of the wage penalty for motherhood in britain, germany, and the united states

This article uses panel data from the Survey of Income and Program Participation (SIPP) and the European Community Household Panel (ECHP) for a comparative analysis of workers' post-unemployment earnings trajectories in the United States and 12 Western European countries. Across the study sample of industrialized countries, results of difference-in-difference propensity score matching show post-unemployment earnings losses to be largely permanent and particularly significant for high-wage and older workers as well as for women. The analyses also show that negative effects of unemployment on workers' subsequent earnings are mitigated through either generous unemployment benefit systems or strict labor market regulation. These effects stem partly from favorable behavioral responses that prevent downward occupational and industrial mobility and partly from changes in the overall structure of labor markets favoring the transferability of worker skills between jobs. These positive effects materialize despite t...

Scar effects of unemployment : An assessment of institutional complementarities

Preface 1. The Transition from School to Work: A European Perspective 2. Educational Attainment in the European Union: Recent Trends in Qualification Patterns 3. Young People and New Entrants in European Labour Markets: The Timing of Gradual Integration 4. The Structure of Labour Market Entry in Europe: A Typological Analysis 5. Learning and Working: Double Statuses in Youth Transitions 6. Returns to Education in Context: Individual Education and Transition Outcomes in Euopean Labour Markets 7. The Integration of Young People into the Labour Market: The Role of Training Systems and Labour Market Regulation 8. Transition Pathways in Italy and Spain: Different Patterns, Similar Vulnerability? 9. Explaining Change in Early Career Outcomes: Labour Market Conditions, Educational Expansion, and Youth Cohort Sizes 10. Conclusions: Explaining Cross-National Differences in School-to-Work Transitions Methodological Index: Using the European Labour Force Survey for Transition Research Bibliography

Transitions from education to work in Europe: the integration of youth into EU labour markets

Originating in econometrics and statistics, the counterfactual model provides a natural framework for clarifying the requirements for valid causal inference in the social sciences. This article presents the basic potential outcomes model and discusses the main approaches to identification in social science research. It then addresses approaches to the statistical estimation of treatment effects either under unconfoundedness or in the presence of unmeasured heterogeneity. As an update to Winship & Morgan's (1999) earlier review, the article summarizes the more recent literature that is characterized by a broader range of estimands of interest, a renewed interest in exploiting experimental and quasi-experimental designs, and important progress in the areas of semi- and nonparametric estimation of treatment effects, difference-in-differences estimation, and instrumental variable estimation. The review concludes by highlighting implications of the recent econometric and statistical literature for sociological...

Markus Gangl

Papers

Assessing bias in the estimation of causal effects: Rosenbaum bounds on matching estimators and instrumental variables estimation with imperfect instruments

Motherhood, labor force behavior, and women’s careers: An empirical assessment of the wage penalty for motherhood in britain, germany, and the united states

Scar effects of unemployment : An assessment of institutional complementarities

Transitions from education to work in Europe: the integration of youth into EU labour markets

Causal Inference in Sociological Research