It is a major achievement of the econometric treatment effect literature to clarify under which conditions causal effects are non-parametrically identified. The first part of this chapter focuses on the static treatment model. In this part, I show how panel data can be used to improve the credibility of matching and instrumental variable estimators. In practice, these gains come mainly from the availability of outcome variables measured prior to treatment. Such outcome variables also foster the use of alternative identification strategies, in particular so-called difference-in-difference estimation. In addition to improving the credibility of static causal models, panel data may allow credibly estimating dynamic causal models, which is the main theme of the second part of this chapter.
Not available in German. After unification the public and private sector spent large amounts of money to retrain the East German labour force in order to ease the transition to the new market economy. Using individual longitudinal and cross sectional data, this book proposes microeconometric methods to evaluate the impact of these training programmes on the participants' situation in the labour market. The contributions concern the issue of how such evaluations could and should be performed as well as the effectiveness of the actual programmes. The empirical findings suggest that the public sector sponsored training programmes were fairly ineffective for the individual participants with respect to their unemployment risk as well as other labour market indicators. In contrast, the training organized and paid by the enterprises caused considerable earnings growth. 1998. X, 204 pp. 73 figs., 29 tabs. (Studies in Contemporary Economics) Softcover DM 85,-; öS 620,50; sFr 77,50 ISBN 3-7908-1091-6 Order via e-mail: orders@springer.de or print out the following form and mail or fax to: Physica-Verlag c/o Springer-Verlag GmbH & Co. KG Postfach 31 13 40 D-10643 Berlin GERMANY Fax: ++49(0)30 / 82787-301
Not available in German. The aim of the paper is to provide the practitioner with easily implementable procedures, both numerical and graphical, to test the specification of the dichotomous, linear-in-coefficents logit model. We discuss the performance of these asymptotic methods in small samples and under various misspecifications on the basis of Monte-Carlo simulations and apply them to a cross-section study of female labor supply in West Germany.
(doi:10.1007/BF01193490)
Not available in German. Recently several studies analysed active labour market policies using a newly proposed matching estimator for multiple programmes. Since there is only very limited practical experience with this estimator, this paper checks its sensitivity with respect to issues that are of practical importance. The estimator turns out to be fairly robust to several matters that concern its implementation. Furthermore, the paper demonstrates that the matching approach per se is no magic bullet solving all problems of evaluation studies, but that its success depends critically on the information available in the sample. Finally, a comparison with a bootstrap distribution provides some justification for using a simplified approximation of the distribution of the estimator that ignores its sequential nature.
Not available in German. Generalized score and Wald tests are proposed to test certain aspects of the specification of panel probit models. The procedures used to detect misspecifications, such as omitted and superfluous variables, heteroscedasticity, non-normality and random coefficient variations, are designed specifically for the case when the model is estimated by sequential methods. The finite sample distributions of the test statistics in correctly and incorrectly specified models are investigated by a Monte Carlo study. Applying the proposed procedures to the analysis of the determinants of self-employment in Germany shows that the suggested specification checks are useful instruments in applied work.
This paper proposes to use dynamic treatment models to analyze the effects of fertility on labor market interactions. It argues that when large data sets are available the dynamic potential outcome model is an interesting modeling framework because it allows the careful consideration of the selection issues coming from the interaction of fertility and labor market decisions at different ages. It allows explicitly considering their dependence on the labor market and fertility history realized up to that period. There is no need to collapse the ‘endogeneity' problem into a static setting since the dynamic nature and timing of the interaction can be explicitly addressed. Furthermore, the paper argues that this approach allows defining relevant parameters of interest in a more precise way. Based on artificial data, the approach is implemented and issues that may come up in practical applications of this approach are discussed. (doi: 10.1007/978-1-4020-9967-0 <A href='www.pickitisbn:978-1-4020-9967-0'><img style='border: 0px none' src='https://www.citavi.com/softlink?linkid=findit' title='Titel anhand dieser ISBN in Citavi-Projekt übernehmen'/></A>_3)
This article reviews inverse selection probability weighting to estimate dynamic causal effects. A distribution theory based on sequential generalized method of moments estimation is proposed and the method is applied to a reevaluation of some parts of the Swiss active labor market policy to obtain new results and discuss several issues about the implementation of the estimation procedure.
Granger and Sims non-causality (GSNC) are compared to non-causality based on concepts popular in the microeconometrics and programme evaluation literature (potential outcome non-causality, PONC). GSNC is defined as a set of restrictions on joint distributions of random variables with observable sample counterparts, whereas PONC combines restrictions on partially unobservable variables (potential outcomes) with different identifying assumptions that relate potential to observable outcomes. Based on a dynamic model of potential outcomes, we find that in general neither of the concepts implies each other without further assumptions. However, identifying assumptions of the sequential selection on observable type provide the link between those concepts, such that GSNC implies PONC, and vice versa.
