Granger and Sims noncasuality (GSNC), a concept frequently applied in time series econometrics, is compared to noncausality based on concepts popular in microeconometrics, program evaluation, and epidemiology literature (potential outcome noncausality or 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 outcome variables to their observable counterparts. Based on the Robins' dynamic model of potential outcomes, we find that in general neither of the concepts implies each other without further (untestable) assumptions. However, the identifying assumptions associated with the sequential selection of the observables link these concepts such that GSNC implies PONC, and vice versa.
Not available in German. This paper addresses microeconometric evaluation by matching methods when the programs under consideration are heterogeneous. Assuming that selection into the different sub-programs and the potential outcomes are independent given observable characteristics, estimators based on dif-ferent propensity scores are compared and applied to the analysis of active labor market policies in the Swiss region of Zurich. Furthermore, the issues of heterogeneous effects and aggregation are addressed. The results suggest that an approach that incorporates the possibility of having multiple programs can be an informative tool in applied work.
This paper compares the inverse-probability-of-selection-weighting estimation principle with the matching principle and derives conditions for weighting and matching to identify the same and the true distribution, respectively. This comparison improves the understanding of the relation of these estimation principles and allows constructing new estimators.
We characterize optimal redistribution in a dynastic economy with observable human capital and hidden ability. We show that the wedge between human capital investment in the laissez faire and the social optimum differs from the wedge for bequests because (i) returns to human capital are risky, and (ii) human capital may change informational rents. We compute the optimal allocation when ability is persistent across generations, as calibrated for the U.S. We show how the allocation can be implemented with student loans featuring contingent repayments. The quantitative results reveal that human capital investment should (i) increase in parental income because of ability transmission across generations, but (ii) decrease in inherited assets because of the negative effect of wealth on labor supply.
This article compares the inverse-probability-of-selection-weighting estimation principle with the matching principle and derives conditions for weighting and matching to identify the same and the true distribution, respectively. This comparison improves the understanding of the relation of these estimation principles and allows constructing new estimators.
The issue of potentially endogenous control variables in causal studies based on the assumption of no selection bias conditional on observables (conditional independence assumption, CIA) is discussed. The paper shows that the standard formulation of the CIA obscures the endogeneity problem. It suggests a CIA based on potential variables together with explicit exogeneity conditions, which allows a separate assessment of the endogeneity bias and the plausibility of the CIA. (doi:10.1016/j.spl.2007.05.019)
Not available in German. This paper explores the potential of an approach suggested by Manski of obtaining nonparametric bounds for treatment effects in evaluation studies without knowledge of the participation process. The practical concern is the effects of continuous vocational training in East Germany. The empirical application is based on a large cross-section that covers about 0.6% of the total population in 1993. The results are rather mixed. The large width of the intervals obtained emphasise the fundamental problem of all evaluation studies without good knowledge of the relationship between potential outcomes and the participation process. However, in some cases suitable exclusion restrictions are indeed capable of bounding the treatment effects strictly away from zero.
Download Supplement: (pdf, 977 kb)
