We present a definition of cause and effect in terms of decision-theoretic primitives and thereby provide a principled foundation for causal reasoning. Our definition departs from the traditional view of causation in that causal assertions may vary with the set of decisions available. We argue that this approach provides added clarity to the notion of cause. Also in this paper, we examine the encoding of causal relationships in directed acyclic graphs. We describe a special class of influence diagrams, those in canonical form, and show its relationship to Pearl's representation of cause and effect. Finally, we show how canonical form facilitates counterfactual reasoning. 1. Introduction Knowledge of cause and effect is crucial for modeling the affects of actions. For example, if we observe a statistical correlation between smoking and lung cancer, we can not conclude from this observation alone that our chances of getting lung cancer will change if we stop smoking. If, however, we als...

This paper builds on the structural equations, treatment effect, and machine learning literatures to provide a causal framework that permits the identification and estimation of causal effects from observational studies. We begin by providing a causal interpretation for standard exogenous regressors and standard “valid” and “relevant” instrumental variables. We then build on this interpretation to characterize extended instrumental variables (EIV) methods, that is methods that make use of variables that need not be valid instruments in the standard sense, but that are nevertheless instrumental in the recovery of causal effects of interest. After examining special cases of single and double EIV methods, we provide necessary and sufficient conditions for the identification of causal effects by means of EIV and provide consistent and asymptotically normal estimators for the effects of interest.

you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at

incumbency advantage that have become widely used in the literature. Recently, however, critics have argued that these measures, as well as less sophisticated measures of incumbency advantage that preceded them, su®er from a failure to consider strategic retirements, and this may lead to as much as a 100in state executive and legislative elections as a natural experiment to correct for strategic retirment. We ¯nd that as an empirical matter strategic retirement is not substantively important. Estimates of incumbency advantages that take account of strategic retirement actually are marginally larger than those that do not. Incumbents ' vote margins in all U.S. state and federal elections have risen dramatically over the last ¯fty years. The estimated incumbency advantage has risen to fully 10 percentage points in the 1980s and 1990s, up from only 1-2 percentage points in the 1940s and 1950s. 1 One line of scholarship argues that the level and increase in the incumbency advantage are much smaller than they appear. All estimates of the incumbency advantage contrast vote margins in incumbent contested seats and vote margins in open seats. Politicians ' strategic career decisions can produce bias in these statistical estimates. Incumbents choose to retire when their electoral prospects look especially dim and staying put when they are assured of reelection. 2 Bias in conventional estimates of the incumbency advantage may result because

A key concern in policy debates over youth smoking is whether preventing children from smoking will stop them from smoking as adults or merely defer initiation into smoking. This paper estimates smoking status in late adolescence viewing smoking at age 14 as an endogenous "treatment" on subsequent smoking, an approach which disentangles causation from unobserved heterogeneity and allows the model to capture the theoretical prediction that addictiveness varies across individuals. Exploiting large tax changes...

This paper uses the 2000 Census 1-in-6 sample to look at the long-term impact of Vietnam-era military service. Instrumental Variables estimates using draft-lottery instruments show post-service earnings losses close to zero in 2000, in contrast with earlier results showing substantial earnings losses for white veterans in the 1970s and 1980s. The estimates also point to a marked increase in schooling that appears to be attributable to the Vietnam-era GI Bill. The net wage e¤ects observed in the 2000 data can be explained by a ‡attening of the experience pro…le in middle age and a modest return to the increased schooling generated by the GI Bill. Evidence on disability e¤ects is mixed but seems inconsistent with a long-term e¤ect of military service on health. This study was conducted while the authors were Special Sworn Status researchers of the U.S. Census Bureau at the Boston Research Data Center. Research results and conclusions expressed are those of the authors and do not necessarily re‡ect the views of the Census Bureau. This paper has been screened to insure that no con…dential data are revealed. Special thanks got to B.K. Atrostic, Jim Davis, and Brian Holly for help with the data used in this study. Thanks also go to Brigham Frandsen and Simone

Abstract This paper demonstrates the extensive scope of an alternative to standard instrumental variables methods, namely covariate-based methods, for identifying and estimating e¤ects of interest in general structural systems. As we show, commonly used econometric methods, speci…cally parametric, semi-parametric, and nonparametric extremum or moment-based methods, can all exploit covariates to estimate well-identi…ed structural e¤ects. The systems we consider are general, in that they need not impose linearity, separability, or monotonicity restrictions on the structural relations. We consider e¤ects of multiple causes; these may be binary, categorical, or continuous. For continuous causes, we examine both marginal and non-marginal e¤ects. We analyze e¤ects on aspects of the response distribution generally, de…ned by explicit or implicit moments or as optimizers (e.g., quantiles). Key for identi…cation is a speci…c conditional exogeneity relation. We examine what happens in its absence and …nd that identi…cation generally fails. Nevertheless, local and near identi…cation results hold in its absence, as we show.

Abstract not found

This paper proposes empirical likelihood based inference methods for causal e¤ects identi…ed from regression discontinuity designs. We consider both the sharp and fuzzy regression discontinuity designs and treat the regression functions as nonparametric. The proposed inference procedures do not require asymptotic variance estimation and the con…dence sets have natural shapes, unlike the conventional Wald-type method. These features are illustrated by simulations and an empirical example which evaluates the e¤ect of class size on pupils’scholastic achievements. Bandwidth selection methods, higher-order properties, and extensions to incorporate additional covariates and parametric functional forms are also discussed.

In this paper we examine the e¤ects of the abolition of the compulsory conscription in France on the demand for education and labour market outcomes. The reform took place in 1997 and a¤ected all men born after 1979. Before the reform, staying on in education was a way to defer the national service and increase the probability of exemption. After the reform, these speci…c incentives to stay on in education have disappeared and the relative cost of education for men has plausibly increased. As a matter of fact, our data reveal that the reform has been followed by a signi…cant decrease in the number of years spent at school by male students, as well as in the proportion of male degree holders. In contrast, the reform had no signi…cant e¤ect on the demand for education for women. We use this exogenous variation in men’s relative demand for education to provide IV estimates of the returns to education. These estimates are larger than standard OLS estimates.