This paper extends the identification results in Nevo and Rosen (2012) to nonparametric models. We derive nonparametric bounds on the average treatment effect when an imperfect instrument is available. As in Nevo and Rosen (2012), we assume that the correlation between the imperfect instrument and the unobserved latent variables has the same sign as the correlation between the endogenous variable and the latent variables. We show that the monotone treatment selection and monotone instrumental variable restrictions, introduced by Manski and Pepper (2000; 2009), jointly imply this assumption. Moreover, we show how the monotone treatment response assumption can help tighten the bounds. The identified set can be written in the form of intersection bounds, which is more conducive to inference. We illustrate our methodology using the National Longitudinal Survey of Young Men data to estimate returns to schooling.

A published version is available here.

Since the introduction of the Tobit framework to perform estimation involving censored dependent variables, practitioners have been facing a clear trade-off between flexibility and theoretical plausibility in modelling consumers’ preferences in the presence of zero consumptions; the Kuhn-Tucker (or virtual price) approach is rigorously based on the economic choice theory but cannot be applied to complex and flexible demand systems, whereas the Tobit-based approach can be applied to any class of demand systems but is deficient in the theoretical foundations on the underlying preferences behind the observed choices. Hence, we assess the performance of three Tobit-based approaches (simple, correlated, and Amemiya-Tobin) and explore the extent of possible biases in elasticity estimates to provide reasonable criteria for model selection. Our analysis concludes that theoretical restrictions implied by the choice theory are essential to the Tobit model and improve its ability to capture the true underlying elasticities and mitigate overrejections. However, the performance of the Tobit models gradually deteriorates as the number of zero consumptions increases; the average rejection rate against the true elasticity values increases substantially as we have more zero consumptions. We illustrate the performance differences among the three Tobit models by applying them to the estimation of demand for fruits and vegetables.

A published version is available here.

The difference-in-differences (DID) method identifies the average treatment effects on the treated (ATT) under mainly the so-called parallel trends (PT) assumption. The most common and widely used approach to justify the PT assumption is the pre-treatment period examination. If a null hypothesis of the same trend in the outcome means for both treatment and control groups in the pre-treatment periods is rejected, researchers believe less in PT and the DID results. This paper fills this gap by developing a generalized DID framework that utilizes all the information available not only from the pre-treatment periods but also from multiple data sources. Our approach interprets PT in a different way using a notion of selection bias, which enables us to generalize the standard DID estimand by defining an information set that may contain multiple pre-treatment periods or other baseline covariates. Our main assumption states that the selection bias in the post-treatment period lies within the convex hull of all selection biases in the pre-treatment periods. We provide a sufficient condition for this assumption to hold. Based on the baseline information set we construct, we first provide an identified set for the ATT that always contains the true ATT under our identifying assumption, and also the standard DID estimand. Secondly, we propose a class of criteria on the selection biases from the perspective of policymakers that can achieve a point identification of the ATT. Finally, we illustrate our methodology through some numerical and empirical examples.

A draft is available here.

This article provides a Stata package for the implementation of the robust difference-in-differences (RDID) method developed in Ban and Kédagni (2023). It contains three main commands: 𝚛𝚍𝚒𝚍, 𝚛𝚍𝚒𝚍_𝚍𝚢, 𝚛𝚍𝚒𝚍𝚜𝚝𝚊𝚐, and we illustrate these commands through simulations and empirical examples.

A draft is available here, and please check out our GitHub page for the installation.

This paper studies identification of the marginal treatment effect (MTE) when a binary treatment variable is misclassified. We show under standard assumptions that the MTE is identified as the derivative of the conditional expectation of the observed outcome given the true propensity score, which is partially identified. We characterize the identified set for this propensity score, and then for the MTE. We use our MTE bounds to derive bounds on other commonly used parameters in the literature. We show that our bounds are tighter than the existing bounds for the local average treatment effect. We illustrate the practical relevance of our derived bounds through some numerical and empirical results.

A draft is available here.

In this paper, we consider identification of causal parameters in the difference-in-differences models in the presence of a binary mediator. Individuals with values 1 or 0 for the mediator are referred to as the "mediated group," and "unmediated group," respectively. Under the no-anticipation effects and some parallel trends assumptions, and monotonicity of the mediator in the treatment, we derive sharp bounds on the controlled direct effects of the treatment for the "always-mediated" and "never-mediated" treated groups, respectively. For the "mediated when-only-treated" treated group, we derive sharp bounds on the total (direct and indirect) effects. To do so, we reformulate the identification question as a mixture problem and use the trimming procedure developed by Horowitz and Manski (1995) and Lee (2009). Our proposed method could also be useful in DiD models in which there is a binary time-varying endogenous covariate. We illustrate the proposed approach through some numerical and empirical examples.  

This paper examines the welfare effects of the United States - Korea Free Trade Agreement (KORUS FTA) on the South Korean consumers. For a more precise analysis, I extend the Almost Ideal (AI) demand model to accommodate household characteristic variables. The demand system is estimated using 14 years of South Korean Household Trend Survey (HTS) data, and the potential endogeneity problem with expenditures in the conditional demand system was solved using an iterative two-stage estimation procedure with the expenditure equation. Based on the theoretically plausible demand system, exact measures of welfare changes are derived, overcoming the disadvantages of using the conventional consumer surplus. The welfare measures are estimated under a counterfactual scenario where all the tariffs have retreated to the pre-FTA level. Under the scenario, the average household would experience a welfare loss equivalent to about a dollar per month, which amounts to 21.7 million dollars per month for all the households in South Korea. Moreover, distributional effects of the FTA are found to be progressive; the trade agreement disproportionately benefits the lower income households more than higher income households.