We estimate the fraction of disease cases, and the fraction of their total medical expenditures, attributable to smoking for two disease groups: (LC) lung and laryngeal cancer and chronic obstructive pulmonary disease, (CHD) cardiovascular disease, stroke and other smoking-caused cancers. We use a generalized additive model to predict the probability of disease; and a semi-parametric, two-part ...

We show that the treatment effect estimated by standard methods such as regression discontinuity analysis or difference-in-differences may contain a transient “learning effect” is entangled with long-term of treatment. This learning occurs when variable interest agents’ efforts, and control correspond to success failure: failure gives agents information about how much their effort matters, cons...

This paper analyzes the initial filing and the subsequent renewal decisions of the international patent applicants. Specifically I use as data input a comprehensive patent documentation collected from the EUropean Patent Office (EPO) that was unprecedented in its level of disaggregation by various characteristics such as nationality of inventor, type of technology and designated country of pate...

In randomized controlled trials, treatment is often assigned by stratified randomization. I show that among all randomization schemes treat units with probability one half, a certain matched-pair design achieves the maximum statistical precision for estimating average effect. an important special case, optimal pairs according to baseline outcome. simulation study based on datasets from ten this...

This paper shows that many estimators of thresholds in ordered response models exist, because binary choice location estimators can be converted into threshold estimators. A new threshold estimator is proposed that is consistent under more general conditions. An extension to random thresholds is provided. JEL codes: C14, C25, C13.

Information-theoretic alternatives to general method of moments (GMM) use over-identifying moments to estimate the data-generating distribution jointly with the parameters of interest. This paper demonstrates how these estimates can be interpreted when the sample is not a random draw from the population of interest. I make explicit the selection probability implied by the empirical likelihood a...

Simulated maximum likelihood has proved to be a valuable tool for fitting the log-normal stochastic volatility model to financial returns time series. In this paper, we develop a methodology that generalizes these methods to more general stochastic volatility models that are naturally cast in terms of a positive volatility process. The methodology relies on combining two well known methods for ...

We propose serial correlation robust asymptotic confidence bands for the receiver operating characteristic (ROC) curves estimated by quasi-maximum likelihood in the binormal model. Our simulation experiments confirm that this new method performs fairly well in finite samples. The conventional procedure is found to be markedly undersized in terms of yielding empirical coverage probabilities lowe...

Modelling stock return generating process as a single factor model, we show analytically that the relation between idiosyncratic volatility measured as variance of the residuals and expected stock return in the cross-section may be represented by a parabola that opens to the left and has horizontal axis. This relation is uncovered for stocks of similar volatility and no abnormal return. The sen...