Search

This note considers the estimation of dynamic threshold regression models with fixed effects using short panel data. We examine a two-step method, where the threshold parameter is estimated nonparametrically at the N-rate and the remaining parameters are estimated by GMM at the root N-rate. We provide simulation results that illustrate the potential advantages of the new method in comparison with pure GMM estimation. The simulations also highlight the importance the choice of instruments in GMM estimation.

This paper develops a specification test for functional form for models identified by moment restrictions, including IV and GMM settings. The general framework is one where the moment restrictions are specified as functions of data, a finite-dimensional parameter vector, and a nonparametric real function (an infinite-dimensional parameter vector). The null hypothesis is that the real function is parametric. The test is relatively easy to implement and its asymptotic distribution is known. The test performs well in simulation experiments.

Many developing countries experience famine. If survival is related to height, the increasingly common practice of using height as a measure of well-being may be misleading. We devise a novel method for disentangling the stunting from the selection effects of famine. Using data from the 1959-1961 Great Chinese Famine, we find that taller children were more likely to survive the famine. Controlling for selection, we estimate that children under the age of five who survived the famine grew up to be 1 to 2 cm shorter. Our results suggest that average height is potentially a biased measure of economic conditions during childhood., 基盤研究(S) = Grants-in-Aid for Scientific Research (S)

This paper develops a specification test for functional form for models identifiedby moment restrictions, including IV and GMM settings. The general framework is onewhere the moment restrictions are specified as functions of data, a finite-dimensionalparameter vector, and a nonparametric real function (an infinite-dimensional parametervector). The null hypothesis is that the real function is parametric. The test is relativelyeasy to implement and its asymptotic distribution is known. The test performs well insimulation experiments.

This paper explores estimation of a class of non-linear dynamicpanel data models with additive unobserved individual-specific effects. Themodels are specified by moment restrictions. The class includes the paneldata AR(p) model and panel smooth transition models. We derive an efficientset of moment restrictions for estimation and apply the results to estimationof panel smooth transition models with fixed effects, where the transition maybe determined endogenously. The performance of the GMM estimator, bothin terms of estimation precision and forecasting performance, is examinedin a Monte Carlo experiment. We find that estimation of the parameters inthe transition function can be problematic but that there may be significantbenefits in terms of forecast performance.

The Great Chinese Famine of 1959-1961 is puzzling, since despite the high death rates, there is no discernable diminution in height amongst the majority of cohorts who were exposed to the famine in crucial growth years. An explanation is that shorter children experienced greater mortality and that this selection offset stunting. We disentangle stunting and selection effects of the Chinese famine, using the height of the children of the famine cohort. We find significant stunting of about 2cm for rural females and slightly less for rural males who experienced the famine in the first five years of life. Our results suggest that mortality bias implies that raw height is not always a good measure of economic conditions during childhood