Consider the simple normal linear regression model for estimation/prediction at a new design point. When the slope parameter is not obviously nonzero, hypothesis testing and information criteria can be used for identifying the right model. We compare the performances of such methods both theoretically and empirically from different perspectives for more insight. The testing approach at the conventional size of 0.05, in spite of being the “standard approach,” performs poorly in estimation. We also found that the frequently told story “the Bayesian information criterion (BIC) is good when the true model is finite-dimensional, and the Akaike information criterion (AIC) is good when the true model is infinite-dimensional” is far from being accurate. In addition, despite some successes in the effort to go beyond the debate between AIC and BIC by adaptive model selection, it turns out that it is not possible to share the pointwise adaptation property of BIC and the minimax-rate adaptation property of AIC by any model selection method. When model selection methods have difficulty in selection, model combining is a better alternative in terms of estimation accuracy.This work was completed when the author was on leave from Iowa State University and was a New Direction Visiting Professor at the Institute for Mathematics and its Applications (IMA) at the University of Minnesota. The fundings from both IMA and ISU are greatly appreciated. The work was also partly supported by NSF CAREER grant DMS0094323. The author thanks Xiaotong Shen and Hannes Leeb for very helpful discussions. The paper also benefited from the questions and comments from the participants at the statistics seminars the author gave at the University of Minnesota and Duke University. The author is very grateful to the anonymous reviewers and the co-editor Benedikt Pötscher for carefully reading earlier versions of the paper, bringing my attention to several closely related previous and current results, and making many very valuable suggestions, which significantly improved the paper in both content and presentation.

Local linear fitting of nonlinear processes under strong (i.e., α-) mixing conditions has been investigated extensively. However, it is often a difficult step to establish the strong mixing of a nonlinear process composed of several parts such as the popular combination of autoregressive moving average (ARMA) and generalized autoregressive conditionally heteroskedastic (GARCH) models. In this paper we develop an asymptotic theory of local linear fitting for near epoch dependent (NED) processes. We establish the pointwise asymptotic normality of the local linear kernel estimators under some restrictions on the amount of dependence. Simulations and application examples illustrate that the proposed approach can work quite well for the medium size of economic time series.We thank Yuichi Kitamura and two referees for helpful comments. This research was partially supported by a Leverhulme Trust research grant, the National Natural Science Foundation of China, and the Economic and Social Science Research Council of the UK.

Adaptations of the classical Wiener–Kolmogorov filters are described that enable them to be applied to short nonstationary sequences. Alternative filtering methods that operate in the time domain and the frequency domain are described. The frequency-domain methods have the advantage of allowing components of the data to be separated along sharp dividing lines in the frequency domain, without incurring any leakage. The paper contains a novel treatment of the start-up problem that affects the filtering of trended data sequences.

In nonstationary heterogeneous panels where the number of units is finite and where each unit cointegrates, a large number of conditions needs to be satisfied for cointegration to be preserved in the aggregate relationship. In reality, the conditions most likely will not hold. This paper takes a closer look at what happens when the conditions are violated. In this case, the question of whether an aggregate relationship is observationally equivalent to a cointegrating equation is of particular interest. We derive a measure of the degree of noncointegration of the aggregate estimates, and we explore its asymptotic properties.We thank Stephen Hall for inspiring discussions and comments during the initial stage of this research. We are grateful to participants in the 9th Conference on Panel Data (Geneva, June 22–23, 2000) and the 2000 Meeting of the Society for Economic Dynamics (San José, Costa Rica, June 29–July 2, 2000) and to colleagues at the University of Bergamo (Italy), London Business School (UK), and Cass Business School (UK) for useful comments. We have enormously benefited from comments and suggestions from two anonymous referees, the editor, Peter C.B. Phillips, and the co-editor, Bruce Hansen. The usual disclaimer applies. G. Urga acknowledges ESRC (Project R022251032) for financial support. Lorenzo Trapani acknowledges financial support from Cass Business School (RAE Development Funds Scheme) and ESRC Postdoctoral Fellowship Scheme (A.N. PTA-026-27-1107).

Dynamic economic theories usually have implications on and only on the conditional mean dynamics of economic processes. Using a generalized spectral derivative approach, Hong and Lee (2005, Review of Economic Studies 72, 499–541) recently proposed a new class of omnibus nonparametric specification tests for linear and nonlinear time series conditional mean models, where the dimension of the conditioning information set may be infinite. The tests can detect a wide range of model misspecifications in mean while being robust to conditional heteroskedasticity and time-varying higher order moments of unknown form. They enjoy an asymptotic “nuisance parameter–free” property in the sense that parameter estimation uncertainty has no impact on the asymptotic N(0,1) distribution of the test statistics. As a result, only the estimated residuals from the null parametric model are needed to implement the tests, and no specific estimation is required.

Although parameter estimation uncertainty has no impact on the asymptotic distribution of the tests, it may have significant impact on the finite-sample distribution, and such an impact may become more substantial as the number of estimated parameters increases. In this paper, we adopt the Wooldridge (1990, Econometric Theory 6, 17–43) device for parametric m-tests to the Hong and Lee (2005) nonparametric tests to reduce the impact of parameter estimation uncertainty. Asymptotic size and power properties of the modified tests are investigated, and simulation studies show that the modified tests generally have better sizes in finite samples and are robust to parameter estimation uncertainty. In the meantime, the size improvement does not cause loss of power against a wide range of alternatives when using the empirical critical values for the tests. These results suggest that the modified generalized spectral derivative tests can be a useful tool in time series conditional mean modeling.We thank Bruce Hansen (the co-editor) and three referees for insightful and helpful comments and suggestions. We also are grateful for financial support from the National Science Foundation Grant SES–0111769 and the Cheung Kong Visiting Scholarship by the Chinese Ministry of Education and Xiamen University, China.

Much of the credit for the outpouring of research on nonlinear models in the 1970s—particularly limited dependent variable models—should go to Takeshi Amemiya. His classic 1973 Econometrica article on estimation of the parameters of the Tobit model [11] set a new standard for mathematical rigor in theoretical econometrics. During his career, which spans nearly four decades, Takeshi's research contributions to econometrics have touched on most of the central models for empirical economics, including linear and nonlinear simultaneous equations models, distributed lag models, heteroskedastic and random coefficient models, qualitative response models, censored and truncated regression and selection models, transformed regression models, choice-based sampling models, and duration models. His 1985 text, Advanced Econometrics [44], became the standard reference for second-year graduate microeconometrics courses at the leading graduate programs in economics.

Takeshi's research accomplishments have been accompanied by numerous professional honors. He is a Fellow of the Econometric Society, the American Statistical Association, and the American Academy of Arts and Sciences and was awarded Guggenheim and Humboldt Fellowships. Takeshi also has a distinguished record of professional service; he served briefly as a co-editor of Econometrica and has been a co-editor of the Journal of Econometrics since 1982.

This interview was conducted in Takeshi Amemiya's office at Stanford University in March 2004. In recent years his interests have turned to comparative study of Greek and Japanese culture and mythology—his scholarship includes a Japanese translation of an English research monograph on Aristotle's Ethics [58]—and the discussion starts with this topic before turning to his unusual education, career as an econometric theorist, and variety of outside interests. His remarks reveal both his careful attention to detail and the dry humor that his colleagues and former students (like me) savor.

This note shows that the orthogonal shocks obtained by imposing a recursive structure on the contemporaneous impacts of the errors in a vector-error correction model (VECM) are related to the orthogonal permanent-transitory shocks in a Gonzalo and Ng (2001, Journal of Economic Dynamics and Control 25, 1527–1546) decomposition provided some of the variables in the system are weakly exogenous. Specifically, when there are as many weakly exogenous variables as common trends and the weakly exogenous variables are ordered first, the orthogonal shocks obtained directly from the VECM constitute a permanent-transitory decomposition, and the permanent shocks are identical to those in the Gonzalo and Ng decomposition.We are grateful to the co-editor Paolo Paruolo and an anonymous referee for useful comments, which have helped to improve the content and exposition of this note. Financial support from the Australian Research Council is gratefully acknowledged.

This note adds to the recent research project on treatment choice under ambiguity. I generalize the Manski (Journal of Econometrics, in press) analysis of minimax regret treatment choice by considering a more general setting and, more importantly, by solving for the treatment rule given finitely many (as opposed to two) treatments. The most interesting finding is that with three or more undominated treatments, the minimax regret treatment rule may assign the same treatment to all subjects; thus, the most salient feature of the two-treatment case does not generalize.I thank Chuck Manski, the co-editor, and especially an anonymous referee for helpful comments. Financial support through the Robert Eisner Memorial Fellowship, Department of Economics, Northwestern University, in addition to a Dissertation Year Fellowship, The Graduate School, Northwestern University, is gratefully acknowledged.