Econometrics Journal最新文献

We develop tests for the presence of deterministic seasonal behaviour and seasonal mean shifts in a seasonally observed univariate time series. These tests are designed to be asymptotically robust to the order of integration of the series at both the zero and seasonal frequencies. Motivated by the approach of Hylleberg, Engle, Granger and Yoo, we base our approach on linear filters of the data that remove any potential unit roots at the frequencies not associated with the deterministic component(s) under test. Test statistics are constructed using the filtered data such that they have well defined limiting null distributions regardless of whether the data are either integrated or stationary at the frequency associated with the deterministic component(s) under test. In the same manner as Vogelsang, Bunzel and Vogelsang and Sayginsoy and Vogelsang, we scale these statistics by a function of an auxiliary seasonal unit root statistic. This allows us to construct tests that are asymptotically robust to the order of integration of the data at both the zero and seasonal frequencies. Monte Carlo evidence suggests that our proposed tests have good finite sample size and power properties. An empirical application to UK gross domestic product indicates the presence of seasonal mean shifts in the data.

A popular approach to factor-augmented panel regressions is the common correlated effects (CCE) estimator of Pesaran (2006). In fact, the approach is so popular that it has given rise to a separate CCE literature. A common assumption in this literature is that the common factors are stationary, which would seem to rule out many empirically relevant cases. Moreover, deterministic factors are typically treated as known, which raises the issue of model misspecification. In the current paper, we show how the conditions placed on the factors in CCE can be made much more general than was previously thought possible. In fact, save for some mild regulatory moment conditions, the factors are essentially unrestricted. One implication of this result is that there is no need to discriminate between deterministic and stochastic factors, but that one can instead treat them all as unknown. This is very convenient for practitioners, because it means that under certain conditions they are spared the problem of having to decide which deterministic terms to include in the model.

In this paper, we consider the semiparametric identification and estimation of a heteroscedastic binary choice model with endogenous dummy regressors and no parametric restriction on the distribution of the error term. Our approach addresses various drawbacks associated with previous estimators proposed for this model. It allows for: general multiplicative heteroscedasticity in both selection and outcome equations; a nonparametric selection mechanism; and multiple discrete endogenous regressors. The resulting three-stage estimator is shown to be asymptotically normal, with a convergence rate that can be arbitrarily close to if certain smoothness assumptions are satisfied. Simulation results show that our estimator performs reasonably well in finite samples. Our approach is then used to study the intergenerational transmission of smoking habits in British households.

Inference based on cluster-robust standard errors in linear regression models, using either the Student's t-distribution or the wild cluster bootstrap, is known to fail when the number of treated clusters is very small. We propose a family of new procedures called the subcluster wild bootstrap, which includes the ordinary wild bootstrap as a limiting case. In the case of pure treatment models, where all observations within clusters are either treated or not, the latter procedure can work remarkably well. The key requirement is that all cluster sizes, regardless of treatment, should be similar. Unfortunately, the analogue of this requirement is not likely to hold for difference-in-differences regressions. Our theoretical results are supported by extensive simulations and an empirical example.

In this paper, we propose a consistent U-statistic test with good sampling properties to detect changes in volatility. We show that the test has a limiting standard normal distribution under the null hypothesis, and that it is powerful compared with various alternatives. A Monte Carlo experiment is conducted to highlight the merits of the proposed test relative to other popular tests for structural changes in volatility. An empirical example is examined to demonstrate the practical application of the proposed testing method.

Gaussian graphical models are recently used in economics to obtain networks of dependence among agents. A widely used estimator is the graphical least absolute shrinkage and selection operator (GLASSO), which amounts to a maximum likelihood estimation regularized using the matrix norm on the precision matrix Ω. The norm is a LASSO penalty that controls for sparsity, or the number of zeros in Ω. We propose a new estimator called structured GLASSO (SGLASSO) that uses the mixed norm. The use of the penalty controls for the structure of the sparsity in Ω. We show that when the network size is fixed, SGLASSO is asymptotically equivalent to an infeasible GLASSO problem which prioritizes the sparsity-recovery of high-degree nodes. Monte Carlo simulation shows that SGLASSO outperforms GLASSO in terms of estimating the overall precision matrix and in terms of estimating the structure of the graphical model. In an empirical illustration using a classic firms' investment data set, we obtain a network of firms' dependence that exhibits the core–periphery structure, with General Motors, General Electric and US Steel forming the core group of firms.

In this paper, we demonstrate that standard central limit theorem (CLT) results do not hold for means of non-parametric, conditional efficiency estimators, and we provide new CLTs that permit applied researchers to make valid inference about mean conditional efficiency or to compare mean efficiency across groups of producers. The new CLTs are used to develop a test of the restrictive ‘separability’ condition that is necessary for second-stage regressions of efficiency estimates on environmental variables. We show that if this condition is violated, not only are second-stage regressions difficult to interpret and perhaps meaningless, but also first-stage, unconditional efficiency estimates are misleading. As such, the test developed here is of fundamental importance to applied researchers using non-parametric methods for efficiency estimation. The test is shown to be consistent and its local power is examined. Our simulation results indicate that our tests perform well both in terms of size and power. We provide a real-world empirical example by re-examining the paper by Aly et al. (1990, Review of Economics and Statistics 72, 211–18) and rejecting the separability assumption implicitly assumed by Aly et al., calling into question results that appear in hundreds of papers that have been published in recent years.

Research on peer effects in sociology has long been focused on social interactions and the associated social influence process. In this paper, we extend a large-network-based game model to a model that allows for the dependence of social interactions on social-influence status. In particular, we use the Katz–Bonacich centrality to measure individuals' social influences, which are obtained directly from the observation of a social network. To solve the computational burden when the data come from the equilibrium of a large network, we extend a nested pseudo-likelihood estimation approach to our large-network-based game model. Using the National Longitudinal Study of Adolescent Health (Add Health) dataset, we investigate the peer effects of dangerous behaviour among high-school students. Our results show that the peer effects are statistically significant and positive. Moreover, students benefit more (statistically significant at the 5% level) from conformity or, equivalently, pay more for disobedience, in terms of peer pressure, if their friends have a higher status of social influence.