
on Econometrics 
By:  Anastasia Semykina (Department of Economics, Florida State University); Jeffrey M. Woodridge (Department of Economics, Michigan State University) 
Abstract:  We consider estimation of panel data models with sample selection when the equation of interest contains endogenous explanatory variables as well as unobserved heterogeneity. We offer a detailed analysis of the pooled twostage least squares (pooled 2SLS) and fixed effects2SLS (FE2SLS) estimators and discuss complications in correcting for selection biases that arise when instruments are correlated with the unobserved effect. Assuming that appropriate instruments are available, we propose several tests for selection bias and two estimation procedures that correct for selection in the presence of endogenous regressors. The first correction procedure is valid under the assumption that the errors in the selection equation are normally distributed, while the second procedure drops the normality assumption and estimates the model parameters semiparametrically. In the proposed testing and correction procedures, the error terms may be heterogeneously distributed and serially dependent in both selection and primary equations. Correlation between the unobserved effects and explanatory and instrumental variables is permitted. To illustrate and study the performance of the proposed methods, we apply them to estimating earnings equations for females using the Panel Study of Income Dynamics data and perform Monte Carlo simulations. 
Keywords:  Fixed Effects, Instrumental Variables, Sample Selection, Mills Ratio, Semiparametric 
JEL:  C23 C24 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:fsu:wpaper:wp2010_10_01&r=ecm 
By:  ERIC HILLEBRAND (DEPARTMENT OF ECONOMICS, LOUISIANA STATE UNIVERSITY,); MArcelo Cunha Medeiros (DEPARTMENT OF Economics, PUCrio Author mcm@econ.pucrio.br) 
Abstract:  We study the simultaneous occurrence of long memory and nonlinear effects, such as structural breaks and thresholds, in autoregressive moving average (ARMA) time series models and apply our modeling framework to series of daily realized volatility. Asymptotic theory for the quasimaximum likelihood estimator is developed and a sequence of model specification tests is described. Our framework allows for general nonlinear functions, including smoothly changing intercepts. The theoretical results in the paper can be applied to any series with long memory and nonlinearity. We apply the methodology to realized volatility of individual stocks of the Dow Jones Industrial Average during the period 1995 to 2005. We find strong evidence of nonlinear effects and explore different specifications of the model framework. A forecasting exercise demonstrates that allowing for nonlinearities in long memory models yields significant performance gains. 
Keywords:  Realized volatility, structural breaks, smooth transitions, nonlinear models, long memory, persistence. 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:rio:texdis:578&r=ecm 
By:  Qing Liu; David Pitt; Xibin Zhang; Xueyuan Wu 
Abstract:  In this paper, we present a Markov chain Monte Carlo (MCMC) simulation algorithm for estimating parameters in the kernel density estimation of bivariate insurance claim data via transformations. Our data set consists of two types of auto insurance claim costs and exhibit a highlevel of skewness in the marginal empirical distributions. Therefore, the kernel density estimator based on original data does not perform well. However, the density of the original data can be estimated through estimating the density of the transformed data using kernels. It is well known that the performance of a kernel density estimator is mainly determined by the bandwidth, and only in a minor way by the kernel choice. In the current literature, there have been some developments in the area of estimating densities based on transformed data, but bandwidth selection depends on predetermined transformation parameters. Moreover, in the bivariate situation, each dimension is considered separately and the correlation between the two dimensions is largely ignored. We extend the Bayesian sampling algorithm proposed by Zhang, King and Hyndman (2006) and present a MetropolisHastings sampling procedure to sample the bandwidth and transformation parameters from their posterior density. Our contribution is to estimate the bandwidths and transformation parameters within a MetropolisHastings sampling procedure. Moreover, we demonstrate that the correlation between the two dimensions is well captured through the bivariate density estimator based on transformed data. 
Keywords:  Bandwidth parameter; kernel density estimator; Markov chain Monte Carlo; MetropolisHastings algorithm; power transformation; transformation parameter. 
JEL:  C14 C15 C63 
Date:  2010 
URL:  http://d.repec.org/n?u=RePEc:msh:ebswps:201018&r=ecm 
By:  André A. Monteiro 
Abstract:  This paper considers the problem of estimating a linear univariate Time Series State Space model for which the shape of the distribution of the observation noise is not specified a priori. Although somewhat challenging computationally, the simultaneous estimation of the parameters of the model and the unknown observation noise density is made feasible through a combination of Gaussiansum Filtering and Smoothing algorithms and Kernel Density Estimation methods. The bottleneck in these calculations consists in avoiding the geometric increase, with time, of the number of simultaneous Kalman filter components. It is the aim of this paper to show that this can be achieved by the use of standard techniques from Cluster Analysis and unsupervised Classification. An empirical illustration of this new methodology is included; this consists in the application of a semiparametric version of the Local Level model to the analysis of the wellknown river Nile data series. 
Keywords:  Clustering, GaussianSum, Kernel methods, Signal extraction, State space models 
JEL:  C13 C14 C22 
Date:  2010–09 
URL:  http://d.repec.org/n?u=RePEc:cte:wsrepe:ws103418&r=ecm 
By:  Li, Feng (Department of Statistics); Villani, Mattias (Research Department, Central Bank of Sweden); Kohn, Robert (The University of New South Wales) 
Abstract:  Smooth mixtures, i.e. mixture models with covariatedependent mixing weights, are very useful flexible models for conditional densities. Previous work shows that using too simple mixture components for modeling heteroscedastic and/or heavy tailed data can give a poor fit, even with a large number of components. This paper explores how well a smooth mixture of symmetric components can capture skewed data. Simulations and applications on real data show that including covariatedependent skewness in the components can lead to substantially improved performance on skewed data, often using a much smaller number of components. Furthermore, variable selection is effective in removing unnecessary covariates in the skewness, which means that there is little loss in allowing for skewness in the components when the data are actually symmetric. We also introduce smooth mixtures of gamma and lognormal components to model positivelyvalued response variables. 
Keywords:  Bayesian inference; Markov chain Monte Carlo; Mixture of Experts; Variable selection 
JEL:  C11 C14 C51 
Date:  2010–08–01 
URL:  http://d.repec.org/n?u=RePEc:hhs:rbnkwp:0245&r=ecm 
By:  Søren Johansen (Department of Economics, University of Copenhagen) 
Abstract:  This paper contains an overview of some recent results on the statistical analysis of cofractional processes, see Johansen and Nielsen (2010b). We first give an brief summary of the analysis of cointegration in the vector autoregressive model and then show how this can be extended to fractional processes. The model allows the process X_{t} to be fractional of order d and cofractional of order db≥0; that is, there exist vectors β for which β′X_{t} is fractional of order db. We analyse the Gaussian likelihood function to derive estimators and test statistics. The asymptotic properties are derived without the Gaussian assumption, under suitable moment conditions. We assume that the initial values are bounded and show that they do not influence the asymptotic analysis. The estimator of β is asymptotically mixed Gaussian and estimators of the remaining parameters are asymptotically Gaussian. The asymptotic distribution of the likelihood ratio test for cointegration rank is a functional of fractional Brownian motion. 
Keywords:  cofractional processes; cointegration rank; fractional cointegration; likelihood inference; vector autoregressive model 
JEL:  C32 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:kud:kuiedp:1028&r=ecm 
By:  Martin Huber; Michael Lechner; Conny Wunsch 
Abstract:  We investigate the finite sample properties of a large number of estimators for the average treatment effect on the treated that are suitable when adjustment for observable covariates is required, like inverse pro¬bability weighting, kernel and other variants of matching, as well as different parametric models. The simulation design used is based on real data usually employed for the evaluation of labour market programmes in Germany. We vary several dimensions of the design that are of practical importance, like sample size, the type of the outcome variable, and aspects of the selection process. We find that trimming individual observations with too much weight as well as the choice of tuning parameters is important for all estimators. The key conclusion from our simulations is that a particular radius matching estimator combined with regression performs best overall, in particular when robustness to misspecifications of the propensity score is considered an important property. 
Keywords:  Propensity score matching, kernel matching, inverse probability weighting, selection on observables, empirical Monte Carlo study, finite sample properties 
JEL:  C21 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:usg:dp2010:201030&r=ecm 
By:  Søren Johansen (Department of Economics, University of Copenhagen) 
Abstract:  There are simple wellknown conditions for the validity of regression and correlation as statistical tools. We analyse by examples the effect of nonstationarity on inference using these methods and compare them to model based inference. Finally we analyse some data on annual mean temperature and sea level, by applying the cointegrated vector autoregressive model, which explicitly takes into account the nonstationarity of the variables. 
Keywords:  regression correlation cointegration; model based inference; likelihood inference; annual mean temperature; sea level 
JEL:  C32 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:kud:kuiedp:1027&r=ecm 
By:  David Pitt (Dept. Economics, University of Melbourne); Montserrat Guillén (Dept. Econometrics, University of Barcelona) 
Abstract:  We present a real data set of claims amounts where costs related to damage are recorded separately from those related to medical expenses. Only claims with positive costs are considered here. Two approaches to density estimation are presented: a classical parametric and a semiparametric method, based on transformation kernel density estimation. We explore the data set with standard univariate methods. We also propose ways to select the bandwidth and transformation parameters in the univariate case based on Bayesian methods. We indicate how to compare the results of alternative methods both looking at the shape of the overall density domain and exploring the density estimates in the right tail. 
Date:  2010–03 
URL:  http://d.repec.org/n?u=RePEc:xrp:wpaper:xreap201003&r=ecm 
By:  Andersson, Michael K. (Monetary Policy Department, Central Bank of Sweden); Palmqvist, Stefan (Monetary Policy Department, Central Bank of Sweden); Waggoner, Daniel F. (Research Department) 
Abstract:  When generating conditional forecasts in dynamic models it is common to impose the conditions as restrictions on future structural shocks. However, these conditional forecasts often ignore that there may be uncertainty about the future development of the restricted variables. Our paper therefore proposes a generalization such that the conditions can be given as the full distribution of the restricted variables. We demonstrate, in two empirical applications, that ignoring the uncertainty about the conditions implies that the distributions of the unrestricted variables are too narrow. 
Keywords:  Central Bank; Market Expectation; Restrictions; Uncertainty 
JEL:  C53 E37 E52 
Date:  2010–09–01 
URL:  http://d.repec.org/n?u=RePEc:hhs:rbnkwp:0247&r=ecm 
By:  Lluís Bermúdez (Departament de Matemµatica Econµomica, Financera i Actuarial, Universitat de Barcelona); Dimitris Karlis (Athens University of Economics and Business) 
Abstract:  When actuaries face with the problem of pricing an insurance contract that contains different types of coverage, such as a motor insurance or homeowner's insurance policy, they usually assume that types of claim are independent. However, this assumption may not be realistic: several studies have shown that there is a positive correlation between types of claim. Here we introduce di®erent multivariate Poisson regression models in order to relax the independence assumption, including zeroin°ated models to account for excess of zeros and overdispersion. These models have been largely ignored to date, mainly because of their computational di±culties. Bayesian inference based on MCMC helps to solve this problem (and also lets us derive, for several quantities of interest, posterior summaries to account for uncertainty). Finally, these models are applied to an automobile insurance claims database with three different types of claims. We analyse the consequences for pure and loaded premiums when the independence assumption is relaxed by using different multivariate Poisson regression models and their zeroinflated versions. 
Keywords:  Multivariate Poisson regression models, Zeroinflated models, Automobile insurance, MCMC inference, Gibbs sampling 
JEL:  C51 
Date:  2010–04 
URL:  http://d.repec.org/n?u=RePEc:xrp:wpaper:xreap201004&r=ecm 
By:  Amélie Charles (Audencia Nantes, School of Management  Audencia, School of Management); Olivier Darné (LEMNA  Laboratoire d'économie et de management de Nantes Atlantique  Université de Nantes : EA4272); Fabien Tripier (LEMNA  Laboratoire d'économie et de management de Nantes Atlantique  Université de Nantes : EA4272) 
Abstract:  This article compares the performances of some nonstationarity tests on simulated series, using the businesscycle model of Chang et al. (2007) [Y. Chang, T. Doh, F. Schorfheide, (2007). Nonstationary Hours in a DSGE Model. Journal of Money, Credit and Banking 39, 3571373] as data generating process. Overall, Monte Carlo simulations show that the efficient unit root tests of Ng and Perron (2001) [Ng, S., Perron, P. (2001). Lag length selection and the construction of unit root tests with good size and power. Econometrica 69, 15191554] are more powerful than the standard nonstationarity tests (ADF and KPSS). More precisely, these efficient tests are able to reject frequently the unitroot hypothesis on simulated series using the best specification of businesscycle model found by Chang et al. (2007), in which hours worked are stationary with adjustment costs. 
Keywords:  unit root rest; DSGE models; hours worked 
Date:  2010–10–18 
URL:  http://d.repec.org/n?u=RePEc:hal:wpaper:hal00527122_v1&r=ecm 
By:  Timo Mitze 
Abstract:  The contribution of this paper is to analyse the role of network interdependencies in a dynamic panel data model for German internal migration fl ows since reunification. So far, a capacious account of spatial patterns in German migration data is still missing in the empirical literature. In the context of this paper, network dependencies are associated with correlations of migration flows strictly attributable to proximate flows in geographic space. Using the neoclassical migration model, we start from its aspatial specification and show by means of residual testing that network dependency eff ects are highly present. We then construct spatial weighting matrices for our system of interregional flow data and apply spatial regression techniques to properly handle the underlying spacetime interrelations. Besides spatial extensions to the BlundellBond (1998) system GMM estimator in form of the commonly known spatial lag and unconstrained spatial Durbin model, we also apply system GMM to spatially filtered variables. Finally, combining both approaches to a mixed spatial filteringregression specification shows a remarkably good performance in terms of capturing spatial dependence in our migration equation and at the same time qualify the model to pass essential IV diagnostic tests. The basic message for future research is that spacetime dynamics is highly relevant for modelling German internal migration flows. 
Keywords:  Internal migration, dynamic panel data; Spatial Durbin Model; GMM 
JEL:  R23 C31 C33 
Date:  2010–09 
URL:  http://d.repec.org/n?u=RePEc:rwi:repape:0205&r=ecm 
By:  Chanont Banternghansa; Michael W. McCracken 
Abstract:  This paper presents empirical evidence on the efficacy of forecast averaging using the ALFRED realtime database. We consider averages taken over a variety of different bivariate VAR models that are distinguished from one another based upon at least one of the following: which variables are used as predictors, the number of lags, using all available data or data after the Great Moderation, the observation window used to estimate the model parameters and construct averaging weights, and for forecast horizons greater than one, whether or not iterated or directmultistep methods are used. A variety of averaging methods are considered. Our results indicate that the benefits to model averaging relative to BICbased model selection are highly dependent upon the class of models being averaged over. We provide a novel decomposition of the forecast improvements that allows us to determine which types of averaging methods and models were most (and least) useful in the averaging process. 
Keywords:  Economic forecasting ; Realtime data 
Date:  2010 
URL:  http://d.repec.org/n?u=RePEc:fip:fedlwp:2010033&r=ecm 
By:  Yaakov Malinovsky; Yosef Rinott 
Abstract:  We study estimation of finite population quantiles, with emphasis on estimators that are invariant under monotone transformations of the data, and suitable invariant loss functions. We discuss nonrandomized and randomized estimators, best invariant and minimax estimators and sampling strategies relative to different classes. The combination of natural invariance of the kind discussed here, and finite population sampling appears to be novel, and leads to interesting statistical and combinatorial aspects. 
Date:  2010–05 
URL:  http://d.repec.org/n?u=RePEc:huj:dispap:dp553&r=ecm 
By:  Larry Goldstein; Yosef Rinott; Marco Scarsini 
Abstract:  We compare estimators of the (essential) supremum and the integral of a function f defined on a measurable space when f may be observed at a sample of points in its domain, possibly with error. The estimators compared vary in their levels of stratification of the domain, with the result that more refined stratification is better with respect to different criteria. The emphasis is on criteria related to stochastic orders. For example, rather than compare estimators of the integral of f by their variances (for unbiased estimators), or mean square error, we attempt the stronger comparison of convex order when possible. For the supremum the criterion is based on the stochastic order of estimators. For some of the results no regularity assumptions for f are needed, while for others we assume that f is monotone on an appropriate domain. 
Date:  2010–07 
URL:  http://d.repec.org/n?u=RePEc:huj:dispap:dp556&r=ecm 
By:  David Azriel; Micha Mandel; Yosef Rinott 
Abstract:  Phase I clinical trials are conducted in order to find the maximum tolerated dose (MTD) of a given drug from a finite set of doses. For ethical reasons, these studies are usually sequential, treating patients or group of patients with the best available dose according to the current knowledge. However, it is proved here that such designs, and, more generally, designs that concentrate on one dose from some time on, cannot provide consistent estimators for the MTD unless very strong parametric assumptions hold. We describe a family of sequential designs that treat individuals with one of the two closest doses to the estimated MTD, and prove that such designs, under general conditions, concentrate eventually on the two closest doses to the MTD and estimate the MTD consistently. It is shown that this family contains randomized designs that assign the MTD with probability that approaches 1 as the size of the experiment goes to infinity. We compare several designs by simulations, studying their performances in terms of correct estimation of the MTD and the proportion of individuals treated with the MTD. 
Date:  2010–09 
URL:  http://d.repec.org/n?u=RePEc:huj:dispap:dp559&r=ecm 
By:  MoralBenito, Enrique 
Abstract:  Fragility of regression analysis to arbitrary assumptions and decisions about choice of control variables is an important concern for applied econometricians (e.g. Leamer (1983)). Sensitivity analysis in the form of model averaging represents an (agnostic) approach that formally addresses this problem of model uncertainty. This paper presents an overview of model averaging methods with emphasis on recent developments in the combination of model averaging with IV and panel data settings. 
Keywords:  Model uncertainty; model averaging 
JEL:  C5 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:pra:mprapa:26047&r=ecm 
By:  Stephen B. Billings (University of North CarolinaCharlotte); Erik B. Johnson (Quinnipiac University) 
Abstract:  We introduce a nonparametric microdata based test for industrial specialization and apply it to a single urban area. Our test employs establishment densities for specific industries, a population counterfactual, and a new correction for multiple hypothesis testing to determine the statistical significance of specialization across both places and industries. Results highlight patterns of specialization which are extremely varied, with downtown places specializing in a number of service sector industries, while more suburban places specialize in both manufacturing and service industries. Business service industries are subject to more specialization than nonbusiness service industries while the manufacturing sector contains the lowest representation of industries with specialized places. Finally, we compare the results of our test for specialization with recent tests of localization and show how these two classes of measures highlight the presence of both industry as well as place specific agglomerative forces. 
Keywords:  Industrial specialization 
JEL:  R12 
Date:  2010 
URL:  http://d.repec.org/n?u=RePEc:ieb:wpaper:2010/10/doc201040&r=ecm 
By:  Christiaensen, Luc; Lanjouw, Peter; Luoto, Jill; Stifel, David 
Abstract:  Tracking poverty is predicated on the availability of comparable consumption data and reliable price deflators. However, regular series of strictly comparable data are only rarely available. Poverty prediction methods that track consumption correlates as opposed to consumption itself have been developed to overcome such data gaps. These methods typically assume that the estimated relation between consumption and its predictors is stable over time—assumptions that usually cannot be tested directly. This study analyses the performance of poverty prediction models based on small area estimation (SAE) techniques. Predicted poverty estimates are compared to directly observed levels in a series of country settings that are widely divergent, but where data comparability over time is not judged to be a problem. Prediction models that employ either nonfood expenditures or a full set of assets as predictors, yield poverty estimates that match observed poverty fairly closely. This offers some support for using SAE techniques especially those based on models employing household assets, to approximate the evolution of poverty in settings where comparable consumption data are absent or settings where price deflators are of dubious validity. However, the findings also call for further validation especially in settings with rapid, transitory poverty deterioration, as in Russia during the 1998 financial crisis. 
Keywords:  consumption prediction, price deflator, poverty dynamics, small area estimation 
Date:  2010 
URL:  http://d.repec.org/n?u=RePEc:unu:wpaper:wp201099&r=ecm 
By:  Oualid Bada; Alois Kneip 
Abstract:  We use a panel cointegration model with multiple time varying individual effects to control for the enigmatic missing factors in the credit spread puzzle. Our model specification enables as to capture the unobserved dynamics of the systematic risk premia in the bond market. In order to estimate the dimensionality of the hidden risk factors jointly with the model parameters, we rely on a modified version of the iterated least squares method proposed by Bai, Kao, and Ng (2009). Our result confirms the presence of four common risk components affecting the U.S. corporate bonds during the period between September 2006 and March 2008. However, one single risk factor is sufficient to describe the data for all time periods prior to mid July 2007 when the subprime crisis was detected in the financial market. The dimensionality of the unobserved risk components therefore seems to reflect the degree of difficulty to diversify the individual bond risks. 
Keywords:  Corporate Bond; Credit Spread; Systematic Risk Premium; Panel; Data Model with Interactive Fixed Effects; Factor Analysis; Dimensionality Criteria; Panel Cointegration 
JEL:  D44 D82 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:bon:bonedp:bgse19_2010&r=ecm 
By:  Søren Johansen (University of Copenhagen and CREATES); Morten Ørregaard Nielsen (Queen's University and CREATES) 
Abstract:  We discuss the moment condition for the fractional functional central limit theorem (FCLT) for partial sums of x_{t}=Delta^{d}u_{t}, where d in (1/2,1/2) is the fractional integration parameter and u_{t} is weakly dependent. The classical condition is existence of q>max(2,(d+1/2)^{1}) moments of the innovation sequence. When d is close to 1/2 this moment condition is very strong. Our main result is to show that under some relatively weak conditions on u_{t}, the existence of q≥max(2,(d+1/2)^{1}) is in fact necessary for the FCLT for fractionally integrated processes and that q>max(2,(d+1/2)^{1}) moments are necessary and sufficient for more general fractional processes. Davidson and de Jong (2000) presented a fractional FCLT where only q>2 finite moments are assumed, which is remarkable because it is the only FCLT where the moment condition has been weakened relative to the earlier condition. As a corollary to our main theorem we show that their moment condition is not sufficient. 
Keywords:  Fractional integration, functional central limit theorem, long memory, moment condition, necessary condition 
JEL:  C22 
Date:  2010–10 
URL:  http://d.repec.org/n?u=RePEc:qed:wpaper:1244&r=ecm 
By:  Stefan Hlawatsch (Faculty of Economics and Management, OttovonGuericke University Magdeburg); Peter Reichling (Faculty of Economics and Management, OttovonGuericke University Magdeburg) 
Abstract:  Copulas erfreuen sich in der Finanzwirtschaft wachsender Beliebtheit. Ursache hierfür ist insbesondere die Möglichkeit, mit ihrer Hilfe nichtlineare Abhängigkeitsstrukturen darzustellen. Ein weiterer Vorteil besteht darin, dass multivariate Verteilungen mit Hilfe von Copulas separat in ihre Randverteilungen und in ihre Abhängigkeitsstruktur zerlegt werden können. Damit ist die Untersuchung der Abhängigkeitsstruktur losgelöst von Annahmen über die Randverteilungen. Diese Flexibilität ermöglicht die Anwendung von Copulas in zahlreichen Bereichen der Finanzwirtschaft, vom Risikomanagement über die Bewertung von komplexen Finanzprodukten bis zur Portfoliooptimierung. Die vorliegende Arbeit dient zum Einen als didaktischer Einstieg in die Copulathematik und stellt zum Anderen die aktuellen Forschungsergebnisse aus den genannten Bereichen vor. 
Keywords:  Copula, Portfoliomanagement, Risikomanagement, Optionspreisbewertung 
JEL:  C16 C51 G11 G12 G15 
Date:  2010–07 
URL:  http://d.repec.org/n?u=RePEc:mag:wpaper:100016&r=ecm 