|
on Econometrics |
| By: | Taisuke Otsu; Luke Taylor |
| Abstract: | In this paper we develop a nonparametric estimator for the local average response of a censored dependent variable to endogenous regressors in a nonseparable model where the unobservable error term is not restricted to be scalar and where the nonseparable function need not be monotone in the unobservables. We formalise the identification argument put forward in Altonji, Ichimura and Otsu (2012), construct the nonparametric estimator, characterise its asymptotic property, and conduct a Monte Carlo investigation to study the small sample properties. Identification is constructive and is achieved through a control function approach. We show that the estimator is consistent and asymptotically normally distributed. The Monte Carlo results are encouraging. |
| JEL: | C24 C34 C14 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:cep:stiecm:/2014/575&r=ecm |
| By: | Liangjun Su (Singapore Management University, School of Economics); Zhentao Shi (Department of Economics, Yale University); Peter C. B. Phillips (Yale University, University of Auckland, University of Southampton and Singapore Management University) |
| Abstract: | This paper provides a novel mechanism for identifying and estimating latent group structures in panel data using penalized regression techniques. We focus on linear models where the slope parameters are heterogeneous across groups but homogenous within a group and the group membership is unknown. Two approaches are considered — penalized least squares (PLS) for models without endogenous regressors, and penalized GMM (PGMM) for models with endogeneity. In both cases we develop a new variant of Lasso called classifier-Lasso (C-Lasso) that serves to shrink individual coefficients to the unknown group-specific coefficients. C-Lasso achieves simultaneous classification and consistent estimation in a single step and the classification exhibits the desirable property of uniform consistency. For PLS estimation C-Lasso also achieves the oracle property so that group-specific parameter estimators are asymptotically equivalent to infeasible estimators that use individual group identity information. For PGMM estimation the oracle property of C-Lasso is preserved in some special cases. Simulations demonstrate good finite-sample performance of the approach both in classification and estimation. An empirical application investigating the determinants of cross-country savings rates finds two latent groups among 56 countries, providing empirical confirmation that higher savings rates go in hand with higher income growth. |
| Keywords: | Classification; Cluster analysis; Convergence club; Dynamic panel; Group Lasso; High dimensionality; Oracle property; Panel structure model; Parameter heterogeneity; Penalized least squares; Penalized GMM |
| JEL: | C33 C36 C38 C51 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:siu:wpaper:07-2014&r=ecm |
| By: | Junhui Qian (Antai College of Economics and Management, Shanghai Jiao Tong University); Liangjun Su (Singapore Management University, School of Economics) |
| Abstract: | In this paper we consider the problem of determining the number of structural changes in multiple linear regression models via group fused Lasso (least absolute shrinkage and selection operator ). We show that with probability tending to one our method can correctly determine the unknown number of breaks and the estimated break dates are sufficiently close to the true break dates. We obtain estimates of the regression coefficients via post Lasso and establish the asymptotic distributions of the estimates of both break ratios and regression coefficients. We also propose and validate a datadriven method to determine the tuning parameter. Monte Carlo simulations demonstrate that the proposed method works well in finite samples. We illustrate the use of our method with a predictive regression of the equity premium on fundamental information. |
| Keywords: | Change point; Fused Lasso; Group Lasso; Penalized least squares; Structural change |
| JEL: | C13 C22 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:siu:wpaper:06-2014&r=ecm |
| By: | W. Robert Reed (University of Canterbury); Raymond J.G.M. Florax; Jacques Poot |
| Abstract: | This study uses Monte Carlo analysis to investigate the performances of five different meta-analysis (MA) estimators: the Fixed Effects (FE) estimator, the Weighted Least Squares (WLS) estimator, the Random Effects (RE) estimator, the Precision Effect Test (PET) estimator, and the Precision Effect Estimate with Standard Errors (PEESE) estimator. We consider two types of publication bias: publication bias directed against statistically insignificant estimates, and publication bias directed against wrong-signed estimates. Finally, we consider three cases concerning the distribution of the “true effect”: the Fixed Effects case, where there is only estimate per study, and all studies have the same true effect; the Random Effects case, where there is only one estimate per study, and there is heterogeneity in true effects across studies; and the Panel Random Effects case, where studies have multiple estimates, and true effects are random both across and within studies. Our simulations produce a number of findings that challenge results from previous research. |
| Keywords: | Meta-analysis, Random effects, Fixed effects, publication bias, Monte Carlo, Simulations |
| JEL: | B41 C15 C18 |
| Date: | 2014–08–13 |
| URL: | http://d.repec.org/n?u=RePEc:cbt:econwp:14/22&r=ecm |
| By: | Ko, Stanley I. M.; Chong, Terence T. L.; Ghosh, Pulak |
| Abstract: | This paper proposes a new Bayesian multiple change-point model which is based on the hidden Markov approach. The Dirichlet process hidden Markov model does not require the specification of the number of change-points a priori. Hence our model is robust to model specification in contrast to the fully parametric Bayesian model. We propose a general Markov chain Monte Carlo algorithm which only needs to sample the states around change-points. Simulations for a normal mean-shift model with known and unknown variance demonstrate advantages of our approach. Two applications, namely the coal-mining disaster data and the real US GDP growth, are provided. We detect a single change-point for both the disaster data and US GDP growth. All the change-point locations and posterior inferences of the two applications are in line with existing methods. |
| Keywords: | Change-point; Dirichlet process; Hidden Markov model; Markov chain; Monte Carlo; Nonparametric Bayesian. |
| JEL: | C22 |
| Date: | 2014–08–07 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:57871&r=ecm |
| By: | Martyna Marczak (University of Hohenheim); Tommaso Proietti (Università di Roma “Tor Vergata” and CREATES) |
| Abstract: | Structural change affects the estimation of economic signals, like the underlying growth rate or the seasonally adjusted series. An important issue, which has attracted a great deal of attention also in the seasonal adjustment literature, is its detection by an expert procedure. The general–to–specific approach to the detection of structural change, currently implemented in Autometrics via indicator saturation, has proven to be both practical and effective in the context of stationary dynamic regression models and unit–root autoregressions. By focusing on impulse–and step–indicator saturation, we investigate via Monte Carlo simulations how this approach performs for detecting additive outliers and level shifts in the analysis of nonstationary seasonal time series. The reference model is the basic structural model, featuring a local linear trend, possibly integrated of order two, stochastic seasonality and a stationary component. Further, we apply both kinds of indicator saturation to detect additive outliers and level shifts in the industrial production series in five European countries. |
| Keywords: | Indicator saturation, seasonal adjustment, structural time series model, outliers, structural change, general–to–specific approach, state space model |
| JEL: | C22 C51 C53 |
| Date: | 2014–11–08 |
| URL: | http://d.repec.org/n?u=RePEc:aah:create:2014-20&r=ecm |
| By: | Trojan, Sebastian |
| Abstract: | A high frequency stochastic volatility (SV) model is proposed. Price duration and associated absolute price change in event time are modeled contemporaneously to fully capture volatility on the tick level, combining the SV and stochastic conditional duration (SCD) model. Estimation is with IBM stock intraday data 2001/10 (decimalization completed), taking a minimum midprice threshold of a half tick. Persistent information flow is extracted, featuring a positively correlated innovation term and negative cross effects in the AR(1) persistence matrix. Additionally, regime switching in both duration and absolute price change is introduced to increase nonlinear capabilities of the model. Thereby, a separate price jump state is identified. Model selection and predictive tests show superiority of the regime switching extension in- and out-of-sample. |
| Keywords: | Stochastic volatility, stochastic conditional duration, non-Gaussian and nonlinear state space model, tick data, event time, generalized gamma distribution, negative binomial distribution, regime switching, Markov chain Monte Carlo, block sampler, particle filter, adaptive Metropolis |
| JEL: | C11 C15 C32 C58 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:usg:econwp:2014:25&r=ecm |
| By: | Sainan Jin (Singapore Management University, School of Economics); Liangjun Su (Singapore Management University, School of Economics); Yonghui Zhang (School of Economics, Renmin University of China) |
| Abstract: | In this paper we propose a class of nonparametric tests for anomaly effects in empirical asset pricing models in the framework of nonparametric panel data models with interactive fixed effects. Our approach has two prominent features: one is the adoption of nonparametric functional form to capture the anomaly effects of some asset-specific characteristics, and the other is the flexible treatment of both observed/constructed and unobserved common factors. By estimating the unknown factors, betas, and nonparametric function simultaneously, our setup is robust to misspecification of functional form and common factors and avoids the well-known “error-in-variable” (EIV) problem associated with the commonly used two-pass (TP) procedure. We apply our method to a publicly available data set and divide the full sample into three subsamples. Our empirical results show that size and book-to-market ratio affect the excess returns of portfolios significantly for the full sample and two of the three subsamples in all five factor pricing models under investigation. In particular, nonparametric component is significantly different from zero, meaning that the constructed common factors (e.g., small minus big (SMB) and high minus low (HML)) cannot capture all the size and book-to-market ratio effects. We also find strong evidence of nonlinearity of the anomaly effects in the Fama-French 3-factor model and the augmented 4-factor and 5-factor models in the full sample and two of the three subsamples. |
| Keywords: | Anomaly effects; Asset pricing; CAPM; Common factors; EIV; Fama-French three-factor; Interactive fixed effects; Nonparametric panel data model; Sieve method; Specification test |
| JEL: | C14 C33 C58 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:siu:wpaper:09-2014&r=ecm |
| By: | Emura, Takeshi; Chen, Yi-Hau |
| Abstract: | Dependent censoring arises in biomedical studies when the survival outcome of interest is censored by competing risks. In survival data with microarray gene expressions, gene selection based on the univariate Cox regression analyses has been used extensively in medical research, which however, is only valid under the independent censoring assumption. In this paper, we first consider a copula-based framework to investigate the bias caused by dependent censoring on gene selection. Then, we utilize the copula-based dependence model to develop an alternative gene selection procedure. Simulations show that the proposed procedure adjusts for the effect of dependent censoring and thus outperforms the existing method when dependent censoring is indeed present. The non-small-cell lung cancer data is analyzed to demonstrate the usefulness of our proposal. We implemented the proposed method in an R “compound.Cox” package. |
| Keywords: | Bivariate survival distribution; Competing risk; Compound covariate prediction; Cox regression; Cross validation; Frailty, Kendall’s tau |
| JEL: | C12 C13 C14 C34 |
| Date: | 2014–05–17 |
| URL: | http://d.repec.org/n?u=RePEc:pra:mprapa:58043&r=ecm |
| By: | Marinho Bertanha; Petra Moser |
| Abstract: | Count data regressions are an important tool for empirical analyses ranging from analyses of patent counts to measures of health and unemployment. Along with negative binomial, Poisson panel regressions are a preferred method of analysis because the Poisson conditional fixed effects maximum likelihood estimator (PCFE) and its sandwich variance estimator are consistent even if the data are not Poisson-distributed, or if the data are correlated over time. Analyses of counts may be affected by correlation in the cross-section. For example, patent counts or publications may increase across related research fields in response to common shocks. This paper shows that the PCFE and its sandwich variance estimator are consistent in the presence of such dependence in the cross-section - as long as spatial dependence is time-invariant. In addition to the PCFE, this result also applies to the commonly used Logit model of panel data with fixed effects. We develop a test for time-invariant spatial dependence and provide code in STATA and MATLAB to implement the test. |
| JEL: | C23 C33 O3 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:nbr:nberwo:20374&r=ecm |
| By: | Steve Gibbons; Henry G. Overman; Eleonora Patacchini |
| Abstract: | This paper is concerned with methods for analysing spatial data. After initial discussion on the nature of spatial data, including the concept of randomness, we focus most of our attention on linear regression models that involve interactions between agents across space. The introduction of spatial variables in to standard linear regression provides a flexible way of characterising these interactions, but complicates both interpretation and estimation of parameters of interest. The estimation of these models leads to three fundamental challenges: the reflection problem, the presence of omitted variables and problems caused by sorting. We consider possible solutions to these problems, with a particular focus on restrictions on the nature of interactions. We show that similar assumptions are implicit in the empirical strategies - fixed effects or spatial differencing - used to address these problems in reduced form estimation. These general lessons carry over to the policy evaluation literature. |
| Keywords: | Spatial analysis, spatial econometrics, neighbourhood effects, agglomeration, weights matrix |
| JEL: | R C1 C5 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:cep:sercdp:0162&r=ecm |
| By: | John Morrow |
| Abstract: | Benford's Law is used to test for data irregularities. While novel, there are two weaknesses in the current methodology. First, test values used in practice are too conservative and the test values of this paper are more powerful and hold for fairly small samples. Second, testing requires Benford's Law to hold, which it often does not. I present a simple method to transform distributions to satisfy the Law with arbitrary precision and induce scale invariance, freeing tests from the choice of units. I additionally derive a rate of convergence to Benford's Law. Finally, the results are applied to common distributions. |
| Keywords: | Benfords Law, data quality, fraud detection |
| JEL: | C10 C24 C46 |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:cep:cepdps:dp1291&r=ecm |
| By: | Angela Gu; Patrick Zeng |
| Abstract: | Factor analysis is a statistical technique employed to evaluate how observed variables correlate through common factors and unique variables. While it is often used to analyze price movement in the unstable stock market, it does not always yield easily interpretable results. In this study, we develop improved factor models by explicitly incorporating sector information on our studied stocks. We add eleven sectors of stocks as defined by the IBES, represented by respective sector-specific factors, to non-specific market factors to revise the factor model. We then develop an expectation maximization (EM) algorithm to compute our revised model with 15 years' worth of S&P 500 stocks' daily close prices. Our results in most sectors show that nearly all of these factor components have the same sign, consistent with the intuitive idea that stocks in the same sector tend to rise and fall in coordination over time. Results obtained by the classic factor model, in contrast, had a homogeneous blend of positive and negative components. We conclude that results produced by our sector-based factor model are more interpretable than those produced by the classic non-sector-based model for at least some stock sectors. |
| Date: | 2014–08 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:1408.2794&r=ecm |