skip to content
 

 


 

  • "A One-Covariate at a Time, Multiple Testing Approach to Variable Selection in High-Dimensional Linear Regression Models", by Alexander Chudik, George Kapetanios and M. Hashem Pesaran, forthcoming in Econometrica, February 2018.

    Abstract: This paper provides an alternative approach to penalised regression for model selection in the context of high dimensional linear regressions where the number of covariates is large, often much larger than the number of available bbservations. We consider the statistical significance of individual covariates one at a time, whilst taking full account of the multiple testing nature of the inferential problem involved. We refer to the proposed method as One Covariate at a Time Multiple Testing (OCMT) procedure, and use ideas from the multiple testing literature to control the probability of selecting the approximating model, the false positive rate and the false discovery rate. OCMT is easy to interpret, relates to classical statistical analysis, is valid under general assumptions, is faster to compute, and performs well in small samples. The usefulness of OCMT is also illustrated by an empirical application to forecasting U.S. output growth and inflation.
    JEL Classifications: C52, C55
    Key Words: One covariate at a time, multiple testing, model selection, high dimensionality, penalised regressions, boosting, Monte Carlo experiments.
    Full Text: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp18/OCMT_CPK_2018Feb23.pdf
    MC Supplement: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp18/MC_Supplement_OCMT_CPK_2018Feb23.pdf

     

  • "Half-Panel Jackknife Fixed Effects Estimation of Linear Panels with Weakly Exogenous Regressors", by Alexander Chudik, M. Hashem Pesarann and Jui-Chung Yang, SSRN Working Paper No. 281, forthcoming in Journal of Applied Econometrics, January 2018

    Abstract: This paper considers estimation and inference in fixed effects (FE) linear panel regression models with lagged dependent variables and/or other weakly exogenous (or predetermined) regressors when N (the cross section dimension) is large relative to T (the time series dimension). The paper first derives a general formula for the bias of the FE estimator which is a generalization of the Nickell type bias derived in the literature for the pure dynamic panel data models. It shows that in the presence of weakly exogenous regressors, inference based on the FE estimator will result in size distortions unless N/T is suffciently small. To deal with the bias and size distortion of FE estimator when N is large relative to T, the use of half-panel Jackknife FE estimator is proposed and its asymptotic distribution is derived. It is shown that the bias of the proposed estimator is of order [code], and for valid inference it is only required that N/T[code], as N, T [code] jointly. Extensions to panel data models with time effects (TE), for balanced as well as unbalanced panels, are also provided. The theoretical results are illustrated with Monte Carlo evidence. It is shown that the FE estimator can suffer from large size distortions when N > T, with the proposed estimator showing little size distortions. The use of half-panel jackknife FE-TE estimator is illustrated with two empirical applications from the literature.
    JEL Classifications: C32, E17, E32, F44, F47, O51, Q43.
    Key Words: Panel Data Models, Weakly Exogenous Regressors, Lagged Dependent Variable, Fixed Effects, Time Effects, Unbalanced Panels, Half-Panel Jackknife, Bias Correction
    Full Text: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp18/CPY_jackknifeFE_15_January2018.pdf
    Readme file: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp18/readme_file_for_data_and_codes.txt
    Data for the Empirical Section: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp18/cpy-data.zip
    Computer codes: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp18/cpy-progs.zip
     

  • "Tests of Policy Interventions in DSGE Models", by M. Hashem Pesaran and Ron P. Smith, forthcoming in Oxford Bulletin of Economics and Statistics, October 2017.

    Abstract: This paper considers tests of the effectiveness of a policy intervention, de…ned as a change in the parameters of a policy rule, in the context of a macroeconometric dynamic stochastic general equilibrium (DSGE) model. We consider two types of intervention, fi…rst the standard case of a parameter change that does not alter the steady state, and second one that does alter the steady state, e.g. the target rate of infl‡ation. We consider two types of test, one a multi-horizon test, where the post-intervention policy horizon, H, is small and fi…xed, and a mean policy effect test where H is allowed to increase without bounds. The multi-horizon test requires Gaussian errors, but the mean policy effect test does not. It is shown that neither of these two tests are consistent, in the sense that the the power of the tests does not tend to unity as H, unless the intervention alters the steady state. This follows directly from the fact that DSGE variables are measured as deviations from the steady state, and the effects of policy change on target variables decay exponentially fast. We investigate the size and power of the proposed mean effect test by simulating a standard three equation New Keynesian DSGE model. The simulation results are in line with our theoretical fi…ndings and show that in all applications the tests have the correct size; but unless the intervention alters the steady state, their power does not go to unity with H.
    JEL Classifications: C18, C54, E65.
    Key Words: Counterfactuals, policy analysis, policy ineffectiveness test, macroeconomics.
    Full Text: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp17/PS_on_PI_12_October_2017_main_paper.pdf
    Supplement: http://www.econ.cam.ac.uk/people-files/emeritus/mhp1/fp17/PS_on_PI_12_October_2017_online_supplement.pdf

     

  • "Estimation of Time-invariant Effects in Static Panel Data Models", by M. Hashem Pesaran and Qiankun Zhou, forthcoming in Econometrics Reviews, June 2016.

    Abstract: This paper proposes the Fixed Effects Filtered (FEF) and Fixed Effects Filtered instrumental variable (FEF-IV) estimators for estimation and inference in the case of time-invariant effects in static panel data models when N is large and T is fixed. The FEF-IV allows for endogenous time-invariant regressors but assumes that there exists a suficient number of instruments for such regressors. It is shown that the FEF and FEF-IV estimators are [code]-consistent, and asymptotically normally distributed. The FEF estimator is compared with the Fixed Effects Vector Decomposition (FEVD) estimator proposed by Plumper and Troeger (2007) and conditions under which the two estimators are equivalent are established. It is also shown that the variance estimator proposed for FEVD estimator is inconsistent and its use could lead to misleading inference. Alternative variance estimators are proposed for both FEF and FEF-IV estimators which are shown to be consistent under fairly general conditions. The small sample properties of the FEF and FEF-IV estimators are investigated by Monte Carlo experiments, and it is shown that FEF has smaller bias and RMSE, unless an intercept is included in the second stage of the FEVD procedure which renders the FEF and FEVD estimators identical. The FEVD procedure, however, results in substantial size distortions since it uses incorrect standard errors. In the case where some of the time-invariant regressors are endogenous, the FEF-IV procedure is compared with a modified version of Hausman and Taylor (1981) (HT) estimator. It is shown that both estimators perform well and have similar small sample properties. But the application of standard HT procedure, that incorrectly assumes a sub-set of time-varying regressors are uncorrelated with the individual effects, will yield biased estimates and significant size distortions.
    JEL Classifications: C01, C23, C33.
    Key Words: Static panel data models, time-invariant effects, endogenous time-invariant regressors, Monte Carlo experiments, fixed effects filtered estimators.
    Full Text: http://www.econ.cam.ac.uk/emeritus/mhp1/fp16/PesaranZhou_Time-invariant-estimation_June-11-2016.pdf
    Supplementary Data: http://www.econ.cam.ac.uk/emeritus/mhp1/fp16/PesaranZhou_Time-invariant-estimation_May-27-2016_supplement.pdf
    Stata Code and Instructions: http://qiankunzhou.weebly.com/research.html