help xtscc                                       version:  1.3                 
                                                also see:  xtscc postestimation
-------------------------------------------------------------------------------

Title

    xtscc - Regression with Driscoll-Kraay standard errors


Syntax

        xtscc depvar [indepvars] [if] [in] [weight] [, options]


    options         Description
    -------------------------------------------------------------------------
    Model
      lag(#)        set maximum lag order of autocorrelation; default is
                      m(T)=floor[4(T/100)^(2/9)]
      fe            perform fixed effects (within) regression
      pooled        perform pooled OLS/WLS regression; default
      noconstant    suppress regression constant in pooled OLS/WLS
                      regressions
      ase           return (asymptotic) Driscoll-Kraay SE without small
                      sample adjustment

    Reporting
      level(#)      set confidence level; default is level(95)
    -------------------------------------------------------------------------
    You must tsset your data before using xtscc.
    by, statsby, and xi may be used with xtscc; see prefix.
    aweights are allowed unless option fe is specified; see weight.
    See xtscc postestimation for features available after estimation.



Description

    xtscc produces Driscoll and Kraay (1998) standard errors for coefficients
    estimated by pooled OLS/WLS or fixed-effects (within) regression. depvar
    is the dependent variable and varlist is an (optional) list of
    explanatory variables.

    The error structure is assumed to be heteroskedastic, autocorrelated up
    to some lag, and possibly correlated between the groups (panels).
    Driscoll-Kraay standard errors are robust to very general forms of
    cross-sectional ("spatial") and temporal dependence when the time
    dimension becomes large. This nonparametric technique of estimating
    standard errors does not place any restrictions on the limiting behavior
    of the number of panels. Consequently, the size of the cross-sectional
    dimension in finite samples does not constitute a constraint on
    feasibility - even if the number of panels is much larger than T.
    However, note that the estimator is based on large T asymptotics.
    Therefore, one should be somewhat cautious with applying this estimator
    to panel datasets with a large number of groups but a small number of
    observations over time.

    This implementation of Driscoll and Kraay's covariance estimator works
    for both, balanced and unbalanced panels, respectively. Furthermore, it
    is capable to handle missing values.


Options

        +-------+
    ----+ Model +------------------------------------------------------------

    lag(#) specifies the maximum lag to be considered in the autocorrelation
        structure.  If you do not specify this option, a lag length of
        m(T)=floor[4(T/100)^(2/9)] is chosen.

    fe performs fixed-effects (within) regression with Driscoll-Kraay
        standard errors.  These standard errors are robust to very general
        forms of cross-sectional ("spatial") and temporal dependence
        (provided that T is sufficiently large). See above.  If the residuals
        are assumed to be heteroscedastic only: use xtreg, fe robust.

    pooled performs pooled OLS/WLS regression with Driscoll-Kraay standard
        errors.  These standard errors are robust to very general forms of
        cross-sectional ("spatial") and temporal dependence (provided that T
        is sufficiently large). See above.  If the residuals are assumed to
        be heteroscedastic only: use reg, robust cluster().  If the residuals
        are assumed to be heteroscedastic and autocorrelated only (i.e.
        there is no cross-sectional correlation): use newey, lag() force.

    noconstant; see [R] estimation options.

    ase returns asymptotic Driscoll-Kraay standard errors. Standard errors
        that are computed this way might be slightly overoptimistic as they
        abstract away from a small sample adjustment.


        +-----------+
    ----+ Reporting +--------------------------------------------------------

    level(#); see estimation options.



Examples

    . sysuse grunfeld

    Pooled OLS estimation:

    . reg invest mvalue kstock, robust cluster(company)
    . est store robust

    . newey invest mvalue kstock, lag(4) force
    . est store newey

    . xtscc invest mvalue kstock, lag(4)
    . est store dris_kraay

    . est table *, b se t

    Fixed-effects (within) regression:

    . est clear
    . xtreg invest mvalue kstock, fe robust
    . est store fe_robust

    . xtscc invest mvalue kstock, fe lag(4)
    . est store fe_dris_kraay

    . est table *, b se t



Reference

    - Driscoll, John C. and Aart C. Kraay, 1998, Consistent Covariance Matrix
      Estimation with Spatially Dependent Panel Data, Review of Economics and
      Statistics 80, 549-560.


Notes

    - The main procedure of xtscc is implemented in Mata and largely follows
      Driscoll and Kraay's GAUSS program which is available from
      http://www.johncdriscoll.net/.
    - The xtscc uses functions from Ben Jann's moremata package.


Acknowledgements

    I would like to thank David M. Drukker and Bill Gould from StataCorp for
    their useful comments and suggestions.


Author

    Daniel Hoechle, University of Basel, daniel.hoechle@unibas.ch



Also see

    Manual:  [R] regress, [TS] newey, [XT] xtreg

    Online:  xtscc postestimation;
             tsset, regress, newey, xtreg, _robust