同求，这个问题困扰我好久了。

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.