{smcl} {* 31mar2005}{...} {hline} help for {hi:xtlsdvc} {hline} {title: Bias corrected LSDV dynamic panel data estimator} {p 8 16 2}{cmd:xtlsdvc} {it:depvar} [{it:varlist}] [{cmd:if} {it:exp}] {cmd:,} {cmdab:i:nitial:(}{it:estimator}{cmd:)} [{cmdab:l:evel:(}{it:#}{cmd:)} {cmdab:b:ias:(}{it:#}{cmd:)} {cmdab:l:sdv} {cmdab:f:irst} {cmdab:v:cov:(}{it:#}{cmd:)}] {p 4 4 2} where {it:estimator} is {p 12 36 2}{cmd:ah}{space 18}Anderson-Hsiao{p_end} {p 12 36 2}{cmd:ab}{space 18}Arellano-Bond{p_end} {p 12 36 2}{cmd:bb}{space 18}Blundell-Bond{p_end} {p 12 36 2}{cmd:my}{space 18}initial values supplied by the user{p_end} {p 4 4 2} {cmd:xtlsdvc} is for use with time-series data. You must {cmd:tsset} your data before using {cmd:xtlsdvc}; see help {help tsset}. However, since {cmd:xtlsdvc} calls {cmd:xtreg} {it:varlist}s may not contain time-series operators; see help {help xtreg}. {p 4 4 2} {cmd:xtlsdvc} shares the features of all estimation commands; see help {help estcom}. {p 4 4 2} The syntax of {help predict} following {cmd:xtlsdvc} is {p 8 16 2}{cmd:predict} [{it:type}] {it:newvarname} [{cmd:if} {it:exp}] [{cmd:in} {it:range}] [{cmd:,} {it:statistic} ] {p 4 4 2} where y[i,t] = {bind: y[i,t-1]a} + {bind:x[i,t]b} + u[i] + e[i,t] and {it:statistic} is {p 8 21 2}{cmd:xb} {space 4} y[i,t-1]a + x[i,t]b, fitted values; the default{p_end} {p 8 21 2}{cmd:ue} {space 4} u[i] + e[i,t], the combined residual{p_end} {p 4 21 2}(*) {cmd:xbu} {space 3} y[i,t-1]a + x[i,t]b + u[i], prediction including fixed effect{p_end} {p 4 21 2}(*) {cmd:u} {space 5} u[i], the fixed effect{p_end} {p 4 21 2}(*) {cmd:e} {space 5} e[i,t]{p_end} {p 4 4 2} Unstarred statistics are available both in and out of sample; type "{cmd:predict} {it:...} {cmd:if e(sample)} {it:...}" if wanted only for the estimation sample. Starred statistics are calculated only for the estimation sample even when "{cmd:if e(sample)}" is not specified. {title:Description} {p 4 4 2} {cmd:xtlsdvc} calculates bias corrected LSDV estimators for the standard autoregressive panel data model using the bias approximations in Bruno (2005a), who extends the results by Bun and Kiviet (2003), Kiviet (1999) and Kiviet (1995) to unbalanced panels {p 4 12 2}y[i,t] = {bind: y[i,t-1]a} + {bind:x[i,t]b} + u[i] + e[i,t] {space 4} i={c -(}1,...,N{c )-}; {space 3} t={c -(}1,...,T_i{c )-}, where {p 4 12 2}a is a parameter to be estimated{p_end} {p 4 12 2}x[i,t] is a (1 X (k-1)) vector of strictly exogenous covariates{p_end} {p 4 12 2}b is a ((k-1) X 1) vector of parameters to be estimated{p_end} {p 4 12 2}u[i] are the individual effects, for which no distributional assumption is made apart being fixed over time,{p_end} {p 4 12 2}and e[i,t] are iid over the whole sample with variance s_e*s_e.{p_end} {p 4 4 2} It is also assumed that the u[i] and the e[i,t] are independent for each i over all t. {p 4 4 2} A more detailed description of {cmd:xtlsdvc} can be found in Bruno (2005b). {title:Options} {p 4 8 2}{cmd:level(}{it:#}{cmd:)} specifies the confidence level, in percent, for confidence intervals of the coefficients; see help {help level}. {p 4 8 2}{cmd:initial(}{it:estimator}{cmd:)} specifies which consistent estimator among Anderson-Hsiao ({cmd:ah}), Arellano-Bond ({cmd:ab}) and Blundell-Bond ({cmd:bb}) is to initialize the bias correction. In alternative, users may want to supply their own values, which can be done by creating beforehand the (1 X (k+1)) matrix {cmd:my}, the i.th element of which serves as an initial value for the coefficient on the i.th variable in {it:varlist}, i=1,...,k and the last element as an estimate for the error variance. This may be useful in Monte Carlo simulations or when the user wants to try initial estimators other than {cmd:ah}, {cmd:ab} or {cmd:bb}. {p 4 8 2}{cmd:bias(}{it:#}{cmd:)} determines the accuracy of the approximation: up to O(1/T) ({cmd:1}); up to O(1/NT) ({cmd:2}); up to O(1/NT^2) ({cmd:3}). {p 4 8 2}{cmd:first} requests that the first-stage regression results be displayed. {p 4 8 2}{cmd:lsdv} requests that the lsdv regression results be displayed. {p 4 8 2}{cmd:vcov(}{it:#}{cmd:)} calculates a bootstrap variance-covariance matrix for LSDVC using {it:#} repetitions. Normality for errors is assumed. This procedure continues to work also in the presence of gaps in the exogenous variables, although in this case bootstrap samples for each unit are truncated to the first missing value encountered. Gaps in the dependent variable, instead, bear no consequence to the bootstrap sample size. {title:Options for {help predict}} {p 4 8 2}{cmd:xb} calculates the linear prediction; that is, y[i,t-1]a + x[i,t]b. This is the default. {p 4 8 2}{cmd:ue} calculates the prediction of u[i] + e[i,t]. {p 4 8 2}{cmd:xbu} calculates the prediction of y[i,t-1]a + x[i,t]b + u[i], the prediction including the fixed component. {p 4 8 2}{cmd:u} calculates the prediction of u[i], the estimated fixed effect. {p 4 8 2}{cmd:e} calculates the prediction of e[i,t]. {title:Remarks} {p 4 4 2}{cmd:xtlsdvc} does not report analytical standard errors. Only bootstrap standard errors are reported, provided that {cmd:vcov(}{it:#}{cmd:)} is given. {p 4 4 2}Bootstrap standard errors are downward biased when values for the unknown parameters are supplied through the matrix {cmd:my}, since the procedure, keeping {cmd:my} fixed over replications, neglects a source of varibility of the bias-corrected LSDV estimator. {p 4 4 2}{cmd:xtlsdvc} saves the following results in {cmd:e()}: {col 5}Scalars {col 9}{cmd:e(N)}{col 23}Number of observations {col 9}{cmd:e(Tbar)}{col 23}Average number of time periods {col 9}{cmd:e(sigma)}{col 23}Estimate of s_e through the within {col 23}residuals from the first stage regression {col 9}{cmd:e(N_g)}{col 23}Number of groups {col 5}Macros {col 9}{cmd:e(cmd)}{col 23}xtlsdvc {col 9}{cmd:e(depvar)}{col 23}Name of dependent variable {col 9}{cmd:e(ivar)}{col 23}Panel variable {col 9}{cmd:e(predict)}{col 23}Program used to implement predict {col 5}Matrices {col 9}{cmd:e(b)}{col 23}Coefficient vector {col 9}{cmd:e(V)}{col 23}Variance-covariance matrix of the estimators {col 9}{cmd:e(b_lsdv)}{col 23}Coefficient vector of the uncorrected LSDV {col 9}{cmd:e(V_lsdv)}{col 23}Variance-covariance matrix of the uncorrected LSDV {col 5}Functions {col 9}{cmd:e(sample)}{col 23}Marks estimation sample {title:Examples} {p 4 8 2}{cmd:. xtlsdvc n w k ys yr1980-yr1984, initial(ah)}{p_end} {p 4 8 2}{cmd:. xtlsdvc n w k ys yr1980-yr1984, initial(ab) bias(3)}{p_end} {p 4 8 2}{cmd:. xtlsdvc n w k ys yr1980-yr1984, initial(ab) bias(3) vcov(50)}{p_end} {title:References} {p 4 8 2}Bruno, G.S.F. 2005a. Approximating the bias of the LSDV estimator for dynamic unbalanced panel data models. {it:Economics Letters,} 87, 361-366. {p 4 8 2}Bruno, G.S.F. 2005b. Estimation and inference in dynamic unbalanced panel data models with a small number of individuals. {it:CESPRI WP n.165} , Università Bocconi-CESPRI, Milan. {p 4 8 2}Bun, M.J.G., Kiviet, J.F., 2003. On the diminishing returns of higher order terms in asymptotic expansions of bias. {it:Economics Letters,} 79, 145-152. {p 4 8 2}Kiviet, J.F., 1995. On Bias, Inconsistency and Efficiency of Various Estimators in Dynamic Panel Data Models. {it:Journal of Econometrics,} 68, 53-78. {p 4 8 2}Kiviet, J.F., 1999. Expectation of Expansions for Estimators in a Dynamic Panel Data Model; Some Results for Weakly Exogenous Regressors. In: Hsiao, C., Lahiri, K., Lee, L.-F., Pesaran, M.H. (Eds.), {it:Analysis of Panel Data and Limited Dependent Variables.} Cambridge University Press, Cambridge. {title:Author} {p 4}Giovanni S.F. Bruno{p_end} {p 4}Istituto di Economia Politica, Università Bocconi{p_end} {p 4}Milan, Italy{p_end} {p 4}giovanni.bruno@unibocconi.it{p_end} {title:Also see} {p 4 12 2} Manual: {hi:[U] 23 Estimation and post-estimation commands},{break} {hi:[U] 29 Overview of Stata estimation commands},{break} {hi:[XT] xtabond} {hi:[XT] xtivreg} {hi:[R] ivreg} {p 4 13 2} Online: help for {help estcom}, {help ivreg}, {help postest}, {help xtabond}, {help xtdes}, {help xtivreg}, {help xtreg}, {help xtregar}, {help xtsum}