XTDYNESTIMB: Stata module implementing dynamic linear panel-data estimators robust to structural breaks, long-T overidentification, and error cross-sectional dependence

xtdynestimb estimates the dynamic linear panel-data model y_it = a_1 y_i,t-1 + ... + a_p y_i,t-p + x_it' b + eta_i [+ f_t] + v_it with three modern estimators that each address a different failure of the textbook Arellano-Bond / Blundell-Bond GMM estimator: o Structural breaks. Mean shifts in the fixed effects invalidate the standard difference and system moment conditions. dd adds the break-robust double-difference moment conditions of Chowdhury & Russell (2017) (both the instruments and the equation are in differences) and offers five nested estimators. o Error cross-sectional dependence. Common factors make the lagged-level instruments invalid. csdgmm removes the common factors by cross-sectional (time) demeaning and, optionally, restricts the instrument set to the regressors only (Sarafidis 2009). o Long-T overidentification bias. When T is large the number of AB moment conditions grows like T^2, biasing the estimator. ablasso uses period-by-period LASSO to keep only the most informative instruments and cross-fitting to remove the over-fitting bias (Chernozhukov, Fernandez-Val, Huang & Wang 2024).