bayespecon.dgp.simulate_panel_sdem_fe¶

bayespecon.dgp.simulate_panel_sdem_fe(N, T, lam=0.4, beta1=None, beta2=None, sigma=1.0, sigma_alpha=0.5, err_hetero=False, rng=None, seed=None, W=None, gdf=None, contiguity='queen', create_gdf=False, geometry_type='polygon', wide=False)[source]¶

Simulate SDEM panel FE data.

DGP¶

u_t = (I-lam W)^(-1) eps_t and y_t = X_t beta1 + W X_t[:,1:] beta2 + alpha + u_t.

param N:

Number of units and time periods.

type N:

int

param T:

Number of units and time periods.

type T:

int

param lam:

Spatial error coefficient.

type lam:

float, default=0.4

param beta1:

Coefficients on X including intercept.

type beta1:

np.ndarray, optional

param beta2:

Coefficients on spatially lagged non-intercept regressors.

type beta2:

np.ndarray, optional

param sigma:

Idiosyncratic noise scale.

type sigma:

float, default=1.0

param sigma_alpha:

Unit effect scale.

type sigma_alpha:

float, default=0.5

param err_hetero:

If True, generate heteroskedastic innovations with observation-specific standard deviations \(\sigma_i = \sigma \sqrt{1 + \|x_{it}\|^2}\) per period.

type err_hetero:

bool, default=False

param rng:

Random state controls.

param seed:

Random state controls.

param W:

Spatial structure input and GeoDataFrame neighbor rule.

param gdf:

Spatial structure input and GeoDataFrame neighbor rule.

param contiguity:

Spatial structure input and GeoDataFrame neighbor rule.

returns:

Includes time-first stacked arrays and panel index columns.

rtype:

dict