groupedpaneldatamodels.models.su_ju

Functions

interactive_effects_estimation(y, x, N, T, ...)

Internal function to estimate the interactive effects model as described by Su and Ju (2018).

groupedpaneldatamodels.models.su_ju.interactive_effects_estimation(y: ndarray, x: ndarray, N: int, T: int, K: int, G: int, R: int, max_iter: int = 1000, only_bfgs: bool = True, tol: float = 1e-06, kappa: float = 0.1)[source]

Internal function to estimate the interactive effects model as described by Su and Ju (2018).

Parameters:
  • y (np.ndarray) – Dependent variable, shape (N, T, 1).

  • x (np.ndarray) – Explanatory variables, shape (N, T, K).

  • N (int) – Number of individuals (cross-sectional units).

  • T (int) – Number of time periods.

  • K (int) – Number of explanatory variables.

  • G (int) – Number of groups.

  • R (int) – Number of common factors (global).

  • max_iter (int, optional) – Maximum number of acceptable iterations, may be too large. Defaults to 1000.

  • only_bfgs (bool, optional) – Only uses BFGS. Defaults to True.

  • tol (float, optional) – Acceptable tolerance for the stopping condition. Defaults to 1e-6.

  • kappa (float, optional) – Kappa penalty parameter. Defaults to 0.1.

Returns:

  • beta: Estimated coefficients for each individual, shape (K, N)

  • alpha: Ordered group-level representative coefficients, shape (K, G)

  • lambdas: Factor loadings, shape (R, N)

  • factors: Common factors, shape (T, R)

  • resid: Residuals of the model, shape (N, T)

Return type:

tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray, np.ndarray]