Generalized Method of Moments Estimator

class mygmm.gmm.GMM(momcond)

GMM estimation class.

Attributes

momcond

Moment function

Methods

gmmest

__init__(momcond)

Initialize the class.

Parameters:

momcond : function

Moment function. Should return:

• array (nobs x nmoms)

  moment function values

• (optionally) array (nmoms x nparams)

  derivative of moment function average across observations.

gmmest(theta_start, bounds=None, constraints=(), iter=2, method='BFGS', kernel='Bartlett', band=None, names=None, **kwargs)

Multiple step GMM estimation procedure.

Parameters:

theta_start : array

Initial parameters

bounds : list of tuples

Bounds on parameters

constraints : dict or sequence of dict

Equality and inequality constraints. See scipy.optimize.minimize

iter : int

Number of GMM steps

method : str

Optimization method

kernel : str

Type of kernel for HAC. Currenly implemented: SU, Bartlett, Parzen, Quadratic

band : int

Truncation parameter for HAC

names : list of str

Parameter names

Returns:

instance of Results

Estimation results

class mygmm.results.Results(opt_out=None, var_theta=None, nmoms=None, names=None)

Class to hold estimation results.

Attributes

theta	Parameter estimate
degf	Degrees of freedom
jstat	J-statistic
stde	Standard errors
tstat	t-statistics
jpval	p-value of the J test
opt	Optimization output