def setup_class(self):
XLISTEXOG2 = 'aget aget2 educyr actlim totchr'.split()
endog_name = 'docvis'
exog_names = 'private medicaid'.split() + XLISTEXOG2 + ['const']
instrument_names = 'income ssiratio'.split() + XLISTEXOG2 + ['const']
endog = DATA[endog_name]
exog = DATA[exog_names]
instrument = DATA[instrument_names]
asarray = lambda x: np.asarray(x, float)
endog, exog, instrument = map(asarray, [endog, exog, instrument])
self.bse_tol = [5e-6, 5e-7]
# compare to Stata default options, iterative GMM
# with const at end
start = OLS(endog, exog).fit().params
nobs, k_instr = instrument.shape
w0inv = np.dot(instrument.T, instrument) / nobs
mod = gmm.NonlinearIVGMM(endog, exog, instrument,
moment_exponential_add)
res0 = mod.fit(start,
maxiter=2,
inv_weights=w0inv,
optim_method='bfgs',
optim_args={'gtol': 1e-8},
wargs={'centered': False},
has_optimal_weights=False)
self.res1 = res0
from results_gmm_poisson import results_addtwostep as results
self.res2 = results
def setup_class(cls):
# compare to Stata default options, twostep GMM
XLISTEXOG2 = 'aget aget2 educyr actlim totchr'.split()
endog_name = 'docvis'
exog_names = 'private medicaid'.split() + XLISTEXOG2 + ['const']
instrument_names = 'income medicaid ssiratio'.split() + XLISTEXOG2 + ['const']
endog = DATA[endog_name]
exog = DATA[exog_names]
instrument = DATA[instrument_names]
asarray = lambda x: np.asarray(x, float)
endog, exog, instrument = lmap(asarray, [endog, exog, instrument])
# Need to add all data into exog
endog_ = np.zeros(len(endog))
exog_ = np.column_stack((endog, exog))
cls.bse_tol = [5e-6, 5e-7]
# compare to Stata default options, iterative GMM
# with const at end
start = OLS(endog, exog).fit().params
nobs, k_instr = instrument.shape
w0inv = np.dot(instrument.T, instrument) / nobs
mod = gmm.NonlinearIVGMM(endog_, exog_, instrument, moment_exponential_mult)
res0 = mod.fit(start, maxiter=2, inv_weights=w0inv,
optim_method='bfgs', optim_args={'gtol':1e-8, 'disp': 0},
wargs={'centered':False}, has_optimal_weights=False)
cls.res1 = res0
from .results_gmm_poisson import results_multtwostep as results
cls.res2 = results
def setup_class(cls):
# compare to Stata default options, onestep GMM
# this uses maxiter=1, one iteration in loop
cls.params_tol = [5e-5, 5e-6]
cls.bse_tol = [5e-6, 1e-1]
exog = exog_st # with const at end
start = OLS(endog, exog).fit().params
nobs, k_instr = instrument.shape
w0inv = np.dot(instrument.T, instrument) / nobs
#w0 = np.linalg.inv(w0inv)
def func(params, exog):
return np.dot(exog, params)
mod = gmm.NonlinearIVGMM(endog, exog, instrument, func)
res = mod.fit(start, maxiter=1, inv_weights=w0inv,
optim_method='bfgs', optim_args={'gtol':1e-8, 'disp': 0},
wargs={'centered':False}, has_optimal_weights=False)
cls.res1 = res
mod = gmm.IVGMM(endog, exog, instrument)
res = mod.fit(start, maxiter=1, inv_weights=w0inv,
optim_method='bfgs', optim_args={'gtol':1e-6, 'disp': 0},
wargs={'centered':False}, has_optimal_weights=False)
cls.res3 = res
from .results_gmm_griliches import results_onestep as results
cls.res2 = results
# Currently statsmodels has two ways of specifying moment conditions.
# The first uses general non-linear functions for the (unconditional) moment condition
# The second version uses an instrumental variables approach with additive error structure
def moment_consumption1(params, exog):
beta, gamma = params
r_forw1, c_forw1, c = exog.T # unwrap iterable (ndarray)
# moment condition without instrument
err = 1 - beta * (1 + r_forw1) * np.power(c_forw1 / c, -gamma)
return -err
endog1 = np.zeros(exog.shape[0])
mod1 = gmm.NonlinearIVGMM(endog1, exog, instrument, moment_consumption1, k_moms=4)
w0inv = np.dot(instrument.T, instrument) / len(endog1)
res1 = mod1.fit([1,-1], maxiter=2, inv_weights=w0inv)
print(res1.summary(yname='Euler Eq', xname=['discount', 'CRRA']))
# We don't need Nelder-Mead in this case, we can use bfgs default directly
# res1_ = mod1.fit([1,-1], maxiter=0, inv_weights=w0inv, opt_method='nm')
res1_hac4_2s = mod1.fit([1, -1], maxiter=2, inv_weights=w0inv, weights_method='hac', wargs={'maxlag':4})
print(res1_hac4_2s.summary(yname='Euler Eq', xname=['discount', 'CRRA']))
def moment_consumption2(params, exog):
beta, gamma = params
