def __init__(self, url, user, token):
self.port = 8000 ## change
if url.find(':') == -1: # Port not included
self.url = url + ':{}'.format(self.port)
else:
self.url = url
self.user = user
self.token = token
self.headers = {'Authorization': '{}'.format(self.token),
'Content-type': 'application/json',
'Accept': 'text/plain'}
shockres = requests.get('http://{}/shock'.format(self.url), headers=self.headers).text
self.shockurl = 'http://{}/'.format(json.loads(shockres)['shockurl'])
self.shock = Shock(self.shockurl, self.user, self.token)
def initialize(self, environment): self.environment = environment self.updater = Updater(self.environment) self.shocker = Shock()
def elast_gen(bs, shocks):
eta = bs[0]
alphap = bs[1]
alphaf = bs[2]
#wage process
wagep_betas = np.array([bs[3], bs[4], bs[5], bs[6], bs[7]]).reshape((5, 1))
income_male_betas = np.array([bs[8], bs[9], bs[10]]).reshape((3, 1))
c_emp_spouse = bs[11]
#Production function [young[cc0,cc1],old]
gamma1 = bs[12]
gamma2 = bs[13]
gamma3 = bs[14]
tfp = bs[15]
sigma2theta = 1
kappas = [bs[16], bs[17]]
sigma_z = [1, bs[18]]
rho_theta_epsilon = bs[19]
lambdas = [1, 1]
#Re-defines the instance with parameters
param = util.Parameters(alphap, alphaf, mu_c, eta, gamma1, gamma2, gamma3,
tfp, sigma2theta, rho_theta_epsilon, wagep_betas,
income_male_betas, c_emp_spouse, marriagep_betas,
kidsp_betas, eitc_list, afdc_list, snap_list, cpi,
fpl_list, lambdas, kappas, pafdc, psnap, mup,
sigma_z)
#The estimate class
output_ins = estimate.Estimate(nperiods, param, x_w, x_m, x_k, x_wmk,
passign, agech0, nkids0, married0, D,
dict_grid, M, N, moments_vector, var_cov,
hours_p, hours_f, wr, cs, ws)
hours = np.zeros(N)
childcare = np.zeros(N)
model_orig = util.Utility(param, N, x_w, x_m, x_k, passign, nkids0,
married0, hours, childcare, agech0, hours_p,
hours_f, wr, cs, ws)
#Obtaining emax instance: this is fixed throughout the exercise
emax_instance = output_ins.emax(param, model_orig)
choices_c = {}
models = []
for j in range(2):
np.save(
'/home/jrodriguez/NH_HC/results/model_v2/experiments/NH/shock.npy',
shocks[j])
models.append(
Shock(param, N, x_w, x_m, x_k, passign, nkids0, married0, hours,
childcare, agech0, hours_p, hours_f, wr, cs, ws))
choices_c['Choice_' + str(j)] = output_ins.samples(
param, emax_instance, models[j])
#Computing changes in % employment for control group
h_sim_matrix = []
employment = []
wages = []
full = []
for j in range(2):
h_sim_matrix.append(choices_c['Choice_' + str(j)]['hours_matrix'])
employment.append(choices_c['Choice_' + str(j)]['hours_matrix'] > 0)
full.append(choices_c['Choice_' + str(j)]['hours_matrix'] == hours_f)
wages.append(choices_c['Choice_' + str(j)]['wage_matrix'])
#Extensive margin
elast_extensive = np.zeros(M)
for j in range(M):
elast_periods = np.zeros(nperiods)
for t in range(nperiods):
elast_periods[t] = (np.mean(employment[1][:, t, j], axis=0) -
np.mean(employment[0][:, t, j], axis=0)) / (
shocks[1] * np.mean(
(employment[0][:, t, j]), axis=0))
elast_extensive[j] = np.mean(elast_periods)
#Intensive margin
elast_intensive = np.zeros(M)
for j in range(M):
elast_periods = np.zeros(nperiods)
for t in range(nperiods):
sample = (employment[0][:, t, j] == 1)
elast_periods[t] = np.mean(
(h_sim_matrix[1][sample, t, j] -
h_sim_matrix[0][sample, t, j]),
axis=0) / (shocks[1] *
np.mean(h_sim_matrix[0][sample, t, j], axis=0))
elast_intensive[j] = np.mean(elast_periods)
return {'Extensive': elast_extensive, 'Intensive': elast_intensive}
def init_shock(self):
if self.shock is None:
shockres = self.req_get('{}/shock'.format(self.url))
self.shockurl = utils.verify_url(json.loads(shockres)['shockurl'])
self.shock = Shock(self.shockurl, self.user, self.token)
