v__0.150

mec/mx.py

@@ -0,0 +1,24 @@
+# Econometrics library
+
+import numpy as np
+import scipy.sparse as sp
+
+
+def iv_gmm(Y_i,X_i_k,Z_i_l, efficient=False, centering = True):
+    def beta_gmm(Y_i,X_i_k,Z_i_l,W_l_l ):
+        ZtildeT_k_i = X_i_k.T @ Z_i_l @ W_l_l @ Z_i_l.T
+        return np.linalg.solve(ZtildeT_k_i @ X_i_k,ZtildeT_k_i @ Y_i)
+    I=len(Y_i)
+    W_l_l = np.linalg.inv( Z_i_l.T @ Z_i_l / I)
+    beta_k = beta_gmm(Y_i,X_i_k,Z_i_l,W_l_l ) # first stage obtained by 2SLS
+    if efficient:
+        epsilon_i = Y_i - X_i_k @ beta_k
+        mhat_l_i = Z_i_l.T * epsilon_i[None,:]
+        mbar_l = mhat_l_i.mean(axis=1)
+        Sigmahat_l_l = (mhat_l_i @ mhat_l_i.T) / I - centering * mbar_l[:,None] * mbar_l[None,:]
+        W_l_l =  np.linalg.inv(Sigmahat_l_l)
+        beta_k = beta_gmm(Y_i,X_i_k,Z_i_l,W_l_l )
+    Pi_i_i = Z_i_l @ W_l_l @ Z_i_l.T
+    XPiY_k_i = X_i_k.T @ Pi_i_i @ Y_i
+    objval = (Y_i.T @ Pi_i_i @ Y_i - XPiY_k_i.T @ np.linalg.inv(  X_i_k.T @ Pi_i_i @ X_i_k ) @ XPiY_k_i )/ (2*I*I)
+    return beta_k,objval

setup.py

@@ -2,7 +2,7 @@
 
 setup(
 	name="mec",
-	version="0.149",
+	version="0.150",
 	authors=["Alfred Galichon"],
 	author_email="[email protected]",
 	licence="",