def cov(self) -> NDArray:
"""Estimated covariance"""
x = self._x
nobs = x.shape[0]
xpxi = inv(x.T @ x / nobs)
eps = self.eps
xe = x * eps
if self._clusters.ndim == 1:
xeex = cov_cluster(xe, self._clusters)
if self._group_debias:
xeex *= group_debias_coefficient(self._clusters)
else:
clusters0 = self._clusters[:, 0]
clusters1 = self._clusters[:, 1]
xeex0 = cov_cluster(xe, clusters0)
xeex1 = cov_cluster(xe, clusters1)
clusters01 = cluster_union(self._clusters)
xeex01 = cov_cluster(xe, clusters01)
if self._group_debias:
xeex0 *= group_debias_coefficient(clusters0)
xeex1 *= group_debias_coefficient(clusters1)
xeex01 *= group_debias_coefficient(clusters01)
xeex = xeex0 + xeex1 - xeex01
xeex *= self._scale
out = (xpxi @ xeex @ xpxi) / nobs
return (out + out.T) / 2
def _xeex(self) -> NDArray:
if self._clusters.shape[1] == 0:
# Heteroskedastic but not clustered
return super()._xeex()
elif self._clusters.shape[1] == 1:
s = cov_cluster(self._moments, self._clusters[:, 0])
if self._group_debias:
s *= group_debias_coefficient(self._clusters[:, 0])
return s
else:
xeex0 = cov_cluster(self._moments, self._clusters[:, 0])
xeex1 = cov_cluster(self._moments, self._clusters[:, 1])
clusters01 = cluster_union(self._clusters)
xeex01 = cov_cluster(self._moments, clusters01)
if self._group_debias:
xeex0 *= group_debias_coefficient(self._clusters[:, 0])
xeex1 *= group_debias_coefficient(self._clusters[:, 1])
xeex01 *= group_debias_coefficient(clusters01)
return xeex0 + xeex1 - xeex01