\ Consider the nonlinear model\

y_(t)=\\\\theta _(1)^(x_(t))+\\\\theta _(2)^(z_(t))+\\\\epsi _(t)

\ where the sample data

(y_(1),x_(1),z_(1)),dots,(y_(T),x_(T),z_(T))

are i.i.d. and\

E(\\\\epsi _(t)|x_(t),z_(t))=0

. We know that the nonlinear least square estimator is\ asymptotically normal, that is\

\\\\sqrt(T)(hat(\\\\theta )_(NL)-\\\\theta _(0))harr^(d)N(0,A_(0)^(-1)\\\\Omega _(0)A_(0)^(-1))

\ To compute the standard errors we need to estimate

\\\\Omega _(0)

,\

widehat(\\\\Omega )_(0)=[[(1)/(T)\\\\sum_(t=1)^T hat(\\\\epsi )_(t)^(2)x_(t)^(2)hat(\\\\theta )_(1)^(2(x_(t)-1)),(1)/(T)\\\\sum_(t=1)^T hat(\\\\epsi )_(t)^(2)x_(t)hat(\\\\theta )_(1)^(x_(t)-1)z_(t)hat(\\\\theta )_(2)^(z_(t)-1)],[(1)/(T)\\\\sum_(t=1)^T hat(\\\\epsi )_(t)^(2)x_(t)hat(\\\\theta )_(1)^(x_(t)-1)z_(t)hat(\\\\theta )_(2)^(z_(t)-1),]]

\

Consider the nonlinear model yt=1xt+2zt+t where the sample data (y1,x1,z1),,(yT,xT,zT) are i.i.d. and E(txt,zt)=0. We know that the nonlinear least square estimator is asymptotically normal, that is T(^NL0)dN(0,A010A01) To compute the standard errors we need to estimate 0, 0=[T1t=1T^t2xt2^12(xt1)T1t=1T^t2xt^1xt1zt^2zt1T1t=1T^t2xt^1xt1zt^2zt1] What is the missing entry in the matrix 0 ? T1t=1T^t2zt2^22(zt1)T1t=1T^t2^1xtzt2^2(zt1)T1t=1T^12xtzt2^22(zt1)T1t=1T^t2xt2+^12(xt1)^22ztT1t=1T^t2xt^12xt1zt^22zt1