The Student Room Group

Implications of contemporaneous exogeneity

For some reason I don't feel confident, but in the context of time-series econometrics, does:
Unparseable latex formula:

E[u_t|x_t]=0 \forall t\]



Where ut u_t is the error term, and xtx_t the matrix of all xjtx_{jt} for all explanatory variables j at time t.

Imply all of the following, via the law of iterated expectations?:
E[ut]=0E[u_t]=0
Cov[ut,xjt]=0j,tCov[u_t,x_{jt}]=0 \forall j,t
E[utxjt]=0j,tE[u_t x_{jt}]=0 \forall j,t
Reply 1
Original post by DQd
For some reason I don't feel confident, but in the context of time-series econometrics, does:
Unparseable latex formula:

E[u_t|x_t]=0 \forall t\]



Where ut u_t is the error term, and xtx_t the matrix of all xjtx_{jt} for all explanatory variables j at time t.

Imply all of the following, via the law of iterated expectations?:
E[ut]=0E[u_t]=0
Cov[ut,xjt]=0j,tCov[u_t,x_{jt}]=0 \forall j,t
E[utxjt]=0j,tE[u_t x_{jt}]=0 \forall j,t


Sounds about right, but it would be good to see your reasoning to check. The different properties seem to be saying
* First one says there is no "constant term" in the error left to model.
* Second says in a linear correlation sense, there is no dependency between the error and the regressors
* Third says the error and regressor signals are orthogonal

Quick Reply

Latest