Monte Carlo Simulation IV/OLS estimator

[hxh]

Hi all,

I made a Monte Carlo Simulation to compare the properties of OLS and IV in a finite sample (n=200).

I have a question to one result I got.

Simple 1 Regressor case

Y_i=X_i * beta + e_i

When choosing the correlation of X and e to be 0.7, the variance of the OLS estimator has decreased compared to the case where there is no correlation.

How can this be explained ?

Bye Bye
greetz.