Question 2: Simulation and Estimation
Assume the true DGP of a random variable y is:
y = α+ βx1 + γx2 +ε (1)
where α = 3, β = 0:5, γ= -0.7 and ε~ N(0; 1).
a. Generate series for the error term and the exogenous variables x1 and x2, all from N(0,1) with
sample size T = 200. Then, estimate the parameters of the model above by OLS (you can use
an appropriate Matlab built-in function or write your own routine { or compare both).
b. Assume now that regressors are highly correlated. For instance, impose Ρx1,x2 = -0:9; and
generate new series with this property (keeping σxi ^2= 1). Then re-estimate the parameters.
c. Simulate N = 100 times model (1) under a. and b. assumptions, estimate each simulated
model by OLS and collect the N estimates. Check consistency and eciency.