Dear all,

I was wondering how the initial observation y_i,0 should look like in the following AR(1) model with covariate(s):

y_i,t = p*y_i,t-1 + b*X_i,t + a_i + e_i,t

where e_i,t is iid N(0,var_e), a_i is constructed following Chamberlain's approach (y_i,0 + linear projection of X_i,t in all time periods,...) and 'p' has an absolute value below 1?

Can I simply set it equal to 0?

Thank you in advance for an asnwer,
Jordi