# How can I estimate parameters (variance) of a linear model with dual error terms using the log mle?

The log likelihood function is then given by:

lnL(α,β,σ_v,σ_u )=-Tln(σ)+T/2 ln⁡(2⁄π)+∑[lnΦ((s.ε_it λ)/σ)-1/2 (ε_it/σ)^2 ]

Where T is the number of time periods,σ=√(σ_u^2+σ_v^2 ) , λ=σ_u/σ_v.

S = -1 for Undervaluation Model
S = 1 for Overvaluation Model

• I have add pictures of the models for a better view. – Ansto Mar 8 '17 at 18:58