Appendix E
Type-1 Tobit Model
(Source: Franses and Paap [1])
For a type-1 Tobit model, the censored variable is 0 if the unobserved latent variable is less than or equal to 0 and if is positive,
(E.1)
(E.2)
with . For observations that are 0, we know only that
Maximum likelihood estimation is used to estimate the Tobit model. The likelihood function consists of two parts: the probability that an observation is censored is given by Equation E.3; and the density of the non-censored observations is a standard normal density. The likelihood function is
(E.4)
where . It is more convenient to reparameterize the model according to and . The log-likelihood function in terms of reads
(E.5)
The first-order derivatives of the log-likelihood function with respect to and are
(E.6)
(E.7)
and the second-order derivatives are
(E.8)
(E.9)
(E.10)