Dear all,

I'm trying to estimate a nested logit myself. Due to the computational burden of the estimation process of the FIML, I'd like to estimate the nested logit model by estimating two sequential logits (LIML) as described in Greene (2002) p.729 onwards or Train(2002) p.97. I couldn't find a handy example neither with or without stata code.

Could someone explain the steps that are needed in order to estimate a model similar to the one of the example (restaurant) of statas nlogit command?

webuse restaurant

nlogitgen type = restaurant(fast: Freebirds | MamasPizza, family: CafeEccell
| LosNortenos | WingsNmore, fancy: Christophers | MadCows)

nlogit chosen cost distance rating || type: income kids, base(family) ||
restaurant:, noconst case(family_id)


RUM-consistent nested logit regression Number of obs = 2100
Case variable: family_id Number of cases = 300

Alternative variable: restaurant Alts per case: min = 7
avg = 7.0
max = 7

Wald chi2(7) = 46.71
Log likelihood = -485.47331 Prob > chi2 = 0.0000

------------------------------------------------------------------------------
chosen | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
restaurant |
cost | -.1843847 .0933975 -1.97 0.048 -.3674404 -.0013289
distance | -.3797474 .1003828 -3.78 0.000 -.5764941 -.1830007
rating | .463694 .3264935 1.42 0.156 -.1762215 1.10361
------------------------------------------------------------------------------
type equations
------------------------------------------------------------------------------
fast |
income | -.0266038 .0117306 -2.27 0.023 -.0495952 -.0036123
kids | -.0872584 .1385026 -0.63 0.529 -.3587184 .1842016
-------------+----------------------------------------------------------------
family |
income | 0 (base)
kids | 0 (base)
-------------+----------------------------------------------------------------
fancy |
income | .0461827 .0090936 5.08 0.000 .0283595 .0640059
kids | -.3959413 .1220356 -3.24 0.001 -.6351267 -.1567559
------------------------------------------------------------------------------
dissimilarity parameters
------------------------------------------------------------------------------
type |
/fast_tau | 1.712878 1.48685 -1.201295 4.627051
/family_tau | 2.505113 .9646351 .614463 4.395763
/fancy_tau | 4.099844 2.810123 -1.407896 9.607583
------------------------------------------------------------------------------
LR test for IIA (tau = 1): chi2(3) = 6.87 Prob > chi2 = 0.0762
------------------------------------------------------------------------------


1.) How to set up the two logit specifications?
2.) How to calculate the Inclusive values?


Best
Julian