Jonathan James, "An MM Algorithm for General Mixed Multinomial Logit 
Models", Journal of Applied Econometrics, Vol. 32, No. 4, 2017, pp. 
841-857.

The programs described below are for estimating a normally distributed
random coefficients mixed logit model applying the MM algorithm. For
comparison, programs for alternative estimation methods are also
provided.

All programs can be run from the file 'main.m', which relies on the 
following functions:

'datagen.m' is a function that generates simulated data for estimation

'standardlogit' is a function that estimates the parameters of a 
standard logit model, which are used for starting values for each of the 
estimators

'vecparm.m' is a function that produces a function handle called 'p2p'. 
p2p either vectorizes the structure of parameters if the input is a 
structure, or structurizes the vector of parameters if the input is a 
vector.

'def_MM.m' is a function that produces a function handle called 
'MMstep', which defines the minorization-maximization step.  The input 
to 'MMstep' a structure of parameters, and the output is the updated 
structure of parameters.

The function 'MMstep' calls the function 'lowerbound_inverse.m' which 
creates the inverse lower bound matrix for updating the fixed 
coefficients.

The parameters are estimated by iterating on the function 'MMstep' until 
convergence.

'em_alg_eps.m is a function that iterates on the function handle
'MMstep' until the parameters converge using the vector epsilon
algorithm in Kuroda and Sakakihara (2006)

'def_EMqn.m' defines the EM step for the EM algorithm using quasi-newton 
for the inner optimization

'def_EMnr.m' defines the EM step for the EM algorithm using 
newton-raphson for the inner optimization

'mslAgrad.m' estimates the parameters using maximum simulated likelihood 
with analytical gradients

'mslNgrad.m' estimates the parameters using maximum simulated likelihood 
with numerical gradients

All of the .m files described above are ASCII files in DOS format.
They are zipped in the file jj-files.zip. Unix/Linux users should use
"unzip -a".