Advertisement

The order of variables, simulation noise, and accuracy of mixed logit estimates

  • Marco A. PalmaEmail author
  • Dmitry V. Vedenov
  • David Bessler
Article
  • 22 Downloads

Abstract

The simulated choice probabilities in mixed logit models are usually approximated numerically using Halton or random draws from a multivariate mixing distribution for the random parameters. Theoretically, the order in which the estimated variables enter the model should not matter. However, in practice, simulation “noise” inherent in the numerical procedure leads to differences in the magnitude of the estimated coefficients depending on the arbitrary order in which the random variables are estimated. The problem is exacerbated when a low number of draws are used or if correlation among coefficients is allowed. In particular, the Cholesky factorization procedure, which is used to incorporate correlation into the model, propagates simulation noise in the estimate of one coefficient to estimates of all subsequent coefficients in the model. Ignoring the potential ordering effects in simulated maximum likelihood estimation methods may seriously compromise the ability of replicating the results and can inadvertently influence policy recommendations. We find that better estimation accuracy is achieved with Halton draws using small prime numbers as it is the case for small integrating dimensions; but random draws provide better accuracy than Halton draws from large prime numbers as it is normally the case in high integrating dimensions. With correlation, the standard deviations have very large fluctuations depending on the order of the variables, affecting the conclusions regarding heterogeneity of preferences.

Keywords

Cholesky Halton draws Maximum simulated likelihood Random draws Simulated noise 

JEL codes

C25 C63 

References

  1. Bhat CR (2001) Quasi-random maximum simulated likelihood estimation of the mixed multinomial logit model. Transp Res Part B Methodol 35(7):677–693.  https://doi.org/10.1016/S0191-2615(00)00014-X CrossRefGoogle Scholar
  2. Bhat CR (2003) Simulation estimation of mixed discrete choice models using randomized and scrambled Halton sequences. Transp Res Part B Methodol 37(9):837–855.  https://doi.org/10.1016/S0191-2615(02)00090-5 CrossRefGoogle Scholar
  3. Calfee J, Winston C, Stempski R (2001) econometric issues in estimating consumer preferences from stated preference data: a case study of the value of automobile travel time. Rev Econ Stat 83(4):699–707.  https://doi.org/10.1162/003465301753237777 CrossRefGoogle Scholar
  4. Cappellari L, Jenkins SP (2006) Calculation of multivariate normal probabilities by simulation, with applications to maximum simulated likelihood estimation. Stata J 6(2):156–189Google Scholar
  5. Chang JB, Lusk JL (2011) Mixed logit models: accuracy and software choice. J Appl Econ 26(1):167–172.  https://doi.org/10.1002/jae.1201 CrossRefGoogle Scholar
  6. Croissant Y (2018) mlogit: multinomial logit model. R Package. https://CRAN.R-project.org/package=mlogit
  7. Drukker DM, Gates R (2006) Generating Halton sequences using Mata. Stata J 6(2):214–228Google Scholar
  8. Geweke J, Keane M, Runkle D (1994) Alternative computational approaches to inference in the multinomial probit model. Rev Econ Stat 76(4):609–632.  https://doi.org/10.2307/2109766 CrossRefGoogle Scholar
  9. Greene WH (2012) Econometric analysis. Prentice Hall, BostonGoogle Scholar
  10. Hensher DA, Greene WH (2003) the mixed logit model: the state of practice. Transportation 30(2):133–176.  https://doi.org/10.1023/A:1022558715350 CrossRefGoogle Scholar
  11. Hess S, Rose JM (2009) Allowing for intra-respondent variations in coefficients estimated on repeated choice data. Transp Res Part B Methodol 43(6):708–719.  https://doi.org/10.1016/j.trb.2009.01.007 CrossRefGoogle Scholar
  12. Hess S, Train KE, Polak JW (2006) On the use of a modified latin hypercube sampling (MLHS) method in the estimation of a mixed logit model for vehicle choice. Transp Res Part B Methodol 40(2):147–163.  https://doi.org/10.1016/j.trb.2004.10.005 CrossRefGoogle Scholar
  13. Hole AR (2007) Estimating mixed logit models using maximum simulated likelihood. Stata J 7(3):388–401Google Scholar
  14. Koop G, Pesaran MH, Potter Simon M (1996) Impulse response analysis in nonlinear multivariate models. J Econ 74(1):119–147.  https://doi.org/10.1016/0304-4076(95)01753-4 CrossRefGoogle Scholar
  15. McFadden D, Ruud PA (1994) Estimation by simulation. Rev Econ Stat 76(4):591–608.  https://doi.org/10.2307/2109765 CrossRefGoogle Scholar
  16. McFadden D, Train K (2000) Mixed MNL models for discrete response. J Appl Econ 15(5):447–470CrossRefGoogle Scholar
  17. Pesaran HH, Shin Y (1998) Generalized impulse response analysis in linear multivariate models. Econ Lett 58(1):17–29.  https://doi.org/10.1016/S0165-1765(97)00214-0 CrossRefGoogle Scholar
  18. Revelt D, Train K (1998) Mixed logit with repeated choices: households’ choices of appliance efficiency level. Rev Econ Stat 80(4):647–657.  https://doi.org/10.1162/003465398557735 CrossRefGoogle Scholar
  19. Sivakumar A, Bhat C, Ökten G (2005) Simulation estimation of mixed discrete choice models with the use of randomized quasi-monte carlo sequences: a comparative study. Transp Res Rec J Transp Res Board 1921:112–122.  https://doi.org/10.3141/1921-13 CrossRefGoogle Scholar
  20. Train K (2000) Halton sequences for mixed logit. Department of Economics, UCB, BerkeleyGoogle Scholar
  21. Train Kenneth E (2009) Discrete choice methods with simulation. Cambridge University Press, CambridgeCrossRefGoogle Scholar

Copyright information

© Springer-Verlag GmbH Germany, part of Springer Nature 2018

Authors and Affiliations

  • Marco A. Palma
    • 1
    Email author
  • Dmitry V. Vedenov
    • 1
  • David Bessler
    • 1
  1. 1.Department of Agricultural EconomicsTexas A&M UniversityCollege StationUSA

Personalised recommendations