New matrix-based methods for the analytic evaluation of the multivariate cumulative normal distribution function

Abstract In this paper, we develop a new matrix-based implementation of the Mendell and Elston (ME) analytic approximation to evaluate the multivariate normal cumulative distribution (MVNCD) function, using an LDLT decomposition method followed by a rank 1 update of the LDLT factorization. Our implementation is easy to code for individuals familiar with matrix-based coding. Further, our new matrix-based implementation for the ME algorithm allows us to efficiently write the analytic matrix-based gradients of the approximated MVNCD function with respect to the abscissae and correlation parameters, an issue that is important in econometric model estimation. In addition, we propose four new analytic methods for approximating the MVNCD function. The paper then evaluates the ability of the multiple approximations for individual MVNCD evaluations as well as multinomial probit model estimation. As expected, in our tests for evaluating individual MVNCD functions, we found that the traditional GHK approach degrades rapidly as the dimensionality of integration increases. Concomitant with this degradation in accuracy is a rapid increase in computational time. The analytic approximation methods are also much more stable across different numbers of dimensions of integration, and even the simplest of these methods is superior to the GHK-500 beyond seven dimensions of integration. Based on all the evaluation results in this paper, we recommend the new Two-Variate Bivariate Screening (TVBS) method proposed in this paper as the evaluation approach for MVNCD function evaluation.

[1]  C. Bhat.  Simulation estimation of mixed discrete choice models using randomized and scrambled Halton sequences , 2003 .

[2]  Rajesh Paleti,et al.  A spatial generalized ordered response model to examine highway crash injury severity. , 2013, Accident; analysis and prevention.

[3]  N. L. Johnson,et al.  Continuous Multivariate Distributions, Volume 1: Models and Applications , 2019 .

[4]  Hea-Jung Kim,et al.  A class of rectangle-screened multivariate normal distributions and its applications , 2015 .

[5]  Peter E. Rossi,et al.  A Bayesian analysis of the multinomial probit model with fully identified parameters , 2000 .

[6]  Michael G. Langdon,et al.  Improved Algorithms for Estimating Choice Probabilities in the Multinomial Probit Model , 1984, Transp. Sci..

[7]  Andrew R. Solow,et al.  A method for approximating multivariate normal orthant probabilities , 1990 .

[8]  Gene H. Golub,et al.  Matrix computations , 1983 .

[9]  Chandra R. Bhat.  A new spatial (social) interaction discrete choice model accommodating for unobserved effects due to endogenous network formation , 2015 .

[10]  Matthew M. Hutmacher,et al.  Extending the latent variable model for extra correlated longitudinal dichotomous responses , 2011, Journal of Pharmacokinetics and Pharmacodynamics.

[11]  Alan Genz,et al.  Bivariate conditioning approximations for multivariate normal probabilities , 2015, Stat. Comput..

[12]  A. Daly,et al.  Analytic approximations for computing probit choice probabilities , 2014 .

[13]  Chandra R. Bhat,et al.  A simulation evaluation of the maximum approximate composite marginal likelihood (MACML) estimator for mixed multinomial probit models , 2011 .

[14]  Chandra R. Bhat,et al.  A New Spatial Multiple Discrete‐Continuous Modeling Approach to Land Use Change Analysis , 2013 .

[15]  Chandra R. Bhat,et al.  A spatial generalized ordered-response model with skew normal kernel error terms with an application to bicycling frequency , 2017 .

[16]  D. McFadden,et al.  The method of simulated scores for the estimation of LDV models , 1998 .

[17]  Chandra R. Bhat,et al.  On Accommodating Spatial Dependence in Bicycle and Pedestrian Injury Counts by Severity Level , 2013 .

[18]  Chandra R. Bhat,et al.  A latent variable representation of count data models to accommodate spatial and temporal dependence: application to predicting crash frequency at intersections , 2011 .

[19]  Chandra R. Bhat,et al.  Simulation evaluation of emerging estimation techniques for multinomial probit models , 2017 .

[20]  Chandra R. Bhat,et al.  A spatial panel ordered-response model with application to the analysis of urban land-use development intensity patterns , 2013, J. Geogr. Syst..

[21]  C. Bhat.  Quasi-random maximum simulated likelihood estimation of the mixed multinomial logit model , 2001 .

[22]  R C Elston,et al.  Multifactorial qualitative traits: genetic analysis and prediction of recurrence risks. , 1974, Biometrics.

[23]  Wagner A. Kamakura,et al.  The Estimation of Multinomial Probit Models: A New Calibration Algorithm , 1989, Transp. Sci..

[24]  Michael P. Keane,et al.  Four essays in empirical macro and labor economics , 1990 .

[25]  Michael Keane,et al.  A Computationally Practical Simulation Estimator for Panel Data , 1994 .

[26]  P. Switzer,et al.  Estimation of spatial distributions from point sources with application to air pollution measurement. Technical report No. 9 , 1977 .

[27]  Karl Pearson,et al.  Mathematical contributions to the theory of evolution. XI.—On the influence of natural selection on the variability and correlation of organs , 1902, Proceedings of the Royal Society of London.

[28]  Dietmar Bauer,et al.  New results on the asymptotic and finite sample properties of the MaCML approach to multinomial probit model estimation , 2016, 1609.03295.

[29]  Zhicai Juan,et al.  Estimation of multinomial probit-kernel integrated choice and latent variable model: comparison on one sequential and two simultaneous approaches , 2017 .

[30]  Chandra R. Bhat,et al.  The maximum approximate composite marginal likelihood (MACML) estimation of multinomial probit-based unordered response choice models , 2011 .

[31]  G. Géniaux,et al.  Approximate likelihood estimation of spatial probit models , 2017 .

[32]  Chandra R. Bhat,et al.  Incorporating Spatial Dynamics and Temporal Dependency in Land Use Change Models , 2011 .

[33]  H. Joe.  Approximations to Multivariate Normal Rectangle Probabilities Based on Conditional Expectations , 1995 .

[34]  John Geweke,et al.  Efficient Simulation from the Multivariate Normal and Student-t Distributions Subject to Linear Constraints and the Evaluation of Constraint Probabilities , 1991 .

[35]  Florian Heiss,et al.  The panel probit model: Adaptive integration on sparse grids , 2010 .

[36]  D. V. Dyk,et al.  A Bayesian analysis of the multinomial probit model using marginal data augmentation , 2005 .

[37]  C. E. Clark.  The Greatest of a Finite Set of Random Variables , 1961 .

[38]  Geoffrey J. McLachlan,et al.  On mixtures of skew normal and skew $$t$$-distributions , 2012, Adv. Data Anal. Classif..

[39]  Manjunath B. G.,et al.  Moments Calculation for the Double Truncated Multivariate Normal Density , 2009, Journal of Behavioral Data Science.

[40]  A. C. Aitken,et al.  Note on Selection from a Multivariate Normal Population , 1935 .

[41]  R. Beaver,et al.  The nontruncated marginal of a truncated bivariate normal distribution , 1993 .

[42]  John P. Rice,et al.  An Approximation to the Multivariate Normal Integral: Its Application to Multifactorial Qualitative Traits , 1979 .

[43]  Rajesh Paleti,et al.  A Spatial Multivariate Count Model for Firm Location Decisions , 2014 .

[44]  C. Varin,et al.  A comparison of the maximum simulated likelihood and composite marginal likelihood estimation approaches in the context of the multivariate ordered-response model , 2010 .

[45]  Florian Heiss,et al.  Likelihood approximation by numerical integration on sparse grids , 2008 .

[46]  Raymond Kan,et al.  On Moments of Folded and Truncated Multivariate Normal Distributions , 2016 .