Simulation Methods for Probit and Related Models Based on Convenient Error Partitioning

Kenneth Train ()

No 95-237, Economics Working Papers from University of California at Berkeley

Abstract: Two probit simulators are described that are conceptually and computationally simple. The first is based on simulating the utilities of the non-chosen alternatives and calculating the probability that the chosen alternative's utility exceeds this maximum. This simulator is apparently new. The second, which is implicit in the discussions of McFadden (1989) and Bolduc (1992), is applicable when the covariance among utilities arises from random parameters and/or error components that are common across alternatives. The parameters and common error components are simulated, and then the probability that the observed event occurs is calculated conditional on these values. Both simulators are unbiased, strictly positive, and continuous. The second is twice-differentiable, while the first has points of non-differentiability. Both are easy to program and can be expected to be very fast computationa- lly.

Date: 1995-06-01

Downloads: (external link)
http://econwpa.wustl.edu/eprints/em/papers/9605/9605001.abs link to EconWPA (text/html)

Related works:
Working Paper: Simulation Methods for Probit and Related Models Based on Convenient Error Partitioning (1996)
Working Paper: Simulation Methods for Probit and Related Models Based on Convenient Error Partitioning
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: http://EconPapers.repec.org/RePEc:ucb:calbwp:95-237

Ordering information: This working paper can be ordered from
IBER, F502 Haas Building, University of California, Berkeley CA 94720-1922

Access Statistics for this paper

More papers in Economics Working Papers from University of California at Berkeley
Address: University of California at Berkeley, Berkeley, CA USA
Contact information at EDIRC.
Series data maintained by Christopher F. Baum ().