SkillAgentSearch skills...

MPECvsNFXP

The code for the paper "Constrained Optimization Approaches To Estimation Of Structural Models: Comment" by Iskhakov, Lee, Rust, Schjerning and Seo. Econometrica, 2015.

Install / Use

/learn @fediskhakov/MPECvsNFXP
About this skill

Quality Score

0/100

Category

Marketing

Supported Platforms

Universal

README

Code for the Comment for Judd and Su 2012, Econometrica

By: Fedor Iskhakov, John Rust and Bertel Schjerning

September 5, 2015

This directory contains the modified code from Su and Judd Econometrica paper The intention of this directory is to carefully monitor all the changes we make to the original code with thorough documentation of each step in clean GIT repository

type: 'git log' in this directory for the log of changes to the original code

Su and Judd code is downloaded from Che-Lin Su website on April 3, 2014 link: http://faculty.chicagobooth.edu/che-lin.su/research/SuJudd_Code_ECMA.zip

To replicate the initial full set of results in the comment (the published version of the comment contains limited results) do:

Table 1: Run run.m with no changes except perhaps changing the AMPL directory and link to KNITRO library Table 2: Run run.m after adjusting the size of the sample and turning off MPEC/ktrlink Figure 1: Run run_nBus.m with no changes to generate the data obtained for Figure 1. Then run plot_cpu_vs_samplesize.m in nBus/ directory.

The folder nBus/ contains all the code for partial likelihood estimation, which required certain edits of the original Judd-Su code.

The MPEC/AMPL code was updated after Che-Lin Su's recentering of the value function, so that logsum does not produce overflow/underflow errors which improved MPEC ability to converge for beta close to 1.

View on GitHub
GitHub Stars18
CategoryMarketing
Updated1mo ago
Forks22

Languages

Matlab

Security Score

90/100

Audited on Feb 3, 2026

No findings