Identification in binary response panel data models: Is point-identification more common than we thought?

Bo E. Honoré, Aikaterini Kyriazidou

Research output: Contribution to journalArticle

Abstract

This paper investigates identification in binary response models with panel data. Conditioning on sufficient statistics can sometimes lead to a conditional maximum likelihood approach that can be used to identify and estimate the parameters of interest in such models. Unfortunately it is often difficult or impossible to find such sufficient statistics, and even if it is possible, the approach sometimes leads to conditional likelihoods that do not depend on some interesting parameters. Using a range of different data generating processes, this paper calculates the identified regions for parameters in panel data logit AR(2) and logit VAR(1) models for which it is not known whether the parameters are identified or not. We find that identification might be more common than was previously thought, and that the identified regions for non-identified objects may be small enough to be empirically useful.

Original languageEnglish (US)
Pages (from-to)207-226
Number of pages20
JournalAnnals of Economics and Statistics
Issue number134
DOIs
StatePublished - Jan 1 2019

Fingerprint

Binary Response
Panel Data
Data Model
Sufficient Statistics
Logit
Binary Response Model
statistics
Conditional Maximum Likelihood
Conditional Likelihood
conditioning
Conditioning
Calculate
Binary response
Model
Estimate
Range of data
Panel data
Sufficient statistics

Keywords

  • Discrete Choice
  • Fixed Effects
  • Identification
  • Panel Data

ASJC Scopus subject areas

  • Statistics and Probability
  • Social Sciences (miscellaneous)
  • Economics and Econometrics
  • Statistics, Probability and Uncertainty

Cite this

Identification in binary response panel data models : Is point-identification more common than we thought? / Honoré, Bo E.; Kyriazidou, Aikaterini.

In: Annals of Economics and Statistics, No. 134, 01.01.2019, p. 207-226.

Research output: Contribution to journalArticle

@article{f9a6f48e5a8f4401b34e4cc153169ac0,
title = "Identification in binary response panel data models: Is point-identification more common than we thought?",
abstract = "This paper investigates identification in binary response models with panel data. Conditioning on sufficient statistics can sometimes lead to a conditional maximum likelihood approach that can be used to identify and estimate the parameters of interest in such models. Unfortunately it is often difficult or impossible to find such sufficient statistics, and even if it is possible, the approach sometimes leads to conditional likelihoods that do not depend on some interesting parameters. Using a range of different data generating processes, this paper calculates the identified regions for parameters in panel data logit AR(2) and logit VAR(1) models for which it is not known whether the parameters are identified or not. We find that identification might be more common than was previously thought, and that the identified regions for non-identified objects may be small enough to be empirically useful.",
keywords = "Discrete Choice, Fixed Effects, Identification, Panel Data",
author = "Honor{\'e}, {Bo E.} and Aikaterini Kyriazidou",
year = "2019",
month = "1",
day = "1",
doi = "10.15609/annaeconstat2009.134.0207",
language = "English (US)",
pages = "207--226",
journal = "Annals of Economics and Statistics",
issn = "2115-4430",
publisher = "Groupe des ecoles nationales d'economie et statistique",
number = "134",

}

TY - JOUR

T1 - Identification in binary response panel data models

T2 - Is point-identification more common than we thought?

AU - Honoré, Bo E.

AU - Kyriazidou, Aikaterini

PY - 2019/1/1

Y1 - 2019/1/1

N2 - This paper investigates identification in binary response models with panel data. Conditioning on sufficient statistics can sometimes lead to a conditional maximum likelihood approach that can be used to identify and estimate the parameters of interest in such models. Unfortunately it is often difficult or impossible to find such sufficient statistics, and even if it is possible, the approach sometimes leads to conditional likelihoods that do not depend on some interesting parameters. Using a range of different data generating processes, this paper calculates the identified regions for parameters in panel data logit AR(2) and logit VAR(1) models for which it is not known whether the parameters are identified or not. We find that identification might be more common than was previously thought, and that the identified regions for non-identified objects may be small enough to be empirically useful.

AB - This paper investigates identification in binary response models with panel data. Conditioning on sufficient statistics can sometimes lead to a conditional maximum likelihood approach that can be used to identify and estimate the parameters of interest in such models. Unfortunately it is often difficult or impossible to find such sufficient statistics, and even if it is possible, the approach sometimes leads to conditional likelihoods that do not depend on some interesting parameters. Using a range of different data generating processes, this paper calculates the identified regions for parameters in panel data logit AR(2) and logit VAR(1) models for which it is not known whether the parameters are identified or not. We find that identification might be more common than was previously thought, and that the identified regions for non-identified objects may be small enough to be empirically useful.

KW - Discrete Choice

KW - Fixed Effects

KW - Identification

KW - Panel Data

UR - http://www.scopus.com/inward/record.url?scp=85069731898&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069731898&partnerID=8YFLogxK

U2 - 10.15609/annaeconstat2009.134.0207

DO - 10.15609/annaeconstat2009.134.0207

M3 - Article

AN - SCOPUS:85069731898

SP - 207

EP - 226

JO - Annals of Economics and Statistics

JF - Annals of Economics and Statistics

SN - 2115-4430

IS - 134

ER -