Simulated maximum likelihood estimator for the random coefficient logit model using aggregate data

Sungho Park, Sachin Gupta

Research output: Contribution to journalArticle

33 Citations (Scopus)

Abstract

The authors propose a simulated maximum likelihood estimation method for the random coefficient logit model using aggregate data, accounting for heterogeneity and endogeneity. The method allows for two sources of randomness in observed market shares: unobserved product characteristics and sampling error. Because of the latter, the method is suitable when sample sizes underlying the shares are finite. In contrast, Berry, Levinsohn and Pakes's commonly used approach assumes that observed shares have no sampling error. The method can be viewed as a generalization of Villas-Boas and Winer's approach and is closely related to Petrin and Train's "control function" approach. The authors show that the proposed method provides unbiased and efficient estimates of demand parameters. They also obtain endogeneity test statistics as a by-product, including the direction of endogeneity bias. The model can be extended to incorporate Markov regime-switching dynamics in parameters and is open to other extensions based on maximum likelihood. The benefits of the proposed approach are achieved by assuming normality of the unobserved demand attributes, an assumption that imposes constraints on the types of pricing behaviors that are accommodated. However, the authors find in simulations that demand estimates are fairly robust to violations of these assumptions.

Original languageEnglish (US)
Pages (from-to)531-542
Number of pages12
JournalJournal of Marketing Research
Volume46
Issue number4
DOIs
StatePublished - Aug 2009
Externally publishedYes

Fingerprint

Aggregate data
Maximum likelihood estimator
Random coefficients
Simulated maximum likelihood
Logit model
Endogeneity
Sampling
Violations
Normality
Markov regime-switching
Endogeneity bias
Maximum likelihood estimation
Maximum likelihood
Train
Randomness
Sample size
Simulation
Market share
Control function
Product characteristics

Keywords

  • Aggregate data
  • Brand choice
  • Endogeneity
  • Heterogeneity
  • Logit model
  • Random coefficients
  • Scanner data
  • Simulated maximum likelihood

ASJC Scopus subject areas

  • Business and International Management
  • Economics and Econometrics
  • Marketing

Cite this

Simulated maximum likelihood estimator for the random coefficient logit model using aggregate data. / Park, Sungho; Gupta, Sachin.

In: Journal of Marketing Research, Vol. 46, No. 4, 08.2009, p. 531-542.

Research output: Contribution to journalArticle

@article{98d2b4823d2f48dc99ae690a972e6413,
title = "Simulated maximum likelihood estimator for the random coefficient logit model using aggregate data",
abstract = "The authors propose a simulated maximum likelihood estimation method for the random coefficient logit model using aggregate data, accounting for heterogeneity and endogeneity. The method allows for two sources of randomness in observed market shares: unobserved product characteristics and sampling error. Because of the latter, the method is suitable when sample sizes underlying the shares are finite. In contrast, Berry, Levinsohn and Pakes's commonly used approach assumes that observed shares have no sampling error. The method can be viewed as a generalization of Villas-Boas and Winer's approach and is closely related to Petrin and Train's {"}control function{"} approach. The authors show that the proposed method provides unbiased and efficient estimates of demand parameters. They also obtain endogeneity test statistics as a by-product, including the direction of endogeneity bias. The model can be extended to incorporate Markov regime-switching dynamics in parameters and is open to other extensions based on maximum likelihood. The benefits of the proposed approach are achieved by assuming normality of the unobserved demand attributes, an assumption that imposes constraints on the types of pricing behaviors that are accommodated. However, the authors find in simulations that demand estimates are fairly robust to violations of these assumptions.",
keywords = "Aggregate data, Brand choice, Endogeneity, Heterogeneity, Logit model, Random coefficients, Scanner data, Simulated maximum likelihood",
author = "Sungho Park and Sachin Gupta",
year = "2009",
month = "8",
doi = "10.1509/jmkr.46.4.531",
language = "English (US)",
volume = "46",
pages = "531--542",
journal = "Journal of Marketing Research",
issn = "0022-2437",
publisher = "American Marketing Association",
number = "4",

}

TY - JOUR

T1 - Simulated maximum likelihood estimator for the random coefficient logit model using aggregate data

AU - Park, Sungho

AU - Gupta, Sachin

PY - 2009/8

Y1 - 2009/8

N2 - The authors propose a simulated maximum likelihood estimation method for the random coefficient logit model using aggregate data, accounting for heterogeneity and endogeneity. The method allows for two sources of randomness in observed market shares: unobserved product characteristics and sampling error. Because of the latter, the method is suitable when sample sizes underlying the shares are finite. In contrast, Berry, Levinsohn and Pakes's commonly used approach assumes that observed shares have no sampling error. The method can be viewed as a generalization of Villas-Boas and Winer's approach and is closely related to Petrin and Train's "control function" approach. The authors show that the proposed method provides unbiased and efficient estimates of demand parameters. They also obtain endogeneity test statistics as a by-product, including the direction of endogeneity bias. The model can be extended to incorporate Markov regime-switching dynamics in parameters and is open to other extensions based on maximum likelihood. The benefits of the proposed approach are achieved by assuming normality of the unobserved demand attributes, an assumption that imposes constraints on the types of pricing behaviors that are accommodated. However, the authors find in simulations that demand estimates are fairly robust to violations of these assumptions.

AB - The authors propose a simulated maximum likelihood estimation method for the random coefficient logit model using aggregate data, accounting for heterogeneity and endogeneity. The method allows for two sources of randomness in observed market shares: unobserved product characteristics and sampling error. Because of the latter, the method is suitable when sample sizes underlying the shares are finite. In contrast, Berry, Levinsohn and Pakes's commonly used approach assumes that observed shares have no sampling error. The method can be viewed as a generalization of Villas-Boas and Winer's approach and is closely related to Petrin and Train's "control function" approach. The authors show that the proposed method provides unbiased and efficient estimates of demand parameters. They also obtain endogeneity test statistics as a by-product, including the direction of endogeneity bias. The model can be extended to incorporate Markov regime-switching dynamics in parameters and is open to other extensions based on maximum likelihood. The benefits of the proposed approach are achieved by assuming normality of the unobserved demand attributes, an assumption that imposes constraints on the types of pricing behaviors that are accommodated. However, the authors find in simulations that demand estimates are fairly robust to violations of these assumptions.

KW - Aggregate data

KW - Brand choice

KW - Endogeneity

KW - Heterogeneity

KW - Logit model

KW - Random coefficients

KW - Scanner data

KW - Simulated maximum likelihood

UR - http://www.scopus.com/inward/record.url?scp=69549117134&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=69549117134&partnerID=8YFLogxK

U2 - 10.1509/jmkr.46.4.531

DO - 10.1509/jmkr.46.4.531

M3 - Article

VL - 46

SP - 531

EP - 542

JO - Journal of Marketing Research

JF - Journal of Marketing Research

SN - 0022-2437

IS - 4

ER -