A Cobb–Douglas type model with stochastic restrictions: formulation, local influence diagnostics and data analytics in economics

Francisco J.A. Cysneiros, Víctor Leiva, Shuangzhe Liu, Carolina Marchant, Paulo Scalco

Research output: Contribution to journalArticle

1 Citations (Scopus)

Abstract

We propose a methodology for modelling and influence diagnostics in a Cobb–Douglas type setting. This methodology is useful for describing case-studies from economics. We consider stochastic restrictions for the model based on auxiliary information in order to improve its predictive ability. Model errors are assumed to follow the family of symmetric distributions and particularly its normal and Student-t members. We estimate the model parameters with the maximum likelihood method, which allows us to compare the normal case with a flexible framework that provides robust estimation of parameters based on the Student-t case. To conduct diagnostics in the model, we use two approaches for studying how a perturbation may affect on the mixed estimation procedure of its parameters due to the usage of sample data and non-sample auxiliary information. Curvatures and slopes used to detect local influence with both approaches are derived, considering perturbation schemes of case-weight, response and explanatory variables. Numerical evaluation of the proposed methodology is performed by Monte Carlo simulations and by applications with two data sets from economics, all of which show its good performance and its further applications. Particularly, the real data analyses confirm the importance of statistical diagnostics in the data modelling.

LanguageEnglish
Pages1-27
Number of pages27
JournalQuality and Quantity
DOIs
Publication statusE-pub ahead of print - 14 Jan 2019

Fingerprint

Influence Diagnostics
Local Influence
Auxiliary Information
diagnostic
Economics
Restriction
Methodology
Formulation
Diagnostics
Perturbation
economics
methodology
Model Error
Symmetric Distributions
Data Modeling
Robust Estimation
Maximum Likelihood Method
estimation procedure
Slope
Monte Carlo Simulation

Cite this

@article{343524c384d941f5b2874eedc38e4514,
title = "A Cobb–Douglas type model with stochastic restrictions: formulation, local influence diagnostics and data analytics in economics",
abstract = "We propose a methodology for modelling and influence diagnostics in a Cobb–Douglas type setting. This methodology is useful for describing case-studies from economics. We consider stochastic restrictions for the model based on auxiliary information in order to improve its predictive ability. Model errors are assumed to follow the family of symmetric distributions and particularly its normal and Student-t members. We estimate the model parameters with the maximum likelihood method, which allows us to compare the normal case with a flexible framework that provides robust estimation of parameters based on the Student-t case. To conduct diagnostics in the model, we use two approaches for studying how a perturbation may affect on the mixed estimation procedure of its parameters due to the usage of sample data and non-sample auxiliary information. Curvatures and slopes used to detect local influence with both approaches are derived, considering perturbation schemes of case-weight, response and explanatory variables. Numerical evaluation of the proposed methodology is performed by Monte Carlo simulations and by applications with two data sets from economics, all of which show its good performance and its further applications. Particularly, the real data analyses confirm the importance of statistical diagnostics in the data modelling.",
keywords = "Likelihood-based methods, Local influence, Mixed estimation, Monte Carlo simulations, R software, Regression models, Symmetric distributions",
author = "Cysneiros, {Francisco J.A.} and V{\'i}ctor Leiva and Shuangzhe Liu and Carolina Marchant and Paulo Scalco",
year = "2019",
month = "1",
day = "14",
doi = "10.1007/s11135-018-00834-w",
language = "English",
pages = "1--27",
journal = "Quality and Quantity",
issn = "0033-5177",
publisher = "Springer",

}

A Cobb–Douglas type model with stochastic restrictions : formulation, local influence diagnostics and data analytics in economics. / Cysneiros, Francisco J.A.; Leiva, Víctor; Liu, Shuangzhe; Marchant, Carolina; Scalco, Paulo.

In: Quality and Quantity, 14.01.2019, p. 1-27.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Cobb–Douglas type model with stochastic restrictions

T2 - Quality and Quantity

AU - Cysneiros, Francisco J.A.

AU - Leiva, Víctor

AU - Liu, Shuangzhe

AU - Marchant, Carolina

AU - Scalco, Paulo

PY - 2019/1/14

Y1 - 2019/1/14

N2 - We propose a methodology for modelling and influence diagnostics in a Cobb–Douglas type setting. This methodology is useful for describing case-studies from economics. We consider stochastic restrictions for the model based on auxiliary information in order to improve its predictive ability. Model errors are assumed to follow the family of symmetric distributions and particularly its normal and Student-t members. We estimate the model parameters with the maximum likelihood method, which allows us to compare the normal case with a flexible framework that provides robust estimation of parameters based on the Student-t case. To conduct diagnostics in the model, we use two approaches for studying how a perturbation may affect on the mixed estimation procedure of its parameters due to the usage of sample data and non-sample auxiliary information. Curvatures and slopes used to detect local influence with both approaches are derived, considering perturbation schemes of case-weight, response and explanatory variables. Numerical evaluation of the proposed methodology is performed by Monte Carlo simulations and by applications with two data sets from economics, all of which show its good performance and its further applications. Particularly, the real data analyses confirm the importance of statistical diagnostics in the data modelling.

AB - We propose a methodology for modelling and influence diagnostics in a Cobb–Douglas type setting. This methodology is useful for describing case-studies from economics. We consider stochastic restrictions for the model based on auxiliary information in order to improve its predictive ability. Model errors are assumed to follow the family of symmetric distributions and particularly its normal and Student-t members. We estimate the model parameters with the maximum likelihood method, which allows us to compare the normal case with a flexible framework that provides robust estimation of parameters based on the Student-t case. To conduct diagnostics in the model, we use two approaches for studying how a perturbation may affect on the mixed estimation procedure of its parameters due to the usage of sample data and non-sample auxiliary information. Curvatures and slopes used to detect local influence with both approaches are derived, considering perturbation schemes of case-weight, response and explanatory variables. Numerical evaluation of the proposed methodology is performed by Monte Carlo simulations and by applications with two data sets from economics, all of which show its good performance and its further applications. Particularly, the real data analyses confirm the importance of statistical diagnostics in the data modelling.

KW - Likelihood-based methods

KW - Local influence

KW - Mixed estimation

KW - Monte Carlo simulations

KW - R software

KW - Regression models

KW - Symmetric distributions

UR - http://www.scopus.com/inward/record.url?scp=85060134716&partnerID=8YFLogxK

UR - http://www.mendeley.com/research/cobbdouglas-type-model-stochastic-restrictions-formulation-local-influence-diagnostics-data-analytic

U2 - 10.1007/s11135-018-00834-w

DO - 10.1007/s11135-018-00834-w

M3 - Article

SP - 1

EP - 27

JO - Quality and Quantity

JF - Quality and Quantity

SN - 0033-5177

ER -