Application of machine-learning models to estimate regional input coefficients and multipliers

ABSTRACT

Due to the unavailability of accurate data and the limitations of the existing methods, reliable input–output tables (IOTs) may not be available for all the regions in a country. This study proposes a novel approach to estimate the regional input coefficients. It harnesses the capabilities of machine-learning (ML) algorithms to estimate the regional input coefficients of one region based on the IOTs of multiple other regions for which reliable data are available. The application of three ML algorithms is investigated using data from Japan. The results highlight the superior performance of the ML models compared with location quotient models.

KEYWORDS:

• machine learning
• input–output tables
• regionalization
• input coefficients
• location quotient

JEL:

• C67
• D57
• R15

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1 The data and codes used in this study, along with brief instructions on using them, are available in the supplemental data online. They are also available at: https://drive.insurer.sharif.ir/index.php/s/TH7eow6iqBdfmK2.