Journal of Scientific Papers


© CSR, 2008-2019
ISSN 2071-789X

91th percentile
Powered by  Scopus

Directory of Open Access Journals (DOAJ)

Strike Plagiarism

  • General Founder and Publisher:

    Centre of Sociological Research


  • Publishing Partners:

    University of Szczecin (Poland)

    Széchenyi István University, (Hungary)

    Mykolas Romeris University (Lithuania)

    Alexander Dubcek University of Trencín (Slovak Republic)

  • Membership:

    American Sociological Association

    European Sociological Association

    World Economics Association (WEA)




Impact of baseline population on credit score’s predictive power

Vol. 12, No 1, 2019

Kamphol Panyagometh,


NIDA Business School,

Bangkok, Thailand,


Impact of baseline population on credit score’s predictive power




Abstract. Credit scoring involves statistical analysis performed by lenders and financial institutions to access person's creditworthiness. It utilizes statistical techniques along with debtor data such as loan application or credit bureau information to measure  debtor’s creditworthiness. When compared with the traditional credit evaluation process, credit scoring has shown less bias, faster speed, and consistent measurement of creditworthiness. For this reason, the National Credit Bureau of Thailand (NCB) has developed the NCB Score, based on credit behavior information collected from its members’ financial institutions, which normally issue a wide variety of credit products. However, problems can arise when this NCB score is applied to a smaller bank that usually offers a few specific types of loans. As a result, the score’s predictive power may deteriorate. In this paper, the impact of baseline population difference on the predictive power of a credit score was studied by separating proprietary data from NCB into two groups. One group represents those who originate personal loans in the seventh month of the study period, and the other group represents those who originate mortgage loans in the seventh month of the study period. The credit score model of each group was developed and their predictive power was compared when used with the same baseline population and a different baseline population to monitor the change in model predictive power


Received: June, 2018

1st Revision: November, 2018

Accepted: January, 2019


DOI: 10.14254/2071-789X.2019/12-1/15

JEL ClassificationC4, C5, G3

Keywords: credit scoring, predictive power, logit model