Data quality assessment in credit risk management by customized total data quality management approach

Thesis Type: Postgraduate

Institution Of The Thesis: Middle East Technical University, Turkey

Approval Date: 2016

Thesis Language: English

Student: Muhammed İlyas Güneş



As the size and complexity of financial institutions, more specifically banks, grow, the amount of data that information systems (IS) of such institutions need to handle also increases. This leads to the emergence of a variety of data quality (DQ) problems. Due to the possible economic losses due to such DQ issues, banks need to assure quality of their data via data quality assessment (DQA) techniques. As DQ related problems diversify and get complicated, the requirement for contemporary data quality assessment methods becomes more and more evident. Total Data Quality Management (TDQM) program is one of the approaches where data quality assessment of banking data is performed since the phases of the program, i.e. definition, measurement, analysis and improvement are well suited for identification of DQ issues. This study presents a customized approach to TDQM for data quality assessment in credit risk management. The study grounds the selection of DQ dimensions for credit risk on identification of data taxonomies for credit risk in accordance with the Basel Accords. Identification of data taxonomies from an IS viewpoint results in determination of data entities and attributes, which enabled the development of DQ metrics based on the DQ dimension selected in the definition phase. DQ metrics are transformed into quality performance indicators in order to assess quality of credit risk data by means of DQA methods in the measurement phase. Analysis of the results of DQA reveals the underlying causes of poor DQ performance in the analysis phase of TDQM. Identification of DQ problems and their major causes is followed by suggestion of appropriate improvement techniques based on the size, complexity and criticality of the problems in the context of credit risk management. TDQM approach customized for credit risk context in this study is implemented for a real bank case. Results of the implementation indicate the significance of and requirement for implementation of such methods sector-wide in order to manage the risks related to poor DQ. Moreover, a survey addressing banks to evaluate validity, applicability and acceptance of the approach as well as their own ongoing data governance activities has been carried out with the participation of senior risk managers. Findings of the survey reveal that the banks surveyed have found the approach to be considerably satisfactory in addressing data quality issues in credit risk management.