Alternative Credit Scoring: A Conceptual Understanding

ACADEMIA Letters Alternative Credit Scoring: A Conceptual Understanding Pardhasaradhi Madasu, Associate Professor, Siva Sivani Institute of Management, Hyderabad, Telangana, India Introduction: The role of credit scoring has been an integral part of the financial system. Credit scoring or credit grading/rating is a mechanism that assesses the creditworthiness of the borrowers. The investors or lenders who wish to park their money in anticipation of a no-risk or low-risk return would decide upon investing or lending based on the creditworthiness of the borrowers or investees (GDS Link, 2020). The process of assessing the creditworthiness of the loan applicants is statistical and is based on the parametric statistical technique of ‘Discriminant Analysis’ (DA). The said analysis is used to differentiate between not-so-good borrowers and genuine borrowers. Ronald A. Fisher, in 1936 has laid a foundation for the usage of discriminant analysis for investment decisions and later, Durand, in 1941, applied the same to categorize good and bad loan applications. The contribution of R.A Fisher was so famous among the researchers that during 1968 Edward I. Altman has proposed the concept of Z-Score for predicting bankruptcy and predicting bankruptcy is correlated with credit analysis. Apart from DA, Linear Programming Models were used to assess creditworthiness (World Bank Group - ICCR, 2019). Even after lot of technological advancement the DA is still considered worth using for credit analysis. Academia Letters, March 2022 ©2022 by the author — Open Access — Distributed under CC BY 4.0 Corresponding Author: Pardhasaradhi Madasu, pardhasaradhimadasu71@gmail.com Citation: Madasu, P. (2022). Alternative Credit Scoring: A Conceptual Understanding. Academia Letters, Article 4926. https://doi.org/10.20935/AL4926. 1 Need for Assessing Creditworthiness: Investors and lenders with surplus money are ready to invest either in equities or debt instruments. In either case, the ability and willingness of the counter-party to repay on a timely basis play a prominent role. Yet, the need to assess the creditworthiness is vital in the case of debt when compared to equity investing. The simple reason for assessing the debt repayment capacity of the borrower when compared to the investment in equity is that equity by definition a high risk investment with expected higher returns. Usually, creditworthiness is assessed based on character, capacity, collateral, capital and conditions if the lender is using the popular 5Cs of Credit and if the lender is using not so popular 7Cs of credit than two more metrics are added i.e. credit and capability. The inputs to quantify the 5Cs or 7Cs of Credit usually are taken from financial statements and past credit behavior of the borrower. However, all the 5Cs may not be readily available for specific categories of borrowers such as micro-credit seekers or Peer-to-Peer borrowers. In the case of business firms who opt for micro-credit, applying the 5Cs framework is not practical (Zhou, Huang, & Wang, 2018). Need for Credit History: Traditionally, the borrowers who have a good credit history and strong financial credentials are considered the ‘Prime’ borrowers, and on the contra, if the credit history is not good and the financial credentials are weak at a point in time, the borrowers are classified as ‘Sub-Prime’. The financial institutions such as banks or NBFCs classify borrowers into buckets based on the empirical financial data and past credit behavior. Previously, financial institutions did the credit assessment on their own using their proprietary credit risk models. Later, the concept of ‘Credit Bureaus’ have seen the light, and CIBIL is the credit bureau in the Indian context. However, the credit bureaus also only base their credit score on empirical quantitative data. Against this backdrop, the question that comes to the foreground is, “How do the persons or business firms such as MSMEs with no verifiable financial history or credit records get a credit score and later the credit?” (Hong Kong Monetary Authority). Even if the credit score is available, it may not be in an acceptable range for the banks or NBFCs. The very thought of not having a credit history or financial track record brings us to the current burning issues of ‘Financial Exclusion’. Nearly 1.2 billion cannot utilize formal financial services such as lending (Intellias, 2021). It is exciting to note that economies like India Academia Letters, March 2022 ©2022 by the author — Open Access — Distributed under CC BY 4.0 Corresponding Author: Pardhasaradhi Madasu, pardhasaradhimadasu71@gmail.com Citation: Madasu, P. (2022). Alternative Credit Scoring: A Conceptual Understanding. Academia Letters, Article 4926. https://doi.org/10.20935/AL4926. 2 are not the only ones where the citizens are ‘invisible’ to the eyes of financial institutions. For example, few industry studies reveal that 25 per cent of US consumers are considered thin-file because their credit histories have records of less than five transactions. Nevertheless, when we get into the data at the global level, 1.7 billion adults are outside the banking zone (Prove, 2021). Changing Face of Credit Risk Assessment: Along with the increase in processing power of computer, the depth and variety of analytical tools also have increased. Today AI & ML are the major data sciences tools that are helping majority of the businesses to take prudent decisions. Even in this era of Machine Learning and Artificial Intelligence (AI & ML), if we talk about financial exclusion but not financial inclusion, it would disrespect the innovations in technology. Now, AI & ML can access, procure and analyze alternative data such as geolocations, spending habits or even the digital networks of the borrowers hence the financial institutions are expected to use these alternative data sources for credit scoring (Shriram City, 2021). This is where the concept of ‘Alternative Credit Scoring’ (ACS) has seen the light. Growing Relevance of ACS: In simple terms, ACS is an approach where varied structured and unstructured data such as e-commerce transactions, travel history, expenditure blueprint and mobile data are modelled using AI & ML technologies (Maji, 2021). ACS requires alternative data for building credit risk models, and the alternative data could be in the form of utility bill payments, rental and lease payments, shopping history, mobile network operator data, behavioural data or even social media data (Kumar, Sharma, & Mahdavi, 2021). In practice, the granular transaction data is the foundation for alternative data sources (World Bank Group - ICCR, 2019). As access to alternative data is becoming more accessible, the popularity of ACS is growing globally. In the Indian context, financial institutions are also showing favour towards using CIBIL score along with ACS such that the loan application assessment is more prudent and covers 360-degree assessment. One of the popular inputs for ACS used by a few Indian banks is the locational data (Data Sutram, 2021). The ACS is a boon to the borrowers as well as lenders. From borrowers’ perspective, even with a weak formal credit history, they would be eligible to get a credit score and subsequent Credit. From the viewpoint of lenders, the quantitative financial data and the qualitative data could also be used for credit assessment. Apart from credit scoring, the ACS would also help Academia Letters, March 2022 ©2022 by the author — Open Access — Distributed under CC BY 4.0 Corresponding Author: Pardhasaradhi Madasu, pardhasaradhimadasu71@gmail.com Citation: Madasu, P. (2022). Alternative Credit Scoring: A Conceptual Understanding. Academia Letters, Article 4926. https://doi.org/10.20935/AL4926. 3 predict bankruptcy. If the number of creditworthy borrowers increases, the global banking sector could gain approximately USD 380 billion in annual revenues with no or low additional credit risk (Intellias, 2021). Financial institutions can now tap unbanked or underbanked potential borrowers such as agriculturalists or gig workers without hesitation. Closing Remarks: Because ACS techniques are highly technologically driven, conventional financial institutions are not ready to fully incorporate the alternative credit scoring methods into their common framework. However, the emerging concept of ‘Open Banking’ enables traditional financial institutions such as banks and NBFCs to use third-party services (in the form of API integrations) concerning ACS. Against this backdrop, along with start-ups in AI & ML space, big credit bureau businesses have also started to provide ACS services to formal financial institutions (GDS Link, 2020). Many verified APIs have entered the financial market, providing a comprehensive credit score based on both traditional metrics and alternative metrics of Credit. The popularity of ACS would undoubtedly help both the borrowers and lenders. Alongside the benefits, ACS also poses a few challenges that must be mitigated over time. Challenges concerning privacy and quality of non-traditional or alternative data are the two significant challenges to be addressed. In addition, privacy in the case of mobile data and social media data is a significant concern. However, we could hope for a balance between the pros and cons of ACS shortly (Koo, Zhou, & Li, 2019). Academia Letters, March 2022 ©2022 by the author — Open Access — Distributed under CC BY 4.0 Corresponding Author: Pardhasaradhi Madasu, pardhasaradhimadasu71@gmail.com Citation: Madasu, P. (2022). Alternative Credit Scoring: A Conceptual Understanding. Academia Letters, Article 4926. https://doi.org/10.20935/AL4926. 4 References Data Sutram. (2021, December 8). CIBIL & Alternative Scoring - How is the Housing Loan Sector Being Affected? Retrieved from Data Sutram: https://www.datasutram.com GDS Link. (2020, March 17). Alternative Credit Scoring Models: Pros and Cons. Retrieved from GDS Link: https://www.gdslink.com Hong Kong Monetary Authority. (n.d.). Alternative Credit Scoring of Micro-, Small and Medium-Sized Enterprises. Hong Kong: ASTRI. Intellias. (2021, December 23). Mobile Data + Machine Learning = Better Credit Scoring for the Understanding. Retrieved from Intelligent Software Engineering: www.intellias.com Koo, J., Zhou, M., & Li, Y. (2019, December 20). Alternative Credit Scoring and Some Challenges. Retrieved from Medium: https://www.medium.com Kumar, A., Sharma, S., & Mahdavi, M. (2021). Machine Learning (ML) Technologies for Digital Credit Scoring Rural Finance: A Literature Review. Risks, 9(192). Maji, P. (2021, June 23). Alternative Credit Scoring: Now Avail Loans Even With a Low CIBIL Score. Financial Express. Prove. (2021, July 12). Six Biggest Problems with Alternative Data. Retrieved from Prove: https://www.prove.com Shriram City. (2021, September). How AI Helps Lenders Assess Loan-Seekers With No Credit Score? Retrieved from Shriram City Articles: https://www.shriramcity.in World Bank Group - ICCR. (2019). Credit Scoring Approaches Guidelines. Washington: The World Bank Group. Zhou, J., Huang, D., & Wang, H. (2018). RFMS Method for Credit Scoring Based on Bank Card Transaction Data. Statistica Sinica. Academia Letters, March 2022 ©2022 by the author — Open Access — Distributed under CC BY 4.0 Corresponding Author: Pardhasaradhi Madasu, pardhasaradhimadasu71@gmail.com Citation: Madasu, P. (2022). Alternative Credit Scoring: A Conceptual Understanding. Academia Letters, Article 4926. https://doi.org/10.20935/AL4926. 5