Perhaps your first question might be: is it not the other way around? That interest is what determines your behaviour, well no, not in the case for usury. So, allow me to briefly explain. The odds are that at this point or perhaps at some point in your future (or even your past) you are going to (have) propose(d) a relationship with a bank to finance your needs. Such needs include a home loan, financing a car, a personal loan, a credit card etc. Thus, your credit worthiness is what determines the terms of the relationship (term of the loan, its structure, the interest rate and additional fees), if the relationship is to exist at all.
The aim of this article is twofold, first it is to uncover what determines your credit worthiness. Secondly, the intersection between increased satiation of consumer demand, the competitive landscape of the banking industry and the proliferation of technology that facilitates this demand-supply relation brings in the moral question of: what matters most between the privacy and the sovereignty of a customer vs servicing their convenience?
To the first point of the article, credit scoring decision models are the tools used by banks to determine a customer’s credit worthiness. The credit scoring models are designed to solve two problems that are analogous to the insurance industry’s problems. The first is the adverse selection problem, how can you predict which customers are financially delinquent from those that are financially stable? An application scorecard model is to ex ante solve this problem. The second problem to solve is that of moral hazard, admittedly this is more relevant for the insurance than it is the banking industry. The point of problem is to derive from an existing customer’s behaviour what is the probability they will move from a state of meeting their financial obligations to them being financially delinquent? A behavioural scorecard model is the tool of choice to predict the possible transition of your existing customers from one default state to the other from a credit risk perspective.
These scorecard models are nested in other models as they are underpinned by models such as the probability of default (PD), loss at given default (LGD), exposure at given default (EAD) models and so forth. The PD model is itself usually underpinned by supervised Machine learning algorithms or statistical models such as the logistic regression and the Markov chain models. Neural networks, random forecasts and support vector machines can be used to ascertain the probability of default but the logistic regression and the Markov chain models are preferred by the industry for their simplicity (Chikukwa and Venter, 2021).
Now this is where it gets interesting because Dash et al (2021) lay out the best practises to be adopted by banks in the industry to lower the cost of acquisition, streamline the customer onboarding end to end process and most importantly to fatten the margins. The expansion of the data sources as well as deep learning techniques employed on existing data to extract a better view of the customer are the noteworthy practices for the moral question given above. Data at the purview of banks can be segmented into four different categories. That is data that is traditional as well non-traditional and also either internally or externally available.
Table 1
Table 1 gives the examples to the four categories of the data. Natural language processes to determine for instance the behaviour of a customer on social media to determine their probability of default. The lines become blurred when a financial institution can track an individual’s sentiments on social media to determine whether they can receive financing from a bank. Thus, at which point do you have the autonomy over the digital footprint you leave behind in the ether and up to which point can banks mitigate their risks against customers that may not be able to meet their debt obligations? Given the high-level perusal of the bank scoring and the technology that facilitates it, the demand side should be looked at before we conclude.
So, to meet the gratification of customer’s demands banks are shifting from a product driven service offering to a tailored customer offering (KPMG, 2019). The latter is not possible without data. However, customers differ with their level of desire for the control of their own data and their knowledge on product offerings. Therefore, customers that want a high level of control and possess a high level of knowledge would prefer to orchestrate their banking experience, customers that trust the bank with their data as well as have a lower level of knowledge on the products offered would prefer to have their experience automated.
Customers that are not as trusting and have lower level of knowledge as well would prefer to have their experience aggregated/integrated with respect to the products offered and the last segment of customer that have a high level of knowledge but still trusts the institution with his/her information would prefer to play a validatory role in their banking experience. The customised product offering presents the potential threat of price discriminatory behaviour by banks to consumers. Such that with the information asymmetry that exists on products offered the pricing may change from being a markup above marginal cost to customer’s reservation price (how badly do you want it?). Competition in the industry as well as transparency using a distributed ledger can be the best ways to circumvent such adverse behaviour by banks, notwithstanding the regulatory technology (regtech) to monitor such activities.
What can be said given the diversity of objectives amongst individuals, is that the pursuit of financial freedom is what is common across the broad range of individual objectives. Therefore, information on how your behaviour can be determined to shape your interest payments is necessary to aid in the fulfilment to that objective.
References.
Biswas, S., Carson, B., Chung, V., Singh, S. and Thomas, R., 2020. AI-bank of the future: Can banks meet the AI challenge. New York: McKinsey & Company.
Gurný, P. and Gurný, M., 2013. Comparison of credit scoring models on probability of default estimation for us banks. Prague economic papers, 22(2), pp.163-181.
KPMG, A., 2019. The future of digital banking. Retrieved, 12, p.2021.
Thomas, L.C., Ho, J. and Scherer, W.T., 2001. Time will tell: behavioural scoring and the dynamics of consumer credit assessment. IMA Journal of Management Mathematics, 12(1), pp.89-103.