Evolving Data Systems in Credit

GICP Expert View / 13 March, 2023

Excerpt from the Global Credit Certificate, Chapter 19: Trends in New Technologies and Their Impact on Credit

Over the last decade, the broader financial industry has seen rapid adoption of new technologies ranging from banking apps over cryptocurrencies and machine learning to electronic authentication and customer scoring. Technology has helped make markets more efficient in a multitude of ways, from achieving cost reductions to the creation of completely new assets. In this way technology in the financial industry has influenced the evolution of economies more broadly. A closer look at its impact reveals that the identification, generation and processing of data is at the core of most recent technological developments, especially in the credit industry.

Consumer finance is the most prominent financial sphere where technology changed the way transactions are made and data is used, with banking apps, cryptocurrencies, automated recommendations and telematics-boxes in cars permeating the lay-user's day-to-day financial life. However, equally transformational is the use of technology at investment management and credit institutions where automation of data sourcing, tools for portfolio management, transaction support and automated trading have become or are becoming commonplace.

Technological Adoption

Institutions with large credit risk exposures, most prominently banks, have rapidly increased the technological sophistication of their risk management in areas such as measurement, reporting and stress testing. The volume and quality of data sources have experienced a corresponding growth, from traditional channels but also from textual data and alternative data sources. With the expanded capture and utilization of data, issues pertaining to the regulation of personal data has arisen and evolved with it.

Financial products inherently involve trust between two or more parties and carry uncertainty about future payment flows, with financial transactions relying heavily on good quality data to support their origination, monitoring and servicing. Modern banks and other financial institutions collect data from a wide array of sources to combine them for pricing, marketing, application processing, decisioning, reporting, risk management and many other operations.

Data Ingestion and Processing

For most financial institutions, the sourcing of data is a fragmented operation. Frequently used and readily available data from external providers are typically ingested via web-applications or application processing interfaces (APIs). Most institutions also collect specific data from customers or trading partners through forms, invoices and legal documentation, with financial institutions worldwide at varying degrees of transforming legacy systems that still require slow processing of frequent but unautomated information. While most organizations aim to maintain a well-catalogued 'data lake' for all data services, technical and procedural necessities lead to some data processing being kept separate from the data lake.

Data Procurement

Financial institutions have long been relying on traditional data such as customer information and macro-economic time-series. Automation to gain efficiencies has led to more and more specialized functions or outright outsourcing of the gathering process. Consumer information, for example, is often no longer collected in a bank branch, but rather at shops (in case of consumer loans) or estate agents (in case of mortgage loans).

In institutional lending and investing, specialized companies focus on the collection of financial data from the companies' websites and the convenient provision of financials via 'spreading tools'. Similarly, specialized data companies, such as CEIC Data or Haver Analytics, pool data from national statistical agencies to make them available in convenient user interfaces or directly via APIs. Market data like bond yields, indices, option prices and equity prices are often sourced from large specialized data redistributors, like Bloomberg, Refinitiv, FactSet or ICE who in turn maintain relationships with the exchanges that generate this information.

More recently, textual data is also sourced and redistributed through the APIs of specialized providers that process raw company filings and other reports. For credit risk assessment, external scores and ratings are provided by specialized institutions like credit reference or rating agencies.

Alternative Data Sources

Data collection has been made easier and cheaper by increased electronification of transactions and communication and the fall in prices for sensors. Such data can be more timely, provide a new perspective or serve as support for traditional data. Credit analysts may therefore be faced with new types of data which must first be evaluated for their meaningfulness and reliability; maybe more so than they were used to with well-established data sources.

Consumer behavior, for example, is now tracked every time the individual is paying by card, mobile phone or online, walking across cell-phone locations, chatting via apps or posting on social media. Analysis of properties and locally active businesses can benefit from granular price information, which is often fed by land registries or surveyor valuations conducted for the purpose of mortgage loans. Satellite images have also proven useful to identify trends in customer footfall and to gauge demand for toll roads, bridges, ports and car parks. During periods of economic stress, such as the global pandemic in 2020/21, such data have been consulted for a more timely update of economic activity and business disruptions and their impact on business disruption or recovery.

Technologies Accelerating Data Integration

Advancements in data technologies are accelerating the pace of integration of the varied data sources above. Textual processing and machine learning methods are transforming the ingestion and analysis of textual information to accelerate news-flow consumption across an increasingly crowded information space for credit analysts, allowing a for more efficient consumption and prioritization of more critical news and data releases to better inform credit analysis. The use of Python coding by a new generation of credit analysts allows for more direct access to primary data series in data lakes to enrich research and analysis.