NatWest Bank is the UK’s largest lender to small and medium-sized businesses, a success story it proudly broadcasts and an achievement the bank maintains through a robust data management strategy that enables it to provide finance to some of the most difficult-to-reach borrowers.
Like other banks, the strength of NatWest’s system has been tested in recent years by the incorporation of ESG data into its data lake and workflows. Unlike other banks, however, NatWest has achieved this to a great extent within a sector that is beset by data quality issues.
The SME data challenge is being managed by Kaushik Ghosh Dastidar, head of ESG data and solution architect at NatWest. The difficulty of the project, however, has prompted the augmentation of Kaushik’s team with machine learning (ML) and artificial intelligence (AI). The bank is testing AI’s ability to unlock the value of huge volumes of data to bring order to the banks’ enormous Back Book, the tens of thousands of loans and mortgages that have been live for some time.
NatWest is piloting AI to match individual customers and property with ESG and other data. Front Book loans, those that are being originated, negotiated and newly introduced, are subject to mapping processes that can link emissions, climate risk and other information to mortgaged homes, commercial property and so on. But older loans are harder to match because they were originated before the bank’s sophisticated mapping techniques were introduced. Automating this processes is fraught with difficulty because information logged about the clients and the loans is incomplete, misspelled or simply absent. Kaushik hopes AI can work around those limitations.
“Our Back Book is massive and unless you have a team of people who are manually typing and searching those loans one by one, it’s difficult to match data to every one of them,” he told ESG Insight. “But we are running a few pilots to see if AI can do this. So far, we’ve seen some promising results.”
The sprawling nature of banks’ business means that they arguably face the greatest challenge of all financial institutions in getting their ESG capabilities in order. Business, retail and mortgage lending require different approaches to their insurance and investment activities.
They also tend to be subject to more regulatory oversight, making it even more important that they get their ESG data strategies in order. On top of that, their exposure to large volumes of retail business means that they must cleave to demands of consumers, a large proportion of whom make their banking decisions according to lenders’ approaches to sustainability. A 2019 study by McKinsey found that 14 per cent of customers choose who to bank with based on the lender’s purpose and values, a proportion that’s expected to have grown since.
Nevertheless, to get it right, banks are increasingly adopting a singular, centralised and often cloud-based data management approach.
AI is helping institutions the world over. Machine learning (ML) and natural language processing (NLP) have been deployed to numerous tasks, especially in data gathering and data scraping from social media and corporate reports. Newer platforms, such as large language programmes, or generative AI software like ChatGPT, are less entrenched in the ESG world and still finding a role.
Kaushik believes that very soon, institutions will be harnessing the full opportunities offered by AI.
“In data matching, we believe AI will eventually be adopted because these matches are one off – once you have made your match you don’t need to keep doing it again.”
Mapping datasets to company and instrument data is harder within the SME realm because data quality is poorer. Unlike publicly listed companies, smaller firms have no regulatory obligation yet to report their ESG performance. For banks such as NatWest, which have established sustainability mandates to which they must work, this is a hurdle they must tackle carefully.
The same is true within its home-lending business. Establishing the financed emissions of a mortgage borrower or a property, for instance, is made doubly hard by privacy laws that prevent institutions obtaining energy meters of homes. The best that banks can hope for is that the properties their clients wish to buy possess an energy performance certificate (EPC), which catalogues the energy efficiency of a building based on the materials and designs used in its construction.
These only tell part of the efficiency story, Kaushik said.
“You could have a home built to extremely high efficiency standards, but if the resident leaves the heating on and the doors open all day, then that EPC counts for little,” Kaushik said.
This is where a revision of privacy laws surrounding energy use data would make a huge difference, he added.
“One of the things we have been trying to work with the government on is having access to the actual electricity usage of a house – that would be a very good parameter to actually understand how much emissions you are driving,” he said.
The challenge is deeper for commercial property, Kaushik said. The difficulty in matching external ESG data with banks’ internal data means that NatWest’s ESG team can establish datasets for about 90 per cent of residential homes, but just 40 per cent of that for commercial properties, which translates to 79 per cent of total customer and corporate lending as of December 2019.
The bank, therefore, must make some estimates. It can take a top-down approach and draw inferences from industry averages, for instance by extrapolating build quality data from other developments on such factors as the proportion of triple- or double-glazed windows in a home. Alternatively, it can adopt a customer-level engagement strategy. In this scenario potential borrowers are quizzed on their decarbonisation plans and that data is used later in broader aggregated datasets. All calculations are run through validation processes to ensure they are accurate.
NatWest also buys in expertise to reinforce its analytics. One provider, for instance, creates synthetic EPCs that NatWest can later augment as direct data becomes available. In the name of transparency, the bank discloses to customers the raw data and the algorithm’s assumptions that generate the estimate.
“More engagement actually enriches the central concept,” Kaushik said.
For SMEs, this process is being achieved partly through the bank’s proprietary Carbon Planner, a freely available app in which clients can input their information and in return receive real-time advice from the bank on improving their carbon footprints. Like all of NatWest’s ESG data management processes, this happens in the cloud – through partnerships with Snowflake and AWS – and the data gathered goes into the bank’s data lake for use across the enterprise. That includes use within the data modelling processes that the banks insurance and investment businesses use.
Customers’ decarbonisation plans are reviewed annually, and validation services are utilized to track progress towards goals. Depending on the lending product, customers may be set key performance targets during the lifetime of the loan which, when met, will earn them improved borrowing rates.
“Our first objective is to see if we can work with the customer to help them decarbonise through a green transformation or through other types of products that we can offer, to get them from point A to point B,” Kaushik said. “The data services all of the NatWest franchises and so it is best placed in the cloud where we can maintain, and distribute from, a golden source.”
Subscribe to our newsletter