Delivering data standards and transforming data collection in financial services

Quarterly Bulletin 2021 Q1
Published on 09 April 2021

By Angus Moir (Data Strategy Implementation).

  • The Bank of England collects data from banks, building societies and insurance companies in the UK. We use this data to monitor and react to risksfootnote [1] in the financial sector, and the economy more widely, and to supply data to the Office for National Statistics.
  • We have been working with other regulators and with industry, including firms, trade associations and third-party providers, to understand how to reform and improve the way firms manage data and how we collect data from them. These reforms may affect the way we analyse and publish data over time.
  • Critical to our proposed reforms is the development and adoption by the finance industry of a set of common data standards. Common data standards can make it easy for us to explain the data we require and for firms to find and source that data. Evidence from our reporting work suggests potential for major long-term benefits to firms, consumers and regulators, if common data standards can be introduced.
  • In publishing our Transformation Plan for data collection, we committed to take action to deliver common data standards and other key reforms. We committed to set up a joint transformation programme with industry and the Financial Conduct Authority (FCA), and to use our powers to help overcome key barriers to delivering reforms (eg problems of co-ordinating collective action), which may prevent adoption of data standards.


Data are critical to our work. They are the raw ingredients we need for monitoring and analysis, and therefore for our work to keep the monetary and financial systems stable. Like many organisations, in recent years our demands on data have grown. We want to make more extensive use of data to guide our decision-making, including making the most of bigger data sets and new techniques. But unlike many organisations, much of the data we use aren’t data we generate internally; instead we collect them from firms.

These new demands on data have put a strain on the processes and systems we use to collect data. In response, since 2016 we have been working with the FCA and with industry to explore ways to improve data collection.footnote [2] Initially some of our focus was on making greater use of technology to improve data collection. But it soon became clear that some more fundamental changes were needed in terms of how data are defined and managed.

Earlier this year we published our Transformation Plan, which outlined this work in greater detail. In this publication, we explained that the development and adoption of common data standards, that consistently describe and identify data within firms, is key to this reform.

Common data standards, and the Bank of England’s interest in them

A famous example of the power of standards relates to the internet. In 1989, when Tim Berners Lee invented the world-wide web he needed a way for large numbers of different computers to be able to understand a web page. To do this he invented a standard called ‘Hyper Text Markup Language’ (HTML) that provided a consistent way for any computer to recognise document parts. Within HTML, he gave each part a standard tag. For instance, a paragraph was tagged with ‘p’ or bold text with ‘b’.

Almost 30 years on and the standard for document data, HTML, has provided huge benefits for society. It not only powers the internet, but it enables thousands of independently created document editors to open, modify and save documents created by thousands of other document editors.

Standards in wholesale markets

A decade on from the invention of HTML, we published an article on how standards could bring similar benefits to the financial sector, and in particular how standards could help automation of trading and processing in wholesale markets. At the time, a lot of trading was still carried out on the phone, and processed using paper, fax machines and proprietary systems that didn’t talk to each other. The creation and adoption of standards, such as Financial products Markup Language (FpML), helped change that. It allowed central clearing and electronic trading to develop and function, and enabled increases in the volume of traded products and falls in the cost of trading them. For example, Chart 1 shows the turnover of over-the-counter derivatives during this period.

Although most people don’t use wholesale financial products on a daily basis, we do benefit when their price falls – for instance, cheaper interest rate swaps lower the costs of mortgages by allowing banks to cheaply hedge interest rate risk, and lower costs of trading shares boost the return on investments.

Chart 1: Turnover of over-the-counter derivatives since 1995

Line shows growth in turnover from 1995 to 2015


  • Source: Bank for International Settlements – derivatives statistics.

Standards and financial stability

The growth in electronic trading was part of a boom in financial services that ultimately lasted until the financial crisis in 2008. The crisis highlighted the problem that regulators at times didn’t have enough visibility of risks emerging in the financial sector. Unsurprisingly, a key post-crisis response from regulators around the world, including the Prudential Regulation Authority, was to increase the amount of data they collect.footnote [3]

Perhaps less visible were steps regulators took to widen the use of data standards. As Andy Haldane, then Executive Director for Financial Stability at the Bank of England, pointed out in a speech in 2012, the financial crisis showed data standards were about more than efficiency: they were important for financial stability. Without them, regulators and other financial sector risk managers were constantly stuck with a partial view of risk. They struggled to combine data from different sources because the data were inconsistent (Box B). It was concerns over financial stability that led the Financial Stability Board to co-ordinate the development of the Legal Entity Identifier (Box A). So data standards do not just make financial services cheaper: they can help safeguard people’s jobs and livelihoods.

Box A: The Legal Entity Identifier (LEI)

In the aftermath of the financial crisis, the financial sector needed a legal entity identifier: an electronic barcode that allowed any computer or programme in the world to recognise data relating to the same legal entity. To make it happen, public authorities gathered under the auspices of the Financial Stability Board and agreed to drive the agenda forward. In 2013, the ‘LEI Regulatory Oversight Committee’ was established to oversee a new data standard, the Legal Entity Identifier (LEI).

The LEI has profound benefits for users and regulators of the financial system (Figure A.1). It allows data from different sources, and on different topics, but relating to a single bank or entity to be easily combined. For reporting specifically, it has two big benefits: it makes it easier for firms to compile the data they send us, and it makes it easier for us to merge data from separate reports – providing a fuller picture of the activity of a firm.

Yet despite the LEI’s benefits, the financial sector still had to overcome challenges to its adoption. Regulatory intervention was needed. In a series of regulatory reporting requirements, regulators required firms to report LEIs for them and their clients alongside their data. In turn, this forced firms to use the LEI in their systems and processes.

The LEI is one example of how data standards can make the financial sector more efficient and safer, and of the role a regulator can play in encouraging their use. There are other examples that have proven the value of data standards in the financial system, such as the creation of FpML that aided the shift to electronic trading in derivatives markets in the 2000s, and the International Securities Identification Number (ISIN) for securities trading.

Figure A.1: Legal Entity Identifiers enable firms and regulators to compile data sources more easily

Graphic uses icons to show how legal entity identifiers make it easier to compile data from different sources

Box B: Consolidating data

Consolidating or merging data from different data sets is a common activity across the financial sector. Financial firms often need to do it to fulfil reporting requirements. And they often do it for their own purposes, for instance to improve data management or build big data sets needed to develop machine learning models.

Currently these tasks often require delicate manual intervention. Although processes in the financial sector are increasingly digital, similar data are often recorded in multiple systems using different formats. Users of the data, like ourselves, often need to consolidate millions of data points, recorded by thousands of systems. The cost to industry of setting up those reconciliations is enormous.

Data standards can support automation of aspects of the process to consolidate or merge data, making it more efficient and the resulting data set higher quality. There are two types of data-merging operations, each requiring a different type of standard.

One operation is to consolidate data from different sources about the same thing. Here we need a standard identifier, such as an LEI (Box A), or an ISIN. The standard identifier tells a computer the data being merged are indeed about the same thing, with no additional intervention needed from a human.

The other operation requires the merging of the same data but about different things (Figure B.1). For instance, creating a single data set with the interest rate and loan amount for two sets of loans from two different systems. To do this efficiently, we need a computer to know whether a concept like ‘loan amount’ should be treated the same as ‘principal amount’, or even whether ‘loan amount’ is the same as ‘LOAN AMOUNT’. In effect, we need an HTML equivalent that allows us to identify data concepts in the financial sector, such as the parts of a financial contract.

Figure B.1: Without common standards, it is harder for banks and regulators to merge similar data sets

Graphic with icons, showing how common data standards make it easier to compile similar data sets from multiple sources

Issues in development and adoption of data standards

When Tim Berners Lee invented HTML, he had the benefit of starting from scratch. He was literally inventing the web, while personal computers were still an uncommon sight in households. The overwhelming majority of digital documents were yet to be created.

For the financial sector, this isn’t the case. Since banks started to use computers in the 1950s, the financial sector has built up a huge legacy of systems. Many of these systems identify the same financial data in their own specific way. These systems record and manage critical information about people/households and businesses: the amount of money they have in their current account, or how much they owe on their mortgage. Adopting a new standard may mean making changes to these critical systems. The consequences of IT changes in banks can be extremely disruptive if things go wrong. So adopting a new standard can be a complex, risky, and expensive process.

In addition to the challenges of transition, firms thinking about adopting a standard want to avoid adopting the wrong standard. Back the wrong standard and you may end up making expensive changes to your systems multiple times. Also, the benefits of a standard increase the more widely a standard is used. This creates a huge collective action problem: everyone wants to move to the right standard, but only once everyone else has moved first (Figure 1).

Collective action isn’t the only issue in standards adoption. Prior to adoption, standards have to be developed. In some respects this should be the easy bit. To create a standard for a loan, we need to write down the key data contained in a loan and how that data should be identified. But loans themselves may not be standard, and people may disagree about what the key components of a loan are, and what they should be called. Is the ‘principal amount’ of a loan the same as the ‘loan amount’ for example? And is a ‘charge’footnote [4] over a property in UK law the same as in French law?

Figure 1: There are challenges to address when introducing new data standards

Four icons that represent the challenges to adopting new data standards, and the need for collective action by firms

These challenges to standards adoption and development may be why, despite their critical importance, the widespread roll out of data standards in the financial sector has been limited. The financial sector has made progress in creating identifiers for some key things, most notably rolling out the LEI to identify legal entities. However, the content of financial data, data on the loan amount and interest of a mortgage for instance, continues to be managed by a huge number of proprietary and isolated standards.

Data standards and our Transformation Plan for data collection

Earlier this year, we published our Transformation Plan for data collection from the UK financial sector. Publishing the plan completed the data collection review that took place throughout 2020, announced in response to the Future of Finance report. Data collection is a foundational activity for us. Without collecting data, we would not be able to publish financial statistics, identify financial sector risks or make informed choices about how to respond to those risks.

But data collection is a costly activity for both the suppliers of the data and us as collectors. Those costs have risen as regulators asked for more data post crisis, and our data needs have changed with the growth of new analytical techniques, such as artificial intelligence. If we can be better at data collection it will bring benefits for us and for our customers – the people of the United Kingdom. For instance, if we spend less time fixing data quality issues, we could spend more time publishing a wider range of statistics.

Our Transformation Plan lays out what we think needs to happen to transform data collection over the next decade (Figure 2).

Figure 2: Our plan prioritises three key reforms to improve data collection

Icons denoting three elements of our plan: common standards, modern reporting instructions and integrated reporting approach

As in our 2000 and 2012 publications, at the heart of our proposed reforms is the need for the financial sector to develop and adopt data standards. We think these standards should sit alongside reforms to how we write our reporting instructions, such as publishing them in code rather than pdf, and integrating our various data collections.footnote [5] Together, we think these changes can help us meet our vision for data collection: ‘we get the data we need to fulfil our mission, at the lowest possible cost to industry’.

But we think delivering these reforms can do more than just improve reporting. As we found out during the review, issues that affect our data collections are in part caused by fundamental data management challenges throughout the financial industry. Fix these, and a whole range of data-driven activities become easier: from better financial reporting, cheaper payments, to more accurate credit scoring so you get the right financial products that suit your needs.

Moreover, in our Transformation Plan, we recognise the role that we, as a central authority in the UK and global financial sector, can play in delivering these reforms, and helping the development and adoption of data standards in particular. We recognise the ability we have to convene industry to develop standards, to raise the importance of data standards at the committees where regulators make the rules for the global financial sector, and, where necessary, to drive the use of standards by including them in our reporting instructions.

We intend to use these powers over the next decade as we work on delivering our Transformation Plan. Our work starts now by launching a multi-phase joint transformation programme with the FCA and industry (Figure 3). During each phase, we will aim to deliver a series of ‘use cases’ focusing on particular data sets. Each use case delivered will add value in its own right, as well as delivering improvements that can then be applied to other data over time. During the first two years of the programme we will look to deliver three use cases, including data standards for commercial real estate data. We expect the standards to be used as part of an industry project to create a commercial real estate database.footnote [6]

As in 2000, and 2012, we hope our Transformation Plan – and our vision for the future – launches another wave of development and use of data standards, and helps reshape financial services for the better.

Figure 3: Our timeline for the first two years of our Transformation Plan

  1. Including risks to our inflation target, to financial stability and the safety and soundness of the financial firms we regulate.

  2. Initially alongside the FCA as part of the Digital Regulatory Reporting programme.

  3. See discussion paper on Transforming Data Collection from the UK Financial Sector.

  4. A ‘charge’ is a legal method created by the Law of Property Act 1925 that gives a lender rights over a property used as collateral to a loan.

  5. We currently collect data for a number of purposes (including regulatory, statistical, and our markets and banking operations). The way we collect data can differ by each purpose.

  6. The database will then be used partly for reporting to us.

Share your thoughts with us at