Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) – using Equifax data to better understand the economy

This article sets out the progress we have made in assessing how data from the Credit Reference Agency, Equifax can improve the coverage, quality and granularity of our financial statistics.

The Equifax data cover both commercial and consumer borrowing. In this article, we explain where we expect to use them to augment and complement our financial account data within the national accounts and what we need to do to make this happen. This follows on from a previous article, which set out our plans to use commercial data within the UK National Accounts.

We have ambitious plans to transform our statistics over the coming years, informed by our Economic Statistics and Analysis Strategy and with the aim of increasing the robustness and quality of UK economic statistics. Working in partnership with the Bank of England, one element of our transformation work is the development of enhanced financial accounts (EFA) – in particular, more detailed “flow of funds” statistics – to meet evolving user needs.

The main aims of the EFA initiative are to improve the coverage, quality and granularity of financial statistics. This includes producing whom-to-whom statistics for the UK National Accounts and identifying the counterparty information for each financial transaction. A possible avenue for these improvements is through the use of data obtained from commercial sources – referred to as commercial data in this article. The benefits of commercial data over traditional surveys include:

• improved consistency of data resulting from the data being collated by a single source

• reduced burden on households and businesses

• increased volumes of data that provide the potential for greater granularity

There are also challenges from using these data, including:

• the potential for reduced alignment with national accounts concepts when compared with survey data

• gaps in the coverage of commercial data may be harder to identify than in a survey

One of our commercial data work strands covers the use of data to improve our financial borrowing statistics within the UK National Accounts, as defined within the European System of Accounts 2010: ESA 2010 framework. Following market engagement and an open tender process, we selected Equifax as our preferred service supplier for these data.

We initially gained access to aggregate data in summer 2017 following a successful presentation to the Steering Committee on Reciprocity (SCOR), who govern access to Credit Reference Agency data, and published a progress update that autumn. This update included a quality assessment that noted the potential for the use of Equifax in the financial account should be considered, provided more granular data be made available to Office for National Statistics. We have spent the first part of 2018 working with Equifax to produce a viable, granular dataset from their raw data. This has included allocating the data from a company level to a specific sector of the economy, definitions for which can be found in the explanatory notes published in November 2017. Our use of this dataset is discussed through the remainder of this article.

Equifax is one of three large Credit Reference Agencies (CRAs) in the UK. They collect various data including lending data from across the country. The Steering Committee on Reciprocity (SCOR) recommend that UK institutions involved in lending share their data with each of the CRAs. The data provided by Equifax cover both consumer (lending to individuals) and commercial lending (lending to other companies).

Although the majority of the data are UK based, there is a small amount of data for the Crown Dependencies of Isle of Man and the Channel Islands that, according to the European System of Accounts 2010: ESA 2010 residency concept, are included in the rest of the world. Figures have been provided for each calendar quarter from Quarter 1 (Jan to Mar) 2015 to Quarter 4 (Oct to Dec) 2017 inclusive. The data supplied include:

• end of quarter balances of amounts outstanding

• type of lending (for example, mortgages, hire purchase)

• any repayments made during the period where possible

• lender and borrower information that enables us to determine their location and type of business

Due to the sensitivity of the data as well as the contents, the data have been divided into different datasets. This ensures that although we are gaining granular information from Equifax, we are not breaching their confidentiality rules regarding the data they are permitted to supply to us. ONS also applies the appropriate set of personnel, physical and information security controls that offer an appropriate level of protection against a typical threat profile as set out in Government Security Classifications, issued by the Cabinet Office.

Part of the enhanced financial accounts (EFA) development work is to improve the quality, coverage and granularity of the UK National Accounts, including producing whom-to-whom data for the financial account. Our intended use of the commercial data sources (including Equifax) falls into three broad categories:

• replacing: we will look to replace existing sources within the accounts, where we can demonstrate that the commercial data are of an appropriate quality to do so

• augmenting: we will look to use the commercial data sources to augment existing sources; for example, this could be through producing whom-to-whom statistics for data that are available from a different source, providing a counterparty relationship dataset or addressing existing gaps in coverage

• complementing: we will look to produce analysis that furthers the aims of EFA and complements the financial account but may not feed directly into the national accounts; for example, this could include geographic information that may not fall into the core national accounts but alongside these data would give a greater insight into the UK financial market

To determine how we can use the Equifax data in the financial account (that is, for replacing and augmenting, as mentioned previously), we must be able to align the data with national accounts concepts and taxonomy, as per the European System of Accounts 2010: ESA 2010 framework. There are three initial stages to this.

Firstly, we must allocate each lender and borrower to the correct part of the economy. This is known as sectorisation. Secondly, we must allocate each type of lending to the correct national accounts financial instrument. Thirdly, we would need to identify the coverage of the Equifax data to implement any of these data into the financial account. In this section, we discuss our approach to achieving this alignment.

Sectorisation

Working in partnership with Equifax, we have been able to match their lender and borrower data to the relevant Standard Industrial Classification 2007: SIC 2007 (PDF, 1.3MB) categories. This classification provides a framework for the main business type for companies and allows us, based on this, to allocate companies to the right sector of the economy.

For the consumer information, we also need to sectorise the data but on the borrower side, this is done using the geography information provided, with all data falling into the households or rest of the world sectors. This has been completed using a combination of information provided by Equifax and the Office for National Statistics (ONS) business register. Both systems have advantages and by using this combination, we have been able to match approximately 90% of the information supplied by Equifax to a relevant SIC, although further work needs to be done to validate this matching.

Using the SIC information, we have allocated each lender and borrower a counterparty relationship to their respective sectors. Where the SIC information is not known, they have been allocated into an unknown sector. We will continue the work to allocate sectors as well as confirming mapped sectors are correct. The mapping of SICs to sectors can be seen in Annex 1.

Allocating lending types

This involves mapping the type of lending to the financial instruments in the national accounts, which are identified within the ESA 2010 framework. This framework classifies all financial transactions into a specific category to allow comparisons of national accounts information between countries. It also identifies the various aspects of the instruments that are required within the accounts for each type of transaction.

Although most of the lending types within the Equifax data would fit into the loans (F.4) category, there are a few lending types that are classified elsewhere, including some to the trade credits and advances (F.81) category. The latter relate to utility bill over- and under-payments and are relatively small in comparison to the loans data. The mapping of Equifax categories to the UK National Accounts concepts is something that we will look to refine to confirm that data are being fed through to the correct area of the accounts. This will include ensuring that we only use data that reflect activity that should be included in the financial account. For example, we will need to make certain that any leasing activity that may be incorporated in the hire purchase component of the Equifax data is removed, as leasing is not a financial transaction. The initial mapping of Equifax lending types to the UK National Accounts categories can be seen in Annex 2.

Table 1 shows a summary of the raw Equifax data after sectorisation and allocation of lending types for Quarter 4 (Oct to Dec) 2017. Comparable figures for all available quarters are included in the dataset. This table only includes sectors that are shown within the Equifax data and have been rounded to the nearest hundred million. It is worth noting that these are raw data that have not been grossed up to produce an overall estimate of borrowing in the UK.

The process of sectorising and allocating lending types enables us to determine where the Equifax data could be used in the financial account. Our view is that the data we are holding on monetary financial institutions (MFI) lending, received from the Bank of England, is of high quality and granularity. Therefore, based on our analysis of the Equifax data and other sources, we feel it is unlikely that we would look to replace it with the Equifax data. However, there may be some scope to use the Equifax data to help provide some analysis such as a geographic breakdown or further lending type analysis.

For lending by other financial institutions (OFIs), we expect the Equifax data to augment the new Financial Services Survey (FSS). OFI’s are institutions mainly involved in the financial sector that do not take the form of a typical bank or building society. The Equifax data could be used to disaggregate these data and produce the counterparty whom-to-whom information. Some work will need to be undertaken to ensure that the population within the Equifax data is representative of the survey population in comparison with other data sources. We hope to complete this work over summer 2018.

We will look to implement the Equifax data on trade credits and advances into the national accounts as a new data source. We detail the work that needs to take place to make this happen in the next section.

Coverage

There are multiple Credit Reference Agencies in the UK, meaning that no single agency (including Equifax) will have complete coverage of the lending market. Our starting assumption is that there is no obvious reason for bias in the coverage of companies within Equifax. This is because lenders are free to choose who they report their information through and whether they report to one or all the agencies. However, we will need to test this to ensure our assumptions are correct – in particular, are there any geographic biases in the coverage, and the coverage of the commercial lending data by different business sizes.

As part of our ongoing work we will compare the Equifax population to the UK population and estimate the coverage of the data. We will also investigate the history of the market to map through any significant changes in coverage and sectors over time. Once we have this information we will be able to gross the data up to full market coverage.

To use the Equifax data to populate the financial account in the UK National Accounts, we must be able to populate all the relevant national accounts concepts within the European System of Accounts 2010: ESA 2010 framework. This means that we require five main elements of the data as well as assessing the coverage of the data. We would also need to project these data back historically to populate the full national accounts time period and to look at how to align the data to the national accounts publications. Our method for estimating each of these areas for the trade credits and advances (F.81) section, related to over- and under-payment of utilities, is detailed in this section.

Levels

For this we require the value of outstanding debt in each category at the end of each quarterly period. The data supplied by Equifax provide this and so can be directly used to populate this element of the accounts.

Flows

This refers to the changes in levels during a period including any repayments made. Equifax have provided the information where they can but unfortunately it is not possible to identify this information in all cases. For those places where this information is not supplied, this could be calculated using the change in levels from the previous quarter to the new quarter minus any interest, revaluations or changes in volume as explained later in this section. In practice, as we are assuming interest and revaluations to be equal to zero, this would be calculated as the change in levels between quarters, minus other changes in volume.

Interest

Within national accounts it is important to note that any interest accrued on an item should be separately identified. Equifax is not able to provide us with this information in their files, as any interest is included in the value of the agreement itself and cannot be removed.

In terms of the interest on the over- and under-payment of utilities, it is unlikely that any interest would be accrued for this type of transaction. Although companies can charge fees and interest, it is rare for this to happen. Therefore, we are assuming that the interest for this category of lending would be zero.

Revaluations

In the context of these data, this would only consist of currency changes. All the data supplied by Equifax are collected in Great British Pounds (GBP) and therefore we cannot calculate any changes in relation to currency within the data. As we are looking at UK transactions in the main, we can reasonably assume that all transactions would be in GBP and therefore the value can be assumed to be zero.

Other changes in volume

In terms of the under- and over-payment of utilities, there may be some changes that would be relevant here. These could include things such as fees or other items but these are not evident in the data. We will work with Equifax as well as national accounts experts to identify any changes that should be captured here.

Historical data

To meet user needs, we aim to create a time series back to at least 1987. To do this, we also need to consider changes in the sector classification of the companies within the data. This will be especially important where companies may have moved between public and private ownership, resulting in a change of sector in national accounts terms. For lenders, we will track their ownership status through time. For borrowers (particularly where data are provided on an aggregate basis), we will develop a rule-based approach to allocation.

Timeliness

The timeliness of new data is currently approximately 11 weeks after the end of the reporting period. We currently publish the national accounts data at the end of the quarter following the reference period (so the Quarter 2 (Apr to June) 2018 release is published at the end of Quarter 3 (July to Sept) 2018). For use within the production of regular statistics, this lag would therefore mean data would be too late to be processed for use within the relevant quarterly National Statistics. We would therefore have to forecast the data for the period and then redeliver in the next period the correct value.

Coverage

We will assess coverage in line with the steps outlined in section 4. For F.81, this will include assessing the market share of those utility companies included in the Equifax data.

In addition to using the Equifax data to augment existing data within the financial account, we intend to use it to produce analysis and statistics that complement the financial account. These analyses are in line with the aims of the enhanced financial accounts (EFA) initiative to improve understanding of how money moves in, around and out of the UK economy. An initial view of such analysis is included in this section.

It is worth noting that the analysis included in this section is based on the raw Equifax data and has not been grossed up in any way to give a full population coverage. This means that any trends shown may, in part, result from a change in the coverage of the Equifax data over time. Moreover, this analysis generally uses Equifax's classifications rather than European System of Accounts 2010: ESA 2010 national accounts classifications.

Trends in the take up of different lending types

The Equifax data contain 28 unique types of lending across the consumer and commercial data, which group into 10 aggregated categories. Annex 2 shows the initial mapping of these lending types to the financial transactions for information.

Figure 1 shows that mortgages are by far the largest category of lending, accounting for over 85% of total lending by value. Figure 2 shows the value of lending per quarter for each lending type, excluding mortgages, from 2015 through to the end of 2017 across both consumer and commercial lending. The hire purchases category has grown in value over the last few years to become the second-largest type of lending, overtaking credit card lending. This has been driven by increases across all three categories of hire purchase (fixed-term deferred (buy now and pay later), hire purchase and non-auto finance, and varying subscription). As per ESA 2010, hire purchase is a type of financial leasing where goods are purchased through making instalment payments over time. Under a hire purchase arrangement, the buyer takes possession of the good immediately but does not obtain ownership of the good until all payments have been made. The other major category is unsecured short-term lending, which includes payday loans. A payday loan is a type of short-term borrowing based on an individual’s income and credit profile normally at a high interest rate.

Figures 3 and 4 provide additional insight on the mortgage market, using the Equifax classifications. Figure 3 shows that standard mortgages (which account for the majority (around 98%) of the value in the Equifax data) have increased in value by 11% over the period. Figure 4 shows that second mortgages have shown a similar increase, while the value of flexible and buy-to-let mortgages have decreased by around one-quarter. However, not all lenders provide this granularity of information. We will work with Equifax to develop a method to estimate the scale of the impact this has.

The Equifax data enable us to differentiate between commercial and consumer lending. The latter accounts for around 98% of the total value of the lending, or 88% of the value if mortgages are excluded.

Figures 5 and 6 show the split of lending types for the Equifax consumer and commercial lending respectively. Mortgages represent the largest lending category within the consumer data, accounting for around 87% of the value. This compares to 79% reported in Blue Book 17. Within the commercial sector, hire purchase represents the largest form of lending, accounting for around two-thirds of the total value of lending.

In this article, we have demonstrated the value the Equifax data offer in improving our understanding of how money moves in, around and out of the UK economy. In summary, we expect these data to augment the financial account through increasing the granularity of data, providing whom-to-whom information, and providing new information on accounts receivable and payable. We will also use the Equifax data to enrich our picture of the economy through producing complimentary analyses, focusing on (for example) trends in lending types.

The next stage of working with the Equifax data is to:

• validate the sectorisation process to ensure that the sectors are appropriate

• refine sectorisation for non-financial companies – the current matching approach does not enable us to distinguish between private and public non-financial corporations; we will use a combination of apportionment and manual sectorisation to produce separate estimates for these two sectors

• refine and validate the mapping of lending types to national accounts concepts

• develop a process for using the data to apportion the financial services survey data into the appropriate whom-to-whom sectors

• investigate the use of the data for geographic or lending type analysis of the Bank of England data

• develop estimates of flows for the trade credit and advances data, in line with the description in section 5

• investigate if any other changes in volume would be appropriate for the accounts trade credit and advances data and if so, develop estimates for this in line with the description in section 5

• develop estimates for the back-history figures for the data, focusing on accounts trade credit and advances data and lending by other financial institutions (OFIs)

• identify the population coverage of the data and what steps need to be taken to gross the data up to the population for the trade credit and advances data transaction

Over the second half of 2018, we will start the processes for incorporating the commercial data into the financial account, and start building the systems that will enable us to do this. During this period, we will publish further analyses using the Equifax data that will showcase how it will complement the financial account.

Having determined that the Equifax data have the potential to augment and complement our current financial account data, we will look to re-engage with other Credit Reference Agencies (CRAs). Through acquisition of data from all CRAs, we would have a higher level of coverage for lending, which would help us to overcome challenges of gaps and enable even more granular analysis and understanding of the market than is possible with one CRA’s data.

Taking a broader look across the enhanced financial accounts (EFA), during 2018 and into 2019 we will be sharing the outcomes of our research and development into using new and improved existing data sources to enhance the financial account. This will include:

• use of the Financial Services Survey

• developing pre-1987 historic estimates of the financial account

• work to improve measurement of financial activity in the monetary financial institutions sector

• use of Bank of International Settlements data for international lending

Towards the end of 2019, we will publish Experimental Statistics to provide a cohesive picture of the accounts as part of the annual update to the flow of funds matrices. In parallel, we will continue to publish analytical work, including updated visualisations of the data and examples of how the data could be used.

We welcome feedback on our plans and work to date. Please email FlowOfFundsDevelopment@ons.gov.uk with any feedback or any questions regarding this article.

The Authors would like to thank the following for their contributions to this article: Kayode Adaramodu, Laurence Day, Khloe Evans, Oliver Haines, Phil Tyrer, Abbe Williams, David Cobbledick, Simen Svenkerud, Philip Davies, David Matthews, Philip Tyrer, Phillip Lee, Daniel Wisniewski, Tomos Davies, and Keith Lai.

Flow of funds archived background information

Explanatory notes:

• Institutional sectors and financial instruments

• AF.2 Currency and deposits

• AF.3 Debt securities

• AF.4 Loans

• AF.5 Equity and investment fund shares or units

• AF.6 Pensions, insurance and standardised guarantee schemes

• AF.71 Financial derivatives

1 May 2018 - Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) – shadow banking introductory article

30 April 2018 - Experimental financial statistics for insurance using Solvency II regulatory data - enhanced financial accounts (UK flow of funds)

15 February 2018 – Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) – historic households and non-profit institutions serving households (NPISH) sectors data on currency and deposits

17 November 2017 – Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) – 2017 matrix update

23 October 2017 – Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) – progress on commercial data use

25 September 2017 – Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) reconciling sources of historic data for households and the non-profit institutions serving households (NPISH) sectors

12 September 2017 – Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) – A flow of funds approach to understanding financial crises

31 August 2017 – Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) estimating the value of other accounts receivable or payable in the UK economy

21 July 2017 – Economic Statistics Transformation Programme: enhanced financial accounts (UK flow of funds) improving the measurement of company quarterly profits

3 July 2017 – Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) progress on financial derivatives data

5 June 2017 – National Accounts articles: The UK Enhanced Financial Accounts; the introduction of the new securities dealers survey data and expansion of financial sub-sector detail

31 May 2017 – Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) commercial data use

31 May 2017 – Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) improving the economic sector breakdown

27 April 2017 – Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) employee stock options

24 April 2017 – Financial intermediation services indirectly measured (FISIM) in the UK revisited

29 March 2017 – Economic Statistics Transformation Programme: Enhanced financial accounts (UK flow of funds) Government tables for the special data dissemination standards plus (SDDS plus)

30 January 2017 – The UK Enhanced Financial Accounts: changes to defined contribution pension fund estimates in the national accounts; part 2 – the data

16 January 2017 – The UK Enhanced Financial Accounts: changes to defined contribution pension fund estimates in the national accounts; part 1 – the methodology

8 August 2016 Article – Economic Statistics Transformation Programme: UK flow of funds experimental balance sheet statistics, 1997 to 2015

14 July 2016 Article – Economic Statistics Transformation Programme: flow of funds - the international context

14 July 2016 – Economic Statistics Transformation Programme: Developing the enhanced financial accounts (UK flow of funds)

10 March 2016 – Identifying sectoral interconnectedness in the UK economy

24 February 2016 – Improvements to the Sector and Financial Accounts

12 January 2016 – Historical Estimates of Financial Accounts and Balance Sheets

6 November 2015 – Comprehensive Review of the UK Financial Accounts including explanatory notes for each financial instrument covered in the article

13 July 2015 – Introduction Progress and Future Work