Skip to main content
  • This website sets cookies on your device. To find out more about how we use cookies please refer to our Privacy and Cookie Policy. By continuing to use the site, we’ll assume that you are content for us to set these on your device.
  • Close
Home > Research > Datasets


The Bank announced in its Strategic Plan our commitment to opening up more of our data to the public in order to crowdsource answers to key research questions and support collaborative research with Bank staff. Today we are publishing six datasets alongside the One Bank Research Agenda. More will follow. Data published today may be helpful in addressing themes and questions set out in the One Bank Research Discussion Paper, as detailed below. Members of the public are also invited to use these data as part of the Bank of England’s Data Visualisation Competition. The six datasets are:


Inflation attitudes survey data

This dataset contains the individual level responses to the Inflation Attitudes Survey, a quarterly survey of households’ attitudes about inflation and other key economic variables. The dataset below covers the period from 2003 to 2016 on a quarterly basis, with annual data for 2001 and 2002. Since February 2016 the survey has been conducted on the Bank of England’s behalf by TNS, previous to that it was conducted by GfK.  The first spreadsheet contains a list of questions asked on the survey and all possible responses. The second spreadsheet contains descriptions for all other variables in the dataset. The final spreadsheet is the full dataset; the numbers in the dataset refer back to the previous spreadsheets. The dataset is also available in STATA format. To access these, please see the Bank of England/TNS Inflation Attitudes Survey page, where updated versions of both the csv and STATA files are available. Email if you have any questions. 

In an inflation targeting framework, inflation expectations are a key measure of central bank credibility and can give an indication of wider economic developments. This micro-level dataset provides greater scope for investigating inflation attitudes, including inflation perceptions and expectations. The Bank of England has previously published summary statistics including medians for these data at a macro-level (for more information see the Bank of England/TNS Inflation Attitudes Survey page). Sample sizes are around 2,000 per quarter, except in the February quarter when around 4,000 people are surveyed. The responses are subject to sampling error as one sample of people will differ from the characteristics of the population as a whole and that will cause small variations in summary results. This new version of the data allows users to further analyse responses split by demographic variables such as age, region and gender.
Some of the One Bank Research Agenda questions which this data may help answer include:
  • How do public communications and disclosure policies affect behaviour and incentives? (Theme 3)
  • How is the reliability of household and corporate data sets affected by survey methodology? (Theme 4)
  • How do demographics and the distribution of wealth and income in society affect the monetary transmission mechanism? (Theme 5)

Download Inflation attitudes survey dataset (22MB)

Bank of England/NMG household survey data

The Bank of England/NMG survey is an annual survey of households conducted by NMG Consulting on behalf of the Bank. It includes questions on households’ balance sheets and spending. The survey has been conducted annually since 2004, during September. Between 2004 and 2011 the survey was carried out face-to-face. The survey has been fully online since 2012. The first workbook contains individual survey responses from the face-to-face survey. The second workbook contains the online survey data. Results from each of the surveys are summarised in an annual article published each year in the Q4 Quarterly Bulletin. See the most recent QB article.
The Bank of England/NMG survey data provide a more timely update of developments in households’ finances than other surveys, which are typically published with a longer lag. They also can be used to address the following One Bank Research Agenda questions:  
  • What determines marginal propensities to consume across households and how they differ across household types? (Theme 4)
  • What determines the distribution of household indebtedness and mortgage arrears? (Theme 4)
  • How is the reliability of household datasets affected by survey methodology? (Theme 4)

Download Bank of England/NMG household survey data - 2004-11 (7MB)
Download Bank of England/NMG household survey data - 2011-15 (15.2MB)

Agents’ historic company visit scores

The Bank of England’s Agents collect economic intelligence from the business community around the United Kingdom. The intelligence is largely qualitative, but Agents also make quantitative assessments in the form of scores. The Bank has published Agents’ macroeconomic scores monthly since 2006. Further information on these scores is available in The Agents' scores: a review (2008). Since mid-2007, the Agents have also assigned ‘company visit scores’ (CVS) based on information gathered from their confidential meetings with individual UK firms. Details of the methods used to assign scores for each variable, and some internal applications, are set out in The Agents' company visit scores (2013). In setting the macroeconomic scores, Agents may have regard to CVS, but may also use other sources of information, such as surveys and sectoral statistics provided by business organisations.
The dataset being published contains historic and anonymised company visit scores assigned from the beginning of 2008 to the end of 2014.
Data are scores assigned by Agents, on a scale of -5 to +5, after visits to individual companies, on eleven economic variables: demand (nominal turnover), exports, investment, capacity utilisation, employment, recruitment conditions, total labour costs per employee, pay (salary per employee), non-labour costs, output prices, and profits (growth in pre-tax profit as share of turnover). 
Most of the scores describe a change in the level of the variables. In most cases, the backward-looking score is for the past three months relative to the same three months a year ago, and the forward score is the expected growth between the past three months and the same three months a year ahead. No change in the level would be scored as zero, and growth broadly in line with its past average rate would be scored as + or – 2, depending on whether positive or negative growth has been observed on average in the past.  Recruitment difficulties and capacity utilisation are both scored (backward) according to the level in the past three months relative to what is considered normal for that firm, and (forward) the expected level for the corresponding three months a year ahead relative to normal; a normal level of utilisation or difficulty would be scored as zero. Although internal guidelines are provided, scores are set judgmentally and not drawn from a fixed or target distribution or a model.
Date range is 2008 Q1 to 2014 Q4.
Reporting frequency: Scores are assigned and input to the internal database continuously. In this dataset they have been aggregated into quarters. Agents visit most companies about once a year, but the intervals may be either longer or shorter. Contacts firms are continuously added and subtracted, so there is no presumption that any firm has been scored more than once, or that a firm scored in a given quarter will also have been scored in the corresponding quarter in the previous or following year. 
Anonymisation: Contacts’ conversations with the Bank’s Agents are confidential and may have contained commercially sensitive information. Consequently the Bank will only publish historic scores, with a lag of at least two years. The following steps have been applied to the dataset to ensure that the identity of the individual firms scored by the Agents cannot be identified:

- no names are provided, no indication of company size (turnover or employment), and no details of ownership;

- within each calendar quarter, company scores have been listed in random order;

- a broad indication of economic sector is provided: 

(i) production, covering Standard Industrial Classification (SIC) categories A to F; 
(ii) distribution, hotels, catering, arts, entertainment, recreation and other services, covering SIC G, I, R and S;
(iii) transport, storage and communications, covering SIC H and J; and
(iv) business and financial services, covering SIC K to N. 
Any scores for companies in SIC O, P, Q, T, and U (public administration, education, health and social work, households as employers, and extraterritorial organisations) have been omitted;

- in respect of any particular visit, some of the variables may not have been scored but left blank, either because they were inapplicable (eg because the company did not export) or because no clear view could be formed from the information provided. Of the dataset provided here, around 25% of the cells were originally blank. To enhance anonymity, values have been imputed to these blank cells in this published dataset. The distribution used for random imputation is that of the actual scores given separately for each of the 22 different variables in each of the seven different calendar years (ie 154 separate distributions in all) without distinguishing either between the quarters within those years or the four broad activity categories.
The full CVS dataset has several features distinguishing it from the Agents’ macroeconomic scores and official economic data:
- forward-looking scores are available, providing a measure of business expectations for the key variables scored;
- inclusion of scores for individual companies allows distributional questions, as well as totals and means, to be examined; for example what proportion of companies were operating above normal capacity at a given period? 
- scores for each variable are ‘co-observed’. This allows behavioural relationships between economic variables to be examined at a microeconomic level, for example whether there is a relationship between firms’ expectations for employment and their profit performance, or between the extent of pressure on capacity and price setting.
The use of the imputation process means that the second and third properties may not be fully preserved in this published dataset. 
The dataset is relevant to the One Bank Research Agenda questions:
  • How can surveys and detailed structured data sets be used to improve understanding of household and corporate behaviour? (Theme 4)
  • What determines the supply potential of the economy? (Theme 5) 

Three centuries of macroeconomic data

This dataset contains most of the series originally used for the 2010 Q4 Quarterly Bulletin article The UK recession in context — what do three centuries of data tell us?. The first edition of the spreadsheet now has been expanded and updated.  The dataset now has a dedicated webpage which can be found at the link below.

Three centuries of data webpage

The Bank of England's historical balance sheet

This dataset currently consists of two spreadsheets.  The first contains annual data on the Bank’s assets and liabilities between 1696 and 2014 that was originally published in February 2015 following the launch of the One Bank Research Agenda.  The second spreadsheet is a higher frequency dataset that contains weekly data on the assets and liabilities of the Banking and Issue Departments of the Bank since September 1844. 

These data have been in the public domain before but only now have they been published in a user-friendly spreadsheet form and as continuous time series.  The expansion of the Bank’s balance sheet in response to the recent crisis has been much discussed.  The spreadsheets are intended to offer historical perspective but also facilitate research into the role of the Bank’s balance sheet as a policy instrument.


QE-related data

These data pertain to the Bank's Quantitative Easing (QE) programme, specifically the Bank's Asset Purchase Facility (APF), that is, data on gilt, commercial paper and corporate bond operations. The first worksheet contains data on APF gilt purchase auctions, including the amount allocated at each (from March 2009 to January 2015 on each operation date). The second worksheet contains data on APF holdings of corporate bonds and commercial paper, including weekly purchases and sales (from February 2009 to February 2015 on a weekly basis). The third worksheet contains data on APF corporate bond purchase auctions, including the amount allocated at each (from March 2009 to June 2013 on each operation date). The fourth worksheet contains data on APF corporate bond sale auctions, including the amount allocated at each (from January 2010 to April 2013 on each operation date).

This workbook brings a prior version up to date. It also contains links to the data which are routinely updated on the Bank’s website. These data may aid research on the following One Bank Research Agenda questions:

  • What has been the impact of central bank asset purchase facilities on bond yields, term premia, and risk taking? (Theme 3)
  • Is the impact of QE-like policies state-contingent? How effective are the interventions under different monetary conditions? What are the lessons for exit strategies? (Theme 3)
  • To what extent does the composition of the central bank's assets matter relative to the absolute size of the central bank balance sheet? (Theme 3)
The earlier workbook also contained information on yield curves. The Bank of England estimates and publishes yield curves for the UK on a daily basis. They are of three kinds. One set is based on yields on UK government bonds (gilts) and includes nominal and real yield curves and the inflation term structure for the United Kingdom. Another set is based on sterling interbank rates (LIBOR) and on yields on instruments linked to LIBOR, short sterling futures, forward rate agreements and LIBOR-based interest rate swaps. These commercial bank liability curves are nominal only. The third set is based on sterling overnight index swap (OIS) rates, which are instruments that settle on overnight unsecured interest rates (the SONIA rate in the UK). OIS curves are nominal only. For each set of yield curves, daily data are published showing estimated zero-coupon yields (both forward and spot) for a wide range of maturities.
Data on yield curves are helpful both in analysing the impact of QE and addressing other research questions. Some of the One Bank Research Agenda questions these yield curve estimates might help answer include:
  • What can we learn from central bank and government crisis interventions? (Theme 3)
  • What accounts for the correlation of economic cycles and asset prices across countries? (Theme 4)
  • How might shifts in demographics and income distribution affect equilibrium real interest rates and the wider economy and financial system?  (Theme 5)
See the Bank of England’s yield curve estimates.
Download QE-related data (3.6MB)