Economic Statistics Centre of Excellence
Addressing the challenges of measuring the modern economy Contents
Foreword 3
Introduction 4
About ESCoE 5
Our story 7 Research overview 9 Bringing together expertise 13 The next generation 17 Creating space for debate 21 ESCoE Economic Measurement Conference 25 Working with ONS 29
Our projects 31
National accounts and beyond GDP 33 Productivity and the modern economy 57 Regional and labour market statistics 81 Communicating and valuing economic statistics 105
What next? 113
Our publications 119
1 2 Foreword Introduction
Today’s economy is more complex than it The Centre’s objectives specifically Economic statistics are fundamental to the ESCoE has experienced in its first has ever been. Global supply chains, digital include supporting the Office for National evidence-based policy making and the years is testament to the impact on intermediation, new goods and services, Statistics, and we are very grateful for decisions that we all make. To serve economic measurement that this collective and constantly evolving business models their efforts. They have helped us as an their purpose they need to keep up endeavour can make. make it ever more challenging for those organisation become more confident, with constantly changing economies, seeking to study the economy, and for open and outward-facing and have drawing on new data sources and the best It is a pleasure to introduce this report. policy and decision makers whose job is to inspired us to look at ourselves differently. available methods. This calls for a creative, It sets out what we’ve achieved together shape it. research-led approach to maintaining and in our first three years. I hope you will I would like to thank all those, working developing them. join us in the years ahead in building the The role of the Office for National Statistics both in the Office for National Statistics expertise, research and collaborations is to describe and measure the economy and in the Centre and its partners, who Established three years ago, the Economic needed to deliver better measurement to help government, policymakers, have contributed to its achievements. Statistics Centre of Excellence (ESCoE) is all and understanding of modern economies. communities, businesses and citizens I look forward to the next stage of our about bringing to the fore this research-led make better decisions. The challenges partnership. approach to measuring the economy. we face in measuring this modern economy cannot be met on our own. The Centre’s ambition is to be an Only by enlisting the support of the wider international point of reference for academic and expert community can we measurement research, supporting hope to better understand and measure cultural change in the delivery of economic the fast-changing and increasingly diverse statistics. economy of the twenty-first century. Our approach is clear. It involves building The Economic Statistics Centre of expertise, increasing collaboration Excellence is the delivery method of our between statistics producers, academia, formal collaboration with that wider policymakers and other data users, and academic and expert community, and its raising the profile of measurement issues, work in these first years has been broad, Jonathan Athow not least amongst economists. Rebecca Riley productive and impressive. Research to Deputy National Statistician and Director date has ranged from conceptual issues, Director General, Economic Statistics We do this by addressing both theoretical Economic Statistics Centre of Excellence such as how we move beyond GDP, Office for National Statistics and practical questions that arise in through to some very specific questions describing the modern economy, relying about how we will use new data sources. on our large and international network of Research Associates and Affiliates, The Centre has successfully developed a and investing in the next generation of strong and active network of world-class experts. We carry out our research in experts. It continues to lead cutting- dialogue with statistics producers and edge work today, with an eye ever on the promote constructive discussion through future of the economic statistics research workshops and conferences. community. But significant challenges remain, and the experience of this initial There are many people involved in period shows the tantalising prospect of delivering the ESCoE - at our partner what could be achieved in the coming institutions, at the Office for National years. Statistics, and amongst our wider network. The goodwill, support and interest that
3 4 About ESCoE
“The establishment of the Economic Statistics Centre of Excellence (ESCoE) was an important step forward in implementing some of the Bean Review’s key recommendations. There’s already been excellent output from ESCoE’s exciting research agenda and we look forward to seeing ESCoE’s research translated into improved economic statistics over the medium-term.”
Andy Haldane Chief Economist Bank of England
5 6 Our story
The UK Economic Statistics Centre of Our work is guided by our Advisory Board, Excellence (ESCoE) is a hub of world- which includes senior figures from key ESCoE Objectives leading expertise built around the stakeholder institutions, and supported by • To be an international focal point analysis of emerging and future issues our Management Team. It is underpinned of reference for world-class research in measuring the economy. It explores by dialogue across stakeholders through on economic measurement and the methods for measuring new and existing regular seminars, workshops, publications, development of economic statistics for forms of economic activity and welfare and presentations concerning economic the modern economy. in a global economy, for meeting the statistics and measurement, as well as needs for local, city and regional statistics, the forum provided by the annual ESCoE • To create a network of world-class and for understanding productivity. Conference on Economic Measurement. experts in economic measurement We use advanced tools from economic to enable the delivery of a cutting- theory, econometrics and data science, edge research programme to improve mathematics and statistics. understanding of the economy through data and statistics. The ESCoE was established in 2017 with • To provide an environment for the the support of the UK Office for National development of a wider research Statistics (ONS, the UK’s National Statistical community with skills and research Institute), following the Independent experience in economic measurement, Review of UK Economic Statistics by building the next generation of Professor Sir Charles Bean, published academic experts in this field. March 2016. This Review recommended that “in conjunction with suitable partners improved economic measurement to meet • To deliver dissemination mechanisms in academia and the user community, ONS user needs, through research, education to bring international expertise to bear on research on economic measurement should establish a new centre of excellence and convening of the wider economic and economic statistics in the UK. for the analysis of emerging and future issues measurement community. This report in measuring the modern economy.” showcases our first three years. • To develop collaboration between the academic research community Established as an independent research Hosted at the National Institute of and statistical agencies, offering centre, the aims of ESCoE are to Economic and Social Research (NIESR) the capacity for fundamental develop collaboration between national in London, ESCoE’s original consortium methodological and conceptual work statisticians and academia to bring a included the University of Cambridge, that will feed into the development research-led approach to addressing the Fraser of Allender Institute at the and improvement of economic challenges and opportunities in measuring University of Strathclyde, King’s College statistics and data-processing the economy. ESCoE is the UK’s first-ever London, innovation foundation Nesta, and methodologies. dedicated academic centre of expertise for Warwick Business School at the University economic measurement and one of just a • To support ONS to deliver world-class of Warwick. This consortium has grown economic statistics that meet user handful around the world. significantly throughout the early years of needs and allow ONS to exert greater ESCoE. Bringing together a partnership of influence over the direction of travel Our ambition is to be an international leading UK and international institutions, of the international standard setting- focal point of reference for economic academics and researchers, we have agenda. measurement research and the created a hub of expertise, enabling development of economic statistics for us to deliver a cutting-edge research the modern economy. We work with programme. ONS and other stakeholders to deliver
7 8 Research overview
years include: priorities and their vision of delivering of this document. To be an international focal point of world-class economic statistics. This aim reference for world-class research on economic measurement and the National accounts and beyond GDP: The runs through all aspects of the ESCoE development of economic statistics for development of a democratic measure of research programme to support the ESCoE Publications Series the modern economy. income growth; the collation of long runs development of UK economic statistics and of National Accounts data on a consistent capability. By bringing together a network Launched in 2017, the ESCoE publications basis; the development of methodologies of internationally recognised academics series includes three types of papers and for bringing tax records into the timely and researchers, each with their own was until recently edited by Professor ESCoE’s research considers the measurement of GDP; and proposals for area of specialisation, we are able to Richard J. Smith, ESCoE’s Academic implications for measurement of economic experimental well-being accounts. deliver progress across a broad range of Director. This series is an excellent vehicle activity associated with the modern economic measurement issues to a high for disseminating research on economic global and digital economy, as well as Productivity and the modern economy: academic standard that provides benefit measurement; we encourage submissions more established limitations of economic A framework for thinking about and impact to economic statistics users from researchers and practitioners statistics. Our research explores the measurement issues arising from and producers in the UK and elsewhere. working in this field at ESCoE and possibilities for using novel data sources digitalisation, including for measuring the Each project is headed by an ESCoE elsewhere. to inform the development of economic value of data; detailed sector estimates professorial lead to ensure that research statistics and new methods for collating of trade-in-value added; a detailed is carried out to world-class standards. The series is a repository for discussion and improving economic statistics from decomposition of the UK productivity Designated ONS staff are associated papers, technical reports and occasional existing data sources. The research puzzle; and a new business survey on with each project to ensure knowledge papers concerned with best-practice is comprised of a core programme of productivity, management and business exchange and to enhance impact and economic statistics and measurement projects as well as additional contractually expectations. benefit within ONS and in economic and research into new methods of funded research that augments and statistics production. measurement in the modern economy. It reinforces our mission. Regional and labour market statistics: The has seen papers submitted on a diverse development of a data-driven taxonomy ESCoE also undertakes work on range of topics; we have published more In addition to ONS our funders to date of skills; the exploration and evaluation communicating and improving public than fifty papers to date. include UK government departments, the of new data sources for measuring understanding of economic statistics, Organisation for Economic Co-operation international migrant populations in supporting our objective to build and Development (OECD) and the UK local areas of the UK; an evaluation of interest and the next generation of Research and Innovation’s Economic and the exposure to changes in international academic experts in the field of economic Social Research Council (ESRC). trading relationships across the UK; and measurement. the development of high-frequency, timely Our ambitious research programme is estimates of regional GVA growth. To date we have completed more than carried out in four broad workstreams, twenty research projects, with many more reflecting the high-level issues that Communicating and valuing economic currently underway. This work has resulted economic measurement research needs statistics: An experiment to understand in numerous seminar, workshop and to address to improve understanding the effects of communicating data conference presentations and discussions; of the economy and meet users’ needs: uncertainty; and an explorative study of ESCoE and other peer-reviewed academic National accounts and beyond GDP, ways of measuring the value of official papers; and social and mainstream media Productivity and the modern economy, economic statistics. interest. Regional and labour market statistics, and Communicating and valuing economic Research projects are developed by You can find out more about our individual statistics. ESCoE Research Associates with a view research projects, including information on to supporting ONS in addressing their methods, findings and recommendations, Research highlights from ESCoE’s first few Economic Statistics and Analysis Strategy as well as impacts, in the ‘Projects’ section
9 10 We’ve presented our research at many UK and international events
American Economic Association ASSA Annual Meeting, San Diego, CA, 2020 - Centre for Applied Economic Research Annual EMG Workshop, University of New South Wales, Sydney, 2019 - Eurostat New Techniques and Technologies for Statistics (NTTS) Conference, Brussels, 2019 - London Business School (LBS) Research Seminar, 2019 - Cambridge-INET Institute Workshop on score-driven time series models, Cambridge, 2019 - King’s Business School PhD Symposium, London, 2019 - Royal Economic Society (RES) Annual Conference, Warwick, 2019 - Society for the Study of Economic Inequality (ECINEQ) 8th Meeting of the Society, Paris, 2019 - European Economic Association (EEA) and European Meeting of the Econometric Society (ESEM) 34th Annual Congress EEA / 72nd Meeting ESEM, Manchester, 2019 - St Louis Fed Applied Time Series Econometrics Workshop, St Louis MO, 2019 - The International Capital Market Association (ICMA), University of Reading Centre Seminar, Reading, 2019 - Society for Nonlinear Dynamics and Econometrics (SNDE) 27th Annual Symposium, Dallas, 2019 - Warwick Business School Conference, 2019 - Organisation for Economic Co-operation and Development (OECD) Workshop on Time Series Methods, Boulogne, 2019 - Office for National Statistics (ONS) Productivity Forum, London, 2019 - Centre for Competitive Advantage in the Global Economy (CAGE) Policy and the UK’s Competitive Advantage: Evidence from 10 Years of Research Conference, Warwick, 2019 - German Industry UK (GIUK) GIUK Roundtable Discussion on Productivity, London, 2019 - Society for Economic Measurement (SEM) Annual Conference, Frankfurt, 2019 - University College London (UCL) Seminar on Measuring and Pursuing Wellbeing, London, 2019 - Bank of Canada 3rd Forecasting at Central Banks Conference, Ottawa, 2019 - Brookings Institution Conference, Washington, DC, 2019 - Bruegel How Should We Measure the Digital Economy Panel Discussion, Brussels, 2019 - National Bureau of Economic Research (NBER) The Macroeconomic Valuation of Health Conference, Boston, MA, 2019 - US Census Bureau International Management Surveys Workshop, Suitland, MD, 2019 - Nesta Education Conference, London, 2019 - Royal Statistical Society (RSS) Conference, Cardiff, 2018 -Government Statistical Service (GSS) Methodology Symposium, London, 2018 - King’s Business School PhD Symposium, 2018 - International Association for Research in Income and Wealth (IARIW) 35th General Conference, Copenhagen, 2018 - Computational and Financial Econometrics Network (CFE Network) 12th International Conference on Computational and Financial Econometrics, Pisa, 2018 - International Institute of Forecasters (IIF) 38th International Symposium on Forecasting, Boulder, CO, 2018 - Bank of England (BoE) 2nd Conference on Forecasting at Central Banks, London, 2018 - World KLEMS 5th Conference, Cambridge, MA, 2018 - International Monetary Fund (IMF) 6th IMF Statistical Forum, Washington, DC, 2018 - French Economic Association (AFSE) 67th Congress, Paris, 2018 - World Economic Forum Annual Meeting, Davos, 2018 - Royal Economic Society (RES) Annual Conference, Brighton, 2018 - Department for Business, Energy & Industrial Strategy (BEIS) Seminar, London, 2018 - European Central Bank (ECB) Roundtable on Digitalisation, 2018 - Centre for European Economic Research (ZEW) Public Finance Conference, Mannheim, 2018 - Middlesex University Seminar, London, 2018 - World Inequality Database (WID) World Conference, Paris, 2017 - Bank of Spain 13th Annual Conference on Real-Time Data Analysis, Methods, and Applications in Macroeconomics and Finance, Madrid, 2017 - International Monetary Fund (IMF) 5th IMF Statistical Forum, Paris, 2018 - Money Macro and Finance Research Group (MMF) Annual Conference, London, 2017 - Royal Economic Society (RES) Annual Conference, Bristol, 2017 - University of Strathclyde Seminar, 2018 - Heriot- Watt University Seminar, 2018 - University of Southern California Seminar, 2018 - Institute for Fiscal Studies (IFS) Seminar, 2018 - Royal Institute of Chartered Surveyors (RICS) Seminar, 2018
11 12 Bringing together expertise
Our partner institutions include: makes an invaluable contribution to our teams and presenting their own work at To create a network of world-class dissemination programme, and especially ESCoE seminars and workshops, each experts in economic measurement to enable the delivery of a cutting- Bank of England to our annual Economic Measurement visitor visits the ONS in Newport. edge research programme to improve Fraser of Allander Institute, University of Conference. Our network includes understanding of the economy through Strathclyde colleagues from organisations such as The To date we have welcomed the following data and statistics. Full Fact Conference Board, Eurostat, OECD, several visitors, some on more than one occasion: Institute for Fiscal Studies national statistical institutes (NSIs) and King’s Business School, King’s College central banks, as well as many academics Ana Aizcorbe, U.S. Bureau of Economic London from universities in the UK and around the Analysis The ESCoE research programme is London School of Economics world. Carol Corrado, The Conference Board and wide-ranging, reflecting the breadth of Massachusetts Institute of Technology Georgetown University economic statistics, and requires bringing National Institute of Economic and Social Erwin Diewert, University of British a very broad range of experts to the task. Research Columbia ESCoE has created a hub of expertise to Nesta Kevin Fox, University of New South Wales support this work in a number of ways. Office for National Statistics Barbara Fraumeni, Hunan University, Queen Mary University of London China, and National Bureau of Economic Royal Holloway, University of London Research ESCoE Research Consortium Royal Statistical Society Wulong Gu, Statistics Canada Stanford University Cecilia Jona-Lasinio, Italian Statistical In delivering our research programme, we University of Cambridge Institute (Istat) and LUISS University of draw upon the expertise of researchers University of Cardiff Rome from a range of institutions, not only University of Essex Dale Jorgenson, Harvard University within the UK but internationally. This University of Groningen Thanks to our extensive network we have Leonard Nakamura, Federal Reserve Bank network has grown (and continues to do University of Nottingham been able to bring many international of Philadelphia so) from the original six main partner University of Sussex experts from North America, Europe, Marshall Reinsdorf, International Monetary institutions to over twenty. ESCoE is University of Westminster Australia and China to the UK to visit Fund (IMF) established at the offices of one of our Warwick Business School, University of ESCoE and engage with our Research Rachel Soloveichik, U.S. Bureau of original partner organisations, the National Warwick Associates. We welcome visits from groups Economic Analysis Institute of Economic and Social Research and individuals and are delighted to share (NIESR), in Westminster, London. NIESR is our ideas with them and learn about the Britain’s longest-established independent International Network work they are doing in their countries and research institute. organisations. We have engaged with the ESCoE has developed and maintains US Bureau of Economic Analysis (BEA), strong links with an international the US Bureau of Labor Statistics (BLS), network of Affiliates who help connect Statistics Canada, the Australian Bureau of us to the global economic measurement Statistics (ABS), Statistics Norway, Statistics community. Working in this way, across Netherlands (CBS), the National Institute of national boundaries, brings our research Statistics and Geography (INEGI) (Mexico) into contact with global best practice and and the National Bureau of Statistics of scholarship, facilitates integrating cutting- China, to name but a few. edge ideas around economic statistics into the UK context, and enhances the We are also delighted to host individuals reach of our research. This network for longer periods. During their visit, as well as working with our relevant research
13 14 conferences, contributing to discussions on the development of the System of National Accounts and measuring GDP in the digital economy.
Kevin is working with the Massachusetts Institute of Technology to measure the value of free digital goods to US consumers. His collaboration with ESCoE has led to an extension of this work to the UK. Carol Corrado
Distinguished Principal Research Fellow in Economics, The Conference Board, Carol visited ESCoE in the autumn of 2017 and in connection with our annual conference in spring 2018. She has since been working with us to develop new methods for measuring education output in GDP, treating schooling produced knowledge as intangible assets. This work was presented at a Brookings Conference on productivity measurement in Washington Leonard Nakamura DC, September 2019. Emeritus Economist at the Federal Reserve Bank of Philadelphia, Leonard has visited ESCoE on several occasions.
During these visits he has presented papers in ESCoE’s Research Seminar Series and in our Measurement in the Modern Economy Series, as well as at the Bank of England and ONS, and co-authored several ESCoE Discussion Papers. Leonard is now working with us on key measurement issues relating to welfare, time use, and intangible investments. Kevin Fox
Professor and Director of the Centre for Applied Economic Research, University of New South Wales, Kevin has visited ESCoE in connection with our annual
15 16 The next generation
emergence of new forms of big data and participate in career events for economics To provide an environment for the the opportunities for economic statistics students. We support students through development of a wider research community with skills and research that these may offer. awards at our annual conference, from the experience in economic statistics, ONS and from the Royal Economic Society, building the next generation of The ESCoE team includes a number of and encourage the upskilling of ONS academic experts in this field. early-career Research Associates and we economists and researchers, promoting play a proactive role in supporting their career development and facilitating PhD progression and research opportunities. opportunities between them and our We are extremely proud that some of partner institutions. Inspiring and developing the next our early-career Research Associates generation of expert economists and statisticians in the field of economic measurement research is crucial to ensuring the continued relevance of economic statistics to decision making, as well as to developing the skills required by the practitioner and policy communities. The lack of focus on measurement within the economics curriculum needs to be addressed, especially in the context of an ever-changing economy, and with the
have subsequently secured positions In addition to our PhD studentships, we “Most economists would probably in high-profile institutions such as the have hosted a number of summer A-level not describe the topic of economic interns at our offices in London. These measurement as being the forefront Bank of England, HM Treasury, the OECD, of their discipline, attracting a lot of University College London and the London students have been able to gain insights attention from academic economists and School of Economics. We are keen to into aspects of professional economic economics students. This is in spite of the support doctoral research in our field, and statistics research. Some interns central importance of data and statistics have gained practical experience; for to the general economics discourse and example, by working on our historic data a long list of Nobel laureates who have repository project. As part of our Public made important contributions to the field Understanding of Economics project, we of economic measurement. Indeed, in have visited A-Level students; through his independent review of UK economic interactive activities and discussions, we statistics, Professor Sir Charles Bean noted improved the students’ engagement with that it is ‘striking how little professional and understanding of economic statistics. economists today are taught about measurement issues’.“
Rebecca Riley, ESCoE Blog ‘Let’s talk about … economic measurement’, July 2019 and currently we are supporting five PhD studentships at King’s College London and the University of Nottingham. We also
17 18 PhD student in the spotlight: Chiamaka Nwosu PhD student in the spotlight: Paul Labonne
Tell us a little about yourself? Tell us a little about yourself? “I’m currently in the third year of my PhD ‘Comparative Assessment of Education Quality in “I’m a PhD student at King’s College London where I have been working on ESCoE projects the UK Over Time’ at King’s College London, supervised by Professor Mary O’Mahony. I worked (mainly ‘Measuring GDP at Different Publication Horizons’) under the supervision of Professor in Washington, DC with NGOs analysing and disseminating international migration statistics Martin Weale. My work investigates how value added tax returns could be used in the before joining the United Nations in Rome as a food security analyst and data consultant. production of GDP figures and replace some ONS surveys. This involves dealing with time I’m really interested in programming and have developed web-based applications which help series techniques and National Accounts concepts. Before my PhD I was an intern at the simplify complex econometric models for non-technical users.” European Investment Bank and at NIESR.”
Where did you first study economics, and what first attracted you to the subject? Where did you first study economics, and what first attracted you to the subject? “I found economics extremely interesting when I was in secondary school. I completed my “I started reading economics during my undergraduate studies in Paris, at the Pantheon- undergraduate in Nigeria and then went to the US for my Master’s degree in Economics.” Sorbonne University. What drew me to economics is the many dimensions it encompasses, such as sociology, history and statistics.” How has being involved with ESCoE benefitted you? “It has been extremely beneficial for my PhD because it has allowed me to be part of a society How has being involved with ESCoE benefitted you? with like-minded researchers, particularly during my first year. I have been able to attend “Through ESCoE I have been able to work closely with ONS staff. This has been a source of various seminars and visit ONS Newport quite regularly to discuss my work and get valuable motivation because our work can have a direct impact on their statistical methods and feedback.” published statistics. ESCoE also provides access to a great network of academics working on similar issues.” Do you have any career ambitions in economics post-PhD? “I’ve had several enjoyable teaching opportunities during my PhD. I would like to remain Do you have any career ambitions in Economics post-PhD? in academia teaching economics at a higher education institution. However, I have a “I would like to continue tackling research questions in economic statistics. That may imply background in data analytics. I’d love to be able to incorporate that into my career also.” working in a research centre, an institution or in academia.”
19 20 Creating space for debate
academics, students, policymakers and projects in consultation with other To deliver dissemination mechanisms practitioners. We record the seminars research groups and statistical agencies. to bring international expertise to bear on research on economic measurement (podcast or video), making these available Providing the space for our network and economic statistics in the UK. online; where possible, we utilise online of academics to come together with conferencing facilities to stream the practitioners and policymakers from presentations for the staff who are based government and international NSIs, we are in ONS’s Newport and Titchfield offices. able to facilitate the sharing of new ideas, methodologies and techniques, helping to bring these into practice. One of ESCoE’s key functions is to provide the space for constructive discussion and dialogue around official statistics and October 2019 ‘Secular Stagnation or Secular Deflation?’ economic measurement. We also aim to disseminate the research findings of our During their week-long visit to ESCoE in team, partners and external colleagues to October 2019, Ana Aizcorbe from the as wide-ranging an audience as possible. U.S. Bureau of Economic Analysis and Arranging a calendar of public events Leonard Nakamura from the Federal is critical to delivering both of these Reserve Bank of Philadelphia kindly took objectives. part in our ‘Measurement in the Modern Economy Series’. We host and arrange four different types Measurement in the Modern of public events: the ESCoE Research Economy Series The session considered why we see Seminar Series held at ONS’s London Workshops have explored: rapid declines in prices of key products throughout the economy, but do not office; the Measurement in the Modern We introduced this ad hoc series of see these price declines reflected in evening panel discussions to promote ‘Management Practices and Productivity: Economy Series, a set of ad hoc evening aggregate measures of inflation or in panel discussion and networking events debate around economic measurement Evidence from Germany, Japan and the UK’ ‘The Conceptual Foundations of the productivity growth. To help address held at NIESR; various ESCoE Workshops and provide an opportunity for networking this, Ana explained how the BEA is Household Costs Indices’ and Conferences focusing on key topics; between our stakeholders. Many of contributing to the measurement of the and an Annual Conference on Economic our visitors and Research Associates ‘Micro-Data Perspectives on the modern economy agenda, in particular Measurement. have participated. Topics considered Productivity Puzzle: evidence for Policy’ its work to measure prices of goods have centred around measuring and ‘What Can We Learn from Business and services related to smartphone examining production and welfare in the Expectations Data?’ usage. Leonard took a long view in his Research Seminars modern economy, in particular the digital ‘Human Capital’ presentation, considering the evidence economy, and how and to what extent ‘Data on Data and Data Flows’ for economic growth acceleration and deflation in the twenty-first century. To date we have held thirty Research we should and might ‘fix’ GDP. This series ‘GDP and Welfare’ has generated a stimulating forum for ‘Towards a Tangible Agenda for Seminars in London, at ONS’s offices These papers stimulated a lively debate in Drummond Gate or at the National discussion and debate. Intangibles’ ‘Imputation of Data into Household around the topic, followed by the chance Institute of Economic and Social Research. to discuss further at a reception. With presentations from ESCoE Research Surveys’ Associates, visiting academics and ONS ESCoE Workshops analysts, these seminars provide an opportunity for research to be showcased, We hold a number of both public and methodologies and findings discussed, invite-only workshops and roundtables and input given from an audience of other throughout the year, to develop research
21 22 Conference Special Sessions April 2019 Royal Economic Society Annual Conference, University of In addition to the above events, ESCoE Warwick has organised special sessions at various external conferences. We have contributed In partnership with ONS’s Data Science special sessions at each of the Royal Campus, we delivered a special session Economic Society annual conferences at the Annual Conference of the Royal Economic Society on ‘Economic Measurement with Big Data’. Chaired by FT Economics Editor Chris Giles, the session considered three novel applications of big data and data science techniques for the purposes of economic measurement.
The session included papers on ‘An open and data-driven taxomomy of skills extracted from online job adverts’, presented by Jyldyz Djumalieva (Nesta, ESCoE); ‘Large-scale real world financial November 2018 ‘Data on Data and Data transaction microdata for national and Flows’ in partnership with the OECD local economic indicators’, presented by Louisa Nolan (ONS); and ‘Making An invited audience were brought text count for macroeconomics: what together by ESCoE and the OECD for a can UK daily newspapers tell us about one-day workshop to discuss how the the future of the economy?’ by Eleni value of new forms of data - generated Kalamara (King’s College London). from digitally enabled activities such as e-commerce, cloud services and the Internet of Things and including private and public, open and controlled data - is not reflected in the national accounts or since our launch in January 2017. We also in international trade statistics. delivered a special session ‘Measuring Aggregate Productivity: How Far Are We This event provided a confidential Off the Mark?’ at the 49th Money Macro forum to discuss the challenges of and Finance Group Annual Conference measuring different types of data 2017 at King’s College London. and data flows and to investigate and better understand, with the help of ESCoE is active on social media. We the private sector, the uses and value maintain a website, escoe.ac.uk, that of these data assets. The overall goal of the meeting was to help develop brings together all our outputs in one recommendations for measuring and accessible online location. We aim to collecting data on data and data flows. make our events and presentations as These recommendations fed into OECD’s widely available as possible and make ‘Going Digital’ measurement roadmap. presentation slides, podcasts and videos available online.
23 24 ESCoE Economic Measurement Conference
measurement and use of economic To develop collaboration between the academic research community statistics. It followed up on the concepts and statistical agencies, offering and challenges raised in the review the capacity for fundamental of economic statistics by Professor methodological and conceptual work Sir Charles Bean and discussed at the that will feed into the development and 2017 ONS International Conference on improvement of economic statistics Economic Statistics. and data-processing methodologies.
The ESCoE annual conference provides a fantastic forum for sharing current leading research on economic measurement from around the world. It is a platform for constructive discussion and dialogue amongst the international community of economic measurement and statistics researchers and practitioners. Papers were invited on areas including Conversations at our conference feed the productivity puzzle, the digital into the development and improvement economy, National Accounts and ‘Beyond of work right across the global landscape GDP’, regional statistics, measurement of economic statistics, data processing, using big data and administrative data, economic research and policymaking. international trade flows and the location of economic activity. Submissions were reviewed by the Scientific Committee, chaired by ESCoE’s Academic Director, Richard J. Smith (University of Cambridge), and by guest reviewers from ESCoE and ONS. Proceedings, including our keynote speakers, were livestreamed and made available through YouTube.
The conference was presented as an ideal learning opportunity for early-career researchers, as well as a chance for Conference 2018 students to network with leading economic experts. To support these aims we ran a We held our first annual conference in competition for economics students at UK 2018, a two-day event hosted by the Bank universities to win a delegate place with of England. Organised in collaboration travel costs provided. with ONS and the Bank of England, the conference covered all aspects of the Keynote speakers in 2018 included
25 26 Nicholas Bloom (Stanford University, ESCoE) and Hal Varian (University of California at Berkeley, Google).
Conference 2019
Following the success of 2018, 2019 saw our conference expand to a three-day, and altogether larger, programme. We were hosted by King’s Business School at King’s College London’s Bush House. In addition to our competition for economics students, the Royal Economic Papers were invited on subjects as per Society provided a cash prize for the the 2018 conference, with this year’s top four PhD papers accepted at the Scientific Committee co-chaired by conference. A selection of papers from the Richard J. Smith and Martin Weale (King’s 2019 conference featured in a special issue College London, ESCoE). Keynote speakers of the National Institute Economic Review on included Vasco Carvalho (University of economic measurement. Cambridge) on production networks in the macroeconomy, Alberto F. Cavallo We have welcomed around 450 delegates (Harvard University) on using online price from around the world to our annual data, and John Fernald (INSEAD) on world conferences. productivity. A panel session led by Ed
What our conference attendees are saying:
“This is a brilliant conference and the best I’ve attended so far. The quality of the talks was really high and I have learnt so many things and made new contacts. Thanks to all the organisers for making this happen.”
“I found that all sessions I went to were very well planned and the speakers were truly experts in their field. There was a Humpherson of the Office for Statistics good environment to ask questions and Regulation considered trust in economic the conference has provided a great statistics. Another panel session led by opportunity for future collaboration.” Peter van de Ven of the OECD considered priorities for developing the System of 2019 conference delegates National Accounts.
27 28 Working with ONS
address the priorities of the ONS as well brought into practice, improving national To support ONS to deliver world-class Jeremy Rowe, an economist at the Bank as maintaining a high academic standard, statistics. We also provide support and economic statistics that meet user of England, was seconded to ONS from needs and allow ONS to exert greater pushing forward the knowledge frontier. expertise to ONS by organising topic- July 2017 to June 2018 and to ESCoE for influence over the direction of travel This research collaboration has resulted specific workshops on areas they are part of this time. Jeremy joined us to of the international standard-setting in jointly authored papers and research currently working on, providing a test bed work on one of our Productivity and agenda. proposals. for new ideas. Modern Economy projects: ‘Granularity in Trade in Value-Added Data for These events, especially our research Key Sectors’. During his time on the Joint working seminars, provide training and encourage project he collaborated with the Bank a research culture at ONS. Alongside our of England, ONS’s trade team and ONS is our primary funder and most ONS’s Data Science Campus, creating important strategic partner. From the ESCoE provides a number of opportunities engagement events, our annual Economic Measurement Conference is a key various linkages and strengthening the outset we have worked closely with them for joint working between academia and collaboration across all the institutions. dissemination platform for ESCoE, ONS and we continue to enjoy a collaborative the ONS. These include secondments; Following the secondment, he moved relationship that benefits the scope, in-depth partnership working; and and the wider economic measurement from the Bank to take on the role impact and reach of our research and opportunities to discuss and jointly community. This event increases dialogue of Senior Data Scientist at ONS’s our engagement activities. Through develop surveys, results, methodologies, between academic economists, national Data Science Campus. Jeremy has the co-production of research projects, and potential strategies to improve statisticians and statistics users, helping to continued to work closely with ESCoE, secondments, research seminars, economic statistics. Since January 2017 we improve economic measurement. Being contributing to our engagement events, seminar series and annual Economic workshops and our annual conference, we have welcomed four ONS staff members to jointly organised, this conference helps Measurement conference. have helped ONS become more influential ESCoE as secondees and two of our ESCoE to bridge the gaps between the academic on the international stage. PhD students have had secondments with research community and statistics the ONS. We have also had several Bank practitioners and between research and of England staff seconded onto ESCoE applicability. Co-production of research projects, encouraging more in-depth “Innovative research on the digital engagement. economy and implications for economic measurement have helped to solidify ONS We deliver research and analysis that Other engagement thinking on the ‘GDP and Welfare’ agenda. helps ONS to meet their vision of “ESCoE research has helped ONS consider In addition to their involvement with In particular, papers on cloud computing delivering world-class economic statistics new and more innovative approaches to ESCoE, several of our Research Associates and the sharing economy provided new and being influential in the international the development of deflators, as well as insights on these difficult topics in the are ONS Fellows and members of ONS’s standard-setting agenda. An example of ensuring that we consider a broader range modern economy, and a paper on time this is ESCoE’s work in helping the UK to of opinions and sources than may have Economics Experts Working Group use helped to identify the increasing win seats on each of the three taskforces happened previously.” (EEWG). The EEWG meets six times a year blurring between work and leisure time. working to develop the System of National to help ONS set its development agenda, This supported ONS staff in publishing Accounts, addressing the key challenges Office for National Statistics as well as offering expert advice to help an ESCoE Discussion Paper on the topic currently facing economic measurement: solve the difficulties of measuring the and designing a roadmap to produce a digitalisation, globalisation and the changing economy. ESCoE is proud to new measure of economic welfare. This was part of an ambitious UK Government measurement of welfare. play a vital role in bringing about a step Collaboration on events change in ONS’s capability, to contribute 2019 Spending Review bid, which will be to the development of a more research- included again for the 2020 Spending Through regular meetings between ESCoE Review.” Research Associates and relevant ONS Through ESCoE’s event programme, we focused culture in the delivery of economic provide a forum where we can bring our statistics, and to help ONS take forward staff, joint periodic reviews of ongoing Office for National Statistics work programmes, and engagement research expertise into the heart of ONS. its transformation into a world-leading events, we ensure that our projects We can have discussions on how ideas and national statistics institute. methodologies can be implemented and
29 30
Our projects
Each of our research projects belongs to one of four core research streams
National accounts and beyond GDP Regional and labour market statistics
Use of Value Added Tax Returns to Produce Monthly Estimates of Industrial Using Administrative Data to Develop New Labour Force and Migration Statistics Output Making Sense of Skills Measuring GDP at Different Publication Horizons Using Administrative Data to Measure New Forms of Working Historical National Accounts Data Regional Nowcasting in the UK UK Historical Data Repository Improving Regional Economic Indicators Democratic Measures of Income Growth The Impacts of Trade on Income, Employment and Inequality in the United Social Transfers in Kind Kingdom and its Regions
Productivity and the modern economy Communicating and valuing economic statistics
Measuring Activity in Services Sectors Modelling and Communicating Data Uncertainty Measurement Issues in the Digital Economy Valuing Economic Statistics Sectoral Productivity Estimates Developing Firm-Level Micro Data for Productivity Analysis Granularity in Trade in Value Added Data for Key Sectors The ‘Rotterdam-Antwerp Effect’ in the Context of UK Trade Statistics
31 32 Use of Value Added Tax Returns to Produce Monthly Estimates of Industrial Output
In the short term, GDP estimates are produced from firms’ turnover data. These have historically been collected by the Monthly Business Survey (MBS). However, companies’ Value Added Tax (VAT) returns, typically made on a rolling quarterly basis, also contain data on business turnover which may be used to improve the accuracy of short-term GDP estimates. This project has addressed the statistical issues needed to compile seasonally adjusted output data from VAT returns. Lags in the collection of VAT data mean that the MBS will probably still be needed for large firms.
Paul Labonne Martin Weale
33 34 Project overview Methods percentage points to GDP. A second a means by which ONS can make full feature of the estimates generated by and efficient use of the VAT data as a Administrative data such as companies’ ONS provided vintages of VAT data for the multivariate model is that they are basis for producing output data. The VAT records provide potential source small firms, industry by industry, with somewhat smoother than those collected work in this project has benefited from material for producing national accounts. the attribution issues resolved as far as by the MBS. It should be remembered engagement at the Royal Economic Society Output figures are currently compiled possible and grossed up to represent that VAT data, being based on close to annual conference and at workshops from turnover data collected by a monthly the whole of each industry. They also full enumeration, ought to be expected to and conferences at the Cambridge INET survey of businesses, the MBS. This provided MBS data for small and large be less volatile than estimates produced Institute, OECD, Eurostat, London Business enumerates all large firms and a sample of firms separately by industry. We used the from a sample survey. For some industries School and the Government Statistical small firms. VAT returns, made by nearly turnover data for large firms to inform the the correlation between the VAT figures Service. all firms with an annual turnover larger temporal disaggregation and nowcasting for small businesses and the MBS figures than £83,000, similarly include data on of the data for small firms. Initially we for large business is not significant; firm turnover. This project has shown how explored least-squares methods of separating the monthly changes from the the VAT returns for small firms can be temporal disaggregation. These, however, measurement errors in these industries is used to produce turnover figures which delivered very erratic estimates. As an therefore more difficult. can be used in the calculation of GDP. alternative, we developed a state-space The data cover about 25 per cent of value model to handle jointly the issues of Key recommendations include: added. temporal disaggregation and seasonal adjustment. 1. The monthly survey of large firms needs There are a number of difficulties to be to be maintained so that timely data can resolved. First, VAT does not necessarily We estimated this model in partnership be produced covering an important part of relate to statistical reporting units; in with one for large firms in the MBS the economy, industry by industry. particular, one return may relate to as a set of seemingly unrelated time turnover from units allocated to different series equations, with correlations in 2. VAT returns for small companies can be Figure 1, Estimates of Seasonally Adjusted industries. Secondly, most VAT returns the disturbances estimated in the VAT used to improve the information content Monthly Output for MBS small firms (bands 1 to 3) of early GDP estimates using the methods generated from VAT and MBS data, January 2012 = relate to rolling quarterly periods rather and MBS models. We accounted for 100 than calendar months. There are a small measurement errors in the VAT figures by described here. number of annual returns and about 10 devising methods to deal with the difficult per cent of firms submit monthly returns. task of separating the irregular changes in 3. Differences in estimates from survey Thirdly, VAT returns accrue over a period the seasonally adjusted figures from these and administrative sources need to of time, with the statutory submission measurement errors. be better understood. PAYE data may date very close to the current publication elucidate these differences and provide date for industrial production. A feature a means for further improving early GDP of the data is that the underlying values Findings and recommendations estimates. for small firms suggest a time profile for output rather different from that shown in Figure 1 shows the results generated by the existing MBS data for the same firms. applying this approach to all seventy-five Impact and engagement In addition, the volatility of the VAT figures relevant industries. The different time suggests that they might contain some profiles of turnover growth in the survey The methods described here can be measurement errors. and administrative data are apparent extended to make use of VAT data and also visible in the smoothed-moving- accruing in real time. They have been average data (not shown). The maximum presented to ONS, having a direct impact discrepancy is 6 per cent, which would on the methods developed and deployed amount to a revision of about 1.5 into National Accounts by providing
35 36 Measuring GDP at Different Publication Horizons
This project involved tracking the state of the economy in real time. We investigated how data sources of different frequencies can be combined to obtain real-time predictive intervals for the first release of quarterly GDP growth. We demonstrated how nowcasts of GDP can be improved: 1) by supplementing standard macroeconomic indicators with big data, and 2) by exploiting advances in Bayesian computational methods to extend standard nowcasting models. The project also found that business confidence in the retail and construction sectors is a good guide to turning points in economic growth.
Ana Galvão George Kapetanios Amit Kara Marta Lopresto Fotis Papailias Ivan Petrella Garry Young
37 38 Project overview Methods vintages to provide evidence of the impact month they reference, although not much of GDP data revisions on turning point earlier than monthly GDP. The methods There is a need for early, reliable estimates The main element of the project used a dating. developed here could be used to evaluate of GDP. For many years, ONS met this bottom-up, mixed-frequency approach the usefulness of these indicators in demand by producing a first estimate of to compute accurate nowcasts for nowcasting and forecasting GDP. quarterly GDP growth, based only on the the first ONS releases of UK quarterly Findings and recommendations output approach, twenty-five days after growth. The empirical results suggested the end of the reference quarter. This that the nowcasting system can deliver The nowcasts developed in this project Impact and engagement first release, which was based on limited well-calibrated short-term forecasts by are competitive in comparison to those survey information only covering output, combining forecasts of GDP components produced by professional forecasters such The nowcasting tools provided by this has since August 2018 been replaced from both the expenditure and the output as the Bank of England and NIESR. They project offer a way for ONS to further with an estimate published forty days approaches. One of the main advantages point to the importance of modelling past develop early estimates of GDP, and after the end of the reference quarter, of this nowcasting system is that it is changes in the volatility of forecasting predictive intervals. The frameworks we incorporating additional information on technically easy to implement in real time errors to achieve well-calibrated predicted develop can be used to incorporate novel the GDP expenditure components. by professional economists. All exercises intervals. Other modelling innovations data sources into early estimates of GDP were carried out using Eviews, suggesting proposed in this project are also capable and to assess the information content While ONS has taken a step forward that an accurate nowcasting system for of delivering substantial improvements in of new indicators of economic activity by producing more soundly based first UK GDP could be used by economists who real-time point and interval nowcasts of derived from novel data sources. NIESR estimates of GDP growth, many users prefer to use menu-based software such GDP. publishes a monthly GDP Tracker that of economic statistics would also like to as Eviews instead of code-based software provides a benchmark for assessing the know what is happening in the economy such as Matlab. Within the frameworks we developed, we economic news in the monthly GDP data in real time. We investigated whether found that through using a basic selection release. The Tracker provides an early reliable nowcasts of GDP growth can As part of this project we also suggested of keywords in Google Trends, we were steer of the GDP first release and is widely be constructed based on real-time a process by which big unstructured able to improve nowcasts of key UK referenced in the news media and by ONS data sources at different frequencies. data, unsuitable for time series analysis, macroeconomic variables. Based on this in their data releases. Some of the new In particular, the project investigated can be translated into structured data example, we predict that employing the approaches developed in this project will the reliability of nowcasts produced at that can then be used in standard ‘high quality’ big data available in statistics be incorporated into the Tracker. horizons of one, two and three months econometric models to produce nowcasts agencies, government departments and before ONS’s first estimate is published. In and forecasts. In a separate exercise we central banks should further improve The methods developed in this project addition to the traditional point forecasts, extended the dynamic factor model - the the accuracy of real-time measures of have benefitted from early feedback as considered by the nowcasting literature, workhorse model used for nowcasting economic activity. from ONS, the Bank of England, central we produced predictive intervals of the economic activity - to account for several banks in Europe and the US and through type popularised by Bank of England key features of macroeconomic time Key recommendations include: engagement with business users via the Inflation Report fan charts, providing a series that the current literature typically Economists and Strategy Council, The guide to the reliability of the nowcasts. treats as nuisance parameters. We put 1. Up-to-date GDP estimates are key to Conference Board. these modelling innovations to the test tracking developments in the economy We also considered how unstructured on UK and US data in a comprehensive and the methods proposed here could be big data sources can be incorporated evaluation exercise that would have been used to enhance early estimates of GDP. in nowcasts of economic activity and infeasible just a few years ago but is now how modelling pervasive features of made possible by cloud computing. 2. Since around the time this project macroeconomic data, including outliers was completed, the ONS Data Science and seasonality, can improve real-time Further, we applied the Bry-Boschan Campus has begun to publish faster assessment of economic activity. Quarterly algorithm for dating turning indicators of economic activity. These points to a set of real UK GDP data have been published soon after the
39 40 Historical National Accounts Data
This project identified the key building blocks necessary for reconstructing historical national accounts data and other useful macroeconomic statistics for the UK. We particularly focused on the last hundred years. A dataset of existing historical macroeconomic statistics was assembled alongside an inventory of existing data sources. This resource has already enabled comparison of the productivity puzzle of the late nineteenth and early twentieth centuries with that since the Great Financial Crisis, which showed many similar measurement issues, and the establishment of a new business cycle chronology for the UK back to 1700.
Jagjit S. Chadha Nicholas Oulton Ana Rincon-Aznar Sylaja Srinivasan Ryland Thomas Martin Weale
41 42 Project overview The Sefton and Weale balanced national of macroeconomic data over horizons that a consistent long-run chronology of accounts dataset covering 1920-1990 was are typically required by macroeconomic record that will be a valuable resource to The ONS has wanted for some time to also made available in spreadsheet form. researchers (for example, the Conference academics, journalists, policymakers and make historical UK national accounts data Board Total Economy Database starts in the wider public. more easily accessible to researchers. This The team also tapped the expertise of 1950). Maintenance of the core national would match the standards of provision many ONS and academic colleagues accounts series since 1948 is a minimum of other statistical offices, such as those who had worked with, and constructed, in this regard. in the Netherlands and the US. Restoring historical data. These colleagues kindly the broader UK historical macroeconomic shared their knowledge and data sources 2. ONS should look to develop methods dataset, along the lines of the discontinued with the project. that would allow revisions to the core Economic Trends Annual Supplement, was historical national accounts series to be also felt to be highly valuable, especially allocated to more disaggregated data at the provision of quarterly and monthly Findings and recommendations sector, product and industry levels. This macroeconomic time series over the past should happen even if only at a basic level hundred years. Reconstructing a set of historical national of disaggregation, for example at the level accounts on a consistent basis across of the Input-Output Supply and Use Tables The first aim of the project was to draw up sectors, products and industries remains a which provide a breakdown of output an inventory of existing historical national considerable challenge. This is despite the into ten industry groups. These would accounts data and other macroeconomic existence of the comprehensive historical need to be applied both retrospectively, and financial time series covering the accounts data compiled by Feinstein (1972) for revisions and methodological changes past hundred years. Where possible, and Sefton and Weale (1995) based on made since ESA95 and ESA10, and also on the data was to be made available in earlier national accounting systems. The an ongoing basis for future revisions. spreadsheet form for researchers to use. challenge springs from the large number With the inventory established, the next of methodological changes that have been aim was to take a view of whether (and implemented since ESA95 was introduced Impact and engagement how far back) it was possible to construct in 1998. These changes were not a comprehensive set of historical national retrospectively applied to many historical Our inventory of macroeconomic data for accounts data consistent with the existing national accounts series prior to 1987, and the last century provides a resource for ESA10 system of national accounts. The in many cases not before 1997. So only a policymakers and economic researchers, vast majority of ESA10 national accounts limited ‘core’ set of national accounts data offering solutions to linking older historical series only began in 1997. exists back to 1948 in the current ESA10 series to current official data across a system. The length of time elapsed since range of macroeconomic variables. The many of the methodological changes project has attracted comments and Methods makes a reconciliation exercise especially questions from academia and policy difficult and our inventory paper revealed institutions and coverage by the BBC and The project built on the previous just how much needs to be done to rectify the Financial Times. knowledge and experience of the project this. However, a good start has been made team. For example, team members had by gathering the necessary raw materials. This project has also led to a number of produced datasets in their previous ‘follow-on’ studies. These include a study research which were then reorganised Key recommendations include: considering improvements to the UK appropriately for dissemination in this national accounts data for the eighteenth project. The Bank of England’s Millennium 1. ONS should continue to support and nineteenth centuries and a study Dataset was used to extract many historic the development of historical national to establish a business cycle chronology macroeconomic and financial series. accounts to facilitate time series analysis for the UK back to 1700, providing
43 44 UK Historical Data Repository
We created a website that is both a free repository of historical UK data and relevant statistical publications, and a hub that links to other data websites. This was a joint endeavour by the Bank of England, ESCoE and ONS. Sourced mostly from the Bank of England, ONS and economic historians, the idea was to create a ‘one-stop shop’ for UK historical statistics by digitising publications and updating the Bank of England’s Millennium of Data spreadsheet. A beta version of the website is live and can be viewed at escoe.ac.uk/historical-data. Curation by the Bank of England will continue for the next few years as we add more publications and update key data spreadsheets.
Sylaja Srinivasan Ryland Thomas
45 46 Project overview NIESR and Bank of England libraries. This the accompanying video demonstrating was supplemented by various Microsoft the content of the website, potentially The UK historical data repository project Excel datasets such as the Bank of reaching over 250,000 people. is intended to provide a valuable and England’s Millennium of Data and the lasting public good that will encourage ESCoE Historical National Accounts Data research into a wide range of areas project. The data are on a best endeavours using time-series data. The provision of basis. Repository items may report historical research datasets will allow an economic figures which are not in line ever-expanding source of historical data with official statistics. A beta website was to be used with new methodologies and launched with the first batch of statistical approaches. documents, and a video demonstrating and launching the site was presented at the ESCoE Conference on Economic Methods Measurement, 2019. More content is continually being added. Key UK historical statistical publications have been digitised to a high quality Key recommendations include: standard (OCR-ready pdfs) so that conversion to machine-readable form can 1. This online repository should be used as happen more easily in the future. Other a key source to facilitate the reconstruction research datasets were sourced through of historical national accounts data. contact with economic historians. The website has nine sections, each of which 2. The repository should continue to be has datasets and a range of supporting curated and developed jointly by the Bank digitised publications, such as Annual Blue of England, ESCoE and ONS. Books and Bank of England Statistical Abstracts. Topics covered include headline macro data, national accounts dating back Impact and engagement as far as 1086, the entire input-output tables through history, financial market Our online repository brings together variables, and overseas trade since the the various materials necessary for the thirteenth century. A key element of the reconstruction of historical national repository is a link to the Bank of England’s accounts, providing a key resource. Millennium of Data spreadsheet, which Since its launch in May 2019, the website will be updated each year to support this has been well-received by academic repository with an up-to-date database. researchers and the wider data community Various rightsholders gave permission to for both its design and its content. The republish copyrighted material. financial media outlet CentralBanking. com acknowledged that the website was “drawing on a huge array of largely Findings and recommendations paper-based sources, as well as several academic estimates for data that was We were able to find a wide range of not collected in earlier years”. The Bank statistical documents, largely from the of England promoted the launch and
47 48 Democratic Measures of Income Growth
It is well known that GDP growth is not a good indicator of welfare change. As an aggregate it does not reflect any changes in the distribution of income. It could be described as plutocratic because a given percentage increase has more impact on GDP growth if it accrues to high earners than if it accrues to low earners. We developed democratic measures of growth which represent the average of each household’s growth experience.
Andrew Aitken Nicholas Oulton Martin Weale
49 50 Project overview income in the Family Resources Survey. our final estimates were produced as the Impact and engagement This allowed for the rapid calculation of a average of a hundred such simulations. GDP was designed as an indicator of democratic measure of household income We have provided ONS with a template economic activity, but it is widely used as growth. for producing democratic measures of an indicator of welfare. It is well known Findings and recommendations growth. ONS views this as a seminal paper that any welfare indicator has to be net Production of a democratic measure of on this topic, significantly informing its rather than gross of depreciation. Equally national income growth was substantially We estimated that over the period 2006- work on welfare measurement. ONS is importantly, from a utilitarian perspective, more time-consuming. It was necessary 2015, real household democratic income, exploring the wider use of our stochastic a single welfare aggregate needs to to attribute public consumption, and the after adjusting for household size, grew imputation methods as a means of represent some aggregate of the welfare consumption of non-profit institutions by 0.2 per cent per annum, while real enhancing its survey data. This work also of the population. GDP, which adds up serving households, to particular household plutocratic income grew by led to further engagement with ONS individual incomes, can do that only if households. We drew on ONS work on 0.52 per cent per annum. Two-thirds to explore the best way of nowcasting welfare rises linearly in income; even then, education and health consumption and of the gap was accounted for by faster median incomes; eliciting interest from it remains silent about the distribution of allocated other consumption on the basis growth in the democratic price index and the OECD, which is implementing related welfare. of household size. We also had to allocate the remainder by a slight change in the methodologies. the income which was not distributed to nominal distribution of income. However, There is a growing body of evidence households, such as retained profits or the picture was reversed when we looked This project involved significant which suggests that welfare rises less pension fund investments, which are not at democratic national income. This grew engagement with academia, and users and and less with income, as income rises. distributed to households. Finally, there by 0.05 per cent per annum while the producers of welfare and distributional This suggests that a measure of welfare was an issue that income is generally plutocratic measure declined by 0.15 per indicators. These groups were engaged constructed from the geometric rather under-recorded in household surveys, cent per annum. through ESCoE workshops, presentations than arithmetic mean of income would be and that for some types of income such at the OECD, and at meetings of the appropriate. The focus, however, needs as dividends, this under-recording is very Key recommendations include: European Economic Association, the Royal to be on the real rather than the nominal substantial. Economic Society, the Society for the Study income growth of each household. 1. That democratic measures of household of Economic Inequality, the International Different households have different Where there was no further information, income growth be published. This might Association for Research in Income consumption patterns. The appropriate we simply scaled the household data to be best done in step with the development and Wealth, and the World Inequality deflator therefore needs to be constructed match the national accounts aggregate. of the household cost indices, as they are Database. The work also received from the average of each household’s Where possible, however, we drew on produced on a democratic basis and are significant coverage in theFinancial Times. expenditure shares rather than the other data sources. These included the therefore not suited to deflating mean shares in total consumption. Democratic Survey of Personal Incomes for dividends, household income. measures of income growth can be and the Wealth and Asset Survey for constructed for several possible definitions information on the distribution of 2. That ONS pursues the production of of income; our focus has been on pension wealth and consequently the democratic measures of national income household income and national income. allocation of pension fund investment growth. If it turns out to be practical to income. We produced stochastic models draw on HMRC and DWP administrative of income from these and used them income data, these would be valuable, Methods to allocate the relevant types of income but they are unlikely to provide a to each household. We took particular comprehensive picture. In the meantime, The study was able to draw on the work care to ensure that very high incomes the imputation methods we have used on democratic price indices carried out were represented, and to represent the are viable and could be employed on a by ONS, who kindly also produced price covariance between different sources of systematic basis. There may also be a case indices with a treatment of housing income. As a consequence, our results for imputing wage income at the top of the consistent with the definition of household had to be based on stochastic simulations; distribution.
51 52 Social Transfers in Kind
We proposed a methodology to impute social transfers in kind of adult social care to individuals, thereby filling a gap in the distributional statistics produced by ONS. The proposed methodology was implemented in ONS’s distributional statistics published on 30 May 2019.
Craig Thamotheram Garry Young
53 54 Project overview financial means to fund their own use of households, such as residential care Impact and engagement care services. In particular, a person will homes and prisons) accounts for just Statistical agencies in different countries have to pay the full cost of their care if under half of total spending on adult social This project has had a direct impact, produce two estimates of household they have more than £23,250 in savings. care in England. However, the effects of through the preferred methodology being income: one through the national accounts Unless they are going into a care home, taxes and benefits on household income fully implemented by ONS into its National using aggregate data, and the other this amount does not include the value (ETB) distributional statistics are based Statistics and published in the ‘Effects of through the distributional statistics using of the person’s property. This financial on the LCF, which is a sample of private Taxes and Benefits on Household Income: micro data. In principle, both aggregate assessment means that those with households. As such, this methodology Financial Year Ending 2018’ ONS statistical and distributional income statistics sufficient financial resources are unlikely does not allocate spending on people bulletin. This has been a key step in the should include social transfers in kind. to be eligible for publicly funded social living in institutional households to people improvement of these data. These are goods and services provided by care, even if they require support, so that in the LCF. government and non-profit institutions the insurance-value approach had to be that benefit individuals but are provided modified somewhat. free or at subsidised prices. The major Findings and recommendations items in the UK are health, education and The consequence of the means-tested adult social care. Prior to this project, arrangement for adult social care The preferred imputation method the UK national accounts included all funding is that detailed information on produces an allocation that is materially of these and other items, whereas the people’s assets is required to understand related to income, reflecting the means distributional statistics made allowance which people are not covered under an testing of the benefit. This is not the only for health and education. The aim of insurance-style allocation. Unfortunately, result found in other countries, where the the project was to suggest a methodology the LCF does not currently contain the allocation across the income distribution for incorporating the social transfers in required level of detail to estimate the is fairly flat. A question remains as to kind of adult social care within ONS’s value of household assets. As such, the whether our preferred result is a genuine distributional statistics. recommended methodology adopts reflection of how adult social care a hybrid approach where 20 per cent spending is allocated across the income of adult social care is allocated via an distribution in the UK, or simply a function Methods insurance approach and 80 per cent via of our chosen imputation method. estimated consumption. Broadly, our approach aimed to allocate Key recommendations include: actual spending on adult social care, using The LCF does not currently contain any data from NHS Digital, to people on the information on whether a respondent 1. To implement our preferred Living Costs and Food (LCF) dataset. Only has been a recipient of adult social care methodology for the allocation of data for England is available, requiring provision. Therefore, the consumption spending on adult social care to individuals further work to gross up to a UK total. element of adult social care spending is in the LCF, which can then be used to allocated to individuals based on the LCF, incorporate it within the ETB distributional While the preferred approach would according to whether they receive any statistics. have been to adopt an insurance-based of the following: Attendance Allowance, allocation, similar to the NHS, the nature Incapacity Benefit, Carer’s Allowance, 2. To include additional questions in the of adult social care provision makes this Severe Disablement Allowance, or LCF, to identify directly the beneficiaries of more complicated. Unlike health care, Disability Living Allowance. spending on adult social care. social care services in the UK are not provided free of charge for everyone. Another consideration in developing this In England, local authorities provide methodology is that residential care (for assistance to adults who have insufficient example people living in institutional
55 56 Measuring Activity in Services Sectors
We investigated deficiencies in current measures of services activities for the UK and how the measures might be improved. The project consisted of four strands. The first covered all services activities and involved an examination of price indices and how the UK compared to international best practice. The others focused on hard to measure sectors: education, insurance and financial services.
Carol Corrado Augustin de Coulon Perry Francis John Lowes Charles Nourse Chiamaka Nwosu Mary O’Mahony Lea Samek Sylaja Srinivasan Martin Weale
57 58 Project overview the sensitivity of the UK methodology quality adjustments, and pointed to the Key recommendations include: for calculating financial intermediation need to consider the education sector as Good measures of services activities services indirectly measured (FISIM) to a whole, rather than separating schools 1. Efforts to improve measurement are crucial to understanding economic alternative choices of reference rate and from further education. Our research also of prices in services should focus on developments. Services are increasingly discussed some of the other important suggested that nominal and real output understanding differences in prices by important as a share of value added and methodological issues with FISIM. FISIM is measures based on increments to lifetime customers, businesses, consumers or employment, and much of the increased a primary source of output produced by earnings that can be attributed to formal exports. productivity growth associated with the financial intermediaries. education are a useful alternative method information technology revolution has to current practice. Such measures could 2. There is a need to take a more holistic occurred in services sectors. However, be incorporated into the national accounts view of education, including both schools there are significant conceptual and Findings and recommendations in a consistent way by treating education and higher education. practical difficulties in compiling accurate as investment in social infrastructure, with statistics on services activities. In this Our audit of SPPIs concluded that it is education of foreign students allocated to 3. The methods proposed for the project we investigated some of the unlikely that errors in measuring services exports. measuring of insurance output should shortcomings of current measures of prices are especially large in the UK be reviewed in 2022 when more data are services activities for the UK and how compared with other countries, suggesting We examined several alternatives to available. these might be addressed. that this aspect of measurement is the way the Bank of England currently unlikely to explain the observed poor calculates the FISIM reference rate. 4. Any future FISIM Task Force should UK performance in productivity growth An issue with current practice is that consider the concept of negative FISIM Methods relative to other major economies in the reference rates mostly reflect charges alongside the approach to measuring past decade. The results did not, however, on intra-group lending rather than the FISIM in real terms. The project built on the previous preclude that some services prices are cost of obtaining funds from the market knowledge and experience of the project badly measured everywhere, which might as a whole. Reference rates that exclude team, including the expertise of the team contribute to an explanation of the global intra-group loans or are based solely on Impact and engagement at the Bank of England who produce the productivity slowdown experienced by rates used for repurchase agreements measure of nominal output of financial many countries. During the period after (repos) turn out to be quite volatile. Our research into the impact of different services by UK monetary financial the financial crisis, the magnitude of the Alternatives such as using three-month methods used to derive SPPIs shows clear institutions (MFIs) for ONS. We compiled potential downward bias in multifactor LIBOR, the Bank Rate or SONIA all produce biases between different approaches. a comprehensive audit of methodologies productivity (MFP) growth arising from negative deposit FISIM intermittently This key evidence was submitted by and data sources and compared growth mismeasurement of SPPIs was estimated from late 2009, because these rates do the UK to the UN’s Voorburg Group on in Services Producer Prices Indices to be significant in sectors such as not accurately capture the actual cost of services statistics. Our work on insurance (SPPIs) in the UK to fifteen competitor professional, scientific and technical funding for MFIs. We propose taking the and financial services has provided countries. In considering education services, and administrative and support weighted average of the non-intra-group a foundation for ONS to launch their services, we explored two avenues for services. However, the implications for loan and repo rates, and then smoothing planned reviews of these sectors as part improvement to current practice: first, aggregate economy MFP growth are this rate by calculating the four-quarter of Blue Book 2021, and beyond; it also quality adjustments applied to volume relatively small. This work also seemed to lagged average, as a way to potentially provides information that should be measures of school enrolments; second, show large differences in prices for some better capture MFIs’ cost of funding. We helpful when the FISIM methodology is output measures based on labour market services between business-to-business also suggest that adjusting FISIM estimates again debated internationally. Our work on earnings outcomes and embedded into a and business-to-consumer transactions. for loan defaults would probably be an education has been widely disseminated national accounting framework. Our work improvement to the National Accounts. and is influencing ongoing work at the on insurance services explored measures Our work on education suggested that the ONS around the measurement of human of insurance output based on the increase use of standardised test scores may not capital and education output more widely. in welfare generated by insurance, and be capturing key aspects of underlying alternatively, on risk carried. We compared skills that should be incorporated in
59 60 Measurement Issues in the Digital Economy
Digital technology and other drivers of structural economic change raise several significant measurement challenges. In this project we considered the implications for economic measurement arising from digital business models, including the trend towards servitisation; frequent incremental innovation and the concepts of ‘quality change’ versus consumer surplus; and the globalisation of online businesses. Our work has helped to solidify ONS thinking on the ‘GDP versus welfare’ agenda and is contributing to the international debate on three key measurement challenges: digitalisation, globalisation and well-being.
Diane Coyle Leonard Nakamura Mo Abdirahman David Nguyen Richard Heys
61 62 Project overview considered quality adjustment of price and 2017; considerably more than the to telecommunications and cloud services indices for telecommunications services current deflator. The market for cloud has been presented at conferences at the Digitalisation raises manifold conceptual and cloud-based IT services. This work was services is large and growing rapidly. OECD and the International Monetary challenges and practical issues for carried out with ONS and the Institution Quality-adjusted prices have fallen by up Fund (IMF), impacting on the international economic measurement. In this project of Engineering and Technology. Here we to 5.5 per cent per quarter. We found that debate as the UN steers towards a revision we developed a taxonomy illustrating considered two methodologically distinct contract manufacturing is more prevalent of the System of National Accounts. the scope of digital changes that raise options to estimate the potential bias in amongst UK firms in the chemicals and Our work has also been presented and measurement issues and explored the the current deflator, informed by both pharmaceuticals sectors, whereas in the discussed at the World Economic Forum at potential magnitudes of some of these economic and engineering perspectives. US it is more prevalent in electronics. Davos, the 2018 meeting of the G7 finance for key statistics and economic concepts. Our work on cloud-based IT services ministers, in UK government departments We then made suggestions for steps involved the construction of both nominal Key recommendations include: and at academic conferences including that could be taken to better measure and quality-adjusted price indices for the Royal Economic Society, the Society for developments in the digital economy. computing and storage products. We 1. We advocate an experimental set of Economic Measurement, the International used prices published online and price time and well-being accounts, with a Association for Research in Income and lists shared with us by Amazon Web particular focus on digitally driven shifts Wealth and the Association of French Methods Services, along with information on the in behaviour. This would require the Economists. number of floating-point operations per collection of time-use statistics as well as We first considered measurement second by central processing unit core, well-being data and direct survey evidence. Our work on telecommunications deflators issues arising from digitally enabled which is a common proxy for the quality has provided options for consideration substitutions in activity across the of a processor. We collated web-scraped 2. Our work on deflators points to the and selection for implementation into conventional production boundary. evidence on the extent of factoryless need for continued research in this area. the National Accounts in Blue Book 2020 Production boundary issues are not manufacturing in the UK and explored this This is particularly relevant with the (on current ONS plans). A presentation of new, as conventionally defined GDP through company case studies based on introduction of double deflation since this work at the UN’s Voorburg Group on statistics account for the monetary cost systematic analysis of annual reports and telecommunications and cloud services measuring services has elicited interest but not the time cost of consumption websites. are important intermediate inputs in from both Statistics Canada and the US and production. This means that changes production. We urge statistical agencies to Bureau of Labor Statistics. Our research on in the way in which time is allocated introduce methods to capture the use and cloud computing has fed directly into the between market and home production Findings and recommendations provision of cloud services by businesses, OECD’s research programme on the digital affect measured growth and productivity, as well as adopting a dedicated category economy. It is shaping thinking on the as well as economic welfare. While it is Statistical agencies do not currently in trade statistics focused on measuring future of the System of National Accounts; impossible at present to know the scale collect the data needed to measure the cross-border use of cloud services. featuring, for example, in a contribution of these substitutions, we examined a scale of the switch from market to home from the IMF at the joint meeting of the range of evidence to establish a picture production due to digitalisation, but the 3. Given the sector-based focus of many Working Party on Financial Statistics and of the potential magnitudes involved. We evidence we considered suggests that it economic policies, our findings point to the Working Party on National Accounts at considered an alternative understanding may be enough to make a contribution to the need for consistent measurement of the OECD in November 2019. of economic progress to real GDP growth. understanding the current productivity factoryless manufacturing through official This combined an extended utility puzzle. ONS is conducting a new surveys. framework - considering time allocation time-use survey which will contribute over paid work, household work, leisure to our understanding. Our findings and consumption - with measures of indicated that the current deflator for Impact and engagement objective or subjective well-being while telecommunications services is upward- engaging in different activities. biased and that telecommunications Our work on the production boundary services prices may have fallen between and cloud computing and on the In a separate stream of this project we 35 per cent and 90 per cent between 2010 measurement of price change in relation
63 64 Sectoral Productivity Estimates
Understanding the productivity puzzle is an important objective of government and policymakers. While there are many explanations in the literature, there is a suspicion that the pronounced reduction in productivity growth after the financial crisis may be due to measurement issues. This is not a criticism of standard practice but reflects the idea that economic activity has increased in relative importance in areas that are poorly measured. We revisited the sector-level productivity estimates in this context and in the light of advances in measurement.
Nicholas Oulton Rebecca Riley Ana Rincon-Aznar Lea Samek Sylaja Srinivasan
65 66 Project overview economy) levels. We used publicly Findings and recommendations by ONS for productivity analysis provide available price indices (deflators) at the a useful new resource for understanding The policy of the ONS is to introduce level roughly of the Supply and Use Tables We found that on average the growth UK productivity growth and should be double deflation into the national accounts to estimate double-deflated real value rate of industry-level real value added maintained and regularly updated. With following the recommendations of the added for seventy-nine industries over was substantially lower under double the implementation of double deflation Bean Review. There are a number of the period 1997-2015. Our price indices deflation than under single deflation, and and the development of new deflators for concerns about the effects of double were the ones employed by the ONS to the year-to-year volatility of growth rates information technology sectors, these data deflation. First, the industry-level estimates produce the official estimate of GDP from was much higher under double deflation. should be reassessed. of real value added may be very different the output side using single deflation. By We also found that with these estimates, from the current official estimates using design the official estimate of real GDP(O) GDP would have grown substantially more single deflation. Second, the double is very close to the preferred estimate of slowly than the official estimate for 1997- Impact and engagement deflation estimates may be more volatile, real GDP from the expenditure side. We 2015. perhaps implausibly so. Third, double aggregated our double-deflated industry- This project challenged the methods and deflation has the potential to change level real value added estimates using the Our analysis of the productivity slowdown delivery of double-deflated estimates of the past. Even using exactly the same same methodology and data as the ONS using detailed industry data pointed National Accounts, contributing to the ONS data, the growth rate of real GDP may be to give an alternative estimate of real GDP to the importance of economy-wide decision to delay full implementation of significantly different from the previously from the output side. and indeed global explanations for the double deflation in Blue Book 2019. The published official figures. We addressed slowdown. Also, the UK productivity project was influential in helping ONS to these concerns using a simple approach In analysing the slowdown in productivity growth puzzle is concentrated in sectors understand the variety of options available relying only on publicly available data. growth, we focused on developments in where productivity is difficult to measure, and methodological challenges as they the UK market sector and in particular on such as information and communication arose. Members of the ESCoE project We also analysed new detailed industry- the post-recession period of productivity and finance, suggesting that measurement team also provided input and challenge level data developed by ONS to re-examine stagnation from 2011 to 2015, compared issues should not be dismissed as part of to ONS work to estimate the intermediate the UK productivity puzzle. We carried to the decade before the financial crisis an explanation of the puzzle. expenditure proportions that feed into the out an accounting exercise that allowed that began in 2007. We considered where Supply and Use Tables and GDP estimates, us to distinguish general macroeconomic in the economy this stagnation was Key recommendations include: following the reintroduction of the Annual patterns from sector trends and located, examining productivity patterns Purchases Survey in 2015. idiosyncrasies, providing a roadmap for across fifty-nine market-sector industries 1. Our theoretical work on double anyone interested in explaining the puzzle. and fifteen broader industry sections. We deflation suggests that it is possible to Our work to analyse the UK productivity We focused on the UK market sector and constructed measures of capital services, implement double deflation in such a way puzzle was covered in the Financial Times found that UK industries that saw the labour composition and total factor that there is no effect on the headline and influenced the way that ONS presents biggest reductions in productivity growth productivity by industry and used standard growth rate of GDP. This requires using the real estate sector in industry analyses tended to be internationally competitive growth accounting techniques to examine a comprehensive set of deflators which of productivity. This work involved and more dependent on global demand the sources of growth across industries. are currently not publicly available. Our significant engagement with policymakers, than other industries. They were also Using internationally comparable data, results, based on a simple analysis relying industry and academia through seminars industries where productivity is difficult to we considered the UK productivity on publicly available deflators, strongly and conferences organised by the measure. growth slowdown compared to those in suggest that this approach should be Institute for Manufacturing, University of the US and EU. Lastly, we noted some seriously considered. It is possible, Cambridge; the European Central Bank; additional patterns that emerged from the however, that even on this approach the Eurostat; the London School of Economics; Methods sectoral composition of the puzzle when industry-level estimates of real value and the Royal Economic Society. contrasted with international trade metrics added may be quite different from the We analysed the effect of double deflation and in light of economic measurement current official ones. on output and hence productivity at issues. the industry and the aggregate (whole 2. The detailed industry data developed
67 68 Developing Firm-Level Micro Data for Productivity Analysis
In collaboration with ONS, we developed and conducted the Management and Expectations Survey (MES), the largest-ever survey of UK management capabilities. The survey revealed the substantial heterogeneity of management capabilities across UK businesses and their strong correlation with productivity. The MES also revealed links between macroeconomic uncertainty and businesses’ expectations of their own performance. The survey provides a new resource for policymakers and researchers to better understand links between management practices and business performance both in manufacturing and, uniquely, in the services sector.
Nicholas Bloom Gaganan Awano Paul Mizen Russell Black Rebecca Riley Ted Dolby Tatsuro Senga Jenny Vyas John Van Reenen Philip Wales
69 70 Project overview through the MES could be matched to the UK and other countries. This evidence core to UK industrial strategy. The ABS data on productivity and economic has already led to interesting findings. For analysis drawn from the MES has been Low productivity is possibly the most performance. example, we have found that management disseminated through the policymaking important challenge that the UK practices correlate more strongly with communities; it has helped formulate economy faces because it affects the The MES questionnaire built on the productivity in the UK than in Germany policy recommendations and ultimately living standards we can sustain. Against experience of the US Management and and Japan. New evidence from the MES shaped a successful bid for an ESRC grant this background, the disappointing Organizational Practices (MOPS) surveys also suggests that lower-productivity firms to carry out another wave of the MES. productivity record of many UK businesses run by the US Census Bureau in 2010 and are more uncertain about their future The final datasets will be made available has provoked significant concern 2015 on around 50,000 US manufacturing turnover, investment and employment to researchers, to further expand on this amongst policy, academic and business establishments. The US surveys were growth than higher-productivity firms, research. communities. The factors contributing designed in partnership with Nicholas even after controlling for a number of to this picture remain the subject of Bloom and John Van Reenen, members other firm-level characteristics including The Industrial Strategy White Paper called debate. Much can be learned from careful of the current ESCoE team. Based on firm size. Further, evidence from the MES for a review of actions for improving the investigation of firm-level micro data. responses to questions concerning suggests that larger and well-managed productivity and growth of small and Indeed, many of the central questions in problem resolution, employment practices businesses in the UK make smaller errors medium-sized businesses. The Business the productivity debate are impossible and the use of key performance indicators in forecasting UK GDP. Productivity Review: Call for Evidence in to answer without detailed firm-level and targets, we retrieved a management Spring 2018 referred extensively to our data. Our objective was to contribute to score for each firm via an identical Key recommendations include: analysis, and later that year the Chancellor the evidence base through developing a methodology to that used in MOPS in of the Exchequer announced a package survey that would allow us to study and other countries, facilitating international 1. The MES has already shed light on of measures to support business to produce firm-level measures of UK firms’ comparison. Using these firm-level scores the relationships between UK firms’ boost skills and growth. With reference management practices, uncertainty and we examined cross-sectional features productivity and management practices. to the Business Productivity Review, forecasting ability. We are also building UK of structured management practices by Further waves of the survey should be these included measures to improve the firm-level micro data capacity by working industry sectors, regions and other firm conducted in order to develop longitudinal management practices of UK businesses. with ONS and data users to improve the characteristics such as firm size and age. firm-level information. This would usability of key ONS business datasets, The MES also collected information on allow understanding to move beyond We have liaised with policymakers and including linked datasets. firms’ subjective expectations of their correlations towards causal links between business bodies on the development turnover, expenditures, capital investment management and productivity. of our work on management practices and employment using a ‘five-bin’ scale, as and expectations as well as on the Methods well as their expectations of future growth 2. Much more could be learned about development of business micro data more in UK real GDP. management practices and productivity generally. The work in this project has In 2017 we developed and conducted by linking data from HMRC on firms’ benefitted from engagement at annual the UK Management and Expectations international trading relationships and conferences of the Royal Economic Society Survey (MES) in partnership with ONS. Findings and recommendations their employees to ONS datasets. and the European Workshop on Efficiency This was executed on a population of and Productivity Analysis (EWEPA), and at 25,000 firms across different industries, The survey dataset from 2017 covers 3. The Annual Respondents Database conferences, workshops and seminars at regions, firm sizes and ages. The MES was employers across most parts of the X should be updated regularly and the University of Chicago, Bank of England, a voluntary survey of businesses with private economy and provides data on developed in conjunction with key users. the US Census Bureau, the German ten or more employees, covering both a wide variety of aspects of how these Embassy in London with German Industry the production and services industries. workplaces operate. The data collected UK, University of Warwick, CASE at the The sampling frame was the ONS Annual show that broad measures of firms’ Impact and engagement London School of Economics and the Business Survey (ABS) for 2016 such management practices correlate highly National Institute of Economic and Social that information on firms’ management with firms’ productivity and enable further Raising productivity is one of the Research. The work has attracted coverage practices and expectations collected comparison within the UK and between government’s key priorities and is in the Financial Times and Die Welt.
71 72 Granularity in Trade in Value Added Data for Key Sectors
We developed experimental estimates of direct domestic trade in value added that would allow identification of the industries in the UK that are most likely to be vulnerable to a loss of Single Market membership. The broad aim was to provide timely estimates of the proportions of direct value added from exports destined for EU and non-EU countries for a range of highly disaggregated sectors, both in services and in manufacturing, by relying on simple methodologies and alternative data sources and surveys.
Giordano Mion Marta Paczos Jack Pilkington Jeremy Rowe Sylaja Srinivasan
73 74 Project overview The direct domestic value added of at the four-digit industry classification Impact and engagement exports is the value added component of level for the period 2012-2015. The same This project aimed to build a detailed exports generated by exporting firms. It is exercise for business services was more This project benefitted greatly from the evidence base for setting priorities in the same concept as GVA but applied to difficult due to data availability. Estimates direct involvement of the Bank of England, negotiating future trade deals with the EU exports, and it can be calculated for each based on the ABS survey offered a fairly ONS and ESCoE Research Associates, and non-EU countries, as well as helping firm as the difference between the value comprehensive view of direct value added facilitating information gathering and to better estimate the impact of leaving of exports and the value of intermediate from non-financial services exports for knowledge sharing. It has generated the EU on the UK economy. In order to set consumption associated with these four-digit industries, although time series interest across government departments priorities for trade negotiations, three key exports. The use of firm-level data to are limited. For the financial services through presentations at workshops questions need to be answered: 1. How calculate detailed industry trade in value sector, alternative data sources had to be organised with the National Institute of important are exports for each industry? 2. added data involves fewer assumptions used. The monetary financial institutions Economic and Social Research and Nuffield How much domestic value added is related than more top-down approaches. component required granular data from Foundation, and by ONS. The analysis has to exports for each industry? 3. To what the Bank of England. We set out the also been presented to a Conference of extent is each industry embedded in a The conceptual approach that we caveats underlying the methodologies and HM Government Trade Analysts and to global supply chain; that is, to what extent applied was common across all sectors datasets in the November 2017 edition of trade advisors at the UK manufacturing does it currently rely on imported inputs? that we analysed. However, its specific the National Institute Economic Review and association Make UK. Furthermore, While broad-brush answers to these implementation varied to account for the provide estimates of trade in value added Recommendation Two has now been questions can be obtained from currently various specificities of the different data for key business and financial services implemented. In September 2019 the available international Supply and Use sources. Our analysis of manufacturing industries and manufacturing industries in ONS published revised Trade in Services Tables, and the trade in value added data companies relied on the individual tax ESCoE Technical Reports. statistics which for the first time included derived from them (World Input-Output returns and customs data from HMRC. data for UK monetary financial institutions’ Database, OECD Trade in Value Added Analysis for a key financial services Key recommendations include: intra-group cost recharges. (TiVA) database), this will not provide a sector utilised Bank of England data. For sufficiently granular evidence base. We the other services sectors we used the 1. There is a trade-off between obtaining developed disaggregated measures of ONS Annual Business Survey (ABS) and simple, granular and more timely exports and domestic value added from International Trade in Services Survey, estimates of trade in value added and exports data for selected industries, in as well as Financial Analysis Made Easy estimates that match National Accounts order to be able to answer these three key (FAME) data from the Bureau van Dijk. levels of aggregation and concepts. questions. Ultimately global input-output tables will be needed to more accurately estimate Findings and recommendations trade in value added. Nonetheless, the Methods information content of the estimates The methodologies developed showed we developed is likely to be useful for Gross value added (GVA) is a key concept that it is feasible to deliver more granular policymakers; this suggests these should in the National Accounts; it is calculated and timely estimates of the direct value be maintained. for each firm or industry as the difference added of exports than currently available. between the value of production (output) However, the quality of those estimates is 2. In developing the methodology for and the value of goods and services closely linked to the underlying data and monetary financial institutions based consumed as inputs to the process of there are important differences between on Bank of England data, we noted that production (intermediate consumption). In the results for manufacturing and services income from intra-group cost recharges is our analysis, GVA can be constructed for industries. not, at present, included in ONS Trade in each firm and then summed together to Service statistics. To fully match the Bank calculate the aggregate GVA for the sector For manufacturing sectors, HMRC data of England data to the ONS Pink Book, we as a whole. were used to calculate yearly estimates suggest that ‘other’ exports be included.
75 76 The ‘Rotterdam-Antwerp Effect’ in the Context of UK Trade Statistics
The ‘Rotterdam-Antwerp effect’ describes distortions in official trade statistics due to misreporting of commodities passing through major world ports en route to their final destination. For example, UK exports for non-EU countries through the Netherlands might be misreported as exports to the Netherlands. We evaluated the extent of the Rotterdam-Antwerp effect in the context of UK trade statistics in a project supported by the Department for International Trade (DIT) and the Department for Environment, Food and Rural Affairs (DEFRA).
Giordano Mion Marta Paczos Ana Rincon-Aznar
77 78 Project overview and exports of goods at the Combined and only considered aggregate trade The methodologies he has developed were Nomenclature (CN) eight-digit product flows. We undertook a more granular presented at the ESCoE Annual Conference The ‘Rotterdam-Antwerp effect’ is a term level. These data are freely available type of analysis, exploiting both product- in May 2019 and are published in a special used to describe potential distortions in through an online download platform. and trader-level data made available by issue of the National Institute Economic official trade statistics occurring as a result Information on imports and exports is the NBB and CBS, which carefully track Review on economic measurement. A of misreporting of commodities passing available for 9,500 products from and to re-exports and quasi-transit flows. We customised dataset was provided by CBS through major world ports (e.g. Rotterdam around 250 partner countries. What is of argue that our method, which to our for the purposes of this work. in the Netherlands or Antwerp in Belgium) crucial importance to our research is that knowledge has not been applied to the on their way to their final destination. these data are recorded and available UK before, allows for a more precise and This effect takes its name from the Dutch under both the ‘Community’ and the comprehensive analysis across different and Belgian ports because they are two ‘National’ concepts. This means that the products, for a number of recent years. of the largest ports in Europe, handling difference between the two can be used as substantial volumes of trade in goods. a proxy measure of quasi-transit trade (or Key recommendations include: However, it can occur at any transhipment the Rotterdam-Antwerp effect). port. 1. Re-export flows are growing in The methodology for the Netherlands importance globally. There are currently In essence, the Rotterdam-Antwerp critically relies on trader-level data no robust estimates of re-export flows effect is an issue of trade data accuracy. as opposed to the product-level data for the UK. The methodologies and Within the context of UK trade with the available in the case of Belgium. It yields estimates in this project provide a useful Netherlands and Belgium, it has been the value of Dutch imports aimed for new resource for understanding UK trade argued that this effect overestimates the re-export by year, country of destination, patterns. importance of these countries (and, by country of origin and commodity code (CN extension, the EU) as trading partners for or HS eight-digit codes). The datasets have 2. This project has illustrated the benefits the UK and therefore underestimates UK been made available to us under Remote of using rich micro data and cross-country trade with the rest of the world. Access to secure data, a fee-based service collaboration to study UK trade patterns. provided by the micro data services of It points to the potential benefits of This project sought to obtain more precise CBS. developing new data collection methods estimates of the extent of transit flows involving HM Revenue and Customs via the Netherlands and Belgium in the (HMRC). UK data, and the characteristics of these Findings and recommendations flows. This is important as the UK seeks to establish an independent trade policy The Rotterdam-Antwerp effect is difficult Impact and engagement outside the EU. to quantify precisely because official UK trade documentation does not capture This study is of direct policy relevance the information required to measure it. at this time, as the UK embarks on new Methods We aimed to further our understanding trade policies outside the EU; it has of the relevance of this effect for the UK, helped inform thinking at the DIT, DEFRA Our research is based on two datasets: estimating the impact on trade statistics in and elsewhere. The work was developed a publicly available dataset from the the context of bilateral trade relationships through discussions with ONS, HMRC, DIT, National Bank of Belgium (NBB) and between the UK and Belgium and the DEFRA and the Trade Remedies Authority. confidential firm-level data from Statistics Netherlands. Previous studies aimed Netherlands (CBS). The Foreign Trade at investigating the prevalence of the The research carried out in this project database compiled by the NBB contains Rotterdam-Antwerp effect have relied on has been possible because of our detailed information on Belgium’s imports strong and often arbitrary assumptions collaboration with Oscar Lemmers at CBS.
79 80 Using Administrative Data to Develop New Labour Force and Migration Statistics
Both net and gross migration to the UK have risen sharply over the past two decades, in particular since EU expansion in 2004. At the same time the nature of migration has changed, with short-term, seasonal and circular migration appearing to become increasingly common. This poses considerable challenges for economic statistics, in particular in relation to the labour market, as well as for population statistics. The availability of administrative data has the potential to make a major contribution to understanding in these areas.
Andrew Aitken Augustin de Coulon Richmond Egyei Michael O’Connor Jonathan Portes Jonathan Wadsworth
81 82 Project overview eligibility to vote in different elections. As samples, and a coherent framework for In addition, this project has identified, such the ER is a potential source of local integrating data, additional covariates, analysed and produced new results There are currently no definitive sources area information on immigrant stocks expert opinion, and measures of which use local authority electoral roll with which to count the number of and flows. We explored whether the local uncertainty. Modelling data for as many data, where different populations have immigrants living in the UK or the area population counts of immigrants countries as possible rather than focusing different legal entitlements to participate associated inflows and outflows to and contained in the ER could help improve the on pairs of countries has the advantage of in local, national and European elections. from the UK. Instead, producers and users accuracy and reliability of published local harmonising variations between different This analysis provides new insights and of immigration data have to rely on survey area migration statistics. countries’ reported figures on migration, reveals significant changes in the way data data to produce immigration estimates and helps to estimate missing data. were collected which make comparisons at frequencies higher than provided by ‘Mirror statistics’ have been used in the over time difficult. This work has provided the decennial Census. Reliance on survey context of international trade to examine Key recommendations include: fresh insights on how a dataset in the data inevitably introduces uncertainty asymmetries in the value of imports and ONS’s possession could be used to give with regard to immigration estimates, exports. The same approach can be used 1. There is sufficient evidence to warrant supplementary data to support survey due to sampling variation, definitions of to examine discrepancies in UK migration further investigation of the usefulness collections. immigration and appropriate weighting data. In the context of migration this of the electoral roll as a supplementary procedures to gross up from sample involves a comparison of two measures measure of local area migration statistics. to population. Administrative data of the same flow of migrants (immigrants sources can provide new information on to one country and emigrants from 2. Existing harmonised and consistent international migration patterns in the UK, another) produced by different countries. methods across OECD countries for often at a highly granular level of detail. We reviewed potential data sources and estimating immigration based on Use of these data sources introduces new methods that might be applied for these household survey data could be exploited issues around coverage and comparability. purposes. to support the reconciliation of estimates We considered the potential for local of cross-country population flows. authority electoral roll data to inform international migration statistics. We Findings and recommendations reviewed differences in migration patterns Impact and engagement arising in surveys that are commonly The ER provides a high-frequency used in UK immigration research, and the alternative data source with which to As international migration to advanced potential for comparisons of international estimate local area EU populations. The economies has increased over time, surveys to improve existing measures of availability of nationality data contained in policymakers have required better data immigration flows. the core data could extend the capability on international migration patterns to of the ER data to measure the wider understand the economy. immigrant population. The ER is not Methods without flaws. There is clear evidence of This project provided a forum for ONS inconsistency in the ER data around the staff and economists in academia to The UK relies on survey data such as the time of changes to the law on registration compare and contrast methods and to Annual Population Survey to produce in 2014 which hinders cross-time test the quality and relevance of different estimates of its local area immigrant comparisons before and after this date. datasets, and the different assumptions populations at higher frequencies than both communities make in relation to generated by the decennial Census. The Bayesian approaches to developing mirror this area. This has fed into the ONS’s electoral register (ER) should be a count statistics should be preferred to other work programme to transform migration of the resident population aged 16 and approaches as they provide flexibility statistics making increased use of over. Respondents are obliged to record in dealing with the limited number of administrative data. their nationality since this affects their observations generally found in survey
83 84 Making Sense of Skills
Skill shortages are costly and can hamper growth, but these shortages are not currently measured in a detailed or timely way. To address this challenge, we developed the first publicly available data-driven skills taxonomy for the UK. A skills taxonomy provides a consistent way of measuring the demand and supply of skills. It can also help workers and students learn more about the skills they need, and the value of those skills.
Hasan Bakhshi Jyldyz Djumalieva Cath Sleeman
85 86 Project overview positive disposition). To generate the data engineering, securities trading, IT Profile Builder which was launched earlier taxonomy, we employed machine-learning security operations, IT security standards this year. The project’s data visualisations Skill shortages arise when there are methods such as word embeddings, and mainframe programming. To date, were shortlisted for an Information not enough people with particular skills network community detection algorithms workers and students have had to decide is Beautiful award in the Science and to meet demand. The Open University and consensus clustering. We modelled between these skills without access to Technology category. The World Economic estimates that skill shortages cost the skills mentioned in job adverts as a graph this information. Skills that attract high Forum report on ‘Strategies for the New UK £2 billion a year in higher salaries, with individual skills as vertices and their annual salaries and for which demand has Economy: Skills as the Currency of the recruitment costs and temporary staffing co-occurrences in job adverts as edges. been growing include data engineering, Labour Market’ cited the skills taxonomy as bills. Once skills were represented as a network, IT security implementation, IT security one of the emerging initiatives for creating we hierarchically grouped them into operations, marketing research, app a skills-based labour market. The research Despite the importance of skill shortages, clusters. The greater the probability of two development and web development. This team received a large number of queries these are not currently measured in a skills appearing in the same advert, the may reflect a shortage of workers who from stakeholders who wished to share detailed or timely way. The best available more likely it was that they ended up in have these relatively new skills. lessons learnt and to explore opportunities estimates come from the Employer Skills the same branch of the taxonomy. for applying the taxonomy and replicating Survey. While the survey can shed light Key recommendations include: the research. These included several on the various causes of skill shortages, international groups, such as the OECD, it is only conducted once every two years Findings and recommendations 1. Our taxonomy gives 143 clusters of the Inter-American Development Bank and and it focuses on broad rather than specific skills. Our methodologies can SkillsFuture Singapore, demonstrating that detailed groups of skills. Looking ahead, The taxonomy was developed in three be used to create skill taxonomies with accurately measuring skill shortages is an skill mismatches may worsen because the layers. The first layer contained 6 broad further layers to give a more granular international challenge. skills we need are changing, for example clusters of skills; these were split into picture of skills demand. due to factors such as automation. The 35 groups, then split once more to give first step to measuring shortages is to 143 clusters of specific skills. Each of the 2. Online job adverts might also be used build a skills taxonomy which would define approximately 10,500 skills lives within one to generate estimates of regional skills accepted groups of skills. Despite having of these 143 skill groups. demand and its composition, for example well-established taxonomies for defining by occupation or industry. occupations and industries, the UK does The taxonomy is much more than a list not have an accepted skills taxonomy. of skills. It provides estimates of the 3. The skills taxonomy is a first step demand for each skill cluster, based on towards a real-time map that links skills, the number of mentions within adverts. occupations and qualifications. Such a Methods Users can search the taxonomy by job map could be used to provide timely, data- title and discover the skills needed for driven and detailed guidance to workers. To build a skills taxonomy for the UK a wide range of jobs. The five clusters we began with a list of just over 10,500 containing the most frequently demanded unique skills that had been mentioned skills are social work and caregiving, Impact and engagement within 41 million UK job adverts, collected general sales, software development, between 2012 and 2017 and provided office administration, and driving and This project used a large database of job by Burning Glass Technologies. These automotive maintenance. vacancies, collated via web scraping, to ‘skills’ are more like job requirements: in produce taxonomies for occupations and addition to skills (e.g. app development) Around 60 per cent of job adverts mention skills. The work on occupations is being they include specific tasks (e.g. insurance a salary. This information can be used used by the Classifications Team in ONS underwriting), knowledge (e.g. biology), to provide estimates of skill values for to inform their update of occupation software programmes (e.g. Microsoft the UK. The five skill clusters with the classification codes. Google Digital Garage Excel), and even personal attributes (e.g. highest median annual salaries are used the skills taxonomy to build their
87 88 Using Administrative Data to Measure New Forms of Working
Business owners have been the fastest-growing part of the UK labour force since at least 2000. This is an important labour market trend and is often hailed as a success because small businesses and start-ups are commonly viewed as the engines of growth. This is questionable in light of evidence that the UK has a long tail of low-productivity firms. However, understanding of business owners has been limited because they are not well captured in traditional survey data sources. We used administrative tax records to learn more about them.
Jonathan Cribb Thomas Crossley Helen Miller Thomas Pope
89 90 Project overview possible using survey data. Specifically, overrepresented at both the top and all questions. With more data, particularly using these data, we documented the the bottom of the income distribution. linked data, researchers could make It has become commonplace to state numbers, characteristics, incomes and A minority of closely held businesses greater progress in understanding the that the labour market is fundamentally business activities of business owners in have substantial costs, make large capital dynamics of businesses and the labour changing. The perceived truism is that the the UK. We tracked the same business investments and will be employing others, market. workforce can no longer be characterised owners over time – something that has not but the majority have low costs and do not by employed individuals with single, full- or been possible before – and used this to invest or employ others. In many cases, part-time, positions, and instead includes analyse patterns of business start-up and profits will mainly reflect a return to the Impact and engagement a significant number of independent closure and to explain substantial falls in owner-manager’s labour. workers with flexible, often intermittent, sole trader incomes since 2008. We held an event in June 2018 at income streams. Yet, despite the growing It is common to describe those who are the Institute for Fiscal Studies, ‘Self- body of anecdotes and examples – Uber Our analysis relied principally on the active in the labour market as either Employment and Entrepreneurship: being perhaps the best known – and universe of self-assessed income tax ‘business owners’ or ‘employees’. But a Lessons from Tax Records and Challenges the concerns about what this means for records (those from 1997/98 to 2015/16 quarter of business owners earn some for Policy’, at which the results from this various individuals’ jobs and welfare, we are available to researchers) and the employment income while running their project were presented and discussed. know remarkably little about the changes available corporation tax records (available business. While the business owner This event was well attended by taking place. to researchers from 2000/01 to 2014/15), population has grown especially quickly academics, civil servants and business which are also linked to companies’ since 2007/08, this growth has been representatives. Although our work has The data collection techniques that accounts. In addition, for a subset of our accompanied by a decline in reported not been explicitly used in further analysis currently underpin the National Accounts, analysis on company owner-managers, capital investment. Sole traders, who at ONS, it acts as a good example of what where measures of activity are largely we utilised a match between directors’ account for 90 per cent of the net growth can be done with sufficient access to based on surveys of businesses, were not personal tax records and the company’s in business owners since 2007/08, have administrative datasets. designed to capture aspects of working corporation tax record. seen large falls in income. These findings patterns such as the volatility of self- highlight that the ‘number of people employed incomes or moves between running a business’ is a poor measure of legal forms, issues which have become Findings and recommendations the economic contribution of that sector. more important in the ‘gig economy’. We used administrative data from tax records Since the turn of the century, the business Key recommendations include: to capture how economic activity is owner population has grown rapidly, changing (whether people are employees, and much faster than the number of 1. Having a better understanding of the sole traders, partners or owner-managers), employees. Owners now number over 6.5 business owner population is important what compositional shifts in these million, up from 4.8 million in 2000/01. for public discourse and policymaking. workforces mean for income growth This growth has been incentivised by a tax A failure to appreciate the diversity of and how well such activity is captured in system that taxes business owners less business owners may lead to poorly current surveys. heavily than employees. targeted policy.
Business owners are a diverse group that 2. This project has shown the benefits of Methods cannot be accurately summarised by a using administrative tax data to study the ‘one size fits all’ description. They operate UK population of business owners. We We used the universe of business owners’ in all sectors of the economy; they include can say much more than was previously administrative tax records provided by some ‘gig economy’ workers, but also, possible about the legal forms people HM Revenue and Customs (HMRC) to among others, partners in law firms, sole use, the activities they engage in and the learn more about business owners and trader construction workers and company ways in which they take their income. their businesses than has previously been owner-manager IT consultants. They are However, this does not provide answers to
91 92 Regional Nowcasting in the UK
We explored and implemented methods for timelier, higher-frequency estimates of economic growth in the UK regions. We developed a model that enables estimation of regional estimates of gross value added (GVA) on a quarterly basis, to approximately the same timetable as the release of UK GVA data. These are currently released through the ESCoE website and are being used by decision makers at a sub-national level. Our estimates have also enabled a better understanding of the pattern of historical regional growth, opening new areas for analysis.
Gary Koop Stuart McIntyre James Mitchell Aubrey Poon
93 94 Project overview Methods additional research is needed to address We are publishing nowcasts as regular these. research outputs. These have generated More timely and higher-frequency regional We used Bayesian Vector Autoregressive media coverage, including in the Financial macroeconomic indicators are essential (BVAR) models, widely used to produce Key recommendations include: Times, and have been cited by many for effective economic policy and decision timelier and higher-frequency estimates sources in the private sector and indeed making. With significant policy emphasis in of a range of economic variables like 1. For many purposes, stakeholders wish by politicians in the UK Parliament and the UK on greater devolution and localism economic growth. In broad terms, greater spatial and sectoral resolution outside. and on promoting growth across the nowcasting methods using these models in economic data than that provided UK, we need timelier regional economic seek to exploit the mixed-frequency nature by aggregate region level estimates. Our work has the potential to inform ONS’s data. While there have been significant of the data and accommodate the differing Econometric methods to extend these estimation approach in their development advances in improving the timeliness of publication timetables of those indicator estimates to greater regional and sectoral of quarterly output indicators for the ONS economic data at the national level, variables chosen for their putative ability disaggregation should be explored. English regions. In the coming years, we these have not, as yet, been reflected in to explain the within-year or quarterly will work closely with ONS to integrate improvements at the regional level. In variable of interest, such as quarterly 2. As ONS start to produce their their newly produced quarterly country parallel to our research though, the ONS economic growth. A range of different own Regional Short-Term indicators, and regional GDP estimates into our has worked to develop a set of Regional econometric methods have been used it is important that these are also model and to develop their ability to Short-Term Indicators in order to improve to ‘nowcast’ macroeconomic variables. accommodated within our model. Further use these methods, enabling new ONS the timeliness of regional growth data by We used two main models to explore research is required to establish the best statistical outputs. several months. But in our work, the data these issues: a stacked VAR model and a way to do this. were as follows. For the UK as a whole, state space VAR model. In order to make the preliminary estimate of quarterly GDP these models suitable for this application, 3. Given the value that timelier, higher- was released around 6 weeks after the end we explored different econometric frequency estimates of regional growth of the quarter; in Scotland, quarterly GVA innovations to meet our needs and in have with stakeholders, it is important data were released, on average, around particular to impose the ‘cross-sectional’ that work is undertaken to make our 100 days after the end of the quarter. constraint that the quarterly regional data model available to ONS to run and develop There are some quarterly short-term add up to the observed UK data. We also themselves. indicators for Wales and Northern Ireland, explored the use of a ‘machine learning’ but these have a limited time series. GVA prior within our BVAR model to enable data for the other regions in the UK were more efficient computation and deliver Impact and engagement available only at the annual frequency. more accurate nowcasts of regional Furthermore, they are released nearly a growth. We have developed methods to produce year after the end of the year to which model-based estimates of regional growth they relate. to a similar timetable to GDP estimates Findings and recommendations for the UK as a whole. This represents a Our research has demonstrated that substantial improvement in timeliness; our empirical macroeconomic methods Our research demonstrated that methods also provide consistent historical are able to produce higher-frequency empirical macroeconomic methods regional growth data back to 1970 estimates of regional economic growth are able to produce higher-frequency enabling current regional growth to be set on a consistent basis across the UK to and more timely estimates of regional into an appropriate historical context. This the same approximate timetable as economic growth. This meets a key meets key demands for both more up-to- ONS estimates for the UK as a whole. demand from stakeholders for more date and more historical information to Importantly, our quarterly regional timely information on the performance help inform policy and decision making estimates exploit and add up to the latest of regional economies. However, several at the regional level, particularly with the quarterly estimates for the UK as a whole. issues remain unsolved at this point, and devolution and ‘City-Region’ deal agenda.
95 96 Improving Regional Economic Indicators
There is increasing focus on regional economic performance and devolution of powers within the UK. A central issue is regional economic integration. We mapped out existing data sources of inter-regional trade, produced a feasibility study of methods to estimate trade flows within the UK, and produced estimates using these methods. We highlighted weaknesses in the data landscape of inter-regional trade within the UK and proposed a framework for future data collection and estimation.
Alastair Greig Katerina Lisenkova Aubrey Poon Graeme Roy Mairi Spowage
97 98 Project overview and transport and communications. England, Northern Ireland, Scotland and Other services were broadly based on Wales, using the best data available. Improved inter-regional trade flow consumption shares, with recreational Further progress should involve a data is vital to understanding regional and tourism spend calculated through coordinated approach between ONS and economic policy and regional fiscal policy the range of UK Tourism Surveys, relying the devolved administrations on trade and to monitoring the principle of ‘no- strongly on Northern Ireland tourism data. collections in goods, a qualitative review detriment’, which is central to devolution Retail was estimated separately based of service sector companies, a quantitative settlements. Robust data to measure the on the distribution of margins across appraisal of methods used for estimating degree of integration between different sectors. Inter-connector data were used to inter-regional service trade, testing the parts of the UK is required to inform the estimate flows of trade in utilities. sensitivity of current approaches, and delivery of the devolution process. For providing recommendations. example, it is crucial for forecasting future regional economic performance and also Findings and recommendations understanding the implications of major Impact and engagement policy shifts. Research in the last few years The results overall suggest a high degree on multi-regional modelling of the effects of linkage between the countries of the Our work in this project can help support of Brexit and the new Fiscal Framework UK on trade, significant in the context regional policymaking in the context has highlighted just how sensitive the of international export partners. The of ever-increasing interest in regional analysis is to data on regional integration. production of these estimates has raised consumption, activity and trade. The In this project we took some first steps a number of issues about the current process of producing the estimates towards the development of better collection of trade data in the UK, and in this project and our work with the statistics on inter-regional linkages. how those data are then used in the devolved administrations and Whitehall production of regional Supply and Use departments have highlighted the need Tables. There are differences in terms of for a coordinated effort to improve Methods approaches taken, consistency, and the the coverage and coherence of inter- data used. There is growing policy interest regional trade information within the This project proposed a framework for in understanding the different linkages UK. This has led to new proposals, estimating the origin and destination between parts of the economy in the UK, which have the support of all these of inter-regional trade between the both in terms of linkages between industry organisations, and have the potential to nations of the UK: England, Northern and links between regions. Devolution of lead to a transformation in the collection Ireland, Scotland and Wales. It explained new powers, and of course Brexit, have and estimation of inter-regional trade where gaps exist in the current UK data heightened the policy requirement for within the UK. This could also lead to a landscape and suggested various ways data such as these. framework for the consistent production in which these could be addressed. of Supply and Use Tables for England, Each sector of the economy has distinct Key recommendations include: Northern Ireland, Scotland and Wales. methodologies for the estimation of inter- Ultimately, these tables are the methods regional trade. Each sector can be seen to 1. We recommend the development of through which these trade estimates are be estimated either through an import or a strategic approach to the collection of reconciled with more established figures export methodology, or both. In general, trade information and estimation of trade for regional production and consumption. given the sources used, export-orientated within the UK, with different approaches methods tend to give more reliable for goods and services. results. Freight data was used to estimate the origin and destination of agriculture, 2. This project considered and estimated other primary goods, manufacturing, inter-regional trade flows between
99 100 The Impacts of Trade on Income, Employment and Inequality in the United Kingdom and its Regions
The nature of international trade is changing. Conventional indicators of the economic significance of trade for countries and regions are much less informative than they used to be. We developed new indicators and investigated opportunities for the construction of high- quality databases required to operationalise these for the case of UK regions. These were also used to analyse the consequences of the changing nature of international trade for the UK and its regions.
Peter Levell Bart Los Matthias Parey Marcel Timmer
101 102 Project overview only be done for very specific products trade patterns with the EU: the case of Outputs from this project were published through detailed case study approaches. a total stop of exports (of goods and in a special issue of Fiscal Studies on The emergence of Global Value Chains The recent availability of several global services) to the European Union and Brexit-related economic research, as well (GVCs) has profound implications input-output databases, such as the World the case where UK goods exports to the as in the NBER Working Paper series. for the assessment of regional trade Input-Output Database (WIOD) and the EU remain unhampered, but where the Our work has also been disseminated performance and the implications of OECD’s Trade in Value Added database UK cannot export services to the EU, and has received feedback via seminars, Brexit for UK regions. Suppose we want to (TiVA), from 2013 onwards has changed for example because it is outside the workshops and conferences, including an quantify the extent to which income and this dramatically. An input-output table Single Market. We found that the risks ESCoE Research Seminar at ONS London, employment in, for example, the West in such databases gives a quantitative vary considerably across UK regions and a workshop on regional modelling in Midlands region are exposed to a sudden description of the world production are disproportionately concentrated in Birmingham, conferences in Cambridge, at reduction in UK exports to continental structure, and its connections to users of Northern regions in the case of a total stop the Bank of England, and at an ESCoE-ONS Europe. The traditional approach would final products. Using these tables a host of exports to the EU, and in London and special session at the annual conference of be to look at the value of sales to EU of indicators related to GVC-trade can southern areas of the UK in the simulated the Royal Economic Society. destinations. However, any effects on be constructed. We adapted our VAX-D case where the UK cannot export services the region will also depend on the extent indicator to focus on export-related to the EU. to which its exporters source parts and employment risks of Brexit for UK regions. components and business services from We used EUREGIO’s global input-output Key recommendations include: suppliers in the region itself, or from tables with inter-regional detail for EU other regions or countries. In the case of countries. This database has WIOD as 1. Understanding the strengths and the West Midlands, much of the export its point of departure but contains inter- weaknesses of the economic relationships value consists of value added generated regional detail for EU countries at the level between the regions of the UK is elsewhere and the hit that the region of NUTS2 (2010 classification). For some important for studies of regional inequality would take from a reduction in exports analyses, we merged these data with and for understanding the effects of to continental Europe would be relatively information on jobs by industry and region macroeconomic policy on different smaller than the value of sales data would derived from ONS’s Business Structure UK regions. High-quality inter-regional imply. Conversely, the West Midlands data. Supply and Use Tables are therefore itself might also be affected in ways that indispensable, since these can provide do not show up in gross exports to EU detailed information on the economic destinations, for example when West Findings and recommendations dependencies of industries within, but also Midlands firms supply intermediate between, regions. inputs to exporters located in other We computed the numbers of jobs per UK regions. The exposure of a region region employed, directly or indirectly, in to trade thus depends crucially on its producing goods and services exported Impact and engagement position in regional and global production to EU regions. Indirectly affected networks. In this project, we developed jobs are associated with activities of Our research provides new insights into more insightful indicators for regional which the output is sold to domestic the role of GVCs in different sub-national trade performance in an era of networked industries, either in the same region or parts of the UK, and a measure of the production. elsewhere in the UK, that export to the international competitiveness of these EU. We consider jobs that are directly or areas. Although this would not form indirectly employed in the production part of a regular statistical production Methods of exports to the EU as jobs that are ‘at process, this has helped ONS to meet risk’. We calculated the percentages of their priorities of better information below Until recently, analyses of the roles of jobs ‘at risk’ in each of the regions under the whole-economy level, and better regions and countries in GVCs could different assumptions about future measurement of the labour market.
103 104 Modelling and Communicating Data Uncertainty
Due to sampling and non-sampling errors, economic data, like GDP, are uncertain. Despite this, economic data are frequently published without any direct quantitative indication of their uncertainty. This project considered the measurement of data uncertainty and its communication. Experimental surveys were used to provide guidance to statistical agencies on how the public and ‘experts’ interpret data, and how uncertainty information might best be communicated when publishing national account estimates.
Ana Galvão James Mitchell Johnny Runge
105 106 Project overview uncertainty information communicated in value’. The majority, whether asked about Impact and engagement different formats, and how communicating this qualitatively or quantitatively, expect This project had two aims: to measure uncertainty affects trust in the data and data uncertainty and are not surprised By showing how communicating data data uncertainty, and then to analyse the the producers of the data. By randomly when data are revised. Importantly, uncertainties can improve the public’s reactions of a range of decision makers to allocating members of the public into one we found that whether uncertainty understanding of data without eroding communication of this uncertainty, with of six groups - a control group presented information is communicated to the public, trust, our research might encourage uncertainty modelled, communicated and with no uncertainty information, and five and how it is communicated, matters. national statistical offices to do more visualised in different ways. A key question treatment groups presented with different Communicating uncertainty information to measure and communicate data was whether or not communicating data uncertainty information - we identified alongside the GDP point estimate affects, uncertainties. It is also hoped that the uncertainty might erode trust in official the causal effects of different ways of and improves, public understanding of work of this project will encourage more economic statistics. communicating uncertainty information. why data revisions happen. It encourages empirical research into uncertainty The results can be used to help inform more of the public, rightly, to view the measurement and communication, such if and how data uncertainty information point estimate as just that: a point within as the Comunikos project at Eurostat. Methods should best be communicated. a range of possible outcomes. Our randomised experiments also revealed Our work has led to interest from central Economic data uncertainty is rarely In complementary work we provided that these quantitative communications banks, including the Bank of England; communicated quantitatively. An an interdisciplinary review of structures of GDP data uncertainty need not reduce Bank of Canada; the Federal Reserve exception is the ‘fan chart’ for historical and summarised current practice and trust in statistical agencies. Banks of Cleveland, Dallas and St. Louis; GDP growth published at the Bank of research in uncertainty communication. and the Reserve Bank of Australia. Our England. We used these fan charts, and This combined statistical and psychological Key recommendations include: work has also benefitted from discussions estimated econometric models of GDP perspectives. at seminar presentations at University data revisions, to measure, track and 1. The ONS already emphasises the College London, Reading, Warwick evaluate the evolving nature of GDP data uncertainty of early GDP data releases by Business School and ONS. uncertainty. We proposed a generic loss- Findings and recommendations indicating that the data will be revised. function-based approach to extract a We recommend the ONS reconsiders if quantitative measure of unforecastable We provided quantitative estimates of and how it communicates the uncertainty data uncertainty from the Bank of unforecastable data uncertainty for UK associated with early GDP estimates to England’s ex ante density forecasts. GDP growth and found strong evidence improve public understanding of these. of changes over time to the GDP growth With the aim of assessing empirically if data revision process. We also found 2. We recommend that uncertainty and how the public and more experienced that data uncertainty rises at the onset information is communicated users interpret and understand GDP data of recessions and is positively correlated quantitatively using intervals, density strips uncertainty, we conducted a randomised with popular measures of macroeconomic and bell curves. These are, in general, controlled online experiment with more uncertainty. Methodologically, we defined preferable to textual descriptions and than 3,000 nationally representative unforecastable uncertainty as a lack certainly better than no communication members of the public and a targeted of knowledge about the future or the at all beyond referring to GDP as an survey of more than 100 experts. To past and showed how our uncertainty ‘estimate’. Our experiments suggest maximise realism, both surveys asked measures can be used as the basis for this will reduce the chance of the public questions about ONS’s latest GDP the construction of tests for calibrating misinterpreting uncertainty information estimates and headline press release. The probabilistic forecasts. given to them. specially designed surveys were used to assess perceptions of the uncertainty in Our experimental results indicated that single-valued GDP numbers, the public’s the majority of the public, like experts, interpretation and understanding of do not take GDP point estimates at ‘face
107 108 Valuing Economic Statistics
This explorative study tested ways of measuring the value of official economic statistics through qualitative research with regular users. It identifies a number of challenges in obtaining quantifiable evidence of value from statistics users and recommends the use of case studies that provide measurable findings. A separate case study explored the economic impact of GDP data revisions. The study suggests that the value of economic statistics is very high. A future study could test the effectiveness of providing such information to the public.
Amit Kara Jason Lennard Rebecca Riley Heather Rolfe Johnny Runge Sylaja Srinivasan
109 110 Project overview exceeding the cost of actually providing have influenced decisions? If so, can we measuring the value of their statistics. In the statistics. Participants attached great measure the consequences of those particular, the methodology behind such In March 2015, the Conference for importance to ONS economic statistics, decisions? We piloted this approach by an estimate must conform to the high European Statistics (CES) established a saying they were invaluable to their work measuring the macroeconomic impact of standards followed by official statistical United Nations Economic Commission for and in informing policy decisions. In terms using vintages of GDP data. We examined agencies; otherwise it could potentially be Europe (UNECE) Taskforce to define the of placing a monetary value on economic this question through the prism of counterproductive and even undermine value of official statistics and develop ways statistics, we tested a ‘stated preference’ monetary policy, restricting our focus to the perceived reliability and independence to measure their value. The Taskforce approach, in which we asked participants two episodes when the ONS introduced of statistics themselves. published its first report in 2017. One of how much they (or their organisation) sizeable revisions to its preliminary GDP its recommendations was for national hypothetically would be willing to pay to data. statistical offices to explore ways of placing have access to ONS economic statistics. Impact and engagement a monetary value on official statistics. In Our study identified several challenges Key recommendations include: this context, ESCoE was commissioned in applying this approach to economic This project tested the viability of by ONS to test these approaches This statistics. One was that most regular users 1. To improve understanding of the value surveying users, alongside using focus was the first time that the UNECE of the statistics do so as employees of of economic statistics, future studies might groups, to understand how those users recommendations had been piloted in an organisation, and they have limited consider how to construct a survey sample value economic statistics. This area is the UK on official economic statistics. knowledge of their organisation’s budget, which includes a mix of people who are complex, and is one the UN has been As such, this study was exploratory in and of how their work contributes to involved and knowledgeable in policy, investigating for the last five years. We nature, focused on identifying promising improving policy decisions or impacts organisational and budgetary decisions, tested some aspects of the UN’s proposed approaches and ways of mitigating individual and organisational outcomes. as well as those who use the data in their methodology, identifying a number of challenges, as well as learning lessons for Another strong theme was that day-to-day work. challenges in using monetary estimates in future studies. participants were somewhat nervous the UK context. Our work has helped the about hypothetical questions about 2. Research involving hypothetical ONS to deliver new insights on this topic to willingness to pay. They said that the questions about users’ willingness to pay further international attempts to develop Methods idea of paying for economic statistics for economic statistics should have a clear valid and efficient methods to elicit this was not that far-fetched, and participants strategy on how to identify and analyse information from users, who may be either The project had two parts. First, we carried repeatedly emphasised the value of ONS ‘protest answers’, in which respondents uncomfortable with this topic, lacking in out qualitative research with regular economic statistics as a public good. simply refuse to accept the hypothetical the necessary data, or fearful the exercise economic statistics users across different premise of the question. is actually seeking evidence to set real stakeholder groups, using a survey and In our broader discussions about how charges. focus groups. Second, we carried out to value economic statistics, participants 3. We recommend the use of case study an empirical case study, illustrating the found it useful to assess their value by evidence to demonstrate the value of The case study evidence we developed macroeconomic cost of ‘early estimates’ thinking about the counterfactual; that economic statistics. Such studies need can be used as a tool for communicating of some key economic variables. We is, what would happen in the absence of to show how economic statistics inform the value of economic statistics to a wide quantified the economic impact of ONS economic statistics? Most participants decision making and to quantify the range of audiences, and for engaging with estimates of GDP data using NIESR’s suspected the data in an open market consequences of those decisions. These students. macroeconomic model, NiGEM. would be less reliable and accountable might be followed up by qualitative than official economic statistics. This led to studies with the public to learn how the The UK, on the back of our research, is one of our main recommendations: to use value of economic statistics could be leading global efforts to develop strong Findings and recommendations innovative case studies that show the cost communicated effectively to the public. methods for valuing all types of statistical of policy decisions based on statistics at data, not just economic. Through our survey and focus groups, we an early stage of the estimates. Such case 4. The final recommendation emanating showed that the value of official economic studies involve addressing two questions: from our research is that national statistics statistics is very high, most likely far Can we show that economic statistics offices should proceed with caution in
111 112 What Next?
113 114 So what next? What are our plans for the us to expand our international network. year. Current research projects future? We intend to continue to meet and expand on the original objectives of Economic welfare in the digital economy Our research funding for the current year We also plan to capitalise on other funding ESCoE; to work towards our ambition of and the next two, has been boosted by opportunities where we can take our being an international point of reference Real-time turning point indicators: a third, due to funds from organisations research out to economics students, both for research on economic measurement international review of current practices other than the ONS. We continue to at schools and universities, to help raise in the modern economy; and to support seek additional funding to improve the the profile of economic measurement cultural change in the delivery of economic Use of administrative VAT and PAYE sustainability of ESCoE. research. statistics to achieve better measurement data to produce industrial output and through research. Our immediate plans earnings data are outlined below. UK Historical Data Repository Bringing together expertise Creating space for debate
Developing sub-national measures of Our Research Associates and international Our research seminar, the ‘Measurement Research overview democratic income and other indicators network are our most valuable assets. in the Modern Economy Series’ and topic of welfare We will continue to support our early- workshop programmes will continue to Our first projects are now coming to career Research Associates and PhDs in create a forum for stimulating discussion fruition and some of the methodologies, Measuring human capital their career aspirations and extend our and debate on new research and ideas models and conclusions that we have expertise by involving more academics bringing together leading academics, developed are being put into practice Measuring digital innovation and institutions in our projects. We will researchers, students, government and incorporated into the ONS’s work continue to invite international experts to statisticians and other practitioners. The impact of offshore profit shifting on and informing international debate. visit ESCoE; as these relationships develop, We plan to make these discussions the mismeasurement of the GDP: the we will encourage them to become even more accessible to a wider audience by Through our current research programme case of the UK we continue to develop economic more involved in the work of the Centre, increasing the number of videos, podcasts measurement in four broad areas: Using firm-level surveys to understand expanding our network. We are also and blogs available on an updated website National accounts and beyond GDP, industrial and regional capacity, strengthening collaboration with the Bank due in 2020. Productivity and the modern economy, investment, productivity and output of England through secondments and joint Regional and labour market statistics, and growth work. Alongside the annual conference we have Communicating and valuing economic plans for a number of themed sessions statistics. Within these, new research Developing firm-level micro data for The appointment of an Academic Director at other conferences, as well as for one- projects strengthen our focus on the productivity analysis from outside the UK to build on the day conferences and forums on specific measurement of welfare and prices and work of our outgoing Academic Director, economic topics and in association with Using administrative data to measure measurement issues arising due to the Professor Richard J. Smith (University other organisations and projects, including income transitions globalisation of production. of Cambridge), will help strengthen with the International Association for Developing experimental estimates international linkages. Research in Income and Wealth and with We have delivered an impressive number of regional skill demand, supply and Rebuilding Macroeconomics. of ESCoE publications in our first years and mismatch this library of papers will continue to grow The next generation rapidly as our first projects are ending. Regional nowcasting in the UK Working with ONS We are delighted to include papers from As well as continuing to support our others in the field, both academics and Improving the quality of regional current ESCoE PhD students at King’s Whilst we seek to expand collaboration practitioners. As ESCoE becomes more economic indicators College London and the University of with international statistics agencies, recognised internationally, we anticipate Nottingham, we are looking to offer academia and new funders, we are Public understanding of economics an increase in the number of external further opportunities at the University of also forging closer ties with ONS. As researchers contributing, in turn helping Cambridge from the 2020/21 academic the impact of our first projects is felt,
115 116 we are expanding our reach within the organisation. We want to continue to support an increased research focus within ONS; one way we plan on doing this is by bringing our expertise to more individuals and teams. Through PhD opportunities with our partnering institutions and further developing our secondments into a two-way programme, we hope to be able to help ONS upskill early-career economists.
ESCoE Economic Measurement Conference
It is anticipated that the annual Economic Measurement conference will continue to go from strength to strength and we are looking forward to welcoming everyone to EM2020 in May 2020 at the fantastic King’s College London Business School. We intend to take the conference to Glasgow for 2021, so look out for further announcements.
117 118 Our publications of Personal Income’ (ESCoE DP 2019-14) by (ESCoE DP 2019-05) Lena Boneva, ‘Can GDP Measurement be Further by Dennis Fixler, Marina Gindelsky and James Cloyne, Martin Weale and Tomasz Improved? Data Revision and David Johnson Weiladek Reconciliation’ (ESCoE DP 2018-15) by Jan P.A.M. Jacobs, Samad Sarferaz, Jan-Egbert Discussion Papers ‘Misreported Trade’ (ESCoE DP 2019-13) by ‘Understanding the Sharing Economy’ Sturm and Simon van Norden Mohammad Farhad, Michael Jetter, Abu by (ESCoE DP 2019-04) Diane Coyle and ‘Measuring Education Services Using Siddique and Andrew Williams Shane O’Connor ‘Regional Output Growth in the United Lifetime Incomes’ (ESCoE DP 2020-02) by Kingdom: More Timely and Higher Carol Corrado, Mary O’Mahony and Lea ‘Who are Business Owners and What ‘Measuring Bilateral Exports of Value Frequency Estimates, 1970-2017’ (ESCoE Samek Are They Doing?’ (ESCoE DP 2019-12) by Added: A Unified Framework’ (ESCoE DP DP 2018-14) by Gary Koop, Stuart Jonathan Cribb, Helen Miller and Thomas 2019-03) by Bart Los and Marcel Timmer McIntyre, James Mitchell and Aubrey Poon ‘Measuring Productivity: Theory and British Pope Practice’ (ESCoE DP 2020-01) by Nicholas ‘Labour Productivity and Firm-Level TFP ‘An Open and Data-driven Taxonomy of Oulton ‘The Welfare Implications of Public Goods: With Technology-Specific Production Skills Extracted from Online Job Adverts’ Lessons from 10 years of Atkinson in the Functions’ (ESCoE DP 2019-02) by Michele (ESCoE DP 2018-13) by Jyldyz Djumalieva ‘Communicating Data Uncertainty: UK’ (ESCoE DP 2019-11) by Fred Foxton, Joe Battisti, Filippo Belloc and Massimo Del and Cath Sleeman Experimental Evidence for U.K. GDP’ Grice, Richard Heys and James Lewis Gatto (ESCoE DP 2019-20) by Ana Beatriz Galvão, ‘Big Data and Macroeconomic Nowcasting: James Mitchell and Johnny Runge ‘The Treatment of Intellectual Property in ‘Towards a Framework for Time Use, Methodological Review’ (ESCoE DP 2018- the National Accounts’ (ESCoE DP 2019-10) Welfare and Household-Centric Economic 12) by George Kapetanios and Fotis ‘Defining and Measuring the by Robin Lynch Measurement’ (ESCoE DP 2019-01) by Papailias Innovativeness of Firms’ (ESCoE DP 2019- Diane Coyle and Leonard Nakamura 19) by Giuliana Battisti and Paul Stoneman ‘Lost in Translation: What do Engel Curves ‘Productivity Growth, Firm Turnover and Tell Us About the Cost of Living?’ (ESCoE DP ‘Cloud Computing and National New Varieties’ (ESCoE DP 2018-11) by ‘Testing for the Presence of Measurement 2019-09) by Ingvild Almås, Timothy K.M. Accounting’ (ESCoE DP 2018-19) by Diane Thomas von Brasch, Arvid Raknerud and Error’ (ESCoE DP 2019-18) by Daniel Beatty and Thomas F. Crossley Coyle and David Nguyen Diana-Cristina Iancu Wilhelm ‘Measuring Data Uncertainty: An ‘Temporal Disaggregation of Overlapping ‘A Firm-Level Perspective on Micro- and ‘Technology, Intangible Assets and the Application Using the Bank of England’s Noisy Quarterly Data Using State Space Macro-level Uncertainty’ (ESCoE DP 2018- Decline of the Labor Share’ (ESCoE DP “Fan Charts” for Historical GDP Growth’ Models: Estimation of Monthly Business 10) by Gaganan Awano, Nicholas Bloom, 2019-17) by Mary O’Mahony, Michela (ESCoE DP 2019-08) by Ana Beatriz Galvao Sector Output from Value Added Tax Data Ted Dolby, Paul Mizen, Rebecca Riley, Vecchi and Francesco Venturini and James Mitchell in the UK’ (ESCoE DP 2018-18) by Paul Tatsuro Senga, John Van Reenen, Jenny Labonne and Martin Weale Vyas and Philip Wales ‘GDP and Welfare: A Spectrum of ‘Variational Bayesian Inference in Large Opportunity’ (ESCoE DP 2019-16) by Vector Autoregressions with Hierarchical ‘Double Deflation: Theory and Practice’ ‘UK Trade in Goods and Productivity: Richard Heys, Josh Martin and Walter Shrinkage’ (ESCoE DP 2019-07) by Deborah (ESCoE DP 2018-17) by Nicholas Oulton, New Findings’ (ESCoE DP 2018-09) by Mkandawire Gefang, Gary Koop and Aubrey Poon Ana Rincon-Aznar, Lea Samek and Sylaja Philip Wales, Russell Black, Ted Dolby and Srinivasan Gaganan Awano ‘No Plant, No Problem? Factoryless ‘GDP is a Measure of Output, Not Welfare. Manufacturing and Economic Or, HOS Meets the SNA’ (ESCoE DP 2019- ‘The Digital Economy, New Products and ‘Classifying Occupations Using Web-Based Measurement’ (ESCoE DP 2019-15) by 06) by Nicholas Oulton Consumer Welfare’ (ESCoE DP 2018-16) Job Advertisements: An Application to Diane Coyle and David Nguyen by W. Erwin Diewert, Kevin J. Fox and Paul STEM and Creative Occupations’ (ESCoE ‘Firms’ Price, Cost and Activity Schreyer DP 2018-08) by Antonio Lima and Hasan ‘Improving the Measure of the Distribution Expectations: Evidence from Micro Data’ Bakhshi
119 120 ‘UK Regional Nowcasting Using a Mixed ‘The Mystery of TFP’ (ESCoE DP 2017-02) by Picture credits Frequency Vector Autoregressive Model’ Nick Oulton (ESCoE DP 2018-07) by Gary Koop, Stuart All photographs © ESCoE 2020, with the following McIntyre and James Mitchell ‘Do-it-yourself digital: the production exceptions: boundary and the productivity puzzle’ P9 Jonathan Athow (courtesy Office for National ‘Below the Aggregate: A Sectoral Account (ESCoE DP 2017-01) by Diane Coyle Statistics) of the UK Productivity Puzzle’ (ESCoE DP 2018-06) by Rebecca Riley, Ana Rincon- P18 PhD students (courtesy Royal Economic Aznar and Lea Samek Occasional papers Society) P19 Chiamaka Nwosu (courtesy Chiamaka Nwosu) ‘Imputation of Pension Accruals and ‘Valuing Economic Statistics: A Case study’ Investment Income in Survey Data’ (ESCoE (ESCoE OP 02) by Amit Kara and Jason P20 Paul Labonne (courtesy Paul Labonne) DP 2018-05) by Andrew Aitken and Martin Lennard Weale P37 Clock, photo by Megan Trace, CC BY-NC 2.0 ‘That’s Your Bloody Post-Truth’ (ESCoE OP licensed ‘Classifying Occupations According to their 01) Edited by Ed Humpherson P49 Street, photo by Ian Halsey, CC BY-NC 2.0 Skill Requirements in Job Advertisements’ licensed (ESCoE DP 2018-04) by Jyldyz Djumalieva, Antonio Lima and Cath Sleeman Technical reports P65 Lab, photo by TRIUMPH Lab, CC BY-NC-SA 2.0 licensed
‘A Data Map of Existing UK Data Sources ‘The Value of Economic Statistics: Baseline P77 Ship, photo by Kevin Hackert, CC BY-NC 2.0 Related to Regional Trade’ (ESCoE DP 2018- Report’ (ESCoE TR 05) by Heather Rolfe, licensed 03) by Alastair Greig, Katerina Lisenkova Johnny Runge and Sylaja Srinivasan and Graeme Roy P81 Passport control, photo by Eric Fischer, CC BY ‘Feinstein Fulfilled: Updated Estimates 2.0 licensed ‘A Democratic Measure of Household of UK GDP 1841-1920’ (ESCoE TR 04) by P85 Taxonomy (courtesy Nesta) Income Growth: Theory and Application to Solomos Solomou and Ryland Thomas the United Kingdom’ (ESCoE DP 2018-02) P89 Courier, photo by Sam Saunders, CC BY-SA 2.0 by Andrew Aitken and Martin Weale ‘A Century of High Frequency UK licensed Macroeconomic Statistics: A Data P93 Angel of the North, photo by Ron Adams, CC ‘Reconciled Trade Flow Estimates Using Inventory’ (ESCoE TR 03) by Jagjit S. BY-NC-SA 2.0 licensed an FGLS Estimator’ (ESCoE DP 2018-01) by Chadha, Ana Rincon-Aznar, Sylaja Thomas Baranga Srinivasan and Ryland Thomas P97 Oil rig, photo by Joe deSousa, public domain
‘A Comparison of Approaches to Deflating ‘Constructing Estimates for Exports, P101 Birmingham, photo by Bs0u10e0, CC BY-SA Telecoms Services Output’ (ESCoE DP Imports and the Value-Added from 2.0 licensed 2017-04) by Mo Abdirahman, Diane Coyle, Exports of the Car Industry and Other Richard Heys and Will Stewart Manufacturing Industries in the UK’ (ESCoE TR 02) by Giordano Mion ‘Measuring the “Free” Digital Economy Within the GDP and Productivity Accounts’ ‘Constructing Estimates for Exports and by (ESCoE DP 2017-03) Leonard Nakamura, the Value-Added from Exports of Monetary Jon Samuels and Rachel Soloveichik Financial Institutions in the UK’ (ESCoE TR 01) by Jack Pilkington and Jeremy Rowe
121 122 Economic Statistics Centre of Excellence National Institute of Economic and Social Research 2 Dean Trench Street, Smith Square London SW1P 3HE United Kingdom
+44 (0) 20 7654 1913 [email protected] escoe.ac.uk
@ESCoEORG
ESCOEOrg
ESCOEOrg
Economic Statistics Centre of Excellence (ESCoE)
© Economic Statistics Centre of Excellence (ESCoE) 2020