Paul Wallace 27 Jul 2017 12:00pm

Is GDP still a valid metric?

As digital business and truly global companies affect the way the Office for National Statistics manages the numbers, Paul Wallace asks if GDP is still relevant

Caption: The task of measuring GDP is getting tougher and tougher.

Gross Domestic Product and the national accounts are central preoccupations of the Office for National Statistics (ONS). GDP is the total value added in an economy over a given period. Usually reported in real terms, it is the figure that captures all the attention. But the full national accounts are also vital, allowing analysts to trace output, spending and income by sector while also showing the financing involved. For both economic policymakers and commercial users it is essential that these figures are as accurate and as timely as possible. That challenge is becoming more daunting as multinationals deploy global supply chains and business goes digital. How is the ONS coping?

Summarising the condition of the complex British economy every quarter and year has never been easy. There is a trade-off between accuracy, requiring as much data as possible, and timeliness, involving the use of available information. But concerns about the quality of macroeconomic figures are longstanding. In 2004 Rachel Lomax, then a deputy governor of the Bank of England, spoke of “the statistical fog surrounding the true state of the economy”.

A persistent complaint has been about revisions to GDP. Some are inevitable, the result of methodological changes. For example, when Eurostat adopted a new international standard for EU member states in 2014, the inclusion of R&D as investment pushed up GDP compared with previously published figures.

Others arise as more information becomes available. Here the ONS’s record is more open to challenge. A study of two decades of quarterly figures by the OECD in 2015 found that Britain’s short-term revisions were not out of line with that of other G7 countries. Even so, the timing and shape of the deep 2008 to 2009 recession now differ markedly from initial estimates and the “double-dip” recession in late 2011 and the first half of 2012 has been revised away.

Indeed, British economic history is often rewritten to a disconcerting extent. A case in point is the recession of the early 1990s, which appeared grave at the time since GDP slumped by 4.3% from peak to trough. But today’s figures for this episode show a much milder decline of only 2%. The scale of such revisions matters for policymakers and analysts as they seek to put more recent reverses such as the 2008 to 2009 recession into historical context.

Revisions are one thing, mistakes are another. Worryingly, mismeasurements have occurred in the figures for trade and construction, both of which feed into the national accounts. In 2014 they lost their status as high-quality statistics, awarded by the UK Statistical Authority, which monitors and assesses all official numbers. The need to improve trade data is now all the more pressing following Britain’s decision to leave the European Union.

Tougher times

Thickening the fog, the task of measuring GDP gets tougher and tougher. Services, which are inherently harder to measure than goods, are becoming ever more important and now account for 80% of overall output compared with less than 50% in the 1950s. Underlying methodologies respond sluggishly to new ways of doing business. The national accounts have, for example, been slow to recognise the intangible investments in knowhow that are so crucial in a services-based economy. Although R&D now counts towards GDP, other forms of intangible capital such as “organisational capital” remain excluded, even though the application of managerial best practices can play a critical role in determining productivity.

National accounting, introduced in the 1930s and 1940s for largely closed economies, struggles to work out the contributions to national output from multinationals using global supply chains and exploiting international opportunities to minimise the taxes they pay. The difficulty was highlighted when in July 2016 Irish national statisticians revised GDP growth in 2015 from
an already rapid 7.8% to an extraordinary 26.3%. The supposed burst of speed had nothing to do with the Irish domestic economy, which expanded by 4.4%. Instead it reflected the decisions of a few multinationals to assign some of their assets including intellectual property to Ireland, causing output in the sectors dominated by such firms to rise by 101%.

The upsurge of the digital economy poses a further profound challenge. People devote far more time to online and mobile activities, yet the consumption of such free services (where in effect they are paying by revealing information about themselves) is not counted because GDP excludes zero-priced products.

Individuals are increasingly making their travel arrangements online rather than through an agent, yet their efforts do not count towards GDP any more than other unpaid work at home. And the national demarcation of output is more problematic as frontiers are erased in a digital world.

On top of these difficulties, the ONS faces increasing demands to provide estimates at regional and local level. Interpreting and producing these figures is intrinsically tricky because bigger businesses operate nationally while people commute across administrative boundaries.

Worries about the ONS’s performance at so challenging a time spurred a recent independent review by Sir Charles Bean, a deputy governor of the Bank of England between 2008 and 2014. His report last year spelled out a variety of shortcomings as well as suggested reforms.

The ONS has set out an ambitious strategy to implement his recommendations.

One overdue change is to exploit administrative data, as is the case in Canada and Scandinavian countries. Now, the ONS relies overwhelmingly on surveys of businesses and households, sending out over 1.5 million paper forms per year. As Sir Charles argued in his report, “given the ubiquity of electronic data today it is incongruous that the production of ONS economic statistics still relies so heavily on the posting of paper forms and knocking on doors”. The number of business surveys will be slashed from 80 to 10, while much more data will be collected online. The shift away from surveys will be facilitated by new legal authority to access administrative data under the recent Digital Economy Act.

Administrative figures should help to make short-term measures of GDP more seaworthy, addressing a key concern for economic policymakers. The immediate plan is to utilise VAT returns for turnover by the end of this year, combining the figures with the results of the regular monthly business survey. This will be particularly valuable in improving coverage of smaller businesses. The new administrative data together with more rigorous appraisal of early GDP estimates should allow national statisticians to do a better job in spotting turning points in the business cycle.

More resources and effort are also being devoted to measuring services. The ONS already produces a monthly index of services activity.

But its coverage of what is now most of the economy remains much less detailed than that of the old industrial economy. A new survey called Servcom (covering all services apart from those from government and banking) will provide a more fine-grained approach covering 274 service sectors with a sample of 40,000, whereas the existing survey covers only 93 with a sample of 20,000. The results should be used fully by 2019 or 2020.

Separate indices

Another vital reform is to introduce “double deflation”, using separate price indices for inputs as well as outputs when estimating value added in real terms. At present the same price index for output is generally used to deflate both gross output and the inputs that must be stripped out to derive value added. This “single deflation” is conceptually inferior to double deflation and can distort GDP estimates. If, for example, input prices are rising faster than output prices then the growth in real value added will be higher with double deflation than with single deflation. The new method is due to be implemented by 2020.

The ONS is also seeking to improve regional estimates of gross value added, which are currently based on income, though experimental production-based statistics are also available. The aim is to produce more robust figures combining production as well as income by the end of 2017 and to provide quarterly figures (already produced in various guises by the devolved administrations in Scotland, Wales and Northern Ireland) for English regions by the end of 2018. These will be supplemented by regional estimates of household spending, due to be published in the summer of 2018.

In a further development, the ONS is enhancing the financial accounts to show the flow of funds. The financial crisis of 2008 exposed the vulnerability arising from sheer interconnectedness while also highlighting the lack of information about this web of connections, many of them affecting the UK because of London’s role as a global financial centre. The reform will provide “who-to-whom” data that set out the counterparties to borrowers, identifying the sectors that are the sources of their financing and the instruments they are using. Much more granular data will also be provided, for example by splitting the catch-all category of “other financial institutions” into more meaningful components such as hedge funds and clearing houses. Experimental statistics are already casting more light on the flow of funds and the aim is to incorporate full data in the national accounts published in 2021.

More generally, the ONS is sharpening up its act and trying to turn the data revolution to its advantage. It may not be fully recording intangible capital in the economy at large but it intends to improve its own knowhow.

A data-science campus was launched this spring at its headquarters in Newport, south Wales, in order to foster new ways of exploiting the plethora of digital information now available through techniques such as “web-scraping”.

It will enable the ONS to explore new sources of data such as supermarket scanning records that should become available through partnerships with the private sector under its new powers in the Digital Economy Act.

The ONS is also tapping external expertise by financing a research centre, run by the National Institute of Economic and Social Research with five partners including the University of Cambridge and Strathclyde Business School. The researchers are exploring knotty issues such as the measurement of digital activities.

More engagement

In another sign that the ONS is changing its ways, it hosted a major conference earlier this year about economic statistics in the digital age. A traditionally rather introverted organisation is striving to engage more with the outside world.

Richard Spencer, head of sustainability at ICAEW, welcomes “the greater openness of the ONS and its increased engagement with outside stakeholders”. That dialogue should prompt, he says, “a wider debate about what constitutes economic success in the 21st century for which there is at present no compelling national vision”.

Some worry that the ONS is now trying to do too much in a rush. One criticism at the conference in February was that it should concentrate upon putting its old house in order before building a new one. There may be a grain of truth in this, but the ONS now has a sense of purpose that was lacking before.