Measuring the Digital Economy

November 16, 2017

The digitalization of the economy, society and politics is the transformation that is marking the XXI century. From blockchain technologies to artificial intelligence, digitalization is making us experience the Gutenberg Revolution and electrification at the same time, and within a compressed time frame. Digitalization is the spreading of a general-purpose technology. It changes everything.

When thinking about national accounts today, what is at stake is no less than the task of “measuring the economy of the XXI century”—to borrow the title from an article by Charles Hulten.

That is my tentative answer to the first question motivating this session: what do we mean by “the digital economy”? The answer is - as the XXI century unfolds - the economy itself. That makes the topic of our session very clearly intertwined with the topic of the first session, which asks whether GDP still tell us what we need to know.

It is possible to draw inspiration from XVIII century Germany. I think that may please our chair – Martin Mulheisen. It is always wise to gain the indulgence of the chair right at the beginning. I will be quoting the Kammerpräsident - director of finances - or, with some liberty, minister of finance, of the duchy of Weimar. The state had a population of about 80000. The economy was in decline. It was not competitive against imports from the lower Rhine or England. The budget was financed on credit as revenues were insufficient. A tiny economic base had to finance an oversized administration and a lavish court. The minister was also an apt writer.

He wrote in memorable verse:

“Happy the man, who still can hope

To swim safely in this sea of error

What we need we don’t really know

And what we know fulfills no need at all”

(Johann Wolfgang von Goethe, Faust Part 1.)

The man was Goethe. The work is Faust. Are we facing a challenge of true Faustian proportions?

I want to start by thanking the organizers for inviting me as a discussant in this exceptional Statistical Forum. It was a learning experience to read from (and listen to) two of the thought leaders in the field: Diane Coyle and Carol Corrado.

More than 15 years ago, the Statistics Department of the IMF invited me to a seminar, which was opened by the Managing Director Horst Kohler, on Getting the Right Numbers and Getting the Numbers Right. That is still very much our concern today. Back then the focus was on the statistical implications from the adoption of inflation targeting in many countries. I am going to use the issue of price index measurement as my main viewpoint for what follows. But I am getting ahead of myself.

I want to organize my remarks as follows: first, I will consider some elements that both papers have in common. Second, I will consider some conceptual challenges in national accounts and, especially, those associated with making the transition from nominal to real GDP. And, finally, I will conclude.

The two presentations by Carol Corrado and Diane Coyle are nicely complementary.

Carol Corrado explores the implications that would follow from treating expenditures on consumer digital stocks as investments. This is the procedure that is already used in national accounts for owner-occupied housing. Carol Corrado, together with her co-author David Byrne, finds that implications for real GDP are quite significant, especially taking into account not only the accumulation of the digital stock but also its utilization rate.

Diane Coyle, follows in the footsteps of a recent Review of UK Economic Statistics (2016) and her own Brief (But Affectionate History) book on GDP. The current System of National Accounts was developed in the 1940s for an economy dominated by the production of goods (with an increased predominance of manufacturing). That leads naturally to the approach used to measure GDP: it is basically the value added created in the market economy. Most importantly, home production is excluded from GDP. However, the public sector is included but measured by inputs rather than outputs. In her book, Diane Coyle explains how that choice was made, against the views of Simon Kuznets. In her presentation today, Diane Coyle shows convincingly that the conventions pertaining to the definition of the boundary of production are quantitatively relevant.

From the two papers one sees that national accounts are the result of practical and pragmatic choices that must be made all the time. From a conceptual viewpoint, the issues that we face today are not new. What is new is the magnitude and the pace of on-going change and transformation: the world is experiencing digital revolutions. In this context, the underlying problems inherent in the measure of GDP have become more acute because of increased pace of change powered by digitalization. The conceptual issues are not new. Their scale is. As economists, you should not ask national accounts to bear conceptual weight that they have not been designed to support. The intention of national accounts is practical and pragmatic and it is not theoretical and conceptual.

This leads to the second part of my comment: conceptual challenges.

Diane Coyle also quotes one of my favorite authors: Thomas Schelling. He wrote in 1958 [1]:

“… only the money magnitudes are real. The “real” ones are hypothetical.”

The practical task is: how to compute the relevant price indices?

Diane Coyle mentions but does not develop the topic. In any case the landscape is well known. It can be covered by mentioned challenges associated with accounting for quality change and completely new goods. The challenge pertains to heterogeneity: how to compare magnitudes in two different periods when the set of goods and services differs?

The issues are intrinsically difficult. Martin Feldstein surveys the field for the Journal of Economic Perspectives [2]. His view on the issue is clear from the title: Underestimating the Real Growth of GDP, Personal Income, and Productivity.

Here I want to follow an alternative and more radical route. A good reference is William Nordhaus [3]. He suggests that the relevant price index should relate to the flow of services that households derive from their purchases of goods and services. I interpret his approach as suggesting that people should provide the invariant standard to compute real magnitudes. It seems to be the only viewpoint that – in the end - matters. In his essay, Nordhaus asks (page 54): “Is it possible that by the very nature of their construction, price indexes miss the most important technological revolutions in economic history?” And he answers (page 55): “By design, price indexes only capture the small, run-of-the-mill changes in economic activity, but revolutionary jumps in technology are simply ignored by the indexes.”

The problem also applies to public services’ provision. The IMF has just put out a book (Digital Revolutions in Public Finance, which contains papers from a conference organized with support from the Gates Foundation [4]). The papers in part III of the book focus on public services’ provision and poverty reduction in developing countries. But digital revolutions are also affecting education and health care in advanced economies. Clearly, the time is ripe to focus on outputs rather than on inputs. That can make a tremendous amount of difference. Just recall the famous comment by Rajiv Gandhi, the former Prime Minister of India, that only 15 per cent of every rupee spent on public food distribution ever reached the poor. [5]

My previous comments hide a more fundamental question that Schelling highlights: whose viewpoint are we taking when determining what is real? In my view the choice is clear: we must take the viewpoint of people, the viewpoint of households. If that is accepted, then Nordhaus’ approach to price measurement and emphasis on outputs when valuing public services are natural options to consider.

So, to conclude: are we facing a challenge of Faustian proportions?

I would say not. The problems that we are facing are not new. They are familiar. And, as the papers by Diane Coyle and Carol Corrado, and others in this Forum, make clear the conceptual framework of national accounts can be used to organize the debate. The system of national accounts is – and as always been – a practical and pragmatic endeavor. The pace and the depth of change add an element of urgency to the task. Data produced by digitalization itself will no doubt be a great assistance.

[1] Thomas C. Schelling, Design of the Accounts, chapter in A Critique of the United States Income and Product Accounts, Princeton: Princeton University Press, 1958. Out of print volume from the National Bureau of Economic Research. The quote is from page 332. Available at

[2] Martin Feldstein, Underestimating the Real Growth of GDP, Personal Income and Productivity, Journal of Economic Perspectives, Volume 31, Number 2, Spring 2017, pages 145-164. Available at

[3] William Nordhaus, 1996, Do Real-Output and Real Wage Measures Capture Reality? The History of Lighting Suggests Not, chapter in Timothy F. Bresnahan and Robert Gordon (eds.), The Economics of New Goods, Chicago: University of Chicago Press, pages: 27-70. Available at

[4] Sanjeev Gupta, Michael Keen, Alpa Shah and Geneviève Verdier (eds.) Digital Revolutions in Public Finance, International Monetary Fund, 2017. Available at

[5] Quoted in Ravi Kanbur, The Digital Revolution and Targeting Public Expenditure for Poverty Reduction, chapter 9, in Digital Revolutions, op. cit., pages 225-237. Available at

IMF Communications Department

Phone: +1 202 623-7100Email: