~~Eurostat - ISTAT Seminar~~

~~Bologna, February 7, 1996~~

Introduction

This paper addresses two questions: What are the major forces shaping the modern economy ? What are the implications of those forces for statistics and statisticians ?

In short, my answers to those questions are as follows. The defining trend of the modern economy is the shift to the intangible. The economic landscape is no longer moulded by physical flows of material goods and products but by intangible streams of data, images and symbols. The source of economic value and wealth is no longer the production of material goods but the creation and manipulation of dematerialised content. We live in the intangible economy.

Non-linear and non-deterministic, the intangible economy raises a whole series of measurement issues. More fundamentally, it changes the role, the function, and the perception of economic measurement data. Because information is its key resource and output, the intangible economy is highly data-sensitive and intrinsically self-reflective. Official agencies no longer have the monopoly of economic data: a lively and diversified measurement and monitoring industry has emerged. Statisticians need to undertake a comprehensive appraisal of their business. Adapting to the new environment will require major changes in three key areas: conceptual foundations, modus operandi and temporal outlook.

Before developing these points in greater detail, let me make some remarks about the scope of the paper and our approach:

- Our perspective is economic. We focus on production, consumption and transaction processes and structures. We shall not discuss here political, social or cultural implications of the emergence of the intangible economy

- Our objective is not a formally rigorous demonstration backed by incontrovertible data. Rather, it is to outline a coherent conceptual framework, which sheds a new light on key economic trends and issues. We seek to stimulate constructive discussion of fundamental assumptions that underlie and guide the work of statisticians.

- This paper relies on a wide assortment of references and data, some of which could be considered somewhat unorthodox. It is however an incontrovertible fact that some of the best information about, and the most penetrating analysis of, the key economic trends come from such unconventional sources as management consultants, journalists, financial analysts or advertising agencies.

- This is not a futuristic “crystal ball” exercise. We shall not venture into scenarios about the world in 2010 or 2050. Forces that will shape the economy of the 21st Century are at work today and therefore we shall focus on the present.

EMergence of the intangible economy

Fundamental change: Incontrovertible yet misunderstood

We should begin by stating the obvious: the fundamental change in the economy has become conventional wisdom. Books, articles, speeches by economists, sociologists, management consultants and politicians about revolutionary change, new paradigms, earthshaking upheavals, megatrends, technotrends... are so numerous that they can no longer be counted let alone read and analysed in depth.

Three key trends

From this great variety of descriptions and analysis, we can highlight three key trends:

· The changing profile of employment and output structure. While the numbers may vary from country to country, the overall trend is definite: the share of industry and agriculture, whether in total output or in employment, has been falling steadily and now represents for instance less than forty percent of the total in the EU countries. Services represent the lion’s share of both employment and output and constitute the principal, for many countries the only, source of employment growth.

· Globalization. Foreign trade has been growing more rapidly than the world’s output for decades. The difference between trade and output growth rates decreased somewhat between 1971 and 1985 (3.7% vs. 3.2%) but since the mid-1980s it has increased again (5.9% for the former and 2.5% for the latter, between 1986 and 1996). The World Bank estimates that world trade will continue to grow at roughly twice the rate of world output [World Bank, 1995].

The international trade of final goods is accompanied by a massive crossborder deployment of production facilities, distribution networks, technologies and people. Global deployment of supply resources has been growing at an even higher rate. According to the World Bank, during the 1980s, total inward foreign direct investment (FDI) stock rose by 14 percent a year in high-income countries and by 11 percent in developing countries.

A key feature of globalization is the increased mobility of people, driven by increased demand for international labour and leisure. The mobility has created a huge travel and tourism (T&T) business. According to an association of leading T&T firms, the World Travel and Tourism Council (WTTC), it is the world’s largest industry. WTTC defines T&T as comprising five groups of activities: transportation, accommodation, catering/retail, recreation/culture and travel related services. With the assistance of WEFA, an economic consultancy, WTTC has estimated that in 1991 T&T generated over 10% of the world’s GDP (3 400 billion dollars), employed over 10% of the global labour force (204 million people) and accounted for 11% of gross investment [WTTC, 1994]. [1]

· The ubiquity of Information Technology. Information Technology (IT), which includes computers, telecommunications, and associated products and services (such as semiconductors or software), is recognised as a structural vector that influences all economic activities. According to Tom Stewart from Fortune Magazine, since 1991 we have entered the Information Age. That year, US capital investment in Information Technology exceeded investment in traditional machinery and equipment [Stewart, 1994a]. [2]

The impact of IT cannot be overestimated or overemphasised. The speed and the magnitude of technical progress are staggering and combine tremendous increases in quality with a continuing decline in prices. According to Gordon Moore, founder of Intel, the dominant supplier of microprocessors, the processing power of a microprocessor doubles every 18 months.[3] Thus, in 1980 a microprocessor chip had 30 000 transistors, in 1995, 3.3 millions. At the same time, the cost of hardware is falling at a rate of about 20% a year. A Pentium-based PC offers today several thousand times as much processing power as the mainframe of the 1970s, at a price that is less than 1% of 1% of the latter.

The development of IT has engendered a huge and rapidly growing new economic domain, whose global size is estimated at between 600 and 1100 billion dollars [EIT0, 1995, Banks, 1994]. [4] Furthermore, it profoundly affects all other economic sectors, no matter how traditional and well-established. IT is markedly different from other vector technologies, such as transportation and energy, which provided enabling infrastructure for the industrial revolution. Pervasive and ubiquitous, Information Technology cannot be reduced to infrastructure: it is simultaneously raw material, processing infrastructure, intermediate good and final product. It penetrates workspace, home and all other areas of social interaction. While transport and energy modified our physical capacity and our mechanical skills, IT touches these and more: it directly affects our mental and sensory capabilities [Solomon, 1989].

Inconclusive controversies

And yet, while there is broad agreement on the existence of these three trends, there is no real consensus on their magnitude, their underlying drivers, their structural causes and, more importantly, on their meaning. Actually, each trend has triggered fierce but rather inconclusive controversies.

For mainstream economists and statisticians, and for many politicians, the shift to services is a puzzling occurrence that is inconsistent with their basic view of the economy. This view, which has historical roots going all the way back to Adam Smith, postulates that the central activity of an economy and the source of wealth is the production of physical goods [Giarini, 1984]. How can technological progress and economic growth result in a shift to a relatively less attractive output and employment mix, weighted toward subsidiary and less productive activities ? There is a vocal and articulate school of thought which asserts that “manufacturing matters, “ that industry remains the true vector of the economy. To buttress their assertion, its proponents marshal data on the lower productivity of services and lower quality of service jobs [Cohen and Zysman, 1992; Taddei and Corniat, 1993]. Yet, other studies and data, based in particular on the recent employment trends in the US, suggest that services can generate high productivity and create high quality jobs [McKinsey, 1992, 1994].

The underlying problem is the tremendous heterogeneity of services. They comprise low paying, low productivity, labour-intensive and local activities such as restaurants as well as highly paid, high productivity, capital-intensive and global activities such as financial trading. Some services are immobile and non-tradable, others are extremely mobile and highly tradable. Certain services are subject to diseconomies of scale (household services), while others are the prime beneficiaries of economies of scale (telecommunications for instance). This heterogeneity makes it difficult to agree on a meaningful definition: a service can be an output, a production factor or an activity. According to an UK economist, Phedon Nicolaides [Nicolaides, 1989]:

" More intensely services are studied, the less certainty there is about how they can be defined and classified."

Service confusion is also a key element of controversies about globalization and Information Technology.

The conventional view of services contends that they are less tradable than physical goods. For instance, the French Statistical Institute, INSEE [INSEE, 1993], characterises services as

" generally not storable [and which] are produced in direct contact with the consumer."

This view is apparently corroborated by international trade statistics, based on IMF balance of payments data, showing that services represent some 20% of the world trade. For OECD, this a relatively stable share, even if market services (excluding government services) grow faster than merchandise trade [OECD, 1993].

This picture is highly misleading. There is compelling evidence that trade in services not only grows much more rapidly than goods but also constitutes the bloodline of globalization. Let us take just one example: international telecommunications. According to the International Telecommunications Union, the international voice traffic has been growing at some 16% a year and in 1992 represented 41 billion minutes, some 70 minutes per subscriber [ITU, 1994]. And the voice ~~communication~~ is only one aspect of international telecommunications. Based on anecdotal information, the data traffic, not only from the Internet but also from corporate data networks, is growing much more rapidly than voice. And let not forget the flood of global images from TV satellite broadcasting.

Finance is another quintessential service activity, yet it is excluded from international services trade data. According to British Invisibles, London-based trade group which promotes international activities of UK based financial institutions and business services, investment income flows approximately equal service trade flows. Taken together, in 1994 services and income flows represented over 35% of the total trade (vs. 32% in 1984) [British Invisibles, 1995].

And this is still an incomplete picture of the relative magnitude of global flows, as it does not take into account financial transactions. Those utterly dominate physical flows: the value of foreign currency trading alone was 1.3 trillion dollars in 1995 a day, more than sixty times greater than the daily physical trade volume of approximately 20 billion dollars. Furthermore, financial transactions experience explosive growth: between 1992 and 1995, the foreign currency trading market has grown by 47%, after growing by 42% between 1989 and 1992 [BIS, 1995]. Yet despite their size, visibility and persistence, global financial flows remain at the periphery of conventional economic measurement framework. Many economists and business decision makers tend to dismiss them as a speculative epiphenomenon, pointing out that only 5% of transactions are connected to “real” commercial operations undertaken by customers [Guth, 1993]. Others seek to introduce a distinction between the “good” investment flows and the “bad” portfolio flows, a rather futile effort, in light of an inextricable intertwining of the two categories. One of the most intensively watched and analysed economic activities, the international finance remains a block box.

Information Technology, on the other hand, appears as a black hole. The IT controversy revolves around the question: where has all the money gone ? This is what Robert Solow called a “computer paradox”: the price of computer equipment has been falling dramatically, businesses have invested massive amounts into computing equipment and yet, the results are not there [Griliches, 1992]. Actually, the period of massive IT investment in the 1980s has coincided with a significant productivity slowdown, which was particularly apparent in services [Baily et al., 1993]. In a 1991 study, based on Department of Commerce data, Stephen Roach, Chief Economist of Morgan Stanley, has calculated that the productivity in services during the 1980s has been growing at an annual rate of 0.5% per year, compared to 2.3% in manufacturing. Yet, during the same period the service sector invested massively in Information Technology, to the tune some 800 billion dollars [Roach, 1991].

The computer paradox prompted a huge variety of studies, many of which were based on detailed and custom-generated data at sectoral and firm levels. Yet, despite the wealth of evidence, opinions remain as polarised at ever. One group of analysts affirms that the computer paradox is simply a by-product of inadequate data and that detailed studies show a significant technology pay-off. Columbia Business School Professor Frank Lichtenberg [Lichtenberg, 1994] studied IT use in several hundred manufacturing and service industries and concluded that IT investments were highly effective. While they accounted for 10% of labour~~labor~~ costs and 15% of capital expenditures, they contributed over 20% of the output of the companies studied. Two MIT researchers, Erik Brynjolfsson and Lorin Hitt, have analysed the impact of IT investment in 400 large firms in the United States between 1987 and 1991. They calculated that the average return on investment was close to 70% [Brynjolfsson and Hitt, 1993]. A survey of French firms, carried by a French computer magazine 01 Informatique published in June 1993, also concluded that the IT impact was highly positive [01 Informatique, 1993]. After reviewing new data, Roach changed his views radically and in 1992 asserted that IT investment in services had significantly improved productivity [Roach, 1992; Gleckman, 1993].

On the other hand, proponents of the computer paradox persevere. In a heavily documented book published in 1995, Thomas Landauer argues that the paradox not only exists but will persist in the absence of radical changes in the way computers are designed and used [Landauer, 1995].

Measurement gap

A common thread runs across the three controversies: the weakness of data and the inadequacy of measurement methodologies. The problem is severe. Zvi Griliches of Harvard University, argues that the share of the economy which is measured by official statistics with a degree of accuracy is declining [Griliches, 1994]. Between 1947 and 1990, the fraction of the economy for which productivity data can be deemed reasonably accurate fell from close to 50% to about 30%. As a result:

“ Our ability to interpret changes in aggregate total factor productivity has declined, and major portions of actual technical change have eluded our measurement framework entirely. “

Furthermore, data weaknesses have been most pronounced in the areas and sectors which are most dynamic and most indicative of new trends such as Information Technology, new services or finance. This measurement gap contributes to the continuing inability of the economic profession to provide answers to such essential questions as:

- What are the determinants of long-term growth ? What is the relationship between growth and inflation ?

- What is the impact of the diffusion of Information Technology on productivity and employment ?

- What is the role of financial markets in the economy ?

Charges of data weaknesses and methodological inadequacy are not limited to official statistics. They are also levelled at financial accounting data and the underlying conceptual framework. The title of an influential book published in 1987 by two Harvard Business School professors, Thomas Johnson and Robert Kaplan, is explicit in this regard: “Relevance Lost, The Rise and Fall of Management Accounting.” [Johnson and Kaplan, 1987]. The main thesis of the book is that the traditional accrual-based systems are at best obsolete and often harmful. For Robert Kaplan, traditional measures for corporate performance are so misleading that they condemn managers to fly blind. In fast-moving sectors, such as computer software or entertainment, accounting statements have simply ceased to be a relevant indicator of business value and performance.

The existence of the measurement gap is not a recent discovery. Problems of official statistics are well known and have been extensively studied. Statisticians have constantly sought to improve the national accounts framework [SNA, 1993].

Nevertheless, the prevailing feeling is that the progress has been painfully slow and uneven. In his 1994 Presidential Address to the American Economic Association, Griliches [Griliches, 1994] asks an obvious question:

“ Why we don’t know more after years of research done by so many good people [?] What is it about our data and data acquisition structure, and possibly our intellectual framework, that prevents us from making more progress on this topic ?

For Griliches, the main problem is the quality of data, which stems from an inadequate attention to their generation methods and to the quality of the sources. But one has to wonder whether this is a sufficient explanation.

It can be argued that the measurement gap is provoked by a tension between the stable statistical apparatus and the dynamic economy. Statistics are meant to facilitate comparisons across space and time and therefore have to emphasise continuity: long time series, stable indices, rigid classifications, taxonomic approach... On the other hand, a major characteristic of the new economy is the permanent shifting of borders and boundaries. Consumption baskets, production functions and sectoral classifications are changing all the time, thus rendering major aggregates obsolete or even meaningless.

In our view, beyond data accuracy and tension arguments, the core measurement issue is the relevance of underlying conceptual models and assumptions. In their 1995 book on Statistics for 21^st century, two well-known economists-statisticians, Joseph Duncan and Andrew C. Gross [Duncan and Gross, 1995], declare:

“ As the 21^st Century approaches, it is increasingly clear that our current conceptual net, designed for earlier realities, does not cover current realities very well.”

The crucial assumptions of national accounts - focus on physical goods production and trade, choice of a nation as the main reference, stable sectoral groupings and classifications, neglect of non-material transborder flows - are grounded in a specific vision of the economy, which has not fundamentally changed since Adam Smith’s time and which postulates the production of physical goods as the main source of value.

These assumptions and the underlying vision can no longer be considered universally valid. While they may be seriously flawed, other conceptual approaches have not gained enough internal consistency and external recognition to provide a credible basis for the development of an alternative measurement framework. Two best known approaches are the service economy and the information economy. They share the same problem. Both terms are well known, to the point of becoming buzzwords. Yet, little progress has been made in making them conceptually more robust and operationally more relevant. They remain largely on the periphery of mainstream economics and statistics.

The service economy

The contrast is particularly striking for the service economy. On the one hand, after years of neglect, services have begun to gain political weight. They have now been officially recognised by statisticians and economists. International organisations such as the IMF, OECD and the World Bank have undertaken several projects to improve measurement of services, particularly in international trade [IMF, 1993, Arkell, 1995]. More importantly, services were a key item on the agenda of the Uruguay round, resulting in the adoption of General Agreement on Trade in Services (GATS), and constitute a priority for the World Trade Organization (WTO).

And yet, on the conceptual level, the service economy has made little progress. This lack of progress can be attributed to two crippling weaknesses:

- The first one is the original sin of a residual approach.

- The second, more~~second,more~~ recent and potentially more worrisome, is the loss of operational relevance.

Work on the service economy started in the early forties, with the seminal work by a prominent British economist, Collin Clark [Clark, 1940]. It was Clark who defined services as a residual category comprising all activities that cannot be classified as either manufacturing or agriculture. This definition entailed a structural heterogeneity of services, which became more pronounced as their range expanded. Such heterogeneity largely explains the difficulties of definition. In turn, these difficulties make it difficult to define boundaries and meaningful classifications. Existing service classifications lack economic intelligibility. In its 1991 survey of service business, INSEE acknowledges that it covers “heterogeneous activities, whose only common point is that they are neither industrial nor financial (banks and insurance).” [INSEE, 1993]. New services classifications, such as the one proposed by the IMF in its 5th edition of the Balance of Payment Manual, suffer from similar flaws [IMF, 1993]. For instance, telecommunications and computer services data appear in different categories, making it difficult to capture the development of telecommunications data networks, such as Internet, one of the most significant technological and economic trends of this turn of the century.

More ominously, product/service output or manufacturing/service activity distinctions appear increasingly irrelevant in the real economy. Services have been industrialised and service has become a key product component [Levitt, 1976, 1981]. Large industrial firms such as General Electric or General Motors generate more than a third of their revenues in financial services. Consumer electronics firms, Philips and Sony for instance, control leading audio-visual entertainment content producers. Acknowledging this interpenetration, Fortune magazine decided in 1995 to integrate its previously separate rankings of Fortune 500 Industrial and Fortune 500 services. It justified this decision by the obliteration of “once valid distinctions between industrial and service businesses and between services”[Stewart, 1995a].

The information economy

The lack of progress on the information economy is puzzling. Nobody would argue that the notion of information is becoming less relevant. Information can be quantified, stored and manipulated. Furthermore, a pioneering work has been carried out by Mark Porat in the mid 1970s seeking to integrate information activities in a statistical framework [Porat, 1977]. This work was published by the Department of Commerce and is widely known. Yet it has spawned little operational follow-up in the United States or elsewhere and thus remains a one-off effort.

One explanation is timing: Porat’s work came too early and therefore did not attract a constituency capable of pushing the work forward. This was aggravated by Porat’s outsider status (he was a Ph.D. student and later went on to other things, such as joining Apple and founding General Magic, an innovative software company).

There is however a more fundamental problem with the information economy. It is the schizophrenic way in which economists define information and approach the fundamental relationship between information and uncertainty. For classical economic theory, the role of information is to reduce uncertainty: more information means less uncertainty. Given its focus on physical goods as store of value, the theory does not address the issue of information as a tradable good. Yet, this is a primary concern for financial economists such as Charles Goodhart [Goodhart, 1975], who seek to understand the functioning of financial markets. From their perspective, the relationship needs to be inverted: it is uncertainty that determines the value of information. The higher the uncertainty, the higher the value of information. This view is consistent with the Communication Theory of Claude Shannon [Shannon and Weaver, 1947], which provides the theoretical underpinning for the development of modern telecommunications and informatics. This theory postulates that uncertainty creates information.

More generally, economists have difficulties coming to grips with the polymorphic and ubiquitous nature of information, simultaneously a good, a production asset and a market attribute.

The continuing spread of Information Technology, which raises a host of new economic issues, has given the information economy a

ABSTRACT