The challenges for GDP, now and in the future

When Professor Sir Charlie Bean published his review of UK economic statistics in March, he warned that current systems and process were starting to show their age. “We need to take economic statistics back to the future or we risk missing out an important part of the modern economy from official figures,” he said.

The problems with current measures are not easily summarised – the Bean review runs to 259 pages. However, as the most widely followed indicator of economic performance, Gross Domestic Product (GDP) provides a useful lens through which to consider the challenges we face in accounting for the wide range of activity that now takes place within most advanced economies.

Many wonder whether GDP is still fit for purpose. After all, says the economist Diane Coyle, it is a measure of the economy best suited to an earlier, production-driven era. In the following extract from her book, GDP: A Brief but Affectionate History, Coyle outlines three issues that explain why we might need to move to a different measure of the economy in time.

- Brian Tarran,
Editor, Significance


1. Complexity

The United States in 1998 offered 185 TV channels, 141 over-the-counter painkillers, and 87 brands of soft drink. These figures all represented big increases since 1970, when there had been five TV channels, five painkillers, and 20 types of soft drink. By 1998, there were 340 kinds of breakfast cereal, up from 160, and 50 brands of bottled water on offer, compared with 16 in 1970.

Variety could be considered one of the key indicators of economic development. To be poor is to have little choice available, and the increase in possibility is the most important aspect of escaping from poverty. On this view, economic development is a combination of increasing individual capacities or skills to be able to take advantage of opportunities, and increasing the range of opportunities and choices available. Economic development is an increase in freedom.

It is surprisingly hard, though, to find statistics on the number of different product types available. The Federal Reserve Bank of Dallas 1998 Annual Report, from which the figures here are taken, is one of the few estimates available even now. The main reason it is so difficult is simply that the statistics are not collected by official agencies. The surveys sent to businesses ask about volumes of output – number of pairs of shoes made by a shoe factory – and prices charged, but not number of styles. So official statistics are published as aggregate categories: “shoes.” The fact that I can choose high-tech walking boots, or running shoes that will cushion my knees and ankles, or vegan shoes, or shoes shaped to exercise my thighs as I walk, or gorgeous red high heel shoes, or ugly but ultra-comfortable sandals, or sneakers I designed myself on the vendor’s website – none of that features in the statistics.

Why does this affect GDP, though? Think about a place setting for a meal. My contribution to GDP is the same whether I manufacture a knife, fork, and spoon, or three spoons. GDP just counts the number of items.

GDP under-records growth by failing to capture fully the increase in the range of products in the economy. It is a poor way to measure innovation and customization, and the extent to which it undermeasures them is extremely large. It also fails to record at all another increasingly important category, namely, preventive goods or services. Take driverless cars for example. One of these will increase GDP by the same amount as any other kind of car, or perhaps by more if statisticians calculate a hedonic price index to adjust for its improved quality – after all, the human can sit back and relax in a driverless car. But GDP will not capture at all the benefit of a reduction in the number of accidents as driverless cars spread, assuming they live up to the high hopes that this will be the result.

Separate statistical headaches arise from the increasing complexity of the economy, due to the fact that most goods are now “made” in global supply chains. The components will be manufactured in a number of countries, shipped around the world to be assembled in one place, and shipped back out to their destination markets. This is true of goods as apparently simple as a shirt or as sophisticated as an iPhone. China, of course, has been the main country of assembly in these global chains, but other Asian countries and their Latin American competitors Brazil and Mexico have been gaining ground.

Price indexes do not, however, capture the large price declines when outsourcing occurs, so import prices have been greatly overstated and import volumes under-recorded. In addition, trade statistics do not net out the intermediate stages: the whole value of the iPhone imported from China to the United States counts toward the US current account of the balance of payments. Value-added trade statistics are now becoming available, and their study is likely to change the big picture we hold in our minds about the shape of the world economy.

2. Productivity

If economists were to play a game of word association, the one that would leap to mind on hearing productivity would be puzzle. But why is productivity puzzling?

It is because of the second increasingly serious issue for GDP as a measure of the economy, namely, that the economy consists less and less of material items. It is relatively straight-forward to measure economic output when you can count the number of cars or refrigerators or nails or microwave meals being shipped from factories. But how do you measure the output of nurses, accountants, garden designers, musicians, software developers, health care assistants, and so on? The only way is to count how many of them there are and how many “customers” they provide with a service, but this entirely overlooks the quality of the service, which is of great importance.

Just as “output” is a concept best suited to an economy made up of products rather than services, and similar, mass-produced products at that, so is “productivity.” The word is used in general terms to mean efficiency or effectiveness. The actual definition used by economist is the amount of output produced per unit of inputs. Inputs are labor, capital, land, and material resources. Usually economists are talking about labor productivity because it is easy to measure the number of workers, and much harder to measure capital. So on this definition, productivity is the amount produced per worker, or GDP per person employed (or per person-hour worked, to be more precise).

This is fine for washing machines or cartons of breakfast cereal. But only a small part of GDP in countries such as the United States and the EU nations consists of physical products.

A related issue is how to account for the value of a specific type of intangible product or service, the purely digital items such as online music, search engines, apps, crowd-sourced encyclopedias or software, and so on. Many traditional activities have ‘de-materialized,’ for instance subscriptions to music or video streaming services rather than buying CDs or DVDs. Sometimes this simply reflects a change of business model and the new services and prices can (in theory) be measured for GDP. Online markets such as Ebay and Marketplace have enabled secondhand sales to grow; again, measurable in theory but much harder to capture in practice. There is also the so-called ‘sharing economy,’ where there is a matching service provided (for example by Uber or Airbnb). These involve prices, to the home- or car owner as well as the software platform, but again these might be harder to measure in practice than to identify in theory.

Often, though, digital goods have a price of zero, and with no market price they are not captured fully in GDP statistics. The electricity Google uses will be counted in GDP, and so will the electricity we use to go online, as will ad revenues, and Google salaries. But how would the value of free search figure in the statistics? The gap between what a consumer pays and the value he or she receives from the purchase is called “consumer surplus,” and the growing prevalence of zero-priced goods and services online seems to be increasing consumer surplus. It is another reason to think the wedge between what GDP measures and aggregate economic welfare is growing uncomfortably large. Even worse, the GDP statistics distort the true picture of the economy. For example, the US Bureau of Economic Analysis estimated that consumption of internet access by Americans declined in real terms from the second quarter of 2011 on. This is absurd. Erik Brynjolfsson of MIT has pointed out that the information sector (software, TV and radio, movies, telecommunications, data processing, publishing) accounts for the same share in official GDP figures today as it did 25 years ago, at about 4 percent. He and his co-author JooHee Oh estimate that there has in fact been a gain to consumers averaging about $300 billion a year for a decade from access to free services online, such as Facebook, Wikipedia, Craigslist, and Google.

Official statisticians need to start thinking about how to measure better the production and consumption of “information” or digital products that clearly deliver value to consumers. Because GDP measures only monetary transactions, the new “free” business models are not being well measured, and neither are the new types of activity with zero market price but of great value to consumers. There have always been free but valuable activities, from public libraries to walks in the countryside; the difference now is that non-monetary activities are extensively intertwined with business, making the concept of the production boundary within which GDP is defined inherently blurred.

3. Sustainability

The third emerging issue for the relevance of GDP – no less tricky than the first two – is that it takes account of the increase in output of goods and services over time without fully accounting for whether or not growth now comes at the expense of growth in the future. GDP statistics do include a measure of the depreciation of physical assets (“capital consumption”), but this is a narrow indicator of how far capital is being used up to consume today by reducing the scope for consumption tomorrow.

One aspect it omits is the need for the physical stock of capital (machines, transportation equipment, buildings) to grow by more than is needed just to make up for depreciation of what is there already. There needs to be additional investment just to keep pace with growth in the population, if consumption per person is to be maintained. This, after all, is what matters, rather than the total size of GDP. This is known in the economics jargon as “capital widening.” In addition, if innovation, technical progress, is taken into account, surely it is important to include some indicator of “required” additional investment in the new kinds of capital, to implement the innovation?

The latest international national accounting standard, SNA2008, has tried to address some of these concerns. The United States is the first country seriously to put into practice its suggested improvements, which include counting spending on research and development as investment rather than a business cost, and estimating as well the value of investment in “artistic originals” such as Hollywood movies and music. Preliminary changes in statistical methodology along these lines led to a one-time jump in US GDP of more than 2 percent in 2007, but a bigger increase of 3.4 percent was announced in mid-2013. The SNA2008 handbook explains that “many of these assets, often seen as a hallmark of the ‘new economy,’ are associated with the establishment of property rights over knowledge in one form or another".

These questions about the treatment of investment in assets are just one dimension of sustainability, however; there are others. More often, the term sustainability refers to the extent to which GDP growth from year to year depletes natural resources or harms the environment in other ways. The most important amendment needed to the existing national accounts statistics is to take account of the balance between investment in new assets and the depletion or depreciation of existing assets. Without this, we can know about the current rate of economic growth but have no information about whether it could be sustained in future.

It should be said that official statisticians have been paying increasing attention to environmental measures, ranging from CO2 emissions and water quality to the extraction of mineral resources. In 2012, the UN Statistical Commission adopted a new international statistical standard with equal status to the System of National Accounts, the System of Environmental Economic Accounting or SEEA. Some countries have been publishing what are known as “satellite accounts” on the environment for a number of years, although it is hard to identify any direct influence they have had on economic policy debates. As long as political contests focus on economic growth, as I think they always will, a set of statistics labeled “satellite” is unlikely to be influential.

If policy decisions are to take account of the environmental impact of growth, and the extent to which current growth comes at the expense of future growth, natural depreciation also needs to be accounted for in GDP, alongside the depreciation of machines and roads.

In conclusion

GDP struggles with measuring innovation, quality, and intangibles, but it does a better job than any currently available alternative. Some economists are concerned that budget cuts at national statistical offices are making it harder to get national account statistics that are of adequate quality, and they regret the diversion of resources to more fashionable indicators such as “happiness”; they would certainly argue against diluting any further the effort that goes into collecting GDP and its related statistics.

At present, we are in a statistical fog, without the information needed about either the negative aspects of growth when it is unsustainable and depletes the natural and other assets available for the future, or the positive aspects, when it delivers innovations and creativity. However, for all its flaws, GDP is still a bright light shining through the mist.

  • Diane Coyle will be speaking on the subject of ‘GDP and beyond’ at the upcoming event, The Bean Review: Making it Happen, to be held at the Royal Statistical Society on 16 May. Details are online here.
  • GDP: A Brief but Affectionate History is out now, published by Princeton University Press. Further information can be found here.