Is it time for national statistics to move away from a framework centered on GDP? If so, how do we go about choosing the new framework?
In today’s post, I turn my attention from anxieties about the potential impact of a new U.S. administration to more fundamental conceptual questions. This will be the first of several posts about papers on economic statistics that were presented at the recent ASSA/AEA conference of economists in Chicago. (I was unable to attend, but preliminary drafts of most of the papers are available on the meeting’s website.)
This first paper is “The political economy of national statistics.” It comes from Diane Coyle, who is professor of economics at the University of Manchester, author of the delightful book, GDP: A Brief but Affectionate History, and writer of a wonderful blog, The Enlightened Economist.
Coyle frames her paper around a growing consensus that GDP has outlived its primacy as a measure of economic success. The influential 2009 report by the Stiglitz-Sen-Fitoussi Commission recommended shifting “emphasis from measuring economic production to measuring people’s well-being,” while also taking account of sustainability. That report was followed with a work program from the OECD and the European Commission known as “GDP and Beyond,” which focuses on supplementing GDP statistics with measures of distribution and well-being. Coyle asks whether the shift from measuring production to well-being can be accommodated by modifying the current measurement framework (the international System of National Accounts 2008, or “SNA”) or if it will require adopting an entirely new framework. If a new framework is needed, she asks how the shift to a new standard could be coordinated to form a new consensus.
Coyle describes several reasons that economists and other data users are increasingly dissatisfied with the current GDP-centric framework for national economic statistics. Since 1980, income inequality has grown dramatically in the U.S., the U.K., and several other countries. Economists increasingly recognize that growth in aggregate income doesn’t necessarily benefit all, or even most households. Traditional GDP also doesn’t account for environmental degradation or climate change, which would require a shift to a measure that reflects sustainability. GDP doesn’t include non-market activities, such as unpaid home production or volunteer work, nor does it include the value of “free” digital goods, such as websites or apps that are paid for by advertising. Finally, although the technology sector is included in GDP, there are concerns that in practice, GDP and productivity measures may understate the sector’s contribution to growth.
The SNA provides the advantage of coordinated, internationally accepted standards for economic measurement, but at the cost of slighting these and other problems. Is there a way we can to come to a consensus on how address these problems while retaining the advantages of an internationally coordinated and accepted set of standards? Coyle mentions the European Union’s Social Progress Index and the OECD’s Better Life Index as examples of measures that have not yet achieved international consensus.
As a recent national accountant, I tend to agree with Coyle that GDP is often misused. Even within the set of statistics that are available in most country’s national accounts, we see GDP being used when other statistics, such as net domestic product or household disposable income would be more appropriate. And if we could agree on a measure of well-being, I think most of us would agree that well-being would be appropriate than GDP as a target for national economic policy.
On the other hand, I don’t see GDP ever being entirely displaced from its appropriate role as a measure of production. For many important purposes, such as analysis of business cycles, productivity, and government revenues, the GDP data provide the appropriate answers to the relevant questions, so I expect GDP will always be an important national statistic.
In some areas of measurement, we’ve seen the consensus shift over time from one statistic to another. An example from economics is productivity statistics, which over the last 40 years have shifted from measures of labor productivity to multifactor productivity. Another example is seen in the sport of baseball, where players were formerly evaluated on their batting averages, but more recently are increasingly evaluated using new, more sophisticated statistics such as “wins above replacement.” In the case of multifactor productivity, the shift in consensus reflected the work of many academics, with Dale Jorgenson particularly playing a leadership role. In the case of baseball statistics, the shift appears to have been driven mainly by statistically sophisticated fans working out new statistics on web forums, then overcoming the resistance of traditionalists in the mainstream media.
A difficulty with achieving consensus on a measure of well-being is that there may be political disagreements on the weight given to certain dimensions of well-being. The OECD has been surveying visitors to its Better Life Index website about the weights they would give to each dimension of that index.
How should move this work forward? To develop a consensus for a measure of well-being, we must first focus on developing a prototype statistic or set of statistics. Just as during World War II, the UK, the US, and a few other countries demonstrated that GDP worked in practice as a measure of production before it was adopted as an international standard, working examples of well-being measures are needed. These measures should then be critiqued and refined, and the measurement framework established, before a measure is proposed as a new international standard. I think a dedicated working party consisting of academics and government statisticians that is open to receiving input from other interested parties would be an appropriate vehicle for accomplishing this work.
Whenever I think about the measures of some concept, be it well-being, economic activity, income, employment, etc. I try separate “How well are we measuring the concept?”, which is often very hard, from “How well are we measuring the movement?”. For example, there are several ways to measure “unemployment”. However when you compare them the levels are very different but they pretty much move in sync.
So maybe well-being is not be measured well, so can make cross-sectional comparisons problematic, but if the measurement goes up or down – across time comparisons – can we say that well-being has improved or worsened?
That’s a good point. Of course, users of national statistics are interested in both questions: How much has economic well-being improved over the last 20 years? and, How does economic well-being in the United Kingdom compare with the United States? But if we can answer the first question and can’t answer the second, it would be worth knowing.
As a heavy user of micro economic data I’m much more interested in our developing much better statistical data on the service sectors of the economy like we have had for years for the goods sector.
For example, advertising is a major economic sector and many of the the new economic activities on line are really in the business of selling ads. But as far as I know the government does not publish any regular data on advertising.
I agree that the coverage of statistical data on service sectors continues to lag behind goods sectors. But the Census Bureau’s relatively quarterly services survey does provide more information than we had in the past. For example, it includes quarterly estimates of revenue for NAICS industry 5418, “Advertising, public relations, and related services.” BEA’s annual and quarterly industry accounts, however, don’t show that level of detail; advertising is included in higher-level aggregates like “Professional, scientific, and technical services.”