Diane Coyle on GDP
What this covers
Diane Coyle and Russ Roberts examine how GDP came to dominate economic thinking and why it obscures as much as it reveals about material welfare. The conversation traces the concept's origins from William Petty in the late 17th century through Simon Kuznets and John Maynard Keynes, who formalized it in the 1930s and 1940s as a tool for measuring the resources available for war. Coyle argues that GDP conflates two distinct purposes: measuring economic activity at market prices—useful for macroeconomic steering—and measuring human welfare, which it does poorly. The core problem is that GDP is not a natural object awaiting discovery but a construction full of judgment calls about what counts as production. Whether housework, volunteering, or government services enter the ledger depends on arbitrary boundaries rather than inherent economic logic.
The discussion spans how GDP systematically misses innovation, product variety, and the explosion of free digital goods. A smartphone's value dwarfs anything hedonic price adjustments can capture; antibiotics that save lives cost dollars; Wikipedia generates enormous consumer surplus yet barely registers in the accounts. Coyle points out that measuring service quality—from tax preparation to medical care—runs into information asymmetry that makes productivity growth in these sectors nearly impossible to track reliably. The conversation also addresses the Consumption vs Investment Distinction, showing how GDP treats a government purchase of tanks identically to household consumption, obscuring questions of sustainability and intergenerational burden. Coyle proposes moving toward a dashboard of indicators—drawing on Australia's model—that would make environmental depletion, leisure time, and sustainability visible alongside growth, enabling democratic debate about tradeoffs rather than treating one number as destiny.
Coyle argues GDP is not a natural measurement but a constructed concept full of judgment calls that systematically misses innovation, variety, service quality, and free digital goods, so it should be supplemented by a dashboard of indicators rather than fetishized as a single welfare measure.
- GDP measures economic activity at market prices, which cannot capture innovation, product variety, or non-market goods like Wikipedia and unpaid housework
- Service and government output quality is inherently hard to measure, undermining productivity concepts
- A dashboard of indicators capturing tradeoffs (environment, sustainability, time horizons) would serve policy better than GDP alone
GDP conflates economic activity measurement with consumer welfare, missing innovation value and quality improvements that matter to people.
- GDP statistics can never properly capture the value of innovation over time; for example, Nathan Mayer Rothschild, the richest man in the world, died of a tooth ulcer because antibiotics had not been invented, yet antibiotics that might cost $10 today represent value that cannot possibly be counted in GDP.
“Nathan Mayer Rothschild, richest man in the world at the time, died of an ulcer on his tooth because antibiotics hadn't been invented. What value could you possibly put on the invention of antibiotic that costs maybe $10 today?”
- GDP conflates measuring economic activity at market prices (useful for macro policy) with measuring consumer welfare; ironically Simon Kuznets, considered the father of GDP, actually wanted a welfare measure and would have argued against including government spending on tanks as well as advertising, criminal lawyers, and police forces, which he regarded as 'consumer bads' rather than goods.
“Simon Kuznets in the 1930s and into the early 1940s... definitely wanted a measure of consumer welfare. He would have argued against including government spending on tanks, but also against all kinds of things. He thought advertising, paying for criminal lawyers and the police forces--they were not consumer goods. They were consumer bads.”
- The criticism that GDP implies gross materialism is largely misplaced, because the great majority of economic activity in leading countries is now in services and intangibles, and because the critique underestimates the value people place on choice and variety; Jerry Hausman's estimate that the apple flavor of Cheerios was worth $200-300 million a year illustrates how much consumers value the ability to shape their own lives through choice.
“Jerry Hausman at MIT famously looks at the consumer benefit of the introduction of the apple flavor of Cheerios. And it was--I think it was $2- or $300 million dollars a year.”
- Statisticians use hedonic regressions of price on product characteristics (camera, Wi-Fi, screen resolution, memory, speed) to separate how much of a price change is due to better characteristics versus normal inflation, which partly addresses innovation, but it cannot fully capture quality change—a smartphone equivalent of a 1946 computer would be the size of half a dozen aircraft carriers, an improvement far beyond what price adjustments reflect.
“they will do a regression essentially of the price on different characteristics to try and work out how much of the change in price... is due to the better characteristics and how much of it is normal inflation.”
- GDP measures are ideal for an economy that mass-produces things where variety doesn't matter, but the modern economy is one where variety is everything and innovation drives growth rapidly; the diversity of products people choose among (different laptops, iPads, cellphones that merge together) is something people genuinely value but gets lost in the statistical shuffle.
“in GDP we have a set of measures that are really ideal for an economy that mass produces things and where variety doesn't matter. And we now have an economy where variety is everything and innovation is driving growth at a great pace these days.”
GDP's production boundary excludes unmeasured goods and services like housework and digital gifts that deliver real welfare and value.
- Many wonderful free digital goods—Wikipedia, Skype, EconTalk and other free podcasts—deliver large consumer surplus and welfare but fall outside the production boundary, contributing almost nothing to GDP; this is just like the housework boundary, where paying someone to clean counts but doing it yourself does not, and some of this activity may move into GDP over time while some will not.
“There are lots of free digital goods, and if all the people producing them found a way of charging for them and making money, that would go straight into GDP... But there are quite large activities that are wonderful that are outside that production boundary. Skype, or--”
- What counts in GDP is determined by a judgment-laden 'production boundary'—unpaid housework is excluded while paid cleaning is included, and volunteering to read at a school does not count while volunteering to build a playground slide does count because there is a physical product at the end—decisions driven mainly by ease of measurement and partly by arbitrary judgment calls.
“if you volunteer to read at your local school, that doesn't count in GDP. But if volunteer to build a slide in the playground, that does count in GDP. Because there's a physical product at the end.”
- Online free goods belong partly to the nonmonetary 'gift economy' tradition of how people transact and relate to each other, and partly to what Adam Smith wrote about in The Theory of Moral Sentiments—the importance of relations between people that forms the bedrock for economic activity—so economic growth should be seen in the context of that wider set of human relationships.
“part of it is more the kind of thing that Adam Smith wrote about in The Theory of Moral Sentiments, about the importance of relations between people and the way that formed a bedrock for the world[?] of creating the economic activities.”
GDP is a judgment-laden concept with no natural referent; its definition and scope change substantially over time and across countries.
- Measuring the economy is not like measuring the height of a mountain or the length of a river; there is no natural object out there. GDP is a concept that involves a lot of judgment and changes substantially over time, as shown by Nigeria's announcement that its GDP was 89% higher after re-basing its statistics to reflect new sectors like Nollywood and mobile phones.
“what this highlights is a misconception, really, that measuring the economy is like measuring the height of a mountain or length of a river. There is no natural object out there. This is a concept and it involves lots of judgment.”
- GDP is a measure of an amount of economic activity measured at market prices, which gives policymakers setting tax, spending, and interest rates a way to gauge how quickly the economy is expanding; it has existed in its present form for just over 60 years.
“GDP (Gross Domestic Product) is a measure of an amount of economic activity, measured at market prices. It's the number that allows policy makers setting the level of tax and spend or setting interest rates to get their arms around how quickly or otherwise the economy is expanding.”
- National income measurement originated with William Petty in the late 17th century, who counted national income to help the British monarch understand the tax base for foreign wars, giving Britain a fundraising advantage over France in its wars; modern GDP only emerged in the 1930s Great Depression through Simon Kuznets and Colin Clark and reached fruition in WWII under Keynes in the British Treasury.
“in the late 17th century, a person called William Petty started to try to count up national income, the purpose being to help the monarch understand what the tax base, how much taxes he was able to raise despite foreign wars.”
- GNP became GDP because policymakers focused on activity they could steer with macroeconomic policy within national boundaries; the difference between the two is income received from or sent overseas, and in recent decades that gap has grown again for countries with large overseas earnings or migrant remittances.
“the reason that GNP became GDP was because the focus switched to what it was that policy makers felt they could steer using macroeconomic policy.”
- Modern GDP dates from 1941, developed by Keynes in the British Treasury with the specific military aim of working out what resources were available for the war effort and what consumption sacrifice civilians would have to make; it then co-evolved with Keynesian macroeconomics, with national accounts built on the same categories Keynes used.
“in the late 17th century, a person called William Petty started to try to count up national income, the purpose being to help the monarch understand what the tax base, how much taxes he was able to raise despite foreign wars.”
Growth is necessary for employment and funds innovation people value, but it is not valuable for its own sake and must be weighed against environmental and distributional costs.
- The intense focus on the single GDP number is almost a fetish; growth is marvelous but not valuable for its own sake. The argument that 2% growth is needed to avoid unemployment stems from population growth, which some critics propose to fix by halting growth and even shrinking the population—a steady-state position Coyle dismisses by asking what such advocates would give up to get the next smartphone they want, since new products require giving up old ones if there is no growth.
“if they want to stop growth, then I want to know when the next smart phone that comes along that they want, what they are going to give up instead. Because if you want new products you've got to give up old ones, if you are not going to have any growth.”
- Growth is genuinely part of what makes people happy—around 2% annual GDP growth is needed to keep unemployment from rising, and unemployment clearly makes people unhappy, plus growth funds the innovation and new services people value—but it is not 'growth at any price' because there are environmental costs (as in China where you cannot see out your window) and other factors, so societies should measure growth and other things and have a public conversation about the tradeoffs.
“we know that if we don't have 2% a year or so, unemployment is going to rise. That makes people unhappy. The evidence on that is really clear. So you need growth for that.”
- When critics call for a limit to growth, what they are really saying is 'let's stop this kind of innovation,' because innovation is growth—that's what growth is all about; trivial-seeming innovations like shampoo brought great consumer benefits, as shown by the 1900 House participant who gave up the show because shampoo hadn't been invented.
“when you think about the criticisms of growth, what people are saying when they are saying, 'Let's have a limit to growth,' is 'let's stop this kind of innovation.' And actually, innovation is growth.”
- Stagnation concerns point to a need to look at a range of indicators and at income distributions rather than just average headline GDP growth, because in OECD economies—most extremely in the United States—people at the lower end of the income distribution have not shared in the gains of growth, with much faster growth in top incomes than at the bottom, a concern reflected in the attention to Piketty's Capital in the Twenty-First Century.
“the people at the lower end of the income distribution haven't been sharing in the gains of growth for some time. Most extreme in the United States, but it's true to some degree in all countries because there's been much faster growth in top incomes than at the bottom.”
- Voters and journalists should judge economic leadership not just on GDP growth but on the prospects given to the next generation—whether leaders merely respond to short-term pressure and headlines or have a vision for the country in ten years, including jobs, future tax burdens, the environment, and education preparing children for the future—even though the chances of anyone getting these hard questions right are low.
“Are they just responding to short-term pressure and headlines, or have they actually had some vision about what kind of country they wanted to be in, in 10 years time?”
Transparent dashboards of multiple indicators enable public tradeoff debate and better policy than focusing on GDP alone or claiming government control over it.
- Australia's consulted dashboard of indicators—quality of local environment, leisure time, convenience getting to work—published annually on a red/green traffic-light system made tradeoffs clear, showing green GDP growth alongside red resource depletion, so Australians know they are digging up mineral resources to have growth now; this kind of transparency enables constructive policy debate.
“last year it was very clear that GDP growth is going great in Australia, and that was green. And resource depletion had worsened, so that was red. And actually, isn't that tradeoff really clear?”
- Governments cannot control GDP in any direct or predictable way, yet politicians are keen to claim they can; Coyle would not hang everything on GDP because its size means nothing, and does not understand why politicians insist on claiming control over it.
“governments can't control it, not in any direct or predictable way. And I don't really understand why politicians are so keen to claim that they can, because its size means nothing.”
- Poorer countries have incentives to stay officially poor even if not really poor, because of the World Bank loans they can access; they also often lack the capacity and resources to collect statistics, and data collection is harder where infrastructure differs, though new technology like smartphone surveys and network-data tracking offers hope of catching up.
“there are incentives for poorer countries to stay poor in an official sense even if not in a real sense, because of the World Bank loans that they can get access to or not.”
- Because GDP is measured at market prices but there is no market in government services, government output was long valued by counting how many people worked for government and how they were paid, which by definition gives zero productivity growth in government services; statisticians now attempt private-sector comparisons, but the very concept of productivity works poorly when quality is inherently part of the service.
“for a long time it was done by counting up how many people worked for the government and how they got paid. And that gave you a figure in pounds or dollars, but it left you with, by definition, serial[?] productivity in government services.”
- Quality is far harder to monitor and measure in services than in physical goods because the professional providing the service often has an information advantage—you must rely on your tax accountant or a computer-system builder to tell you whether they did a good job, and even afterward you may not know how efficient the result was—so the measurement problem with services is the same as with goods but much more severe.
“Many of them, the professional providing the service has an information advantage. Is my tax accountant doing a good job? Well, I have to rely on him to tell me, because I'm not a tax accountant myself.”
Granular microeconomic happiness research reveals actionable policy targets like unemployment and commuting, unlike macro happiness debates.
- Microeconomic happiness research is useful, unlike the macro happiness debate, because it identifies things policymakers can actually tackle: unemployment makes people unhappy, unhappiness is strongly correlated with mental ill health, and noise and commuting depress self-reported happiness; the macro happiness bandwagon is largely driven by people who dislike economics and growth.
“We know that unemployment makes people unhappy; we know that unhappiness is strongly correlated with mental ill health; we know that noise and commuting are definitely bound[?] in people's self-reported happiness. And these are useful because these are things that policymakers can have some hope of tackling.”
- The Easterlin Paradox—that within a country richer people are happier but over time rising GDP per capita doesn't raise happiness—is bizarre because you would not expect happiness to rise in proportion to GDP per capita; it compares a stationary time series (happiness) with a non-stationary one (GDP), so over some period the correlation is bound to look like zero, just as you wouldn't expect height to rise in proportion to GDP, and more recent work using GDP growth shows a tight correlation.
“In technology terms you are comparing a stationary time series and non-stationary time series, and over some time period the correlation is bound to look like zero.”
GDP cannot capture tradeoffs between current consumption and future investment, obscuring sustainability questions that matter for intergenerational fairness.
- GDP does not distinguish whether spending is for consumption today or investment for tomorrow—they count the same—so it cannot capture the tradeoff of using resources now versus investing for future consumption; this matters for any kind of sustainability (environmental, financial, or other) and has been confused in the macroeconomic debate of recent years.
“it doesn't enter GDP whether or not you are looking at something that is for consumption today or investment tomorrow. They count the same in GDP. And it doesn't allow us to think about the tradeoff... about using resources now or investing them for consumption in the future.”
- If put in charge, Coyle would move toward an internationally agreed dashboard of indicators to enable cross-country benchmarking, take effort away from making GDP statistics ever more complicated because there are real diminishing returns to that, and refocus on the hard-to-notice tradeoffs—especially time-horizon and broad sustainability questions like how much burden is placed on future taxpayers or resources used up.
“I'd take some of the effort away from making GDP statistics ever more complicated. Because I think there's real diminishing returns to what's happening with the GDP statistics. And I'd refocus effort on dashboards.”