GDP is a flawed but magical indicator

All the back and forth about how GDP is calculated is only possible because despite all the flaws, the measure somehow ends up feeling right

Economists have long argued that the gross domestic product has many flaws as a measure of well-being and policy success. Yet there’s a good reason it’s still being used: There’s a certain magic to it, despite its science being somewhat iffy.

On Monday, the National Bureau of Economic Research published a paper by Harvard economist Martin Feldstein detailing an argument he has been making for years—that GDP calculations underestimate actual growth and productivity. This optimistic argument is based on the difficulty of measuring changes in the quality of products and services, and therefore of life. Feldstein points out, for example, that official measurements, for the most part, only catch quality improvements if a product or service requires more expensive inputs: “If it doesn’t cost more to produce a product or service this year than it did last year, there has been no improvement.”

That way, for example, leaps in the quality of healthcare—when a patient who used to need a week in hospital to recover from a cataract operation is now discharged on the day of the procedure—are not measured. The way official statistics measure the introduction of new products, too, doesn’t account for their actual contribution to consumers’ well-being or to the economy as a whole.

According to Feldstein, government messaging should be more optimistic to make sure people understand that their savings will buy more in the future. Goods and services are improving lives more than price increases would indicate.

Nobel laureate Joseph Stiglitz has long held the opposite view—that the GDP as measured today may overestimate well-being. For example, it counts any increase in government spending as positive, even though these increases may be inefficient or even counterproductive.

And as for those improvements in healthcare quality that form the basis of Feldstein’s argument, they, too, can be overestimated in the US because healthcare spending there is higher than other countries while the outcomes are the same or worse.

Some recent work also argues against the theory, supported by Feldstein, that the recent productivity slowdown is due to a failure of measurement. Last year, Chad Syverson of the University of Chicago pointed out that even the most generous estimates of the value added by the growth in digital technology aren’t big enough to bring productivity growth to its pre-2004 trajectory.

Another analysis by International Monetary Fund (IMF) economist Marshall Reinsdorf found that their unmeasured effect on productivity could only be small. Statistics fail to record some of the added value because of the tech sector’s use of tax havens, he wrote. But even the “free” Internet services provided now are counted through the advertising they attract. And some of the improvements that tech created for consumers don’t belong in the GDP calculation in the first place: If they save a user some personal time, that stays in the home and doesn’t affect economic activity (even if it did, it might be cancelled out by the time our digital addictions take out of our productive workday).

All the back and forth about how GDP is calculated is only possible because despite all the flaws, the measure somehow ends up feeling right. The distortions often end up cancelling themselves out.

In 2013, Nicholas Oulton of the London School of Economics’ Center for Economic Performance wrote a paper to disprove the notion that the UK’s economic growth had been overestimated because official calculations overstated the contribution of banking to GDP. He showed that “if banking output has been overstated, then the output of some other industry or industries must have been understated”.

Earlier this year, a team of IMF economists attempted to figure out how GDP numbers would have changed for a number of developed countries had they used an outdated deflation method, still used by China and India. It turned out that the effects wouldn’t have been consistently negative or positive for most countries; for Western European countries, on aggregate, the effects would have been small.

The team’s recommendation was that more countries adopt the more progressive deflation methods now used by most of the G20—but their research made it clear that in some cases the difference in the results would be tiny.

As much as GDP calculation isn’t an exact science, the results usually make sense. That’s why per capita GDP is one of the strongest predictors of happiness measured through people’s subjective perceptions of their well-being.

It’s fine to argue for better measures of well-being. These measures, however, add even harder-to-measure indicators, such as levels of social support, freedom and generosity. For many countries, these data are either unavailable or subjectively coloured.

The choice is between engineering and science: The former will accept an imperfect approximation, while the latter will always strive for perfection.

As Federal Reserve chair Janet Yellen recently pointed out, GDP is “a pretty noisy indicator” at best. Yet it remains extremely useful as a reference