GDP Is Blind

My phone replaced hundreds of pounds of cameras, maps, and stereos, and everything it removed came off GDP. The dashboard is blind to abundance, it was built for a different world, and policy steers by it anyway.

When I bought that phone I spent a few hundred pounds once, and in return I stopped spending, year after year, on film, developing, maps, CDs, and long-distance calls. I got more capability than I’d ever had, and the national accounts recorded the event as the economy shrinking. The measurement didn’t malfunction. It did exactly what it was designed to do, and that’s the problem.

What GDP can’t see

GDP [gross domestic product, a tally of the money spent on final goods and services in a country over a year] counts what people pay for, plus a named list of unpaid items that the statisticians price by rule.¹ It still doesn’t count value, because every one of those rules is built from money someone paid or from what something cost, never from what the thing is worth to the person who gets it. What is paid for and what it’s worth used to travel together, and technology has been pulling them apart.²

GDP can’t see value that stops being paid for. When photography meant film, developing, and postage, every step was a purchase, so all of it showed up in GDP. Now a photo costs nothing at the point of use, so the whole chain of film, developing, and postage has fallen out of the accounts, while you take more photos than you could ever have afforded on film. Your benefit went up and the spending went down, and only the spending is counted. Economists call the gap consumer surplus [the value you get above what you pay], and GDP’s entry for it is zero. Where the capability is folded into a service you still subscribe to, part of it can enter the accounts when the statisticians mark that service down as better value for its price.³ Where it arrives from someone you pay nothing, nothing enters.

GDP can’t see time. When a free video call replaces a flight and a hotel, GDP falls by the price of the flight and the hotel, and it records nothing for the two days you got back. In an economy where more and more output is information, freed time is the actual wealth being created, and the headline accounts [the accounts GDP comes from] have no entry for it. Hours worked are counted, because the productivity figure is output divided by hours.⁴ Unpaid hours are counted too, in side accounts, at what it would cost to hire someone to do them.⁵ So an hour you no longer have to spend makes those side accounts smaller, not larger.

GDP can’t tell spending apart from creating value. A factory that installs robots and makes the same shoes with half the inputs is genuine productivity, and it can show up as GDP falling, because the shoes now cost less. A company that borrows to build a showroom nobody visits adds to GDP, because money changed hands. So the money total is flattered by borrowing and insulted by efficiency. The figure that counts how much was made is flattered by borrowing just the same, because the showroom counts as output whether anyone visits it or not. That figure takes a price fall differently. A pure price fall leaves it where it was, and a price fall that sells more shoes raises it. Where efficiency removes the priced step altogether, both figures lose the entry. That’s backwards for judging whether life is getting better.

The statisticians do try. They use hedonic adjustment [statistical change to prices to reflect quality improvements], and inside the official measure of prices it can only adjust the price of things that still have prices.⁶ It has no entry for the capabilities that stopped carrying a price of their own. The method itself reaches further wherever something that’s sold carries the same features. That’s how the UK house price index is built.⁷ It works out what buyers paid for each feature by comparing the prices of houses that have it with the prices of houses that don’t. No market prices the features of a free service, so the people who try to measure those reach for something else. They ask people what they would need to be paid to give the service up, or they value the service from the equipment people already own.⁸

Why it was built that way

There are two answers, and it matters to keep them separate.

The first is practical, and here I’m reaching into general history, so hold it loosely. The accounts were designed in the 1930s and 1940s to answer depression and wartime questions, how much steel, how many tanks, how much output could be mobilised.⁹ In a physical economy, what people traded was scarce, so almost all of it carried a price. Counting spending was then a rough stand-in for the value moving through the market. It wasn’t a stand-in for all value, because work done at home was free of charge then too, and it has never been in the headline accounts. The tool matched the world it was built to measure. Its main architect even warned it wasn’t a measure of welfare. The world then changed and the tool didn’t.

The second answer is a large part of why the tool was never replaced. Our money is loaned into existence. Banks create most new pounds when they make loans, so the economy carries a debt load that must be repaid in fixed pound amounts.¹⁰ A system built on debts fixed in pounds can take a slow fall that comes with better tools. In that fall it’s the cost of making things that drops, and pay drifts down slowly behind it, so the money to meet a fixed payment is still there. What it can’t take is a sudden fall in the prices credit pushed up, houses first, because then the losses spread through the banks. In that fall it’s the lending that has stopped, so incomes and the house fall together and at once, and the fixed payment is hit from both sides in the same year. Houses come first because people buy them with borrowed money. The debt takes a bigger bite of every income, yours, your employer’s, the government’s. And where the loan was written against a house, the house is worth less too, so the bank stands to lose more on that loan. A dashboard that measures spending, plus a standing target of making that number grow, is exactly the instrument panel that system requires.

Nothing has displaced GDP at the centre of policy, and a large part of why is that it measures the thing a debt-based system needs to keep rising, the total spent on goods and services. The total doesn’t rise every year, though. It fell in 2009 and again in 2020, and each fall was met with rescue rather than collapse.¹¹ Whether it measured wellbeing well was beside the point. And once a number becomes the target, it gets managed to hit the target, so it ends up telling you more about the policy than about your life.

What happens when policy steers by it

Technology succeeds, so the cost of making things falls and people need to spend less to live the same life. The dashboard doesn’t say why spending fell. A fall because things got cheaper and a fall because people had less to spend look the same on it, and the standing response to a shortfall is to make borrowing cheaper. Where the cause is plain and the fall is judged temporary, that response can be held off for a time. Where the response isn’t held off, policy treats the fall as genuine failure. The central bank cuts interest rates, and the government and the banking system create more money and credit to push spending back up.

So every time a sudden fall in the prices credit pushed up threatens, the system is rescued with new money, and the rescues keep getting bigger. In between the rescues, policy targets rising prices of about 2% a year, forever.¹² Central banks publish why they aim there. They want room to cut interest rates in a downturn, they know wages resist falling, and they say the official measure of prices overstates the rise.¹³ Those are the reasons they give, and the debts aren’t among them. The debts show up in the rescues, not in the everyday 2%. The starting line for that 2% isn’t zero, because technology keeps pushing prices down. So when a central bank targets 2% inflation, it doesn’t need two points of push. It first has to cancel whatever better tools would have taken off prices that year, then add two on top. How big that fall would be is about how fast output per hour of work grows, which is what the productivity figures measure. On the published figures that’s a point or two a year.¹⁴ No statistic records that fall, because it never happened.

Technology keeps compounding, and the tools are improving faster than they did, so what policy has to cancel keeps getting bigger. What the record shows growing is the stock of debt and the rescue that each downturn gets, not the yearly pace of new money, which hasn’t been getting faster.¹⁵ Central bank holdings jump in a crisis and come only part of the way back.¹⁶ Rescues went from hundreds of billions of dollars in 2008 to trillions in 2020.

Instead of recreating the spending that fell, the money that was created to push it back up flows into whatever can soak it up, mostly houses and shares. So asset prices inflate while wages lag, and the person who owns little watches the cost of living rise faster than their pay.¹⁷ Rising prices and falling real wages are the same event seen from two sides. The gain from the better tools is real. Part of it reached you as higher pay, late and unevenly.¹⁸ The rest went to whoever got the new money first, through house and share prices.

Meanwhile the debt keeps growing faster than the growth it buys. Between 2000 and early 2018 the world added roughly 185 trillion dollars of debt to buy about 46 trillion of measured growth.¹⁹ Each new pound of borrowing produces less real output than the last, which means more and more of the “growth” on the dashboard is just the debt itself being spent.

Then the loop feeds itself. The new money pushes firms’ costs up, so they automate faster to protect margins, which deepens the very price falls the policy is fighting, which calls for more new money. Investment goes to the wrong places, because borrowing made cheap by policy makes projects that destroy value look profitable on paper. Firms that should fail are kept alive by cheap credit. The companies closest to the new money compound their advantage, so the biggest companies end up holding more of the economy. Renters fall further behind owners, people feel cheated without being able to say why, and politics turns into a fight where one side’s gain is the other’s loss. And on a finite planet, a system that needs total spending to keep rising keeps us buying things that efficiency would otherwise have made unnecessary.

The worst of it is that we lose the ability to see abundance at all. Your phone made you richer, and the accounts counted that as the economy shrinking. The dashboard kept reading growth anyway, because the money households moved to other things, together with borrowed spending, more than covered the gap. So your gain is buried inside a total that has gone on rising in all but a handful of years, and every one of those was a slump or the end of a war.²⁰ That’s why the “productivity slowdown” looks so mysterious, and why the inflation versus deflation debate goes in circles. Both sides are measuring with a unit that’s being stretched, and every calculation built on a false measure inherits the error.

Four reasons to trust the number anyway

“GDP is still the best single indicator we have.”

Best available isn’t harmless when the errors all point one way. GDP systematically undercounts abundance and overcounts spending paid for with borrowed money, so steering by it tilts policy toward forcing spending up.²¹ A compass that’s always off to the same side is worse than no compass, because you trust it.

“People still pay. The phone itself wasn’t free.”

The device is one purchase where there used to be several, each with its own replacement cycle. The accounts lost the film and the developing year after year, and the camera, the maps, the CDs, and the player each time one would have been replaced. Hardware cycles exist, but the cost per use of what the phone does has collapsed toward zero, and per-use cost is what your life runs on.

“If GDP fell, we’d be in a depression.”

In this system, yes, if the fall came suddenly. That’s the trap, not a defence of the measure. In a system this loaded with debt, a sudden broad fall really would be a disaster. But that’s an indictment of the debt design, not of cheaper goods. Falling prices from better tools raise living standards.²² Falling prices from a credit collapse destroy them.²³ As a rule, the current system responds to both the same way, because the dashboard it steers by can’t tell them apart.

“Measured productivity growth is weak, so this tech-deflation story must be overdone.”

Part of that measured weakness is the gap between value and spending. Output that becomes free drops out of the numbers. The slowdown is partly genuine, because cheap credit keeps low-productivity firms alive, and partly the ruler failing.

What to measure instead

The way out of the fog starts with measurement. Judge the economy by purchasing power [what your money actually buys] and by time, not by the flow of spending. The flight and the hotel are the price, and the two days are the wealth. Ask how many hours of work a home costs now versus a generation ago, and the picture looks very different. The official measure of prices can tell you roughly how much the pound stretched against the goods it prices. What it can’t show you is how far those goods would have fallen under a unit nobody can expand, because it does all its sums in the same stretched pound. So to read that fall you eventually need a unit nobody can stretch, because you can’t measure a plank with a ruler that changes length. That’s the role I’d argue bitcoin plays in this story, but nothing I’ve said here needed it. What the phone does got cheaper. The dashboard counted that as a loss. Policy fought your raise. That much you can verify from your own pocket.