How to redesign GDP for the 21st century

Value judgments.
Value judgments.
Image: REUTERS/Pichi Chuang
We may earn a commission from links on this page.

For decades, economists have tried to come up with better methods of measuring the economy than gross domestic product. But the dominance of GDP—a single figure, universally understood, and widely used for almost a century—has endured. This is despite its shortcomings gauging wellbeing and other factors that are important to a nation’s economic health. Now, though, some economists have created an ambitious new measure with the potential to challenge the preeminence of GDP.

This new metric doesn’t directly measure happiness or wealth or development, but calculates something else GDP neglects—free digital goods and services. These 21st-century technologies have basically broken 20th-century practices for measuring the economy. MIT’s Erik Brynjolfsson, along with Avinash Collis, W. Erwin Diewert, Felix Eggers and Kevin J. Fox, have developed “GDP-B”. It’s designed to capture the numerical value of the things that we don’t pay for but still have plenty of value, such as online maps, photos taken on smartphones, Wikipedia, and social media.

Quartz spoke to Brynjolfsson, who has been studying technology and the economy for two decades and co-authored The Second Machine Age with Andrew McAfee, about the creation of GDP-B. The conversation has been condensed and lightly edited for clarity.

Quartz: This is a really ambitious project. Why is it necessary?

Erik Brynjolfsson.
Erik Brynjolfsson.

Brynjolfsson: All economists should understand that GDP is not a measure of welfare, even though that’s the way people traditionally use it. It’s just a measure of production.

I was working with lots of digital goods like music and Wikipedia that had zero price. By definition, if something has zero price it doesn’t show up in GDP. It’s like it doesn’t exist. And so it’s been something I’ve been worrying about for quite a while.

Why were you researching digital goods?

For one of my first projects as a PhD student at MIT, a professor asked me to chart the growth of computer power in every industry. It was around 1989 and I made these charts, which had all these exponential curves taking off, in finance and manufacturing and retailing. I guess you see steep exponential curves a lot these days but back then it wasn’t so common and I had this moment of shock and awe. You didn’t have to be very creative to see that this is going to continue getting steeper and steeper. At that moment I decided I was going to devote my research career to understanding the implications of that because we were just in the early days of the computer revolution.

In 2009, I wrote an article with Adam Saunders called what the GDP gets wrong and why you should care. It was all the problems with productivity and measurement. But then I had this “aha!” moment about four years ago, which was that we could do online choice experiments to come up with a better way of fixing this. For a long time, I knew there were problems and, like other people, I was pointing out the problems. It’s easy to be the critic. I wanted also to do something better and come up with an alternative approach.

Why is it so important that we change the way GDP is measured?

There’s probably no more basic economic question than: How are we doing?

Are we better off than we were last year? Is the economy doing better? Are children better off than their parents? Are the economic policies pointing us in the right direction? Are these innovations that companies are developing helping us at all? Those are really basic questions that GDP is used to answer, even though it isn’t quite the right tool.

GDP is used is for two reasons. One, because we figured out a way to measure it. Simon Kuznets and his team invented GDP and productivity metrics back in the 1930s, and an enormous amount of effort has gone into refining them. It was one of the greatest inventions of the 20th century, this way of measuring everything in real detail. The second reason is that it is usually somewhat related to welfare. Even though it’s not identical it is often correlated. For instance, if you have twice as many cars in a country, all else being equal, you’re probably going to be maybe twice as well off.

The problem is that that relationship of more GDP leading to more welfare is not true for digital goods in the way it is for physical goods. And that’s because digital goods have zero price. So if there’s twice as many people reading Wikipedia, it doesn’t really change GDP at all. Or if there’s twice as many articles there. If they’re all free, GDP is effected by zero.

When you say goods with zero price, you mean things that have no price at the point of consumption like Wikipedia articles and YouTube videos?

For example, the music industry, as a share of GDP, has shrunk even though we’re listening to more and better music than ever before. Or, to take it even more broadly, according to the Bureau of Economic Analysis, there’s a category of all information goods that was about 4.6% of the economy back in the early 1980s. Today, it is 4.6%—it hasn’t even changed. Well, it bounced around a little bit, but it basically hasn’t changed. It’s pretty obvious to you and me that we have so much more information as we sit here with our smartphones in front of us with access to basically all the world’s information.

But the statistics say nothing has changed.

It hasn’t changed as a share of GDP. It’s not that GDP is wrong, it’s that GDP is measuring something different. It’s measuring what we spend.

If GDP is just measuring one thing—what we are spending—what does GDP-B measure?

The benefits we get. The “B” is for benefits.

For some purposes, you care what you’re spending and maybe the Federal Reserve wants to know how much money is in circulation. It’s not that production and spending aren’t important, but they aren’t everything. To measure the economy you need a dashboard with different metrics. What we’re measuring are the benefits you get even when you spend nothing on the good.

You’ve used smartphones to explain the sorts of benefits that GDP-B measures. 

When you get a smartphone today, most of the value comes from all of the software and the apps that are on it. GDP is measuring the hardware costs, even though a phone today has vastly more value than a smartphone 10 years ago. Blackberry is a good example. It’s out of business not because its hardware was so much worse—well, you could debate that—but it had more to do with the app ecosystem. If we don’t measure this, we’re going to completely misunderstand what we’re dealing with. There’s a saying in management, “you can’t manage what you don’t measure.” That’s true for businesses but it’s also true for societies. Our GDP-B measure is a way to capture the benefits of these online goods.

And although the original motivation and the main focus has been on digital goods, it also works for non-digital goods.

What would be an example?

Jet travel. You get value from jet travel, but the amount that you spend on it isn’t exactly equal to the value you get from it.

There’s a third category that we haven’t yet begun working on, but our framework can easily be extended to address. That’s public goods like government healthcare or park services. They are also valuable but the benefits aren’t properly accounted for in GDP.

If you don’t measure these things, people sort of implicitly assume the value is zero. With a moment’s reflection you know that’s not true but it doesn’t show up in any statistics anywhere. I want to briefly mention how this relates to another kind of research, which is the work on wellbeing indexes. They are also very important but are often done by surveys of how happy you are in different countries. They are also good supplements to GDP but they tend to be very qualitative. We see our approach as being a lot more precise, where we look at individual goods and put dollar values on what people would pay for them.

How do you work out the dollar values?

We do hundreds of thousands of online choice experiments. We get people to compare their preferences between two goods. Would you rather give up online music for a month or Facebook for a month? Wikipedia or Twitter? We do these comparisons with lots of different goods and lots of different people. You start getting a ranking of all the goods.

Have you done the experiments across different countries?

We’ve done some in the Netherlands and the United States and we’re beginning to see some differences. For instance, I was amazed how much more valuable WhatsApp is in the Netherlands than the United States.

It’s striking how much you need to add to GDP once you start doing this. Your paper shows that the welfare gains from Facebook would have added between 0.05 and 0.11 percentage points per year to US GDP since 2004.

Our estimate was the median user would have to be paid $48 to give up Facebook for one month. These numbers get quite large for all the many kinds of digital goods. Things like email and maps were valued fairly high. Music was not as high as I thought.

[Editor’s note: Brynjolfsson et al expanded the experiment to value other free goods in a lab experiment mostly using students in the Netherlands. They found that WhatsApp, Facebook, and digital maps on phones were highly valued, requiring a median compensation for losing one month of access of €536, €97, and €59, respectively (that’s $602, $109, and $66). Other applications such as Instagram (€6.79), Snapchat (€2.17), and LinkedIn (€1.52) have lower valuations, with Skype (€0.18) and Twitter (€0.00) the lowest of all. Their research was published in the peer-reviewed journal Proceedings of the National Academy of Sciences of the United States of America this week.]

What would be the impact of people using GDP-B alongside GDP?

We’ll have a much more realistic idea of what creates value in society and what doesn’t. A lot of digital goods we’ll find are creating a ton of value that people hadn’t been measuring before. That will change the ways governments invest in digital infrastructure and digital education. As we expand into other things, like public goods, I think people will start really putting time, energy and resources into the things that actually make us better off as opposed to just this narrow slice of the things that show up in GDP.

This is likely to raise more questions about how value is distributed.

We’ll start seeing where the benefits are occurring. It may be that some types of inequality are not as bad as we thought because digital free goods can provide you with something valuable at no cost. But other kinds of inequality may be worse. Certain policies may be helping or hurting in ways that we hadn’t accounted for before. There will be an opportunity to be much more realistic about which groups are benefiting and which ones aren’t, and how we are allocating our resources to benefit as many people as possible.

This fits into research such as Mariana Mazzucato’s work on understanding value creation and William Nordhaus’s work on measuring the environment and climate change. We’ve known that there’s a problem for years, so is this the moment when there are finally viable solutions?

I’ve been one of the many people who for a long time has been pointing out the flaws with GDP, but it’s not enough just to point out the flaws. The reason that people continue to use it is no one’s developed a better supplement. Now, we have one that we’ve shown can work. I think we’ll soon see more and more people taking this broader view.

I’m very excited about scaling this up. Our hope would be that, just as GDP and its related metrics revolutionized measurement for the 20th century, we can work to do a similar thing for the 21st century.