What is the value of free services provided by the internet? In this presentation, Stanford Professor and Director of the Digital Economy Lab as well as the Senior Fellow at Stanford Institute for Human-Centered AI (HAI), introduces his latest work into measuring the New Economy.
This Dialogue intends to provide a forum for prominent scholars, policymakers and industry experts to exchange views and collaborate on possible new and better measurement approaches.
Luohan Academy has been in close collaboration with some of the world's foremost social scientists, policy experts, and business practitioners to document and analyze society's and the economy's ongoing digital transformation worldwide. Our very first Frontier Dialogue focuses on the measurement of the new economy, a crucial step in the direction of better understanding the impact of digital transformation.
Erik Brynjolfsson: Well, thank you so much for giving me the chance to share some of the research. I wish I could be there in person with all of you, but maybe it's appropriate that we're talking about free digital goods that zoom is allowing us to interact digitally this way very conveniently from all around the world. And I want to build very much on the insightful comments that Dirk just made and that Mike Spence made a little bit earlier. As Mike said, it's rare that there's one good metric of the entire economy, unless it's perfectly correlated with the others. And so what I'm going to argue is that we could use some new metrics in particular to introduce this concept of GDP-B, where the B stands for benefits to help account for the value of new and free goods in the digital economy.
Now, as the others have noted, GDP is an incredibly valuable set of tools. Some have said it's one of the most valuable inventions of the 20th century. But it is often stretched beyond what it really can be used for. In particular, it's not a good way of measuring the welfare that we are experiencing. In fact, Simon Kuznets himself, the father of GDP, said that we shouldn't measure the welfare of a nation by GDP or the other national accounts. That's because GDP is a measure of production, not well-being. Furthermore, the other statistics that we rely on such as productivity are based on GDP, simply GDP per hour's work. So while GDP is very useful for understanding what's being produced in the economy, if we want to understand the value that consumers are getting in the welfare, it can at times be misleading.
Let me dive a little deeper on that briefly to compare GDP versus consumer welfare. There are many goods for which actually it's not a bad metric. For typical ordinary goods like cars and haircuts, GDP is not a bad approximation. It's correlated, as Michael said. Things can be, on this chart, the blue rectangles represent the spending on a good, the price times the quantity. And the green triangles represent the consumer surplus that we get from those goods. And as you can see, if you double the spending, you roughly double the consumer surplus. In other words, if GDP, if you get twice as many cars or haircuts or apples and oranges, it's not a bad approximation to say that your welfare has roughly doubled. And I think that was the heuristic that we used in much of the 20th century. We continue to use extensively, but it's increasingly misleading for digital goods.
Digital goods, like the free maps that we use or zoom here, or, Mike mentioned the digital diagnostics for the dermatologist now have available that could reach potentially billions of people through smartphones. Those kinds of free goods have zero marginal costs and don't contribute noticeably to GDP the way it's measured, but they could create enormous consumer surplus. So as we get more of those goods, we're not seeing it reflected in the GDP statistics. In fact, ironically, if you substitute a digital good for physical goods, let's say Wikipedia for encyclopedia Britannica, you may see GDP goes down, even as consumer surplus goes up. So when these things are not well correlated, it can be misleading for us to look at the GDP statistics as a measure of our well-being.
So what we've done and I've been working with Avinash Collis, Erwin Diewert, Felix Eggers and Kevin Fox to develop an alternative approach that looks directly at consumer surplus. And it's especially important for free goods and for new goods as I'll show in a minute. Fortunately, there are a set of tools that are now available in the digital economy that make it much easier to measure consumer surplus than it was even about 10 or 20 years ago. So we're using those tools to get at the value of these goods and incorporating them into a new metric called GDP-B. There's been an absolute explosion of these free goods in China and the United States. I don't have to list all the ones that we are using. A large percentage of our hours during the day are consuming these free goods. But if you look at the official US statistics that include digital services, as well as books, movies, newspapers, the share of GDP accounted for by what they call information goods was 4.7% back in the early 1980s. And then it bounced around a little bit. But recently it's been more or less still around four to 5%. So it's almost like this explosion of digital goods didn't exist. And that of course reflects the fact that GDP was not designed to capture these particular digital goods.
So if we want to measure them, we need alternative metric. And in the case of new goods, we have a similar problem where new goods like, say, smartphones can contribute enormously to our well-being, but it's well known that those also are not captured well in the national accounts. To give you an example, the number of photos taken, according to Varian, went from about 80 billion to 1.6 trillion by 2015, the 20 fold increase. At the same time, the price dropped to roughly zero from 50 cents. And those two facts mean that we weren't capturing the benefits of those new goods in our national statistics. They also, we also weren't capturing the benefits of the hardware itself. The cameras have largely disappeared.
20 years ago, a lot of people buy digital cameras. Hardly anybody buys digital cameras anymore because they've been incorporated into smartphones. In fact, there are so many things that are now incorporated into our smartphones and into some of our other tools. Here's an ad from the 1990s and virtually every one of these several dozen items that you used to buy separately are now available in a smartphone, whether it's a camera, clock, music player, game, machine movies, video cameras, and a lot of new things that never existed before, like GPS maps and payment systems. So what we need is a new framework for measuring these welfare changes. And we can derive an explicit term for the value of not only new goods, but also for free goods. And this NBR paper goes into more detail, but let me now show you some of the empirical implementation of this, about how we go about doing it.
One of the tools that we can use now are very large scale, massive online choice experiments, where we are able to reach literally hundreds of thousands or millions of people very cheaply with surveys like this one, asking them whether or not, at the bottom of the screen here, asking them whether or not they would give up Facebook and exchange for $5. And other people, we send a similar ad, but we asked them whether they would give it up for $20 or $50. And then we have a parallel set of experiments where we enforce incentive compatibility, which means that we only pay them if we can monitor and find that they have in fact given up Facebook for that month. And that makes sure that they're answering truthfully. We did this for a whole set of different goods, both with incentive compatibility and also like this survey, simply asking them. And we get now data on how people are behaving.
So for instance, here's the data with incentive compatibility for Facebook. So these are people who actually had to give up Facebook in order to get their payment. And the black dots there show the values they had in 2016. And then the white circles are in 2017. We've continued to do that. I should update this chart with more recent data. But what you see is that the median value about 37 or 38 dollars per month. That's how much we had to pay people in order for them to stop using Facebook. And it varied in systematic ways. And it might be consistent with your intuitions, people who spent more time on Facebook, who had more friends or posted more frequently tended to have higher valuations, as you might expect. We also found that women and older people tended to have higher valuations. I think some of the older people maybe didn't have as many alternatives like Instagram and Snapchat and other social tools.
And we found that some substitution that those people who use some of the stuff, Instagram and YouTube, had lower valuations. So it's very consistent. We've now done this for many different types of goods, on search, email, video. And what we find is that for each of them, we can do these demand curves of what people are willing to pay. And we get a sense of the value of many of these different kinds of goods. So the paper goes into more detail of what people's valuations are for a broad variety of different goods. So we can now do it for digital goods, but we can also do it for non-digital goods. We also get, of course, consumer surplus from breakfast cereal and cars and jet travel and an [inaudible] to a Bob Borden who I think is going to be speaking shortly.
I went ahead and we asked about toilets in the home, as Bob knows, six or seven years ago, we were giving the opening talk for the Ted Conference and Bob challenged the audience, whether they would rather give up indoor plumbing versus internet. And we did a vote there. And so I decided to scale that up to our population of a few hundred thousand respondents. And indeed, Bob's right that most people value indoor plumbing more than internet access. So I think I'm happy with that result. But you can see that there are many other kinds of valuations there that we can measure for all sorts of goods. Some of them digital, some of them non-digital. And these are the median valuations. Of course, there are some people who have much higher evaluations or lower valuations for any one of these particular items.
So ultimately, as Mike Spence noted, we want to have the dashboard of metric. And Dirk mentioned this as well and showed quite a few different options there. And I think one way to think about it is that there's a spectrum going from things like GDP productivity, which we can measure with great precision to seven or eight or nine significant digits, and that's great for the national accounts. At the other end of the spectrum are the well-being metrics, which typically are on a scale of, say, 1 to 10 where people ask about how happy are you today? Maybe it's an important question, but it's also one that we don't measure very precisely. And I see GDP-B as being somewhere in between a bit more precision because we can actually go good by good and see what people's valuations are, but also a little bit more relevant to overall well-being. As I mentioned, measuring the benefits of goods, not simply the cost of them. It’s getting closer to what we mean when we want to measure the welfare of a nation.
And this is described as well in one of the papers that, RG and I wrote. So let me conclude by saying that the GDP which was coming up almost a century ago has been the de facto metric of economic growth and almost all the economic studies that you see of countries. But as economists know, it really is not the measure of consumer well-being, and it's especially problematic as the economy becomes more and more digital and more and more of what we consume each day has zero price. So that means we need a new metric. Fortunately there are a set of tools where we can do massive online choice experiments to supplement these.
And we've already been using them quite extensively. We're scaling it up, not only for digital goods, but for non-digital goods to create a suite of metrics. And these are describing our papers where we are introducing this concept called GDP-B to measure the benefits. And we see it as a parallel tool alongside traditional GDP. And as the economy becomes more digital, I think it'll give us a more accurate sense of where the value is being created in our economy. I'll just point you to my website, where you can read more about this. We have one just focused on measuring the economy, the new digital economy lab that I started at Stanford now has a website. And I also have my own website where you can read some of the related papers. I'm looking forward to a lively discussion of all this. Thanks.
For more information, please visit Luohan Academy's youtube channel: Luohan Academy