27 May 2019

An economist explains how to value the internet


It is one the most commonly used measures of economic activity: gross domestic product (GDP), defined as the total market value of all final goods and services produced within a country in a given period.

But GDP misses out on huge chunks of value in the digital economy. When digital goods, whether Google Maps or Wikipedia, are available free of charge, they make no impact on GDP despite the value to their users.

This has major consequences. Without a valid tool to measure the value of the digital economy, policymakers are left scratching their heads over how to manage it. That led a group of economists at MIT to develop a new tool to measure the benefits of the digital economy.

At the World Economic Forum’s Annual Meeting in January 2019, we spoke to Erik Brynjolfson, Director of the MIT Initiative on the Digital Economy and Professor at MIT Sloan School about this new measure. Below is an edited transcript.

How does GDP count up the value of digital goods at the moment?

GDP is one of the great inventions of the 20th Century but it also has some weaknesses. In particular, anything with zero price has precisely zero weight in GDP: whether that's Wikipedia, or the apps on your phone, or the air that we breathe. This was a design choice but it's becoming a problem in the digital age.

Digital goods may have zero price but companies are getting an awful lot of value out of them. How do you explain that paradox?

Even though digital goods have zero price, we as consumers can get a lot of value from them. I benefit greatly from access to Wikipedia, or Google, or the apps on my phone. Likewise companies can make a lot of money from them. Some of them are volunteer organizations like Wikipedia, but others make billions or even hundreds of billions of dollars of valuation from selling advertising, or selling things directly on the internet. So clearly there's a lot of value being created, even if the goods themselves don't always show up in GDP.

Is that a problem?

The reason this is a problem is that these digital goods and these free goods are becoming a bigger and bigger share of the value we get in the economy. To some extent, we've always been missing parts of this, but with the advent of smartphones and the internet, this has become a bigger share of how we spend our time and where we get our value.

Take one example: if you replace Encyclopedia Britannica with Wikipedia, GDP actually goes down because you don't have to pay for anything anymore, but our value goes up. So if we're using GDP as a measure of our well-being, then we're going to be misjudging where the value is being created and we need some new metrics to capture it better.

Counting the cost: A table showing what people would least want to live without, from top to bottom
Image: PNAS, Erik Brynjolfsson, Avinash Collis, and Felix Eggers

How do you define a digital good? If I order something over the internet, is it necessarily a digital good?

The internet is affecting all kinds of goods. It's affecting the goods that we've always bought like food; you can buy that over the internet. But there's also new classes of goods, that are purely made out of bits, not atoms.

These digital goods have three characteristics that are very different from previous goods. They're virtually free, perfect and instant. What that means is the cost of making an additional copy of a digital good is basically zero; each copy is an identical, perfect replica of the original; and they can be distributed anywhere in the world instantaneously, or at the speed of light.

Those three characteristics – free, perfect and instant – were never used to describe earlier goods like apples, or cars, or haircuts, but they're ubiquitous for digital goods. And they present lots of value creation opportunities, but also some new measurement challenges.

So, for example, photography used to be made with an expensive chemical process. It cost about 50 cents each. Now there are about 100 times as many photos being taken but they can be distributed for zero cost, so they're disappearing from GDP. Similarly, music was on vinyl records or CDs, now can be distributed as bits.

Overall, if you look at the information share of the economy, which includes music, data, software, news, all those different kinds of information goods: in 1983 it amounted to 4.6% of GDP in the United States; now, with this explosion of digital goods, it's still… 4.6%!

Basically our official GDP measures have completely missed the information explosion. So, if we want to measure it, we need a new metric and that's what we're developing with my team at MIT.

How do we go about valuing these things?

We call the new metric GDP-B and the B stands for benefits. So, while traditional GDP measures the production cost of things, GDP-B measures the benefits we get.

The way we do that is we ask people how much we would have to pay them to give up a particular good. We've surveyed hundreds of thousands of people now and compared what their valuations are for goods like Twitter and Facebook, but also breakfast cereal, jet travel and many others.

In the case of Facebook, we offered people various different prices: some people $1, some people $10, some people $50, some people $100. Some people require even more to give up Facebook for one month. What we found was that the median person needed to be paid $48 to give up Facebook for one month. That suggests that there's an enormous amount of value from this free good.

I'm not surprised at how big that number is because the average Facebook user spends almost one hour per day using it. Some of us use it very little, others use it a great deal. But if you add up all the different valuations, it adds up to tens of billions of dollars of value that some people are getting collectively from Facebook that is completely missed in our conventional measures of GDP.

What are the implications of finding out that value?

The implications for this are profound because the old saying is: you can't manage what you don't measure. We haven't been measuring big chunks of the economy and understanding where people are getting real value.

We still need conventional measures of GDP for measuring the dollars that are flowing through the economy. If the Federal Reserve wants to know what's happening to wages and interest rates they really have to look at GDP. So we've got to keep traditional GDP.

But if we also want to understand where the value is coming, conventional GDP isn't going to be as helpful as it once was. That's why we need this new metric of GDP-B, to see where the value is being created, especially in the digital economy.

Is there a flip side to the research? You’re looking at the value people get from the internet, what about the dark side of the internet?

One of the things that we've discovered when we do this research is that when people put a value on something, you really think a bit harder about what that really means.

People say they will pay $48 for Facebook but is that because they're addicted? Maybe they're unhappier in the long run despite being very attached to their Facebook accounts. There are some reports that people in the long run end up being less happy when they use social media. It really starts suggesting that there might be a divide between what people are willing to pay and their true happiness.

To be fair, this was true of many goods in the past too. Whether it's cigarettes, or automobiles, or luxury handbags, the valuation that people are willing to pay isn't always what you might think is, or at least should be, their true valuation.

In research going forward, we're trying to get more deeply at those distinctions. The initial research is just looking at what people are actually willing to pay. But ultimately I think it's important, if you want to understand society's well-being, to consider their deeper valuations and the effects on their peers and on society as a whole. It's a very interesting research agenda going forward.

Do you think GDP will become obsolete at some point in the future?

Definitely not. GDP is still very useful. It measures something conceptually different from GDP-B.

Traditional GDP measures the production of the economy. But Simon Kuznets, who is one of the inventors of GDP back in the 1930s, essentially said, ‘Please do not use my measure as a measure of wellbeing’. It's for production not wellbeing and those are two different concepts.

We now need a new measure that focuses on wellbeing, especially in the 21st Century digital economy. I liken it to a car's dashboard. I've got one gauge for the speed that I'm going; there's another fuel gauge; there's the temperature; the air conditioning; maybe the oil pressure. It would be silly to try to roll those up into one aggregate number and say, what is my average “car-ness”? Instead, you need separate metrics for different concepts.

So, we still need traditional GDP to measure production, but now we have GDP-B to measure the benefits we're getting from production: a different concept.

What’s next on the research agenda?

Now that we've done the research for American consumers, we're hoping to ramp it up to other countries and see what the difference in valuations is in France, or South Africa, or Korea, or Brazil, or other countries. And we would welcome partners who would like to work with us to come up with this new, expanded measure of the benefits we're getting not just from digital goods but from all goods.

No comments: