How can the digital economy move forward when the GDP underestimates the value of digital goods and services?
The Information Age has revolutionized how we shop, travel, and entertain ourselves — and yet we know little about how digitization has impacted the economy. That’s because the main gauge of economic growth, Gross Domestic Product (GDP), doesn’t capture much of the value created by “information goods.”
Current research by MIT professor Erik Brynjolfsson, Adam Saunders, and Avi Gannamaneni shows why GDP underestimates the value of digital goods and proposes methods to account for what’s missing that will be the basis for future work. GDP limitations were also discussed in the 2009 book, Wired for Innovation, and in earlier research reports.
Just how bad is the problem? According to the research team, the Bureau of Economic Analysis, which tabulates GDP, calculates that the information sector accounts for the same share of the U.S. economy as it did 30 years ago — between 4 and 5%. That hardly seems possible at a time when consumers and businesses have access to a vast and growing storehouse of information, apps, and other online tools. Better understanding of the GDP’s limitation is also useful as executives develop their own digital business capabilities and as they try to assess the value of those efforts.
How to Value Digital Goods
The researchers report that the root of the problem is prices. GDP mainly focuses on the market value of goods and services. That made sense when economists developed the measure during the Great Depression of the 1930s, when the world was mainly concerned with how many tons of steel or bushels of corn were produced. But digital products like Wikipedia, Google, Facebook, and YouTube — by their nature — are often free. So are many social marketing tools in use by businesses today. That makes them virtually invisible in terms of consumer purchases, though not in terms of value delivered.
But it’s not just a matter of adding these goods into the count: The researchers identify two theoretical issues that must be tackled in order to include digital goods and services in GDP. The first is the “production boundary” — the line between those activities that should be considered when adding up the total production of the economy and those that ought to be excluded because they’re just part of everyday life. The researchers note that if you buy a cup of coffee at a cafe, the sale clearly falls within the boundary. But if the same barista who whipped up the java at the store makes a cup at home for himself or his family, it’s outside the boundary.
Excluding digital goods from the production boundary is especially problematic in cases where consumers can swap free digital goods for market goods. Consider newspapers: As consumers shift to free, online news sources, GDP decreases — because nobody is handing over money at the newsstand for a bundle of paper.
The second issue identified by the researchers is more daunting: Even if the government wants to include something in GDP, it’s often left out if no one has a clear picture of its value. For example, black-market transactions aren’t counted, which means a business that wants to sell a good or service in a country with a thriving underground economy may have trouble assessing actual demand.
That’s not to say that the government only counts things with price tags. The researchers document a host of items counted in GDP that are not subject to market transactions or have no price. The government gets around that hurdle by “imputing” their value — usually by comparing the good or service in question to something with similar characteristics that does have a market price. These items — municipal services, for example — make up a big chunk of our existing GDP measure: Almost one out of every six dollars of GDP was imputed in 2014.
The federal government regularly updates how it calculates GDP to include new things and often starts with “satellite” accounts that allow it to experiment on gathering and analyzing information about a slice of the economy not previously part of GDP. In 2013, for instance, the government added a new category of private, fixed investment called “intellectual property products,” which includes research and development, artistic originals, and software.
Despite a variety of ways to value digital goods, each has weaknesses. Some researchers have tried estimating the value of Internet access, for instance, but that doesn’t address the growing value of the digital goods themselves. Others have tried gauging their value based on advertising revenue, but that means ignoring digital goods that are subsidized or not supported in any way by ads — and even for those that are, advertising revenue can be completely unrelated to the value of the associated content.
A variety of economics researchers have also tried imputing the value of digital goods. The value of a free news site, for example, would be based on the price of a news site with a paywall. But it’s hard to find a parallel for many digital goods and services. There are no products similar to a Google search that charges users, for instance. Yet another method is to estimate what it costs to produce digital goods. The problem with that is that many digital goods and services are produced by free labor, such as the millions of people who write Wikipedia articles. In order to use that approach, the government would have to assume the value of those articles is roughly equal to the hypothetical wages that would be needed to create and edit the pages.
The Choice Experiment Approach
In light of the difficulties with all these approaches, Brynjolfsson and his fellow researchers are exploring a more direct methodology — choice experiments, and specifically a technique called “conjoint analysis.” This is a tool often used by marketers who want to find out how buyers make tradeoffs among competing products and suppliers. The idea is to use this approach to determine how people value different features or functions of various free goods. Subjects for this research could be recruited through online tools, keeping cost per subject low and allowing results to be gathered from hundreds of subjects within an hour. A key challenge will be getting consumers to attach value to goods and services that they perceive as available for free or at very low price.
If the economists are able to study the full extent of the information economy, they could potentially identify hundreds of billions of dollars in benefits not measured in current GDP statistics. This measure could offer the added benefit of identifying which cities, industries, products and services are generating the largest share of this previously hidden value, and whether growth is slowing or accelerating. They could also finally test whether the information sector has held steady at 4 or 5% of the economy for the last 30 years — or if it has actually vastly grown once all of its free output is included. And that data would be a big boon for businesses deciding which areas of the digital economy are worth their investment dollars — and which aren’t.
A full version of the research brief can be found online here.