Quantifying Digital Technology’s Impact on the Economy
Although digital technology has become a ubiquitous presence in society — with the Internet, artificial intelligence (AI), data, and machine learning now underlying how we watch TV, eat, exercise, and more importantly how we floss, brush our hair, and toast bagels — measuring its precise impact on the economy is difficult. However, in spite of this, every year dozens of national agencies and research institutes attempt to capture this metric. Why? Because although quantifying the impact of digital technology on the economy may be challenging, it’s a critical data point, known to have a significant impact on everything from GDP to employment to labor productivity.
Unlike estimating the impact of sectors dominated by physical goods, such as the agriculture, food, or automobile industries, the digital sector is harder to quantify because a large portion of it is not physical. It is also frequently an input providing efficiencies for other sectors, and many digital products are free to the user.
But there’s no question that digital technology is critical to economic activity; one estimate reports that the “Internet sector” was responsible for approximately $966.2 billion, or 6% of real GDP, in 2014 alone. As such, these estimates can inform investments, government policies, and regulations. And, what’s more, understanding digital technology’s current impact on the economy may help predict (and therefore prepare for) its future impact.
Consequently, what has resulted is a multitude of institutions and agencies attempting to quantify the economic impact of digital technology, using a multitude of methods to do so.
As shown in the table, the U.S. Bureau of Economic Analysis (BEA) looks at U.S. imports and exports of services as classified by the North American Industry Classification System (NAICS) codes, as well as foreign investment by U.S. companies; the Census looks specifically at transactions completed online, also according to NAICS codes; whereas other entities examine merger and acquisition information.
And, while there are clearly discrepancies in the tactics used to measure the digital economy, there appears to be a relative consensus on their limitations.
In the BEA’s “Digital Trade in North America” report the bureau concedes that it may be underestimating the extent of digital trade in North America since only trade involving monetary exchange is accounted for — meaning many cross-border, zero-dollar data transactions, despite being valuable to businesses and consumers, do not show up on the BEA estimates.
The U.S. International Trade Commission (ITC) echoes these concerns, arguing that the BEA’s methods fail to account for the “many commercially important aspects of the Internet [that] lack a set price.”
Imprecise Information Collection
Critique of these measurements also centers on how the data/information for these reports was collected which, in the U.S., as in many other countries, is collected primarily through surveys.
The ITC has characterized these surveys as “imprecise,” and specifically criticized them for failing to capture “inter-company communications within a global corporate group and their associated data flows across national borders.” As a result, the ITC states that this information “often [has] no reported value” (i.e. it isn’t helpful) when estimating the value of the digital economy.
Ed Gresser, Assistant United States Trade Representative for Trade Policy and Economics, offers a similar critique; he asserts that few governments collect data and report on “the way in which services move from buyer to customer.” As a result, Gresser asserts that it’s difficult to know “how much of our exports of services are done on-line.”
Gresser also points out that “no report offers data on the amount of information transferred in the course of trade.” In other words, current information collection methods fail to capture the scale of information (i.e., data) flows. Instead, Gresser states, information on data flows should be similar to the information on imports and exports collected by the Census — information on the weight and value of imports and exports is recorded.
Overly Broad Categories
Classification codes, like the North American Industry Classification System (NAICS), were created to help agencies collect, analyze, and publish statistical information about different industries in the economy. The BEA and Census (as shown in the table above) rely on NAICS codes, in particular, while the OECD relies on its own categorization method, BPM5.
Although these classification systems were designed to help agencies more easily quantify different sectors in the economy, they have become a focal point for criticism due to their usage when determining the economic value of digital technology.
The ITC, for one, argues that estimates like the BEA’s may be invalid because they rely on NAICS codes, which they regard as “a broad categorization of types of services.” They also assert that “the industry categories used in the BEA services data [NAICS codes] are based on the affiliate’s primary output, not taking into account other items produced by that affiliate or the parent company’s industry category.”
The Internet Association, a D.C.-based trade association representing Internet companies, issued a similar critique, stating that that NAICS has yet to develop a comprehensive range of classification codes that account for the the entire spectrum of activities carried out and enabled by the Internet.
The OECD also asserts that “the choice of components [i.e the categories used] and their weightings may emphasize some points and obscure others,” and that the data used to measure the digital technology’s impact on the economy lacks the high-level granularity “needed to measure how businesses and individuals use ICTs.”
Variety of Measurements
What’s more, the variety of estimation methods currently in use is, itself, identified as a key limitation to accurately assessing the value digital technology has added to the economy.
The NTIA points to the lack of standardization in definitions as one reason for this:
“A 2012 report from the Bureau of Economic Analysis (BEA) published estimates of the value of international trade in “digitally-enabled” services. In 2014, the Department’s Office of the Chief Economist and the Brookings Institution used a common definition to estimate the value of “digitally-deliverable” international trade. The European Centre for International Political Economy (ECIPE) estimated trade in “data-intensive services sectors.” Most recently, in 2016, the BEA released a report with estimates of “potentially ICT-enabled services trade.” In other reports, USITC categorized firms as “digitally-intensive” to study how they contribute to the economy and rely on the Internet, while McKinsey Global Institute (MGI) ranked sectors on their level of ‘digitization’ [emphasis added].”
A better approach to quantifying the industrial significance of digital technology may lay in recognizing that it is not an “industry” at all, but a general purpose technology that adds value to the entire economy. Stated otherwise, Internet-based sales are a measurable component of Internet economic activity, but does the existence of Walmart.com mean Walmart is in the Internet industry? If so, arguably everyone is in the Internet industry, since large enterprises and small businesses alike all utilize Internet services to increase productivity.
In other words, rather than describing a new “industry” that encompasses the Internet, AI, data, and other digital goods, resources might be better invested in calculating what portion of industrial activity can be attributed to them. Assessing value-add in this manner would be more consistent with these general-purpose technologies, and would ultimately provide policymakers more accurate and useful information.
The ITC seems to agree with this approach, recommending “measure[ing] the degree to which firms in a given industry category have adopted Internet technologies in their businesses—their ‘digital intensity.’” For example, this could include: “the proportion of online sales (e-commerce) to total sales; the share of total input purchases that are information technology (IT)-related; the proportion of employees in IT digital occupations; and the share of total IT spending directed to cloud services.”
The OECD offered similar recommendations: “take into account the indirect impact of the Internet” and “look at the impact that the Internet might have on all industries, and hence on rates of productivity growth and eventually GDP growth.”
AUSTR Gresser believes the key to capturing this impact lies with making “services trade figures… far more detailed than they now are.” To do this he suggests collecting information about “how much of [an industry’s] exports of services are done on-line,” and creating more nuanced industry delineations. For example,“scientific research in chemistry, aeronautics, and computer science rather than simply for “research and development,” or for bank lending, securities trading, and real estate transactions as opposed to “financial services.””