Staff Discussion Paper/Document d’analyse du personnel—2023-20

Last updated: September 20, 2023

Digitalization: Definition
and Measurement
by Guyllaume Faucher and Stephanie Houle

Canadian Economic Analysis Department

Bank of Canada
[email protected]

Bank of Canada staff discussion papers are completed staff research studies on a wide variety of subjects relevant to
central bank policy, produced independently from the Bank’s Governing Council. This research may support or
challenge prevailing policy orthodoxy. Therefore, the views expressed in this paper are solely those of the authors
and may differ from official Bank of Canada views. No responsibility for them should be attributed to the Bank.

DOI: https://doi.org/10.34989/sdp-2023-20 | ISSN 1914- 0568 ©2023 Bank of Canada

Acknowledgements
We offer a special thank you to Anna Shatalova and Ron Rautu for producing many of the
charts and tables in this report. Our work benefited greatly from comments provided by the
Economic Analysis Division at Statistics Canada. We also thank the entire Digitalization
Overview series project team: Tatjana Dahlhaus, Christopher Hajzler, Vivian Chu, Alexander
Chernoff, Gabriella Galassi, Jeffrey Mollins, Temel Taskin, Pierre-Yves Yanni and Alexander
Ueberfeldt for comments and feedback on this paper.

i
Overview
How has digitalization shaped the landscape of Canada’s economy? In this paper, we seek to
grasp the breadth of digitalization. We define digitalization and distinguish it from other related
concepts, such as information and communication technologies (ICT) and the digital economy.
We then assess the scope of digitalization in Canada by looking at measures based on outputs
and inputs as well as the challenges that come with trying to estimate it. Finally, we raise
important questions for future research about the impacts of digitalization.

Key messages:
• Digitalization is transforming processes and interactions between economic
agents. The output of digital products was 5.5% of gross domestic product (GDP) in
2019, and the number of digital-related jobs was 7.1% of GDP in 2015. These metrics
reveal similar proportions of the digital economy in Canada’s overall production
process.
• Challenges remain when it comes to accurately measuring all aspects of the
effects of digitalization:
o Free or reduced-price digital products are increasingly driving a wedge
between GDP and consumer welfare that was estimated to represent 2.3% of
GDP for Canada in 2019.
o Intangible capital represents a growing share of the capital holdings of
Canadian firms. The national accounts still do not capitalize many forms of
intangibles, such as:
 product development, brand equity, human capital and organizational
capital (augmented by digitalization)
 data, which were estimated to represent about 5% to 8% of gross fixed
capital formation in 2018
o The adoption of cloud computing technology means that many Canadian firms
are substituting away from in-house investment and toward purchasing cloud
services. This substitution may help explain some of the under-investment seen
in Canada relative to that in the United States.
o The gig economy may be leading to the underestimation of the labour force
participation rate or work hours. Most recent estimates suggest gig workers
represented about 8.2% of Canadian workers in 2016.

1
1. How we define digitalization
The digital revolution—which involves shifting production processes from using predominantly
mechanical technologies to using analogue and digital technologies—began in the latter half
of the 20th century. This transformation led to the fourth industrial revolution, which included
automation, computerization and the shift of investment from tangibles to intangibles. This
revolution continues to impact and reshape the Canadian economy. In this paper, we explore
the technologies that drive digitalization, how we measure them, the ambiguities they entail
and how Canada’s adoption of them compares with other countries.

Digitalization’s boundaries are often unclear relative to ICT, digital adoption and the digital
economy. We define economic digitalization as the process through which the use of data,
digital platforms and advanced analytics (such as machine learning and artificial
intelligence) transforms the production process and interactions with other economic
agents. 1 This definition captures both the drivers and the economic impacts in accordance with
the existing literature. Since most of the related technological advancements are in their early
stages, we anticipate that this definition will evolve over the coming decades.

We differentiate digitalization from ICT and the digital economy. According to the fourth
revision of the International Standard Industrial Classification of All Economic Activities, ICT is
more formally known as the sector in which “the production of goods and services of a
candidate industry must primarily be intended to fulfill or enable the function of
information processing and communication by electronic means, including transmission
and display” (UNSD 2008, 278). Hence, digitalization encompasses some parts of ICT, such as
ICT services, but not others, such as ICT hardware. 2

Digitalization also goes beyond the pure digital economy as defined by the common
framework of the Organisation for Economic Co-operation and Development (OECD). Here, the
digital economy “incorporates all economic activity reliant on, or significantly enhanced
by the use of digital inputs, including digital technologies, digital infrastructure, digital
services and data. It refers to all producers and consumers, including government, that
are utilising these digital inputs in their economic activities” (OECD 2020, 5). Once again,

1
In a similar spirit, the European Central Bank (2021, 9) defines digitalization as “including, inter alia, a wide range of
information and communication technologies enabling automation and robotisation, and technologies related to
the processing and analysis of digital data, including big data, such as artificial intelligence and machine learning,
and edge and quantum computing.”
2
While ICT hardware is technology that enables digitalization, it is not part of it—just as the electricity network and
power plants enabled electrification technologies but were not part of that technological revolution.

2
digitalization captures some components of the digital economy that include digitally delivered
products, but not the infrastructure or telecom services associated with it.

Digitalization comprises all the processes enhanced by the use of digital inputs, which could
be digital labour or digital capital. However, some aspects of digitalization are difficult to
quantify, and the impact of these extends far beyond traditional economic measures. For
example, the field of health services sees benefits of digitalization that do not fit conventional
measures. These benefits include the use of personalized genomics to customize cancer
treatment for patients or handheld scanners that use machine learning to easily detect a
cancerous melanoma. So while ICT and the digital economy capture the number of computers
in the economy, digitalization also considers all the technologies they enable. Important to
consider are the diffusion effects of these technologies to other sectors that may not be directly
measured as part of the digital economy or ICT measures. Some of these benefits can be
difficult to measure in national accounts.

In the remainder of this section, we touch on the scope for digitalization to change the
economic future of Canadians. In section 2, we go over some of the broader characteristics of
digitalization and how we measure its impact. Section 3 discusses particular challenges
associated with measuring digitalization. We conclude in section 4 by outlining important
future trends and open questions.

1.1 Digitalization’s role as a general purpose technology

Some technologies are more important than others, and the potential of a technology in its
early stages is hard to assess. Considering historical precedence, economists have identified
three characteristics that are critical for a technology to have a major impact (Jovanovic and
Rousseau 2005; Bresnahan 2010):

• pervasiveness, meaning that the technology diffuses to other sectors

• the ability to generate ongoing technical enhancement or, more generally, to

improve over time

• the tendency to enable innovation complementarities, meaning the technology can

precipitate the creation of other new products and processes

A general purpose technology (GPT) is one in which all three features come together. GPTs can
drive entire eras of technical progress and economic growth (Bresnahan and Trajtenberg 1995).
Well-known and generally accepted examples of GPTs over history include the steam engine,
electricity and, more recently, computers. These technologies reshaped the world once they
were widely adopted.

3
Evidence in the literature suggests that digitalization is a GPT because it is pervasive, it
improves over time and it spawns innovation. An open question is what aspects of
digitalization can be considered GPTs. Because of how rapidly digitalization continues to
evolve, assessing which of its underlying technologies will be key remains difficult.

Many aspects of digitalization could be considered GPTs, including artificial intelligence,

machine learning, big data and cloud computing. Recent research by Goldfarb, Taska and
Teodoridis (2023) presents a systematic approach to determining which of these should be
considered GPTs. They analyze and rank 21 different emerging technologies using a
quantitative approach.3 Other studies look at particular technologies to assess whether they
are GPTs, such as robots (Dixon, Hong and Wu 2021) and artificial intelligence (Brynjolfsson,
Rock and Syverson 2021). The latter study also shows evidence that the introduction of a GPT
will likely cause an initial underestimation of productivity growth, since early adoption of
technology is typically slow, with productivity quickly accelerating after the technology is
adopted.

Recently, Petralia (2020) proposed using patent data to identify GPTs. Using patent counts as
a GPT indicator follows the existing literature on that topic, such as Hall and Tratjenberg (2004)
in the context of ICTs and Moser and Nicholas (2004) in the context of electricity. However,
Petralia (2020) offers a more systematic framework. He implements a three-dimensional index
that uses patenting growth rates to measure improvement, applies a text-mining algorithm to
identify the range of uses and, finally, uses co-occurrence of technological claims in patents to
assess complementarity with other technologies. This approach allows him to evaluate the
pervasiveness of each GPT technology and to produce a ranking.

Given digitalization’s GPT features, we argue that it should be considered general

purpose. Historical examples show similar technologies transforming processes and reshaping
the structure of the economy. While assessing the current stage of digitalization is difficult, its
importance is clearly increasing and an adoption wave of digital technologies is underway. In
another paper in this Digital Overview series, Mollins and Taskin (2023) discuss the likely
productivity potential of digitalization as a GPT. In the next section, we highlight the importance
of measuring digitalization’s impact and its evolution.

3
See Goldfarb, Taska and Teodoridis (2023) for the complete list of technologies they consider in their analysis as well
as their ranking. To quantitatively rank the emerging technologies, they use information related to job postings and
map emerging technologies to skills needed along the three common characteristics of a GPT. They find that
machine learning, alongside a set of complementary technologies (business intelligence, big data, data mining, data
science and natural language processing) consistently ranks at or near the top, suggesting it is likely a GPT. However,
during the period they consider, most of the other emerging technologies in their dataset are unlikely to be on the
path to becoming GPTs.

4
2. How we measure digitalization
Assessing digitalization’s size and impact is challenging because of its ubiquitous nature.
Nonetheless, measuring it remains important since its adoption affects many aspects of the
economy, such as productivity, labour markets and inflation.

Digitalization can be measured in different ways. In this section, we focus on three approaches
to estimate the scope of digitalization in Canada:

• using output-based measures, which estimate the value of products or outputs from
the digital economy
• measuring the capital type used in production, both tangible and intangible

• using employment-based measures, focusing on the jobs associated with digital

activity

What we know about digitalization in Canada depends largely on the work done by Statistics
Canada. Statistics Canada’s satellite accounting for the digital economy uses the output-based
approach to measure digitalization. 4 Its approach to estimating software, research and
development (R&D) and intellectual property products measures the capital associated with
digitalization. And its estimates for data, database and science exploit the labour cost approach.
These approaches focus on the more easily quantifiable aspects of digitalization at the expense
of leaving out aspects that are potentially sizable but less cleanly attributable. We augment
Statistics Canada’s measures with related measures from the literature.

2.1 Measuring output of the digital economy

The digital economy satellite accounts were released in 2019 to present estimates of the
output value, GDP and jobs associated with the digital economy (Statistics Canada 2019a).
These satellite accounts were replaced by the digital supply and use tables in 2021. 5 They do
not represent a change in the core national accounting framework but simply present a
reorganized breakdown of the supply and use tables (SUTs) into new categories considered to
be part of the digital economy. 6 Other countries currently estimate the digital economy but
without a common definition, making cross-country comparisons challenging (Moulton,
Tebrake and Tovar 2022). Statistics Canada adopted a modified version of the framework

4
Satellite accounting involves rearranging the classification of data to better understand an activity or sector, such as
digital transactions (Statistics Canada 2019a).
5
Statistics Canada, “Digital supply and use tables, 2017 to 2019,” The Daily (April 2021).
6
The purpose of the SUTs is to trace the production by domestic industries (including imports) through their use as
intermediate inputs, final consumption products, investments or exports. In this case, satellite accounting is used to
disaggregate and recompile the information from the SUTs into new categories (Statistics Canada 2019a).

5
outlined by the OECD (2020). It identifies “full” and “partial” digital products in the SUTs to
measure the value of digital activity (output, GDP and jobs) in Canada by province and industry.
These are broken down into three categories:

• Digitally enabled technology. This includes computer hardware, software, telecom

equipment and services, support services, structures used to produce digital economy
goods and services and the internet of things (devices and objects that can be
controlled through the internet).
• Digitally ordered transactions (e-commerce). This is when the sale of goods or
services is done through the internet.
• Digitally delivered products. This is when content is transmitted and consumed in a
digital format.

Nominal GDP associated with digital economic activities in Canada totalled 5.5% of the
total economy in 2019, or $118 billion (Statistics Canada 2021). From 2010 to 2019, the
digital economy grew roughly 40% faster than overall GDP. Chart 1 shows the size of the
digital economy by industry in 2018 according to use of these digital products. Unsurprisingly,
services industries—information and cultural industries, wholesale trade, retail trade and
professional services—dominate production in the digital economy.

Chart 1: Size of the digital econom y output by business sector industries in 2018

Information and cultural industries

Wholesale trade
Retail trade
Professional, scientific and technical services
Manufacturing
FIREL
Construction
Administrative and support, waste management and…
Other services (except public administration)
Educational services
Transportation and warehousing
Arts, entertainment and recreation
Accommodation and food services
Utilities
Mining, quarrying, and oil and gas extraction
Management of companies and enterprises
Health care and social assistance
Agriculture, forestry, fishing and hunting
0 20 40 60 80 100
Can$ millions
Note: The chart shows output of digital products as defined in Annex A in Statistics Canada, "Measuring digital economic activities in Canada: Initial estimates," Catalogue No.
13-605-X (May 2019). For our calculations, we have used supply-use table 2018 (15-602-X)—supply of digital products by sector. FIREL stands for finance,insurance, real
estate and leasing.

Sources: Statistics Canada and Bank of Canada calculations

6
One of the drawbacks of using these digital economy satellite accounts is that they were only
recently developed and therefore do not cover a long enough time horizon to analyze the
evolution of the digital economy from its origin. 7 For example, the digital economy satellite
account data start only in 2010 for Canada and in 2005 for the United States (see Chart 2 for a
growth comparison), but the digitalization process started earlier. Also, the SUTs that these
accounts use are often released with a lag of a few years and only annually. The data they
provide are also available only in nominal terms, and therefore researchers need to make
additional assumptions when using them to track real activity. We discuss further challenges
related to output-based measures in section 3.

Chart 2: Grow th in the digital econom y's gross dom estic product, Canada and the United
States

%
9

3
2011 2012 2013 2014 2015 2016 2017
Canada United States
While both Canada and the United States follow the framework provided by the Organisation for Economic Co-operation and Development, their measures
have some methodological differences.
Sources: Statistics Canada and Bureau of Economic Analysis Last observation: 2017

2.2 Measuring the capital type used in production

Since the early days of the internet, statistical agencies have been adjusting how they create
the national accounts to incorporate the digital economy into their measurements. In this
section, we first outline Statistics Canada’s incorporation of capital goods in investment
measures, what Statistics Canada has incorporated (software and R&D) and what it is currently
working on incorporating (data). Then we look at additional measures of capitalization, such as
the digital capital stock, robots and technology adoption.

7
The way Statistics Canada compiled estimates also changed between when the initial estimates were published
(Statistics Canada 2019a) and when they were released in SUTs (Statistics Canada 2021).

7
Early days: Capitalizing software and research and development
The 1997 revision of the Canadian System of National Accounts, implemented in 2001, was the
first step to incorporating the intensifying digital landscape. The machinery and equipment
investment category of capital expenditures was expanded to include three types of software:
pre-packaged (licensed), custom-designed and own-account. Software was previously treated
as a current expense, meaning that it would be consumed within the same period as an
intermediate input, which is not included in value added. As a result, GDP was raised by the
amount of software investment done by businesses, net investment in hardware that was
already counted. Estimates showed that the new accounting treatment for software raised
GDP by $10.2 billion in 2000 (Jackson 2003). This new accounting also revealed a higher share
of GDP over time, from 0.3% of the revised GDP in 1981 to 1.0% in 2000.

The R&D satellite account was launched in 2008 following the recommendations of the United
Nations’ System of National Accounts (SNA) (United Nations 2008). It was initially launched to
explore conceptual issues around R&D’s capitalization and inclusion in core accounts. Before
this, expenditures in these R&D categories were not capitalized in Canada or internationally,
other than for software R&D. Barber-Dueck (2008) developed estimates from 1997 to 2004 and
found that additional R&D capitalization raised GDP by $20.4 billion in 2004 and that total
R&D capitalization represented 2.9% of GDP in 2004.

The 2012 and 2015 SNA revisions brought additional changes. Software and R&D were
reclassified under a new investment category: intellectual property products. Also, following
the publication of the R&D satellite accounts in 2008, expenditures on R&D were reclassified
from intermediate consumption to gross capital formation (investment—as recommended by
SNA 2008). This meant that R&D changed from being used in production in the same year it
was purchased or built to being capitalized over many years. This led to upward revisions of
both the expenditure-based GDP, due to business investment increasing, and income-based
GDP, due to gross operating surplus increasing. Figure 1 presents an evolution of Statistics
Canada’s efforts to include capital-based measures of digitalization.

Figure 1: Timeline of Statistics Canada’s inclusion of digital capital

Capitalization of software Creation of the intellectual property

(2001) investment category (2012)

Research and development Data, databases and data science experimental

satellite account (2008) estimates (2019)

8
Modern quandary: Incorporating data
More recently, Statistics Canada released a conceptual framework to estimate the value of data,
databases and data science (data-related products) in the Canadian economy (Statistics Canada
2019c). Data-related products can be either produced by the firm (own-account) or sold on the
market. Since Statistics Canada does not have comprehensive information on market sales of
data-related products, it imputes the value of own-account data-related products using the
cost of production (labour costs with a return to capital markup). It provides lower- and upper-
bound estimates to account for the fact that even though workers’ jobs may be related to the
production of data-related products, workers likely do not spend all their time on this. Total
gross fixed capital formation for data-related products in 2018 was between $29 billion
and $40 billion at current prices (5% to 8% of total gross fixed capital formation for
Canada in 2018 prices).8 Using the perpetual inventory method, Statistics Canada also
estimated the net capital stock of data-related products to be between $157 billion and
$217 billion as of the end of 2018. However, it noted challenges with estimating the data-
related capital stock, which we discuss in section 3.

Services sectors: The most intensive in digital capital

The stock of digital capital is currently ill-defined, and estimating its value is difficult. Examples
of challenges include the definition and value of intangible capital and data-related products,
which are both discussed in the section on challenges in measurement. In this section, we follow
the definition and measure developed by Liu and McDonald-Guimond (2021). Specifically, they
use both the National Accounts Longitudinal Microdata File (NALMF) and Statistics Canada’s
capital, labour, energy, material and service (KLEMS) datasets to build an ICT capital metric. This
measure—the ICT capital intensity metric—represents the ratio of the volume of digital capital
over the volume of productivity-enhancing capital across industries. Using this metric, Liu and
McDonald-Guimond (2021) determine which industries use digital capital more intensively and
assess the evolution of its use over time. By using productivity-enhancing capital as the
denominator instead of non-residential capital stock, they can control for the fact that some
sectors have a higher share of capital in non-residential buildings than others.

Liu and McDonald-Guimond (2021) find that, from 2013 to 2015, ICT capital intensity
was higher in the services-producing sectors than in the goods-producing sectors, with
the ratio of ICT capital over total capital averaging 0.19 and 0.08, respectively. In terms of
evolution, compared with 2000−02, the ICT intensity metric increased in all industries but at a
faster pace in the services-producing sectors. Nonetheless, the results are driven mostly by a

8
We used Statistics Canada Table: 36-10-0108-01.

9
few industries with a high intensity, as shown by the median at only 0.03 and 0.09, respectively,
much lower than the average, across all industries for the two time periods considered. 9, 10

Automation: Increasing robot adoption

Robots are not new. Their adoption started to reshape the economy and labour supply before
the beginning of the fourth industrial revolution around the start of the 21st century. A
distinctive feature of this fourth revolution highlights the difference between automation and
robots. Automation is a series of technological processes that allow machinery to do tasks
that would otherwise be performed by humans, while robots are physical machines that
execute predetermined programs. The complexity of automation increased significantly over
time and now includes technology’s the ability to learn and make decisions in more
sophisticated ways. In another paper in the Digital Overview series, Chernoff and Galassi (2023)
discuss labour market implications and the impact of robots and automation on labour
reallocation across occupations and sectors. In this section, we focus on the measurement of
the robot stock.

The number of robots in the economy could be a proxy to assess the evolution of automation
in an economy. As mentioned, robots and automation are not interchangeable, and some
challenges exist in directly measuring automation adoption. Dixon, Hong and Wu (2021)
estimate the Canadian aggregate robot stock from 1996 to 2017 using data from the Canadian
Border Services Agency that identify robots imported by Canadian firms. 11 Their paper finds
that the value of aggregate robot stock in Canada was $1.6 billion in 2017. One of the
drawbacks of using import data is that these data do not provide a full picture of where the
robots are actually used within the country. For example, robot imports may be handled by
robot wholesalers or a parent firm that in turn ships the robots to plants elsewhere.

Liu and McDonald-Guimond (2021) also examine the evolution of robot intensity in Canada.
They extend Dixon’s (2020) work with the Canadian Border Services Agency data by creating a

9
From 2013 to 2019, the industry with the highest ICT capital ratio in goods-producing sectors was computer and
electronic product manufacturing (ratio of 0.222), and in services-producing sectors, it was advertising, public
relations and related services (ratio of 0.478).
10
Liu and McDonald-Guimond (2021) also use other metrics, such as digital labour and robots, to build a composite
index. They find that the high intensity goods-producing sectors are computer and electronic products
manufacturing; machinery manufacturing; transportation equipment manufacturing; and utilities. The highest
intensity services-producing sectors are design, computer systems, management services; architectural, legal,
accounting, engineering, and other professional services; information services; broadcasting and
telecommunications; and advertising.
11
Dixon et al. (2021) argue that Canada is not a meaningful producer of robotics hardware domestically and therefore
the quantity imported, measured by Canadian Border Services Agency data, is a relevant proxy. To estimate the
capital stock, the authors use a depreciation rate that assumes a useful life of 12 years, as stated in International
Federation of Robotics guidance.

10
metric for the intensity of adoption across industries. Specifically, this metric represents the real
value of robot stock in a given industry divided by employment from the Labour Force Survey
in the same industry. Comparing two samples (2000–02 with 2013–15), they find that robot
adoption increased in all industries analyzed except for wholesale trade, where it
remained mostly unchanged. 12, 13
The International Federation of Robotics also provides data on the stock of robots by country
and by industry. Using the IRF data, the European Central Bank (2021) shows that global robot
diffusion, measured as the number of robots per million hours worked, is only at about 0.75.
Canada is well behind other regions such as Japan, the United States and the euro area. In fact,
Canada outperforms only the United Kingdom among the 10 regions considered. The robot
density, measured by the number of robots installed in the manufacturing industry per 10,000
employees, increased during the COVID-19 pandemic. The advantage of the International
Federation of Robotics data is that the federation, unlike most other sources of data, directly
measures the adoption of digital technologies instead of relying on some indicators that
measure adoption of technologies only indirectly. However, as mentioned in Aghion et al.
(2022), the International Federation of Robotics data have some limitations—for example, they
are available only at the country level (which does not allow for provincial comparisons in
Canada) and from only 13 manufacturing industries.

Latest advances: Prevalent technology adoptions

In its Survey of Digital Technology and Internet Use, Statistics Canada provides estimates to
assess the impact of digital technologies on businesses. This survey includes measures of the
value of e-commerce, the adoption of technologies and the use of the internet. The 2021
release highlighted the implications of the COVID-19 pandemic by comparing the evolution of
these indicators relative to 2019.

This survey also assesses trends in the adoption of various emerging technologies. Each firm
provides a list of the ICTs they use, including cloud computing, robotics, 3D printing, big data
and blockchains. Between 2019 and 2021, the largest increases were in cloud computing
(from 39% to 45%) and industry-specific software (from 40% to 46%). Overall, the
number of businesses that mentioned they do not use any ICT decreased from 20% to
15% (Statistics Canada 2022). In terms of emerging technologies, the adoption of artificial

12
Liu and McDonald-Guimond (2021) measure robot intensity in millions of chained 2007 dollars per 100 employees.
13
Wholesale trade had the second highest level of robot intensity between 2000 and 2002, after machinery
manufacturing. Most of the robot adoption in the wholesale trade sector could have taken place before 2000.

11
intelligence and 3D printing increased by 1.4 and 0.3 percentage points (pps) respectively, while
blockchain, big data and advanced robotics were relatively unchanged.

2.3 Measuring digital work in the labour force

A growing digital workforce

A third way to measure digitalization is by looking at the number of employees working in
digital occupations based on Canadian SUTs. Statistics Canada determines a digital output ratio
for each product in the digital economy (described in section 2.1). It then multiplies this ratio
by total jobs to estimate a proxy of the number of digital economy jobs. According to this
metric, close to 900,000 jobs were associated with digital economy activities in 2017. 14 This
represents about 4.7% of all the jobs in Canada responsible for producing 5.5% of
nominal GDP in that year. Between 2010 and 2017, jobs in the digital economy grew at a
pace of 37%, much faster than the 8.6% growth rate of the total economy for the same
period. Using data up to the end of October 2021, Bellatin and Galassi (2022) show that this
trend has continued during the pandemic period. They find that job postings in digital
infrastructure outpaced the postings in the rest of the economy after the first wave of the
pandemic. In fact, the gap between the two growth rates exceeded 40 pps by May 2021.

Lamb and Seddon’s (2016) results, based on a slightly different definition, point in the same
direction, whereby the tech sector represented 5.6% of employment and 7.1% of GDP in
2015. They define the tech sector more broadly than just the companies operating in ICT and
include industries such as aerospace manufacturing and pharmaceutical and chemical
manufacturing. They also find that 55% of employees in this sector are working in ICT.
Architecture, engineering and design accounted for most of the remaining non-ICT workers,
with a share of 25%. Tech employees tend to be among the most highly educated members of
the labour force, with over 50% of them having a university degree. They also earn the third
highest annual wage compared with employees in other industries, just below mining, oil and
gas, and utilities (Lamb and Seddon 2016).

Digital workforce intensity

Liu and McDonald-Guimond (2021) build a measure of the intensity of the digital workforce
that allows us to compare various industries in Canada. They define this measure as the ratio
of workers employed in the digital occupations over the sum of all workers. They find that the
industries with the highest digital intensity are computer and electronics manufacturing,

14
This metric includes both paid and self-employed jobs associated with the production of digital output. However, it
doesn’t consider employees working in a digital job but in an industry without any digital output. For example, it
would not include a web designer who is hired by a bakery and then develops a website for online orders.

12
professional services, and information, culture and telecommunications (Liu and
McDonald-Guimond 2021).

The challenge in creating such a metric is to determine the list of occupations that could be
considered digital. This difficulty could lead studies to exclude people who perform some
digital tasks due to the fact that they are employed in an occupation that is not classified as
digital. The literature shows clear consensus on classifying some jobs as “digital,” while others
remain ambiguous (Table 1). To obtain their workforce-intensity metric, Liu and McDonald-
Guimond (2021) use a definition of occupation that is slightly narrower than Lamb and
Seddon’s (2016) mentioned above, but they add occupations not included in previous studies.

Table 1: Selected digital occupations

Brookfield
OECD McKinsey Liu and
(Lamb
NOC (Calvino (Manyika McDonald-
and
code et al. et al. Guimond
Seddon
2018) 2015) (2021)
Digital occupation included in each study 2016)
Telecommunications carrier managers 131 Yes Yes Yes Yes
Engineering managers 211 No No Yes Yes
Computer and information systems managers 213 Yes Yes Yes Yes
Statistical officers and related research support 1254 No No No Yes
Mechanical engineers 2132 No No Yes Yes
Electrical and electronic engineers 2133 Yes Yes No Yes
Industrial and manufacturing engineers 2141 No No No Yes
Metallurgical and materials engineers 2142 No No Yes Yes
Aerospace engineers 2146 No No Yes Yes
Computer engineers 2147 Yes Yes Yes Yes
Mathematicians, statisticians, actuaries 2161 No No Yes Yes
Information systems analysts and consultants 2171 Yes Yes Yes Yes
Database analysts and data administrators 2172 Yes Yes Yes Yes
Software engineers and designers 2173 Yes Yes Yes Yes
Computer programmers and interactive media developers 2174 Yes Yes Yes Yes
Web designers and developers 2175 Yes Yes Yes Yes
Industrial and manufacturing technologists 2233 No No No Yes
Electrical and electronic technologists 2241 No No No Yes
Technical occupations in geomatics and meteorology 2255 No No No Yes
Computer network technicians 2281 Yes Yes No Yes
User support technicians 2282 Yes Yes No Yes
Information systems testing technicians 2283 Yes Yes No Yes

13
 Graphic art technicians 5223 No No Yes Yes
Graphic designers and illustrators 5241 No No Yes Yes
Supervisors, electronics manufacturing 9222 No No No Yes
Note: NOC is the National Occupational Classification; OECD is the Organisation for Economic Co-operation and Development. New unit
groups have since been created for emerging occupations, such as data scientists and cybersecurity specialists. See Employment and
Social Development Canada, “Changes to the National Occupational Classification,” (June 2023) for details. Data for this table are from
H. Liu and J. McDonald-Guimond, “Measuring digital intensity in the Canadian economy,” Statistics Canada Economic and Social Reports,
Catalogue No. 36-28-0001 (February 2021).

2.4 Comparing other advanced economies

Assessing the size of the digital economy could also be done by directly analyzing the diffusion
of digital technologies. Researchers across different institutions have created many
international indexes and indicators to measure specific aspects of digitalization, allowing for
cross-country comparisons. For example, the European Commission tracks data on broadband
access and coverage, the size of e-commerce and employment in the ICT sector. 15 Some of
these metrics are used to construct the International Digital Economy and Society Index
(I-DESI), an international index that ranks countries along many dimensions related to
digitalization. While many indicators exist in the literature, we focus on three: the I-DESI,
the ICT Development Index (IDI) and the Network Readiness Index (NRI).

Some common conclusions emerge from these indicators. One is that digital public services—
which include e-government, legislation and open public data—are a strength in Canada.
Some differences also emerge—for example, the divergence between the NRI and I-DESI on
the human skills dimension—but these are due mainly to the methodology and choice of
variables. Table 2 provides a brief overview of each indicator, with additional details in the
Appendix.

15
For a complete list of the indicators, see the “ICT sector” in the Key Indicators of the I-DESI (European
Commission and Tech4i2 2020).

14
Table 2: Summary of international indexes
Index Description Geography Canada’s ranking Canada’s strengths Canada’s weaknesses
Canada has a high
citizen use of internet
score for its number of
internet users and
Measures digital
activities performed
economy
online, such as: Canada performs poorly in
performance based  video calls the human capital
on 24 indicators
2018 normalized  social networking dimension for its relatively
across 5 different Total of 45
score for Canada: 55  banking low:
dimensions: countries:
Canada’s rank: 11  shopping  share of employees
 connectivity  27 EU
I-DESI  human capital working in
member
2017 normalized Canada performs well telecommunications
 citizen use of states
score for Canada: 57 in the digital public  proportion of graduates
internet  18 non-EU
Canada’s rank: 18 services dimension for in information and
 integration of
its digitalization of communications
digital technology
public services, such technology (ICT)
 digital public
as:
services
 eGovernment
 online service
completion
 open data

Canada scores poorly in

Measures the Canada performs well
IDI (2017) some of its ICT access
evolution of ICT in the ICT skills
Canada’s score: 7.8 dimension, particularly for:
development using dimension, in
Canada’s rank: 29  fixed-telephone
11 indicators (2017) particular for its:
IDI 176 countries subscriptions per 100
across 3 sub-indexes:  tertiary enrolment
IDI (2016) people
 ICT access ratio
Canada’s score: 7.6  mobile-cellular
 ICT use  mean number of
Canada’s rank: 26 subscriptions per 100
 ICT skills years of schooling
people

15
 Canada performs poorly in
For each of the 4
its impact dimension (20th
dimensions and 12
place), specifically in the:
Measures the impact sub-dimensions,
 economy sub-dimension:
and application of ICT Canada scores higher
o growth rate of GDP
with the use of 60 than the high-income
NRI (2021) per person engaged
selected indicators group average.
Canada’s rank: 11 o ICT services exports
across different
NRI 130 countries
dimensions: Among the 4
NRI (2020)  sustainable development
 governance dimensions, Canada
Canada’s rank: 13 goals contribution sub-
 technology performs best in:
dimension:
 people  governance
o affordable and clean
 impact (6th place)
energy
 technology
(9th place)

Note: I-DESI is the International Digital Economy and Society Index; IDI is the ICT Development Index; and NRI is the
Network Readiness Index.

3. Challenges in measuring digitalization

In the previous section, we discussed several ways to measure digitalization and its
components. However, challenges remain in accurately measuring all its elements. This section
explores first how digitalization introduces difficulties in measuring output due to free digital
products. Then we discuss digitalization’s impact on measuring capital in light of the rise in
intangible digital forms of capital, followed by how it changes the measurement of labour
because of the gig economy. Finally, we investigate how digitalization affects the determination
of prices.

3.1 Missing “free” digital products widen the gap between

GDP and surplus
The rise in digitalization has enabled many digital products to be easily accessible to and
customized for every consumer. An output-based measurement of digitalization is difficult
mainly because of free or reduced-priced digital products. It is also exceedingly difficult to
distinguish between what should be measured as increasing value added and what should be
measured as enhancing welfare. Many online platforms offer free products to users in exchange
for database information or advertising revenues. These products are often offered at no
charge or a reduced price, with companies funding their production through different methods.
Some digital products are freely accessible through crowd-production schemes where users
consume and build the product, such as Wikipedia and Stack Overflow. Others are funded
through building customer databases or advertising revenue, and often both, such as most

16
social media websites (e.g., Facebook, Instagram and TikTok) and many Google products and
other streaming services (e.g., Maps, Calendar, YouTube and Spotify). 16

Offering free versions of a product is also a common strategy for sellers of high-quality, low-
marginal-cost products, such as computer software and certain digital applications and
services. Free versions serve as a partial information signal of the quality of the product because
price is not always a sufficient indicator (see Bourreau and Lethiais 2007). This is consistent with
online financial information markets: coupling high-quality services with a proportion for free
allows consumers to assess the quality of the service.

Since consumers spend little or nothing to use these platforms, these services are inherently
challenging to value, and the national accounts may be underestimating them. This can
become especially problematic in measuring output or value added if consumers are
increasingly substituting away from paid, tangible products to free or quasi-free digital
products. Such a development leads to lower consumer expenditures even though there is no
change in the demand for the goods or services—only in how they are delivered to consumers.

Free digital products bring two challenges to measuring production. First, these products may
be missing from our measurement of GDP since there is no or a reduced-price market
transaction between the final user and the producer. Second, their contribution to household
and even worker production is likely understated—for example, delivery drivers using Google
Maps to find the shortest path between deliveries or programmers resolving small bugs using
Stack Overflow. 17 These would lead to a higher consumer or producer surplus without
necessarily affecting GDP measurements. 18 The full implications of this for productivity
measurements are discussed in other papers in the Digital Overview series (see, for example,
Mollins and Taskin 2023).

Researchers can take two approaches to impute a value for these digital products: using a
money-based measure such as advertisement revenues or using a time- or use-based measure
such as time spent on certain apps. The latter approach more closely reflects the value these
products add to consumer surplus. 19

Nakamura, Samuels and Soloveichik (2017) use a money-based approach and find that
including free media has little impact on GDP. They explore adding advertising-supported

16
Some of these platforms, such as Spotify and YouTube, also offer ad-free versions for a price.
17
However, some argue that many of the free digital products, such as social media, may also have a negative impact
on production (Schimmele, Fonberg and Schellenberg 2021; Braghieri, Levy and Makarin 2022).
18
This would be because a worker’s use of free digital products would be counted as an intermediate input.
19
Bourgeois (2020) mentions that one could assign free digital products a monetary value through the value of the
data generated with the use of the platform. However, the scarcity of good statistics on the value of data, mentioned
in greater detail in the next section, makes this approach less feasible.

17
media to GDP from 1998 to 2012 by including it as both a final expenditure and a business
input. However, advertising can also be seen as a form of investment in a brand rather than an
intermediate input (Bourgeois 2020; Nakamura 2015).

On the welfare side, by using a time- or attention-based measure of the value of free digital
goods, Brynjolfsson and Oh (2012) claim that measuring advertising expenditure understates
the welfare gains these products give to households. The use of these goods has been rising
as consumers have substituted away from other, costly non-digital products and toward free
digital goods. This highlights the wedge digital products are driving between traditional
measures of GDP—which are expenditure-based—and consumer welfare measures (Heys,
Martin and Mkandwire 2019). We explore the welfare effects of digitalization in Box 1, focusing
on consumer surplus.

18
Box 1

A shift toward consumer surplus

Since expenditures may not capture the full benefits of “free” digital products, some studies instead
impute the value of the extra surplus these products can yield to consumers.

o For the United States, Byrne and Corrado (2019) estimate that accounting for digital innovation
adds $2,000 in consumer surplus per connected user per year—the equivalent of an extra
0.6 percentage points per year—to growth in US real gross domestic product (GDP) from 2007
to 2017. Goolsbee and Klenow (2006) and Syverson (2017) measure consumer surplus by
considering the time spent using these digital goods and services.

o For Canada, Bellatin and Houle (2021) construct a similar measurement and find the increase in
technology use by households yielded Can$26.17 billion more in consumer surplus from 2005 to
2019. They measured total consumer surplus from digital goods to be approximately
$48.15 billion in 2019, representing about 2.3% of Canada’s GDP in 2019.

Overall, these free digital goods and services appear to yield an extra benefit, but this benefit is not
large relative to the size of the economy.

Another approach to assess the value of free goods is based on large-scale online choice experiments
or surveys to understand how much money people would require in exchange for giving up access to
free digital goods—the so-called willingness to accept (WTA) loss. WTA is a proxy for the consumer
surplus, which tends to be higher than the market price. For example, Brynjolfsson et al. (2019) find that
in 2016, the sample’s median WTA loss of one month of Facebook was $48.49, and this valuation
dropped to $37.76 in 2017. Coyle and Nguyen (2020) find significant increases in valuations of digital
goods and services (such as online groceries, online learning, WhatsApp, Netflix, Facebook) between
February and May of 2020.

We can either capture consumer surplus as a way of imputing the value of these free digital products
or estimate their full contribution to economic surplus using a modified calculation of GDP. To get at
the latter aspect, studies have suggested tracking alternative measures of GDP that would better
encompass these free digital products. Brynjolfsson et al. (2019) and Hulten and Nakamura (2017; 2019)
propose GDP-B and expanded GDP, respectively. These measures aim to better capture improvements
in living standards from free digital platforms.

19
3.2 Accounting for digital capital
Digitalization has allowed for the rise in value and use of intangible forms of capital. From using
data in machine learning to targeted online advertising algorithms, digitalization enables the
expansion of intangible capital. Though not all forms of intangible capital are digital, those that
are have a greater likelihood to be located across multiple platforms and are difficult to value
because they are usually produced in house. This section explores the challenges in properly
valuing the various forms of intangible capital, including data and databases.

Uncapitalized intangibles understate GDP

Intangible capital represents a growing share of firms’ balance sheets in Canada (Gu and
Macdonald 2020). Most forms of intangible capital are different from tangible capital because
they are non-rival but can also be firm-specific. For example, the knowledge of a new drug
innovation can be used across different firms (once the patent expires). In contrast, intangibles
like supply chain knowledge and brand awareness (e.g., Walmart’s strategy to gain market
share in the early 2000s [Holmes 2011]) cannot be easily replicated across firms. Improvements
in intangible capital are made not only through R&D but also through making supply chain
innovations, hiring workers with new knowledge, and implementing marketing campaigns, all
of which are facilitated by digitalization.
Not all intangible assets are related to digitalization. Those directly related to it include own-
account software, databases and systems infrastructure, while those indirectly related could
include patents or organizational capital.20 Exactly how much intangible capital is directly
related to digitalization remains unclear, but Tambe et al. (2019) combine firm-level data with
LinkedIn profiles and estimate that about 25% of US firms’ holdings of intangible capital is IT
intangible capital. Table 3 lists the different types of intangible capital and whether they are
currently included in Canada’s national accounts.

20
Organizational capital may or may not embed digitally driven processes.

20
 Table 3: Types of intangible capital in Canada
Categories of intangible capital Business intangible item Included in national accounts?
Computer software Yes

Computerized information
Computerized databases Partially

Science and engineering R&D Yes

Mineral exploration Yes

Innovative property Copyright and licence costs Yes
Other product development, design and No
research expenses
Brand equity No

Firm-specific human capital No

Economic competencies

Organizational capital No

Note: Details in this table are based on work from C. Corrado, J. Haskel, C. Jona-Lasinio and M. Iommi, “Intangible
Investment in the EU and US before and since the Great Recession and Its Contribution to Productivity Growth,” Journal
of Infrastructure, Policy and Development 2, no. 1 (2018): 11–36 and W. Gu and R. Macdonald, “Business Sector Intangible
Capital and Sources of Labour Productivity Growth in Canada,” Statistics Canada Catalogue no. 11F0019M—No. 442
(2020).

Measuring intangibles is especially difficult for a few reasons. First, this type of capital is typically
created within the firm and tends to be under-reported as an asset on firms’ balance sheets.21
Second, depreciation rates are hard to establish for assets whose reduction in value is not
physical (through wear and tear) and whose economic lifespans vary hugely. 22 Finally, it is
difficult to differentiate firm from industry value-added improvements. Often, intangible
investments such as marketing or brand equity lead to gains in market share at the firm level
rather than higher value added at the industry level.
In the United States, the term “broader investment puzzle” was coined by Crouzet and Eberly
(2019) to designate the separation of firms’ investment from their market valuation. In essence,
the firms with the top valuations no longer necessarily have the highest level of investment.
Some papers (Ewens, Peters and Wang 2020; Crouzet and Eberly 2019) explain this puzzle by
pointing to firms’ under-reporting of intangible capital. Corrado, Hulten and Sichel (2009) find

21
Ledoux and Cormier (2013) find that the reporting of intangibles in Canadian firms’ financial statements was lower
than firms’ voluntary disclosure about innovation from their corporate websites.
22
For example, the value of data used for advertisement targeting may be short-lived, while genomics data stay
relevant for at least the individual’s lifetime.

21
that more than $3 trillion of business intangible capital stock is missing from investment data
as of 2003. Kogan et al. (2017) use patent grants and resulting stock market valuation changes
to improve the estimation of innovation at the firm level. They find substantial improvements
when they use their measure to explain the relationship between firm growth and market value.

The consequences of mismeasuring intangible capital are twofold. First, if intangible

investments are missing, then firms may appear to have much higher profits relative to the
amount of capital they use as an input. This is exacerbated by the finding that production
functions tend to have constant returns to tangible capital but increasing returns to intangible
capital (Corrado et al. 2022). Second, if expenditures on intangibles are no longer counted
as being consumed in the same period, as intermediate inputs, and instead are
capitalized, measured GDP would be understated. These impacts on productivity are
discussed in another paper in the Digital Overview series (Mollins and Taskin 2023).

In Canada, Gu and Macdonald (2020) reclassify many new forms of intangibles from
intermediate inputs and capitalize them as an investment. Many of these new categories are
directly or indirectly related to the digital economy (advertising and brand equity, financial
innovation, architectural design, purchased non-R&D science, own-account non-R&D science,
firm-specific human capital and own-account and purchased organizational capital). Gu and
Macdonald ‘s estimates suggest that intangible capital represents about 36% of gross fixed
capital formation as of 2016 and that the intangible assets category with the largest
expenditure is own-account organizational capital. However, these estimates are derived from
firms’ balance sheet reporting and are likely to be underestimates of the true value.

Internationally, a key implication of the non-rivalry of digital capital is the ease of relocation
across countries. If a capital input is intangible, firms are more likely to choose its country of
residence based on tax advantages. Lipsey (2008) compares multinational firms’ reporting of
the relative size of total assets with their reporting of the size of their tangible assets. He tallies
the amount of labour and tangible capital (plants and equipment) relative to the total amount
of reported assets (tangible plus intangible) and identifies locations with the greatest
imbalance. He finds the ratio of assets per employee of US multinationals in tax haven
Caribbean islands to be $16 million per employee in 2005, relative to $4 million per employee
in Europe. In particular, US multinational affiliates held $150 million per employee in Bermuda.
However, the intangibles typically targeted in these overseas strategies are usually more
finance-related and not necessarily directly tied to digitalization, though digitalization does
enable their displacement.

In Canada, Caribbean tax havens (Bermuda, Barbados, Cayman Islands and Bahamas) receive
11.4% of Canada’s foreign direct investment. 23 Collectively, they were the second largest

23
Our calculations are based on Statistics Canada’s Table no. 36-10-0008-01.

22
recipient of Canadian foreign direct investment after the United States in 2021. Since this
investment is largely not going into tangible assets (such as factories or machinery and
equipment), much of it is done under the umbrella of intangible capital. This makes it
complicated to identify the origin and the location of use of intangible capital for Canada.

Reallocation from investment to cloud computing services

Powered by digitalization’s ease of movement across locations or platforms, cloud computing
has transformed companies’ ability to outsource computational tasks. Servers can reside in
areas where cooling services, electricity and land are more affordable, provided the location is
well integrated in the global network. Cloud computing is relatively simple to integrate within
a firm and typically much less expensive than buying the computing capital and hiring the
skilled labour to operate it. Otherwise, a firm would have to invest in the maximum amount of
information technology capacity it needs even if it did not always use it. By purchasing cloud
services, a firm can rapidly scale production up or down and reduce the amount it needs to
invest in ICT infrastructure.24 The adoption of cloud computing has blurred the boundaries for
both investment and production accounting.

On the investment side, the increasing returns to scale in this industry make this market highly
concentrated. In 2021, the biggest providers of cloud computing worldwide—Amazon Web
Services, Microsoft Azure, Alibaba, Google Cloud Platform and Huawei—represented 80.2% of
the global market share.25 These companies are headquartered in a small number of countries.
Firms are increasingly moving toward these cloud services and away from hosting the
infrastructure necessary to run software, computations and data in house. This means that much
of the investment in software and ICT hardware to run cloud computing services is often not
made in the country where the firm purchasing these cloud services is located. Instead, these
services are recorded as a service import and do not contribute to the country’s capital
stock. This may help explain some of the under-investment seen in Canada relative to the
United States.
On the production side, the national accounting problems created by cloud computing can be
explained through a simple example, outlined in Baer, Lee and Tebrake (2020). Consider a
situation where the cloud services are located outside of Canada. In this case, an import of
cloud computing services may be made in Canada, while the servers (where the
production activity is done) are in another, or multiple other, countries. This makes it

24
However, this gain in efficiency for the firm is done at the expense of stability of production. If the connection to the
cloud platform goes down, it will significantly impact the firm’s cloud-based production.
25
“Gartner Says Worldwide IaaS Public Cloud Services Market Grew 41.4% in 2021,” Press release, Gartner, June 2,
2022.

23
difficult to establish where the production of cloud services is taking place and the proper
production structure between the involved countries.

Baer, Lee and Tebrake (2020) also show how cloud computing is often misclassified in various
categories of the Canadian Annual Survey of Manufacturing and Logging Industries (ASML).
Cloud computing involves trade in digital services captured through enterprise surveys rather
than through customs transaction data. This makes tracking its development and impact on
the Canadian economy difficult. As Chart 3 shows, based on results from Statistics Canada’s
Survey of Digital Technology and Internet Use, a large portion of Canadian firms reported
purchasing cloud computing services, especially large firms. Almost 80% of Canada’s large
firms and almost half of all firms purchased cloud-based services. Large firms spend a
significantly larger amount on cloud computing services—$558,000 per year on average—
whereas small business spend on average $8,800 per year (Statistics Canada 2022). Additionally,
the share of firms using cloud computing services increased by 6.4 pps across all firm
sizes between 2019 and 2021. 26

Chart 3: Com parison of Canadian firm s’ purchase of cloud com puting by size

Percentage of firms

%
80

60

40

20

0
Large enterprises Medium-sized Small enterprises Total, all enterprises
enterprises
2019 2021

Note: Data are from Statistics Canada's 2019 and 2021 Survey of Digital Technology and Internet Use and from Statistics Canada Table 22-10-0117-01.

Source: Statistics Canada

Missing capitalization of data

Data are a critical component of digitalization. Data and datasets act as input, intermediate
goods and output in production. Many challenges are associated with estimating and
capitalizing the value of digitalization. Some datasets have significant value as investment

26
It is worth nothing that the Survey of Digital Technology and Internet Use covers fewer firms than the ASML.

24
goods, while the value of others is short-lived. As a result, measuring the value of data in
the Canadian economy and being able to capitalize it accurately are challenging.

Data are non-rival goods, meaning that the same data point can be copied and used
simultaneously across different platforms without diminishing the quantity available to
a single user. Data also do not physically depreciate the same way buildings, machinery and
equipment do, but they may depreciate in their economic usefulness. For example, the value
of customer data to advertisers can be very short-lived, whereas the value of journal archives
can be long-lasting.

Another challenge for valuing data-related products in Canada arises when using firms’ own
estimations from their tax statements. The firms’ own valuations of their data and databases
might provide good information to establish the value of their stock. However, few firms in
Canada report this number on their balance sheets, or they tend to group it with other
categories of intangible capital, such as goodwill.

Statistics Canada (2019b) outlines the main method for estimating the stock of data-related
products and its associated challenges. As recommended by the SNA 2008, most countries
have adopted a perpetual inventory method to measure the stock of data-related products in
the economy. 27 Using this method requires making assumptions about data’s depreciation
profile. Estimates for the stock of data-related products presented in the previous section
assume a useful life of 25 years, 5 years and 6 years for data, databases and data science,
respectively. Currently, data and some parts of databases are not included in the Canadian
System of National Accounts.

3.3 Assessing the impact of the gig economy

Digitalization has allowed the expansion of platforms for gig work in Canada. A gig worker is
someone who works one or more temporary jobs as an independent contractor or freelancer.
The digitalization of these platforms can now match these gig workers with firms or consumers
instantaneously and globally. The rise in e-commerce allows gig workers to sell their products
directly to consumers almost anywhere in the world. As well, peer-to-peer rideshare and
accommodation apps have streamlined the procurement and payment of these services on one
platform.

Gig workers benefit from nonpecuniary perks such as flexible schedules. However, they tend to
have less predictable earnings and lack the legal rights, health benefits and retirement plans
afforded under conventional employment contracts. For firms, gig workers can lower their

27
Other possible methods include treating data-related products as assets or discounting the future stream of revenue
(treating it as a natural resource). These methods are found to be less feasible for data-related products than the
perpetual inventory method (Statistics Canada 2019c).

25
labour costs and allow them to respond to rapidly changing market conditions. By using
platforms such as Amazon Mechanical Turk, firms can compartmentalize small components of
their projects and outsource them to workers. If the project ends up getting cancelled, they can
halt the outsourcing of these tasks without having to make direct modifications to their own
labour force.

Measuring the size of the gig economy in Canada is challenging, with different surveys and tax
data yielding different estimates. The key issue is that these gig workers are difficult to
distinguish from self-employed workers, given that gig work is not likely to be their main work
activity. Kostyshyna and Luu (2019) use a special edition of the Bank of Canada's Canadian
Survey of Consumer Expectations that asks households about their participation in gig work to
estimate the size of the gig economy in Canada. They find that approximately 30% of Canadians
participated in gig work in 2018. Overall, they find that properly accounting for gig workers
would represent the equivalent of approximately 700,000 full-time jobs or about 3.5% of the
labour force in the third and fourth quarters of 2018. In its Labour Force Survey, Statistics
Canada (2017) found only 0.3% of Canadians participated in rideshare services and 0.2% in
shared accommodations.28, 29 However, using Canadian administrative tax data, Jeon, Liu and
Ostrovsky (2022) find that gig workers represented 8.2% of workers in 2016. 30, 31 If these
workers are not all accounted for in the labour force, then their labour participation rates
or contribution to work hours may be underestimated.

3.4 Tracking prices in a digital world

This section explores how digitalization affects the measurement of prices. Challenges around
this impact occur in four main ways:

• the increased difficulty in tracking quality improvements of digital goods

• the delayed introduction of new products to price indexes

• the increase in “free” digital products

• the rise in e-commerce and online data collection

28
Statistics Canada (2017) used a smaller definition of gig work, surveying respondents’ participation in only th e
platforms Uber, Lyft, Airbnb and FlipKey.
29
A newer release of the Labour Force Survey in December 2022 found 1.5% of those employed that month provided
either rideshare or food or goods delivery through a digital platform.
30
Jeon, Liu and Ostrovsky (2022) define gig workers as unincorporated, self-employed workers (sole proprietors) who
report business, professional or commission self-employment income on their T1 tax returns and attach at least one
T2125 form without a business number.
31
Jeon, Liu and Ostrovsky (2022) consider this number to be a lower bound because they may be missing gig workers
with a gross income above $30,000. They find an upper bound of 10.3% of workers.

26
We also discuss some benefits, such as improvements in how prices are observed in real time.

The first three challenges to tracking prices, as covered in Reinsdorf and Schreyer (2019), are
particularly problematic for digital products. In new iterations of digital products, the product’s
technological characteristics change when the new version replaces the old. But the quality
improvements are more likely to be understated for digital products. Also, novel products may
be replacing existing varieties but these new products are too different from their predecessors
to be assigned a proper quality adjustment. These quality adjustments require significant
revisions before they are added to the price index. Finally, the free digital products discussed
in section 2.1 also contribute to this mismeasurement of prices, which is especially problematic
when they replace more expensive products (e.g., Wikipedia replacing encyclopedias).
However, it is difficult to disentangle what portion of these free products should be added to
consumer welfare and what portion contributes to revisions to aggregate deflators.

The fourth way in which digitalization makes it more difficult to measure prices is through the
rise in e-commerce. People increasingly turn to e-commerce platforms, where prices tend to
be lower, especially for products listed solely online. If the prices on these platforms are not
adequately captured in consumer price index (CPI) calculations, the cost of living may be
overstated. An important consideration is that e-commerce statistics from Statistics Canada’s
business survey do not include products sold by Amazon (Chernoff 2019).32 Leaving Amazon
Canada’s sales out of the estimates clearly represents a challenge in the measurement of e-
commerce, since the estimates would not provide a global view of the e-commerce market.

In some circumstances, e-commerce also allows sellers to charge different consumers different
prices for the same product. Recent developments in marketing technologies allow firms to
collect and link information about consumers. As a result, they may charge different prices to
different consumers based on their perceived willingness to pay.33 Using Canadian price data,
Mitchell (2019) finds that products sold solely online have a higher price variance than products
sold only in stores and products sold both online and in stores.

On the positive side, digitalization has the capacity to improve how we collect price data.
Traditionally, statistical agencies relied on people visiting physical stores to collect price data

32
However, Amazon activity is included in other statistical measures through imports by households. As well, any
online seller with a physical headquarters located in Canada that sells through Amazon Marketplace would be
included in Statistics Canada’s Non-store Retail Survey and therefore would be represented in Statistics Canada’s
estimates of e-commerce sales.
33
This type of price differentiation is not possible for all online platforms. It requires access to customer data and th e
capacity to link their information to an account or IP address. Google, Facebook and Amazon’s privacy statements
say they can collect information on users’ views, searches and interactions on their page and third-party pages.
Some studies find evidence of this price differentiation by various firms at one point or another (Hannak et al. 2014;
Mikians et al. 2013; Valentino-Devries, Singer-Vine and Soltani 2012; Viswanathan et al. 2007) while others do not
observe this price discrimination, for example, in the airline industry (Vissers et al. 2014).

27
on the items in the basket of goods and services contained in the CPI. Obtaining, or scraping,
online prices and product information can improve the timeliness and reduce the labour costs
associated with data collection. The Billion Prices Project by Cavallo and Rigobon (2016) collects
a vast amount of online prices to evaluate pricing dynamics in different countries. This has
yielded the PriceStats database, which is used by statistical agencies around the world and is
comparable to many price indexes produced in advanced economies. Details of this database
are further explored in the Digitalization Overview series paper by Chu, Dahlhaus and Hajzler
(forthcoming) on the topic of prices and inflation.

Online price collection leads to new ways of incorporating alternative data in Canadian CPI.
Statistics Canada (2021) describes the changes to its CPI data collection, which include the use
of web-scraped and application programming interface data. The main challenge with
collecting online prices is that consumers may be buying products from all over the
world, and it is difficult to discern exactly which and how many of these products
Canadians are buying.

4. Trends and questions for the future

Given digitalization’s broad impact on businesses, households and markets, measuring it is
difficult. Our discussion highlighted that digitalization is truly pervasive, relating to automation,
ICT, technology adoption, and technological developments in machinery and equipment. The
various measures reviewed here strike a balance between leveraging available data and
focusing on areas where digitalization is thought to be most transformative. However, some
important developments in digitalization and their impacts may not be included.

Section 3 discussed many challenges, but an important open question relates to the future
impacts of the COVID-19 pandemic on digitalization. Most studies agree that the pandemic
accelerated the pace of digitalization. Examples of this include the increased capacity for
employees to work from home, the surge of e-commerce when businesses had to temporarily
close their physical stores and the increase in digital skills. It remains to be seen if these were
only a front-loading of future improvements or if this pace will be maintained.

Increasing resilience to future pandemics or supply disruptions could lead firms to even higher
levels of adopting and investing in ICT. Firms and consumers have so far retained their attitudes
toward online shopping and working from home, despite the reopening of the economy.
Assessing and measuring the long-term structural implications of these trends is important
because of their impacts on inflation, productivity and the labour market.

Another outcome of the pandemic is the enhanced use of timelier high-frequency indicators.
Examples include the use of credit card data, mobility data, online bookings and Indeed job
postings, all coming from our digitalized economy. These were particularly useful given the

28
publication lags in official statistics and suspensions in data collection. In addition, concerns
about the inaccuracy of the CPI basket weights due to rapid changes in consumers’ behaviour
led price collection to move almost entirely online. While these metrics are valuable in filling
information gaps, their timeline does not extend far into the past and the series are volatile,
making it difficult to disentangle true signals in the data from noise. Nonetheless, we should
continue efforts to develop such indicators and to leverage their use in policy decisions.

A trend with potentially large economic consequences is the rise in intangible capital. In this
paper, we discussed how challenging it could be to measure the value of firms’ intangible
capital. However, an important characteristic of such assets is their sensitivity to monetary
policy. The literature shows that, compared with tangible investment, intangible assets tend to
be less sensitive to interest rates since they have higher depreciation rates and are less usable
as collateral (see Crouzet and Eberly 2019; Döttling and Ratnovski 2021). Given the implications
of this for monetary policy, we should aim to better understand the size and evolution of
intangible capital in Canada and the effect of interest rate shocks on firms with a greater
proportion of intangible assets.

Lastly, another important issue is the potential impact of digitalization on welfare. Most agree
that GDP is an imperfect measure of welfare. It remains to be seen how much digitalization will
improve the well-being of consumers compared with how much it will improve productivity
and output. We discussed the impact of digitalization on consumer surplus in Box 1, but more
work should be done. In this paper, we focused on direct metrics such as output, capital and
labour. However, analyzing impacts on welfare by using measures similar to the expanded
measures of GDP presented in Box 1 could lead to a better understanding of the ubiquitous
nature of digitalization.

Key open questions

• How will the COVID-19 pandemic and the related new behaviours of consumers and
businesses affect the adoption of new technologies? Will more timely indicators help
measure the economy better?

• How will the challenges in measuring intangible capital evolve and how sensitive is this
type of capital to monetary policy?

• Will digitalization have a bigger impact on productivity or on the well-being of consumers?

What metrics could be developed to assess the impact of digitalization on welfare instead
of on GDP?

29
Appendix: How digitalization in Canada compares
internationally
The pace of digital adoption varies across countries. We present several international indicators
and indexes and compare how Canada ranks relative to other advanced economies. In this
appendix, we refer to the International Digital Economy and Society Index (I-DESI), the ICT
Development Index (IDI) and the Network Readiness Index (NRI). However, a full range of other
indexes exists (see Csonto, Huang and Tovar 2019 for other examples).

International Digital Economy and Society Index

The European Commission publishes the Digital Economy and Society Index (DESI) along with
its international version, the I-DESI. The indexes focus on 24 indicators across five dimensions
to assess the degree of digital adoption in each economy. Compared with the 44 other
countries assessed, Canada’s score (54) is above both the EU average (48) and the non-EU
average (49), but the scores vary across dimensions of the index, particularly among European
countries (European Commission 2020). Canada’s score remains below some other G7
countries, such as the United States (62) and the United Kingdom (58). Canada scores highly
overall in terms of digital adoption. This reflects a strong relative performance in three of the
dimensions (connectivity, citizen use of internet and digital public services), while the human
capital dimension shows significant growth potential. In the following, we consider each of the
five dimensions of this index and explain how Canada performs in them.

• Connectivity measures the deployment and quality of broadband. It includes fixed

and mobile broadband coverage, as well as speed and affordability. Canada’s score
for connectivity is 60, slightly above the non-EU average of 59 but under the EU
average of 62 and well below the US score of 70. The affordability indicator in this
category is one where progress could be made. Nonetheless, in terms of the
evolution of connectivity, Canada improved between 2015 to 2017 but plateaued
in 2018.

• Human capital examines the skills needed to take advantage of the possibilities
offered by a digital society. This dimension includes two sub-dimensions: internet
user skills and advanced skills and development. The first measures the number of
users having various levels of digital skills in word processing, using spreadsheets
or coding. The second refers to the share of employees working in
telecommunications and the proportion of people graduating in ICT. For this
dimension, Canada performs relatively poorly with a score of 36.5 compared with
the EU average of 41.8 and the non-EU average of 43.0.

30
 • Citizen use of internet considers the variety of activities performed by citizens
already online. Such activities include the use of internet for video calls, social
networks, banking and shopping and, more generally, the number of internet users
as a share of the total population. Canada significantly outperforms the average of
EU (47.0) and non-EU (51.8), with a normalized score of 61.6.

• Integration of digital technology assesses the digitalization of businesses and

development of online sales. This dimension is grouped into two sub-dimensions.
The first measures the availability of the latest technology and technology
absorption. The second captures the proportion of small and medium-sized
enterprises selling goods or services online and the number of secure internet
servers per one million people. For this dimension, Canada ranks above the EU
average of 41.1 and the non-EU average of 46.2 with a score of 55.7.

• Digital public services measures the digitalization of public services, focusing on

eGovernment. It comprises three specific indicators: the proportion of the
population accessing government services online (eGovernment), online service
completion (availability of online information and online public consultation) and
the OECD indicator for open data. Canada performs well in this dimension with a
score of 70.2, well above the EU average of 56.0 and the non-EU average of 60.4.

To obtain the total I-DESI index, scores for these five dimensions are aggregated using a weight
of 25% for the first two, 15% for the third and fifth and 20% for the fourth. While Canada is
above average across most of the dimensions, it is never among the top five countries and
remains behind its closest neighbour, the United States. The area with the most potential for
improvement is in the human capital dimension, particularly the sub-dimension associated with
the share of employees working in telecommunications and the proportion of people
graduating in ICT, where Canada ranks close to the bottom of the list. While improvement is
observed overall between 2015 and 2017, a plateau was reached in most dimensions by 2018.
More work could be done to assess the underlying causes of this slowdown.

ICT Development Index

The IDI is a composite index published by the UN’s International Telecommunication Union. Its
main focus is to monitor and compare development in ICT. The 2017 version comprises 11
indicators combined in three sub-indexes: ICT access, ICT use and ICT skills. Each of these sub-
indexes includes various indicators to measure the evolution of ICT development. Examples
include the percentage of households with computer or internet access, the secondary and
tertiary enrolment ratio, and fixed or mobile broadband subscriptions.

31
Overall, Canada ranks 29th out of 176 countries with a score of 7.8 in the 2017 edition of the
IDI, three positions lower than in the 2016 report. By comparison, the top five countries
averaged a score of 8.8 while the United States ranked 16th with a score of 8.2. In terms of sub-
indexes, Canada performs well in the ICT skills dimension, particularly in the tertiary enrolment
ratio, as well as in the mean of years of schooling. It might seem surprising that Canada
outperforms in this category since human capital was the weakest sub-dimension in the I-DESI,
but this is due to the broader indicators in the IDI used to proxy ICT-related skills. The lowest
indicators are the fixed and mobile telephone subscription per 100 people, both averaging a
normalized score of only 6.4 out of 10.

Network Readiness Index

The Network Readiness Index (NRI), launched in 2002 by the World Economic Forum, is another
composite index to compare the impact and application of ICT. Similar to the I-DESI and the
IDI, this index comprises different dimensions: technology, people, governance and impact.
Each dimension is then decomposed in three sub-dimensions, and, overall, 60 indicators are
selected to compare 130 different economies. An interesting feature of this index is the fact
that the latest 2021 edition can be used to examine the impact of the COVID-19 pandemic,
which is currently not possible with the IDI and the I-DESI.

Overall, Canada ranks 11th in the 2021 report for the aggregate index. In term of sub-
dimensions, Canada performs well in governance (6th place) and technology (9th place). In
particular, Canada ranks as a leader in a few indicators, such as e-commerce legislation (1st),
good health and well-being (1st) and publication and open data (2nd). However, there is scope
for improvement in the other two dimensions, particularly in the impact dimension (20th). The
weakest indicators include energy intensity (111th ), ICT services exports (66th ) and ICT regulatory
environment (60th ). Nonetheless, Canada’s score is higher than the group average of high-
income countries in each of the four dimensions and 12 sub-dimensions. Canada’s NRI score
relative to its GDP per capita is above the trend line determined by other high-income-group
countries. Canada is second in the Americas regional group with a score of 76.5, behind the
United States with a score of 81.1.

32
References
Aghion P., C. Antonin, S. Bunel and X. Jaravel. 2022. “The Effects of Automation on Labor
Demand.” In Robots and AI: A New Economic Era, edited by L. Y. Ing and G. M. Grossma,
15–39. London: Routledge.

Baer, A., K. Lee and J. Tebrake. 2020. “Accounting for Cloud Computing in the National
Accounts.” IMF Working Paper No. 2020/127.

Barber-Dueck, C. 2008. “The Canadian Research and Development Satellite Account, 1997 to
2004.” Statistics Canada Income and Expenditure Accounts Technical Series No.
2007056.

Bellatin, A. and G. Galassi. 2022. “What COVID-19 May Leave Behind: Technology-Related Job
Postings in Canada.” Bank of Canada Staff Working Paper No. 2022-17.
Bellatin, A. and S. Houle. 2021. “Overlooking the Online World: Does Mismeasurement of the
Digital Economy Explain the Productivity Slowdown?” Bank of Canada Staff Analytical
Note No. 2021-10.

Bourgeois, A. 2020. “Free Services from the Digital Economy: Do We Need to Measure Their
Value and How?” Economie et Statistique / Economics and Statistics 517-518-519:
157–172.

Bourreau, M. and V. Lethiais. 2007. “Pricing Information Goods: Free vs. Pay Content.” In Internet
and Digital Economics: Principles, Methods and Applications, edited by E. Brousseau and
N. Curien, 345–367. Cambridge: Cambridge University Press.

Braghieri, L., R. Levy and A. Makarin. 2022. “Social Media and Mental Health.” American
Economic Review 112 (11): 3660–3693.

Bresnahan, T. 2010. “Chapter 18: General Purpose Technologies.” In Handbook of the Economics
of Innovation, Vol. 2, edited by B. H. Hall and N. Rosenberg, 761–791. Oxford: Elsevier.
Bresnahan, T. and M Trajtenberg. 1995. “General Purpose Technologies ‘Engines of Growth’?”
Journal of Econometrics 65 (1): 83–108.

Brynjolfsson, E., A. Collis, E. Diewert, F. Eggers and K. Fox. 2019. “GDP-B: Accounting for the
Value of New and Free Goods in the Digital Economy.” National Bureau of Economic
Research Working Paper No. 25695.

Brynjolfsson, E. and J. Oh. 2012. “The Attention Economy: Measuring the Value of Free Digital
Services on the Internet.” Presented at the International Conference on Information
Systems 2012.

33
Brynjolfsson, E., D. Rock and C. Syverson. 2021. “The Productivity J-Curve: How Intangibles
Complement General Purpose Technologies.” American Economic Journal:
Macroeconomics 13 (1): 333–372.

Byrne, D. and C. Corrado. 2019. “Accounting for Innovation in Consumer Digital Services: It Still
Matters.” National Bureau of Economic Research Working Paper No. 26010.

Calvino, F., C. Criscuoloi, L. Marcolin and M. Squicciarini. 2018. “A Taxonomy of Digital Intensive
Sectors.” Organisation for Economic Co-operation and Development (OECD) Science,
Technology and Industry Working Papers No. 2018/14.

Cavallo, A. and R. Rigobon. 2016. “The Billion Prices Project: Using Online Prices for
Measurement and Research.” Journal of Economic Perspectives 30 (2): 151–178.
Chernoff, A. 2019. “Amazon Effects in Canadian Online Retail Firm-Product-Level Data?” Bank
of Canada Staff Working Paper No. 2019-42.

Chernoff, A. and G. Galassi. 2023. “Digitalization: Labour Markets.” Bank of Canada Staff
Discussion Paper No. 2023-16.

Chu, V., T. Dahlhaus and C. Hazjler. Forthcoming. “Digitalization: Prices of Goods and Services.”
Bank of Canada Staff Discussion Paper.

Corrado, C., J. Haskel, C. Jona-Lasinio and M. Iommi. 2018. “Intangible Investment in the EU and
US before and since the Great Recession and Its Contribution to Productivity Growth.”
Journal of Infrastructure, Policy and Development 2 (1): 11–36.

Corrado, C., J. Haskel, C. Jona-Lasinio and M. Iommi. 2022. “Intangible Capital and Modern
Economies.” Journal of Economic Perspectives 36 (3): 3–28.

Corrado, C., C. Hulten and D. Sichel. 2009. “Intangible Capital and U.S. Economic Growth.”
Review of Income and Wealth 55 (3): 661–685.

Coyle, D. and D. Nguyen. 2020. “Free Goods and Economic Welfare.” Economic Statistics Centre
of Excellence Discussion Paper No. 2020-18.

Crouzet, N. and J. C. Eberly. 2019. “Understanding Weak Capital Investment: The Role of Market
Concentration and Intangibles.” National Bureau of Economic Research Working Paper
No. 25869.

Csonto, B., Y. Huang and C. E. Tovar. 2019. “Is Digitalization Driving Domestic Inflation?”
International Monetary Fund Working Paper No. 2019/271.

Dixon, J. 2020. “How to Build a Robots! Database.” Statistics Canada Analytical Studies: Methods
and References, Catalogue no. 11-633-X.

34
Dixon, J., B. Hong and L. Wu. 2021. “The Robot Revolution: Managerial and Employment
Consequences for Firms.” Management Science 67 (9): 5586–5605.

Döttling, R. and L. Ratnovski. 2023. "Monetary Policy and Intangible Investment." Journal of
Monetary Economics 143: 53–72.

European Central Bank. 2021. “Digitalisation: Channels, Impacts and Implications for Monetary
Policy in the Euro Area.” European Central Bank Occasional Paper Series No. 266.
Ewens, M., R. Peters and S. Wang. 2020. “Measuring Intangible Capital with Market Prices.”
National Bureau of Economic Research Working Paper No. 25960.

European Commission. 2020. “International Digital Economy and Society Index 2020.” SMART
Paper No. 2019/0087.

Goldfarb, A., B. Taska, F. Teodoridis 2023. “Could Machine Learning Be a General Purpose
Technology? A Comparison of Emerging Technologies Using Data from Online Job
Postings.” Research Policy 52 (1): 104653.

Goolsbee, A. and P. J. Klenow. 2006. “Valuing Consumer Products by the Time Spent Using
Them: An Application to the Internet.” American Economic Review 96 (2): 108–113.

Gu, W. and R. Macdonald. 2020. “Business Sector Intangible Capital and Sources of Labour
Productivity Growth in Canada.” Statistics Canada Catalogue no. 11F0019M—No. 442.

Hall, B. H. and M. Trajtenberg. 2004. “Uncovering GPTs with Patent Data.” National Bureau of
Economic Research Working Papers No. 10901.

Hannak, A., G. Soeller, D. Lazer, A. Mislove and C. Wilson. 2014. “Measuring Price Discrimination
and Steering on E-Commerce Web Sites.” In IMC ’14: Proceedings of the 2014
Conference on Internet Measurement Conference, 305–318. New York, NY: Association
for Computing Machinery.

Heys, R., J. Martin and W. Mkandawire. 2019. “GDP and Welfare: A Spectrum of Opportunity.”
Economic Statistics Centre of Excellence Discussion Paper No. 2019-16.

Holmes, T. J. 2011. “The Diffusion of Wal-Mart and Economies of Density.” Econometrica 79 (1):
253–302.

Hulten, C. and L. Nakamura. 2017. “Accounting for Growth in the Age of the Internet: The
Importance of Output-Saving Technical Change.” National Bureau of Economic
Research Working Paper No. 23315.

Hulten, C. and L. Nakamura. 2019. “Expanded GDP for Welfare Measurement in the 21st
Century.” National Bureau of Economic Research Working Paper No. 26578.

35
Jackson, C. 2003. “Capitalization of Software in the National Accounts.” Statistics Canada
Income and Expenditure Accounts Technical Series, Catalogue no. 13-604-MIE.

Jeon, S.-H., H. Liu and Y. Ostrovsky. 2022. “Measuring the Gig Economy in Canada Using
Administrative Data.” Canadian Journal of Economics / Revue canadienne d'économique
54 (4): 1638–1666.

Jovanovic, B. and P. L. Rousseau. 2005. “General Purpose Technologies.” In Handbook of

Economic Growth, Vol. 1, Part B, edited by P. Aghion and S. N. Durlauf, 1181–1224.
Amsterdam: Elsevier.

Kogan, L., D. Papanikolaou, A. Seru and N. Stoffman. 2017. “Technological Innovation, Resource
Allocation, and Growth.” Quarterly Journal of Economics 132 (2): 665–712.
Kostyshyna, O. and C. Luu. 2019. “The Size and Characteristics of Informal (‘Gig’) Work in
Canada.” Bank of Canada Staff Analytical Note No. 2019-6.

Lamb, C. and M. Seddon. 2016. “The State of Canada’s Tech Sector.” Brookfield Institute Report,
July.

Ledoux, M.-J. and D. Cormier. 2013. “Market Assessment of Intangibles and Voluntary
Disclosure about Innovation: The Incidence of IFRS.” Review of Accounting and Finance
12 (3): 286–304.

Lipsey, R.E. 2010. “Measuring the Location of Production in a World of Intangible Productive
Assets, FDI and Intrafirm Trade.” Review of Income and Wealth 56: S99-S110.

Liu, H. and J. McDonald-Guimond. 2021. “Measuring Digital Intensity in the Canadian

Economy.” Statistics Canada Economic and Social Reports, Catalogue no. 36-28-0001.

Manyika, J., M. Chui, P. Bisson, J. Woetzel, R. Dobbs, J. Bughin and D. Aharon. 2015. “The Internet
of Things: Mapping the Value beyond the Hype.” McKinsey Global Institute Report,
June.

Mikians, J., L. Gyarmati, V. Erramilli and N. Laoutaris. 2013. “Crowd-Assisted Search for Price
Discrimination in E-Commerce: First Results.” In Context ’13: Proceedings of the Ninth
ACM Conference on Emerging Networking Experiments and Technologies, 1–6. New York:
Association for Computing Machinery.

Mitchell, T. 2019. “E-commerce and the Consumer Price Index: Measuring Inflation in a Digital
Economy.” Statistics Canada Price Analytical Series, Catalogue no. 62F0014M.

Mollins, J. and T. Taskin. 2023. “Digitalization: Productivity.” Bank of Canada Staff Discussion
Paper No. 2023-17.

36
Moser, P. and T. Nicholas. 2004. “Was Electricity a General Purpose Technology? Evidence from
Historical Patent Citations.” American Economic Review 94 (2): 388–394.

Moulton, B., J. Tebrake and M. Tovar. 2022. “Experimental Indicators of Digital Industries in
Select Countries: Definitions, Methods, and Results.” International Monetary Fund
Working Papers No. 2022/197.

Nakamura, L. 2015. “Advertising, Intangibles, and Unpriced Entertainment.” In Intangibles,

Market Failure and Innovation Performance, edited by A. Bounfour and T. Miyagawa,
11–26. Springer International Publishing.

Nakamura, L. I., J. Samuels and R. H. Soloveichik. 2017. “Measuring the 'Free' Digital Economy
within the GDP and Productivity Accounts.” Federal Reserve Bank of Philadelphia
Working Paper No. 17-37.

Organisation for Economic Co-operation and Development (OECD). 2020. A Roadmap Toward
a Common Framework for Measuring the Digital Economy: Report for the G20 Digital
Economy Task Force.

Petralia, S. 2020. “Mapping General Purpose Technologies with Patent Data.” Research Policy
49 (7): 104013.

Reinsdorf, M. and P. Schreyer. 2019. “Measuring Consumer Inflation in a Digital Economy.”

OECD Statistics Working Paper No. 2019/01.

Schimmele, C., J. Fonberg and G. Schellenberg. 2021. “Canadians' Assessments of Social Media
in their Lives.” Statistics Canada Economic and Social Reports, Catalogue no. 36-28-
0001.

Statistics Canada. 2017. “The Sharing Economy in Canada.” The Daily, February 28.

Statistics Canada. 2019a. “Measuring Digital Economic Activities in Canada: Initial Estimates.”
Statistics Canada, Catalogue No. 13-605-X.

Statistics Canada. 2019b. “Measuring Investment in Data, Databases and Data Science:
Conceptual Framework.” Statistics Canada Latest Developments in the Canadian
Economic Accounts, Catalogue No. 13-605-X.

Statistics Canada. 2019c. “The Value of Data in Canada: Experimental Estimates.” Statistics
Canada Latest Developments in the Canadian Economic Accounts, Catalogue No. 13-
605-X.

Statistics Canada. 2021. “Digital Supply and Use Tables, 2017 to 2019.” The Daily, April 20.
Statistics Canada. 2022. “Digital technology and Internet use, 2021.” The Daily, September 13.

37
Syverson, C. 2017. “Challenges to Mismeasurement Explanations for the US Productivity
Slowdown.” Journal of Economic Perspectives 31 (2): 165–186.

Tambe, P., L. M. Hitt, D. Rock and E. Brynjolfsson. 2019. “IT, AI and the Growth of Intangible
Capital.” Social Sciences Research Network.

United Nations. 2008. “International Standard Industrial Classification of All Economic Activities
(ISIC), Rev. 4.” Department of Economic and Social Affairs, Statistics Division, Statistical
Papers Series M No. 4/Rev.4.
Valentino-Devries, J., J. Singer-Vine and A. Soltani. 2012. “Websites Vary Prices, Deals Based on
Users’ Information.” Wall Street Journal, December 24.
Vissers, T., N. Nikiforakis, N. Bielova and W. Joosen. 2014. “Crying Wolf? On the Price
Discrimination of Online Airline Tickets.” In Privacy Enhancing Technologies, xx–xx.
Proceedings of the 14th International Symposium, Privacy Enhancing Technologies
(PETS) 2014, July 16–18, Amsterdam, The Netherlands: Springer Cham. Paper presented
at the 7th Workshop on Hot Topics in Privacy Enhancing Technologies, July 18.
Amsterdam, The Netherlands.

Viswanathan, S., J. Kuruzovich, S. Gosain and R. Agarwal. 2007. “Online Infomediaries and Price
Discrimination: Evidence from the Automotive Retailing Sector.” Journal of Marketing
71 (3): 89–107.

38