Measuring The Economic Value of Data and Cross-Border Data Flows

MEASURING THE
ECONOMIC VALUE OF
DATA AND CROSS-
BORDER DATA FLOWS
A BUSINESS PERSPECTIVE
OECD DIGITAL ECONOMY

PAPERS
August 2020 No. 297
2 | MEASURING THE ECONOMIC VALUE OF DATA AND DATA FLOWS
OECD Working Papers should not be reported as representing the official views of the OECD or of
its member countries. The opinions expressed and arguments employed are those of the authors.
Working Papers describe preliminary results or research in progress by the author(s) and are
published to stimulate discussion on a broad range of issues on which the OECD works. Comments
on Working Papers are welcomed, and may be sent to Directorate for Science, Technology and
Innovation, OECD, 2 rue André-Pascal, 75775 Paris Cedex 16, France.
Note to Delegations:
This document is also available on O.N.E. under the reference code:
DSTI/CDEP/MADE(2019)4/FINAL
This document, as well as any data and map included herein, are without prejudice to the status of
or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to
the name of any territory, city or area.
© OECD (2020)
The use of this work, whether digital or print, is governed by the Terms and Conditions to be found
at http://www.oecd.org/termsandconditions.
OECD DIGITAL ECONOMY PAPERS

MEASURING THE ECONOMIC VALUE OF DATA AND DATA FLOWS |3
Foreword
This paper investigates how the economic value of data can be conceptualised and
measured from a business perspective. It first discusses data monetisation as a strategy for
developing new business models, as well as enhancing “traditional” business models.
Secondly, taxonomies for data are reviewed and a new taxonomy proposed with a specific
focus on measuring the business value of data. Here discussion is centred on four stylised
‘data monetisation strategies’ that are commonly used by companies to generate new
streams of revenue, or to improve existing business processes or products. The role of
cross-border data flows as a key enabler of our global economy is also looked at, leading
to the concept of a ‘global data value chain’ based on the idea that digitalisation enables
the physical detachment of data collection, analysis, storage, and monetisation. Finally, the
most promising avenues for measuring the economic value of data are summarised and
discussed.
The report was initially prepared to inform an expert workshop on measuring data and data
flows convened by the OECD and ESCoE in London in November 2018. It benefits from
the detailed discussions which took place at that workshop as well as further input and
feedback from delegates to the OECD Working Party on Measuring and Analysing the
Digital Economy and members of the OECD Secretariat.

Measuring the Economic Value of Data and Cross-Border

Data Flows: A Business Perspective
David Nguyen and Marta Paczos
UK Economic Statistics Centre of Excellence
Abstract
The process of collecting, aggregating and analysing data for the purpose of successful
operation is nothing new for companies. However, the amount and variety of data they use
has increased dramatically in recent years. In fact, data have often become a central
element in business models, posing fresh challenges to researchers and policymakers alike.
In this paper, we investigate how the economic value of data can be conceptualised and
measured from a business perspective. We first discuss data monetisation as a strategy for
developing new business models, as well as enhancing “traditional” business models.
Secondly, we review taxonomies for data and propose a new taxonomy with a specific focus
on measuring the business value of data. Here our discussion is centred on four stylised
‘data monetisation strategies’ that are commonly used by companies to generate new
streams of revenue, or to improve current business processes or products. We also discuss
how different data characteristics and types affect economic value. Next, we examine the
role of cross-border data flows as a key enabler of our global economy. We discuss how
and why businesses transfer data across borders, as well as the broad scale and value of
cross-border data flows. To do so we present the concept of a ‘global data value chain’,
based on the idea that digitalisation enables the physical detachment of data collection,
analysis, storage and monetisation. Finally, we summarise and discuss the most promising
avenues for measuring the economic value of data and consider their feasibility in the short
and long-term.
Keywords: digital, science and technology
Acknowledgements
This research paper has been funded by the Organisation for Economic Co-operation and Development
(OECD) within the Going Digital project and as part of the research programme of the Economic Statistics
Centre of Excellence (ESCoE). Our work has benefitted greatly from discussions at the joint OECD-
ESCoE workshop “Data on Data and Data Flows” which took place in November 2018 at the National
Institute of Economic and Social Research (NIESR) in London. We are grateful for comments and
feedback received from participants at the WPMADE meeting in May 2019 at the OECD in Paris and the
AEA/ASSA 2020 Annual Meeting in San Diego. Further, we thank a number of individuals for very
helpful comments and suggestions on earlier drafts of the paper. This includes Nadim Ahmad, Sarah Box,
Alessandra Colecchia, Carol Corrado, Diane Coyle, David Gierten, Daniel Ker, Wendy Li, Michael
Mandel, Christian Reimsbach-Kounatze, Rebecca Riley, Rachel Soloveichik, Peter van de Ven and
Andrew Wyckoff. All remaining errors are our own.

Executive summary
The process of collecting, aggregating and analysing data for the purpose of successful operation is nothing
new for companies. However, the amount and variety of data they use has increased dramatically in recent
years. In this paper, we investigate how the economic value of business data can be conceptualised and
measured.
In the first step towards developing a taxonomy with which to categorise data and conceptualise and
measure its economic value, it is helpful to discuss the concepts of data-enabled and data-enhanced
businesses. This distinction focuses on the core function of data within each business model: data-enabled
businesses are companies that have developed revenue generation strategies fully reliant on data and that
would not exist without access to large amounts of data and advanced data analytics. On the other hand,
data-enhanced businesses exploit data to better coordinate pre-existing business operations, facilitate
decision-making and to introduce new goods and services; data does not alter or determine their core
business models.
However, the distinction between both is not dichotomous and we discuss four related revenue generation
strategies: data-enabled businesses will likely generate revenue from (1) selling or licensing data, or (2)
selling entirely new data-related products, whereas data-enhanced businesses are more likely to use data
to (3) improve existing products, or (4) overall productive capacities and business efficiency. Examples of
such companies and relevant data monetisation strategies are discussed in detail. These four strategies are
then put together in a framework of data monetisation strategies across different business models, which
can be used to assess the degree to which businesses are relying on data to generate revenue (and how this
changes over time).
Next, to further address the data value-generation process we elaborate on the concept of the ‘data value
chain’. It is composed of four stages: i) data collection, ii) data aggregation, iii) data analysis, and iv) data
use and monetisation, all of which are underpinned by data storage and (cross border) data flows. However,
the economic value of data is a combination of many factors, including the information content that it
carries, demand for data, and its actual or intended use. Despite obvious empirical challenges in measuring
data characteristics in a standardised way, we review key properties that are associated with economic
value from the perspective of businesses. These include linkability, accessibility, timeliness,
trustworthiness, and scarcity.
To provide a broader context we review most commonly used data taxonomies and typologies. It should
be noted that a universal taxonomy covering all types of data does not exist. Rather, different policy
questions and considerations around data use by businesses call for different classifications and typologies.
We review several ways to distinguish data types, based on the funding for data generation or collection,
data usage rights, data subjects, method of data generation, and data source.
We extend our discussion to include the international nature of many business models, by looking at the
role of cross-border data flows and barriers to these. While one might argue that these flows do not
influence the value of data per se, they are important as they critically enable the value creation processes
within the ‘global data value cycle’. We discuss different types of cross-border data flows and conclude
that the volume of data transferred is not necessarily helpful when trying to establish their value in an
economic sense. At the same time, the formation of barriers to cross-border data flows can have a material
impact on the ability of companies to monetise their data as the creation of economic value is often very
much dependent on the ability to move and aggregate data across a number of locations scattered around
the globe. Different types of barriers are discussed, in addition to studies estimating the magnitude of their

economic impact, and indicators of restrictiveness such as the Digital Services Trade Restrictiveness Index
(DSTRI).
The paper concludes with an overview of the most promising avenues for measuring the economic value
of data and a discussion on their feasibility in both the short- and long-term. Here we also discuss what
aspects of data and data flows are already measured (e.g. use of cloud computing services, data
management & processing and e-commerce). We assess different measurement approaches, including the
valuation of data based on market prices (e.g. based on data brokers, data insurance, or data breaches),
valuation of data as a knowledge-based or intangible asset (e.g. using cost-based methods), valuation of
data based on the business model (i.e. income-based approach as in the key framework developed in this
paper), or input-output tables of the economy.

Table of Contents
Foreword ................................................................................................................................................ 3
Measuring the Economic Value of Data and Cross-Border Data Flows: A Business
Perspective ............................................................................................................................................. 4
Abstract ................................................................................................................................................ 4
Acknowledgements .............................................................................................................................. 4
Executive summary.............................................................................................................................. 5
1. Introduction ....................................................................................................................................... 9
2. The use of data in new business models: towards a taxonomy .................................................... 10
2.3. The use of data: a taxonomy of business models and business types ......................................... 14
3. A taxonomy of data types for the purpose of economic measurement ....................................... 19
3.1. Review of data taxonomies or classifications ............................................................................. 19
3.2. Data characteristics ..................................................................................................................... 20
3.3. Data types ................................................................................................................................... 21
4. Cross-border data flows .................................................................................................................. 24
4.1. Creating value by transferring data across borders ..................................................................... 24
4.2. The scale and scope of cross-border data flows .......................................................................... 26
4.3. Barriers to cross-border data flows ............................................................................................. 28
4.4. Measurement challenges arising from cross-border data flows .................................................. 29
5. Approaches to measuring the economic value of data ................................................................. 31
5.1. What aspects of data are already measured?............................................................................... 31
5.2. Review of approaches to measuring the economic value of data ............................................... 32
6. Discussion and implications for measurement agenda ................................................................ 37
References ............................................................................................................................................ 39
Notes ..................................................................................................................................................... 44
Tables
Table 3.1. Overview of different data types .......................................................................................... 21
Figures
Figure 2.1. The Global Data Value Cycle ............................................................................................. 13

Figure 2.2. Businesses and data............................................................................................................. 15
Figure 2.3. The use of data: data-enabled vs. data-enhanced business models ..................................... 16
Figure 2.4. Data monetisation across business models and sectors....................................................... 17
Figure 3.1. Data categories as set out in ISO/IEC standard 19944 “Cloud services and devices: Data
flow, data categories and data use” ............................................................................................... 20
Figure 4.1. Total used cross-border bandwidth, 2005-2017 .................................................................. 25
Figure 4.2. Global Submarine Cable Map, 2018 ................................................................................... 27

Figure 4.3. Global data centre traffic, by type and Consumer Internet Protocol (IP) traffic, by sub-
segment, 2015-22 .......................................................................................................................... 30
Boxes
Box 2.1. How data enable business functions across different business models .................................. 10
Box 2.2. Valuing data in business models based on providing ‘free’ digital services .......................... 12
Box 4.1. Types of barriers to cross-border data flows........................................................................... 28
Box 5.1. Bankruptcy case studies highlighting the value of business data ........................................... 34

1. Introduction
In their most basic form, data can be described as the unordered and unprocessed
representation of any types of observations that are quantified and stored in symbols
(OECD, 2013a). As part of the digital transformation, the use of data has become
increasingly valuable to businesses and has spurred the creation of entirely new business
models. Data, or more precisely the information data hold, often function as a critical input
into the production processes of goods and services. For some firms, data are the most
valuable asset they own. Businesses leverage data as business intelligence, use it for the
purpose of process optimisation, improvement of products and services, and in research
and development (R&D) activities (Magalhaes and Rosiera, 2017).
The economic literature looking specifically into the economic value of data for business
operations and business productivity is limited. 1 This also means that neither economists
nor statisticians have, as of yet, developed a consensus on the best way to conceptualise,
classify, and value different types of data and data inputs into the production process. At
the same time, data are becoming ubiquitous in business operations and ever more actions
are leaving digital “footprints”. Nevertheless, its economic “value” is often ambiguous,
making conceptualising and measuring it both theoretically and practically difficult.
These measurement issues are further exacerbated by the international orientation of many
business models, which involve considerable cross-border data flows within and between
companies. However, without the proper measurement and valuation of ‘data’ it is difficult
to assess its role in terms of firm performance or product market structures. Conceptually
and empirically, measurement problems arise at the company, industry and country levels.
They potentially undermine the accuracy of current economic statistics and, as a result, the
development of effective policies aimed at fostering growth and inclusive development in
the digital era.
The aim of this paper is to investigate how to measure and value data with a focus on
business data. We first discuss the role of data monetisation as a basis for new and
traditional business models (Section 2). We then review existing data taxonomies or
classifications and propose a new framework focused specifically on measuring the
economic value of data used by businesses (Section 3). Next, we examine the increasing
importance of cross-border data flows for international business in a global economy
(Section 4). Here we also discuss different drivers of cross-border data transfers and
barriers to them. Finally, we describe the most promising avenues for measuring the
economic value of data (Section 5) and discuss their feasibility in the short- and long-term
and implications for the economic measurement research agenda (Section 6).
Naturally, there are aspects related to the value and measurement of data that are beyond
the scope of this paper. This includes the development of detailed ‘ready-to-go’ options to
empirically measure different types of business data. While this is a necessary and desirable
outcome of future research in the area, the main aim here was to conceptually place the
business model related to the monetisation of data at the centre when thinking about how
to measure the value of data. Nevertheless, the second part of the paper discusses a number
of empirical measures that we could find based on existing studies or borrowed from related
fields. Finally, while our discussion focusses on the private returns generated from data,
we are aware that this has clear implications for discussions on under-priced or under-
served goods or services. For example, businesses deciding to invest in (or acquire) data

likely do not “price-in” all potential externalities of that data being shared more widely
(e.g. with researchers or the government). This could be due to a lack of awareness or
business incentives, when data monetisation is not feasible or profitable. 2
2. The use of data in new business models: towards a taxonomy
2.1. Data-enhanced and data-enabled business models
Companies have long been involved in the process of collecting, aggregating and analysing
data for the purpose of running their businesses. However, it is only in recent years that the
scale and scope of data used by companies has changed and that data have moved to the
core of many business models. 3 Some even suggest that “so much of the world is
instrumented that it is difficult to actually avoid generating data” (OECD, 2016b: p.52).
On the one hand, data help to better coordinate existing business operations (e.g. supply
chains), facilitate decision-making and enable the improvement or introduction of new
goods and services. In Box 2.1 we provide a number of practical examples of the type of
processes concerned. Crucially, this ‘incremental’ digitalisation does not alter or determine
the core business models of these companies. We refer to these types of companies as
“data-enhanced businesses”. 4
For data-enhanced businesses, data facilitate the creation of new value within an
established business model. This is closely related to the concept of “data-driven innovation
(DDI)”, as outlined in OECD (2015). 5 This intuition is confirmed by an empirical study
based on 18,000 manufacturing firms in the US that shows how establishments that rely on
data-driven decisions exhibit higher productivity and higher output in terms of value-added
(Brynjolfsson and McElheran, 2016). More recently, Hughes-Cromwick and Coronado
(2019) discuss in detail how the automobile, energy and financial services sectors rely on
data for their short-term and long-term business decisions. By contrast, some businesses
could be thought of as fully digital, or ‘data-native’ companies. Data are the lifeblood of
their operations and the key enabler of their core revenue-generating activities. This
includes online platforms that rely on data and data analytics to match users and providers
of goods or services. We refer to these as types of companies as “data-enabled
businesses”. 6
Box 2.1. How data enable business functions across different business models
Potential uses of data flows to improve business functions:

• Automation of supply chains
• Consolidation of back-office operations
• Scalability of software via the cloud
• M2M communication (e.g. sensors, Internet of Things)
• Digital collaboration between teams (e.g. in R&D, sales, procurement)
• Online purchases of goods and services
• Use of mobile apps to deliver products and services
• Use of online platforms as intermediaries
• Analysis of big data
Examples reported by the Information Technology & Innovation Foundation, ITIF (2015):
1. Mining: To enhance efficiency and cut costs, Rio Tinto aggregates data from its
laboratories, control systems, maintenance logs and surveillance cameras, located in
different mining sites globally. It is reported that every day around 30GB of data are

MEASURING THE ECONOMIC VALUE OF DATA AND DATA FLOWS | 11
transferred between its global sites and a data processing centre in Brisbane, where the
information is analysed.
2. Manufacturing: Volvo and Scania can aggregate real-time vehicle information such
as location and diagnostics data to improve driver safety, environmental impact and supply
chain management. Volvo aggregates data from all countries in a centre in Sweden.
Similarly, Unilever is operating two global data centres with a total of 4 000 servers, which
enables it to use global consumer analytics based on a very large dataset.
3. Oil & Gas: Royal Dutch Shell aggregates data from around 10,000 sensors placed
in oil wells using three cloud-enabled global data centres. These sensors have been
co-developed with HP and enable the company to find new resources based on high-
resolution seismic data.
4. Aerospace: Boeing aeroplanes transmit information in-flight which is analysed in
near real time to identify problems early on and enable maintenance crews on the ground
to be prepared for when the plane arrives. This reduces turn-around times and delays.
Reportedly, a single engine of a Boeing 737 produces 20 TB of data per in-flight hour
(though it is not clear what proportion of this data is actually transmitted).
5. Healthcare: Hospitals increasingly send medical images to doctors located in other
countries for the purpose of diagnosis. This can reduce waiting times for patients. For
example, the Nation al Health Service (NHS) in England outsourced the processing of MRI
scans using the company Alliance Medical which has around 200 imaging sites across
Europe. The Swedish company Hermes Medical Solutions offers cloud-based software
applications to share medical images across 30 countries, though 95% of patient data are
stored in Sweden.
Sources: ITIF (2015), Business Roundtable (2015), US Department of Commerce (2016).
Data-enabled businesses are companies that have developed fully digital business models,
that would not exist without access to large amounts of data and advanced data analytics
(e.g. Amazon, Uber, Twitter, Booking.com and Airbnb). 7 In addition, many start-ups are
founded without a clear initial revenue model but with the expectation of future data-related
revenue streams. For example, TravelPerk is a start-up using a freemium 8 business model
based on taking affiliate commissions on bookings. However, “down the road, it also has
its eye on generating a data-based revenue stream via paid-tier trip analytics”
(TechCrunch, October 2018). A more mature transport and mobility app, CityMapper,
monetises some of its data on users’ travel patterns by providing a transport network
analysis tool. It has also attempted to run its own bus service to compete directly with other
forms of transport and is developing new software for buses to transmit real-time data to
scheduling systems.
For both data-enabled and data-enhanced businesses, the value-added related to data is
often a by-product of data analytics and data itself. Moreover, as highlighted by the
examples above, the ways in which firms monetise the data-data analytics nexus often
depend on the specific business model they adopt (Li et al., 2019). 9 Therefore, when
discussing the value of data, it is useful to consider a data value chain or data value cycle
(OECD, 2013b; Li et al., 2019).
The recent literature on measuring ‘free digital services’ holds some further insights for
how businesses turn data in economic value. Some studies are reviewed in Box 2.2.

Box 2.2. Valuing data in business models based on providing ‘free’ digital services
• Digitalisation now affects almost all aspects of the economy and new digital goods pose
great challenges to our current measures of consumer welfare and GDP (Coyle, 2017). One
challenge is that new digital services are often free at the point of use, especially for data-
enabled businesses that rely on network effects for gathering data (i.e. social networks).
However, this does not mean that the data have no value. For example, the market
capitalisation of Facebook, which also owns WhatsApp and Instagram, exceeded USD 600
billion in mid-2018 illustrating the considerable value the market believes that revenue
streams based on data collected via “implicit” or “barter” transactions can generate.
• The “professional network” LinkedIn which is operating a data-enabled business model,
was acquired by Microsoft in 2016 for a total of USD 26 billion, which makes it possible
to link data on individuals with other Microsoft products such as Office and Outlook and
in turn increases their value. Shapiro and Aneja (2019) estimate that revenues associated
with monetising Americans’ personal data collected by the major Internet search engines,
social media platforms, data brokers, credit card companies and healthcare data businesses
amounted to more than USD 78 billion in 2018.
• One attempt to derive the value of ‘free’ digital services, e.g. via massive online choice
experiments as done by Brynjolfsson, Collis and Eggers (2019). Similarly, Nakamura et
al. (2018), treat free content as a barter transaction where onsumers and businesses receive
content in exchange for exposure to advertising or marketing, and ultimately, “households
are treated as active producers of viewership services that they barter for consumer
entertainment.” (Nakamura et al., 2018; p.2). Since consumers provide their data for free
in exchange for free services this approach is directly related to the value of data. Ahmad
and Schreyer (2016) describe these triangular transactions between the users of digital
products, advertisers (as data users) and the service providers, but they also highlight the
conceptual issues that these types of valuation approaches entail. For example, while it is
difficult to interpret the magnitudes of valuations, Coyle and Nguyen (2020) show that
changes and rankings in valuations of free digital products can be very insightful.
• Another method is to use massive online choice experiments as done by Brynjolfsson,
Collis and Eggers (2019), who estimate that online services such as Wikipedia or
WhatsApp generate annual consumer surplus in the range of billions of US dollars. Since
people are not directly paid for their data by social media platforms, it is difficult to say
how valuable it is. However, considering the large amounts of welfare that people seem to
get from these free services can serve as an indication. Future extensions of these types of
studies might consider asking people more directly about how they would value their data
in monetary terms.
2.2. The global data value chain

To gain some further insights into how exactly data can generate value for businesses we
build on the concept of a ‘data value chain’ (see e.g. Rayport and Sviokla, 1995; Visconti
et al., 2017; OECD, 2015). We distinguish four stages in the process of deriving value from
data: i) data collection, ii) data aggregation, iii) data analysis, and iv) data use and
monetisation. All four stages are underpinned by data storage and (cross border) data flows
which occur throughout (see Figure 2.1).

The collection of raw data (stage 1) can take place in a single place or across many places
and countries simultaneously. However, the additional value creation is limited when
businesses are faced with barriers to data transfers and aggregation (stage 2), particularly
when these involve cross-border transactions. The next stage is data analysis of aggregated
data or multiple combined datasets. This can take place in yet another location and hence
involve (further) data transfers across borders. 10 Finally, the monetisation stage generates
additional data that can feed into the data value chain to make it a global ‘data value cycle’.
When it comes to the storage and computing that underpins data collection, aggregation,
analysis and monetisation, businesses can benefit from economies of scale by centralising
all of their data in one or a few locations. Others might find it beneficial to store copies in
data centres across the globe, e.g. to protect it from disasters and to reduce access times
(latency). Cloud computing is a key enabling technology of this process.
Figure 2.1. The Global Data Value Cycle
Source: Authors’ own elaboration.
The concept of the ‘data value cycle’ can also pose challenges to the standard distinction
between process and product innovation. For instance, for many automobile companies,
data serve as an input in the production process, constituting a process innovation. In a
similar vein, gathering more data can also fuel the improvement of the algorithms used for
data analytics. Examples include Facebook’s process of ranking of all available posts to be

displayed on a user’s news feed (McGee, August 2013) or Amazon’s A9 algorithm used to
match users’ search queries to products they are most likely to purchase (Wikipedia, s.d.).
On the other hand, the business models that involve data collection coupled with advanced
data analytics could eventually lead to the introduction of new data-based services. Here
we are thinking of manufacturing firms that extend their product range as well “servitise”
their business model – making a gradual transformation from a traditional goods
manufacturer into a (data-driven) services provider. One example is the car manufacturer
BMW that offers “CarData” – a set of individually tailored customer services based on car-
generated telematics (BMW, s.d.). Similarly, John Deere, a manufacturer of agricultural
machinery, uses IoT solutions to offer data-fuelled and AI-driven agritech platforms (John
Deere, s.d.). It should be noted that these data-driven improvements often allow for
collecting further data and, in turn, broaden the potential for even more new or improved
products. The examples above also relate closely to economies of scale and scope
experienced by both data-enabled and data-enhanced businesses, and their implications for
economic measurement. Manufacturing firms that add services to their portfolio are not
new, and earlier approaches may indeed offer insights into how to integrate data-enhances
business models into economic statistics. The key difficulty here is to separate the value-
added from the physical product from the data-enabled service.
Demand economies of scale as demonstrated by constant improvement of Facebook’s
algorithm thanks to the increasing number of its users, or data-driven solutions for
manufacturing companies where accuracy improves with more data inputs, further
complicate the assessment of the value of underlying data that is exploited by the algorithm.
Similarly, measurement is difficult because data-reliant companies can offer different-but-
similar customised varieties of new products at a very low marginal cost (economies of
scope). 11
Considerations of economies of scale versus scope, product versus process innovation, and
servitisation of manufacturing, demonstrate that answering the question of “What
determines the value of data?” is a non-trivial task. It is fair to state that there is no
standardised approach for measuring the ‘value’ of data and its importance from a business
perspective. However, there seems to be a consensus that simple measures of data volumes
are not good enough to approximate the economic value of data, or data flows (OECD,
2019a; HMT, 2018).
Straightforward data quantity measures, such as bits and bytes, have little connection with
the information contained within each data ‘unit’. As an illustration, the transfer of a new
car design intuitively carries a different value potential than an individual’s purchase
history or clickstream. Cisco (2018) show that video accounted for 75% of all Internet
Protocol (IP) traffic in 2017, the greatest single category of online data flow. Considering
data volume versus value is further complicated by the use of ever-improving compression
techniques widely applied to flows of data. Understanding the context in which data is used,
combined with a categorisation of different data types and characteristics, is essential for
the purposes of economic measurement of the value related to data.
2.3. The use of data: a taxonomy of business models and business types
The economic value of data is a combination of many factors, including the information
content that it carries, demand for the data, and its actual or intended use. 12 This is closely
related to the idea of relevance – the ways in which data can be exploited within a particular
business model. In a survey conducted by the Economist Intelligence Unit in 2015, close

to 90% of almost 500 executives stated that data were now used by “the majority or all
parts of the business”. (The Economist Intelligence Unit, 2015). This serves as an indication
of how relevant data have become across constituent parts of many businesses.
Interestingly, 65% of respondents stated that their company had assigned a monetary value
to the data they store (see Figure 2.2). Furthermore, 83% reported that their companies
made use of data to increase profitability of existing products and 70% saw a clear business
case to develop new goods or services based on their data. Finally, more than half saw a
business case for selling data owned by their company.
Figure 2.2. Businesses and data
Do you agree or disagree with the following statements about your organisation?
Agree Neither agree nor disagree Disagree Don't know
Have used data to make exis ting

products or services more profitable
See a clear business case for starting

a new business dedicated to developing
new products or services from our data
Have given a monetary value

to the data that we store
0 10 20 30 40 50 60 70 80 90 100
% respondents
Note: Agree summarises “strongly agree” and “somewhat agree” responses. Disagree summarises “somewhat
disagree” and “strongly disagree” responses. Based on a survey of 476 senior executives worldwide.
Source: OECD based on “The Business of Data”, The Economist Intelligence Unit report (2015).
Some companies have developed internal data valuation methods and prepare internal
measures of performance (or key performance indicators, KPIs) based on such data
characteristics. For instance, the asset valuation models summarised in Laney (2011)
include methods to approximate the market value of information by trying to capture the
income that can be generated by selling, renting or bartering data and the loss value of
information, which aims to measure the cost to replace the data and the financial impact to
the business if the data were lost. Gartner (2016) propose two sets of firm KPIs related to
the level of (i) firm digitalisation, as well as the (ii) direct improvement of current business
functions based on data and new business models.
Based on the previous two paragraphs it becomes clear that the monetisation of data can
come in different shapes and forms. This includes selling data directly, improve operations
by sharing data, use data to develop new products and services, and improve existing
products and services. However, what they have in common is that they are directly related
to a particular business model.
The taxonomy developed and presented here focuses on the role of data across different
business models and within businesses that combine multiple business models. A key factor
is the actual or intended use of data to generate revenue now or in the future. Below we

also discuss several data characteristics associated with the value of data for businesses,
though we acknowledge the empirical challenge in measuring those in a standardised way.
Taken together we believe this can provide the conceptual basis for the economic
measurement and valuation of data since data only become valuable to businesses if it is
(or is intended to be) used in a profit-generating way.
In our taxonomy, the ‘business model’ takes centre stage as it describes the ways in which
a company generates and captures economic value, i.e. new or improved revenue streams.
In our understanding, the adopted business model seeks to answer central questions,
including what to offer to which customer group as well as how to deliver economic value
and for what price. Hence, the business model seems an appropriate starting point when
thinking about how businesses rely on data – directly or indirectly – in order to create value,
develop a competitive edge and, ultimately, to generate streams of revenue. Certainly, a
company can adopt multiple models at the same time.
In Figure 2.3, we illustrate how data could be used in different business models within a
company. We distinguish four categories of data-related business models and illustrate how
they relate to the distinction between data-enhanced or data-enabled businesses:
Category 1. Selling or licensing raw or aggregated data;
Category 2. Developing and selling new data-related products;
Category 3. Use data to improve existing products;
Category 4. Use data to improve production processes or business efficiency.
Figure 2.3. The use of data: data-enabled vs. data-enhanced business models
Figure 2.4 provides a schematic illustration of our taxonomy that relates stylised business
types to the four different data-related business models outlined in Figure 2.3 based on the
use of internal and external data. The key metric we propose to measure the economic
‘value’ of data for a business is the share of total revenue that is derived from the
monetisation of data (in some form or another). 13 For the distinction of data-enabled and
data-enhanced businesses, we refer to their ‘core’ business model, defined here as the
‘main’ source of revenue generation (i.e. more than 50% of total revenue). This share is
illustrative as estimates do not appear to be readily available. Nevertheless, developing such
a metric in the future seems highly relevant. For example, findings based on the McKinsey
Global Survey (2018) suggest that data monetisation is an increasingly important driver of
revenue growth. It is reported that the monetisation of data contributes to 10% or more of
the total revenue for 32% of high-performing businesses and 9% of all other businesses. 14

Figure 2.4. Data monetisation across business models and sectors
Source: Authors’ own elaboration
The coloured bars show – in a highly stylised way – how much the different revenue
streams of a business potentially contribute to its overall revenue. In addition to the four
data-driven business models (Figure 2.3) we separate ‘traditional’ sources of revenue,
which in simplified terms relates to ‘non data-driven’. Whenever data-driven revenue
contributes to more than half of total revenue (the dashed line), we speak of a ‘data-enabled’
business (as opposed to a data-enhanced business). In addition, for each type of business
we illustrate what types of internal and external data it may have access to for monetisation
purposes. Again, this is a highly stylised representation for the simple purpose of
classifying different types of businesses with regards to how they can generate economic
value from data. By using internal or external sources of data, or a likely a combination of
both, businesses can find new ways to create economic value for themselves and their
customers. Note that external data, despite possibly being free (or non-rival), carries value
in the production process when integrated with the overall business model. The last row in
Figure 2.4 provides examples of such data monetisation strategies.
The first column refers to the example of a manufacturer of transport goods (e.g. Airbus,
Boeing); while the insights would be analogous for other manufacturing businesses. For
instance, the bulk of total revenue is likely driven by the sales of aircraft and parts as well
as providing maintenance and training services. However, based on sensors that can be
installed in those aircraft, the business is able to reduce operational interruptions, maximise
aircraft utilisation and flight operations (data monetisation category 3) as well as offer a

new additional service of real time remote access to aircraft data (category 1) (Airbus,
October 2018) and (Boeing).
Similarly, based on the second column of Figure 2.4, the main stream of revenue for an
energy company would be the provision of electricity. However, using customer energy
usage data combined with freely available external data on energy prices, it can design new
pricing schedules, e.g. peak usage surcharge (category 4). Alternatively, it may decide to
monetise those data by directly selling them, if compatible with data privacy regulations
(category 1).
On the other end of the spectrum, data-enabled businesses would only be able to offer a
minimal fraction of their goods or services if they could not collect the data (or rely on
externally acquired data and data analytics) to improve their services, or to license the data
access. In this sense, their revenue streams are completely data-dependant. However, there
exists a degree of differentiation among those businesses in terms of their reliance on a
specific business model. Li et al. (2019) show how this is the case for different online
platforms (column 4 in Figure 2.4). For instance, the revenue of the online platform
Amazon Marketplace predominantly comes from the supply of a data-enabled service – a
buyer-seller matching platform (category 2). However, Amazon also uses data to offer
sellers the opportunity to promote their products to some individuals (category 3) 15, and
licenses access to internally collected customer behaviour data (category 1). The company
also constantly uses data to improve its algorithms (category 4).
In addition, based on a data-driven understanding of customer needs, Amazon also offers
its own products that directly compete with independent sellers on its platform. This
includes ‘Amazon Basics’, a private-label for home goods, office supplies and tech
accessories, launched by Amazon in 2009. Based on the vast amounts of data it can access,
these products can be designed to suit the taste and price range of specific groups of
consumers. 16 At the extreme, almost all of the revenue of data brokering companies such
as Experian or LexisNexis is derived from licensing the access to data (category 1) and the
delivery of analytical data-based services (category 2).
With advancements of artificial intelligence (AI) new business models will emerge and
firms will undergo further transformations. It is possible to imagine that, in the near future,
data-enhanced businesses such as car manufacturers will see major changes in their core
business models as they shift from offering ‘traditional’ goods (i.e. sale of vehicles) to
‘digital’ services (e.g. sale of ‘mobility solutions’ like on-demand or subscription-based
rental of vehicles via the Internet). Some examples of firms that already undertook this shift
include German car manufacturer BMW, which started to offer mobility services, and
Finland-based Konecranes – a manufacturer of cranes that uses data analytics to offer a
new service called “TrueConnect”. The service allows continuous monitoring of cranes and
their operators (e.g. to predict wear and tear or maintenance requirements) with the final
goal of selling crane “uptime or number of lifts” rather than just cranes. 17

3. A taxonomy of data types for the purpose of economic measurement
3.1. Review of data taxonomies or classifications
A broad taxonomy that distinguishes economic and social dimensions (i.e. personal vs.
non-personal data, open vs. closed data) from the technical dimensions of data (i.e. user-
vs. machine-generated data) is developed in OECD (2013a). These are mainly discussed
within the context of ‘big data’, which prima facie refers to data volume, making it difficult
to link it to its economic value. Various classifications of personal data are reviewed in
OECD (2013b), including a discussion of methods for assigning monetary values to
personal data. It should be noted that a universal taxonomy covering all types of data does
not exist. Rather, different policy questions and considerations around data use by
businesses call for different classifications and typologies.
The Swedish National Board of Trade (2015) provides a definition and taxonomy of
personal data in order to exemplify what kind of data is used by companies and what
categories of personal data are relevant for international data transfers. Their report
differentiates between volunteered, observed and inferred personal data. Based on a small
survey of Swedish firms, the report relates different categories of data transfers to the
specific business functions they enable or complement (e.g. technical data to upgrade
software). Beyond personal data, it also proposes a classification of different types of data
based on how data are used:
• Corporate data;
• End-customer data (business-to-consumer; B2C);
• Human resources data;
• Merchant data (business-to-business; B2B); and
• Technical data.
The U.S. Department of Commerce (2016) provides a categorisation of four types of data
flows for the purposes of discussing how their cross-border transfer may be captured by
economic statistics. The aim of their taxonomy is to scope the potential for measuring the
economic impact of cross-border data flows and to put forward recommendations to
improve available economic metrics for them. The four types are:
• Purely non-commercial data traffic (e.g. government and military
communications);
• Transaction data flows between buyers and sellers at a market price (e.g. online
banking or advertising);
• Commercial data and services exchanged between or within businesses or other
related parties at zero market price (e.g. design information);
• Digital data and services delivered to and from end-users at zero market price (e.g.
free email, free maps and navigation, social media).
In the advent of the EU’s General Data Protection Regulation (GDPR) enforcement date in
May 2018, many businesses searched for support in classifying their data with the goal of
coming up with GDPR compliant data use statements. Many have relied on the existing
ISO/IEC standard 19944 on “Data flow, data categories and data use” 18, which, most

importantly for our purposes, contains a well-developed section discussing different data
categories (Figure 3.1). The four main categories are customer content data (incl. biometric
and contact information), derived data (identifying either end-users or organisations), data
by cloud service providers, and account data (payment or admin data). This is attractive
from the point of view that companies are likely to both understand and use this
classification. However, the primary focus of this typology is data privacy in the context of
individual and personal data rather than its value in an economic sense.
Figure 3.1. Data categories as set out in ISO/IEC standard 19944 “Cloud services and
devices: Data flow, data categories and data use”
Source: Reproduced from ISO/IEC 19944:2017. Copyright ISO/IEC. All rights reserved.
3.2. Data characteristics
Data have a number of distinctive features, which contribute to their unique nature as an
intangible asset. These are an important determinant of usability within different business
models. Most importantly, data is non-rivalrous in nature (OECD, 2013a; Mandel, 2017)
and it is not subject to the standard wear and tear of tangible assets. This means that data
can be used multiple times without inherently diminishing their value. In principle, data
can be exploited and re-exploited infinitely at low marginal cost. Usually, it is data
infrastructure and analytics that mainly determine the costs of data re-use. 19
A simple Google search for ‘data monetisation’ returns almost 18 million results discussing
countless data characteristics that are potentially associated with value. In an early
contribution on this topic, Laney (2011) discusses a number of factors that affect data
‘utility’ and, ultimately, economic value from a business perspective - providing a list of
‘objective’ and ‘subjective’ data characteristics. Although they could prove difficult to
measure in a standardised way, those data features are fundamental to how data can be
utilised to generate value for both data-enabled and data-enhanced businesses. Data are
most likely to provide a basis for monetisation and value-creation if they are:
• Linkable 20 – can be merged with the other datasets;
• Accessible – easily retrieved and/or integrated into business processes;

• Disaggregated – at the desired level;

• Timely – updated with sufficient frequency to meet the business requirements, e.g.
annually, daily or in real time;
• Trustworthy – deemed credible by those using it; data are unbiased and impartial,
and do not depend on the judgment, interpretation, or evaluation of individuals (see
Open Data Watch, 2018);
• Representative - records do not contain missing fields, data are representative
enough to meet business requirements;
• Scarce – proprietary or secret, difficult to come by.
There are also some data characteristics that are often discussed in the context of big data
and its features – known as the “3 V’s of big data”: volume, variety, velocity (Gartner
2011). For example, data volume, understood as data being a collection of a sufficient
number of observations (closely related to the statistical power of data), could impact the
data’s value-generating potential. However, as noted above, volume alone will not be a
sufficient characteristic that determines economic value. When considering the costs of
storing and processing large amounts of data (though those have decreased drastically in
recent years), ‘hoarding’ large a volume of irrelevant data could even be detrimental to
business performance.21
3.3. Data types
By building on the OECD (2013a) classification, we can consider data ‘types’ based on
ownership or right to use, data source, funding of data collection and maintenance, data
access, and methods of collection. Note that the categories, listed in Table 3.1, are not
mutually exclusive as some datasets (acquired or generated) will fall into more than one
category. Crucially, it is often the case that there is a continuous mapping of data types
within a given criterion. For example, the distinction between what is personal and what is
non-personal data has become increasingly difficult with new data mining techniques or
data analytics solutions. It could be the case that new (or newly combined) data sources
enable the identification of an initially anonymised individual (e.g. comparing data on IP
address, cookies, geo-location, social media profiles, and clickstreams), thus making non-
personal data, personal. This process of “browser fingerprinting” is discussed in great detail
by Cyphers and Gebhart (2019) and often based on triangulating information on screen
resolution, language setting, time zone, and software version. 22
Table 3.1. Overview of different data types
Key aspect Data Type Description
Data that is funded, created, maintained and held

Funding of Private sector by the private sector, e.g. car company in-house
data collection data generated data on their production processes, or a
and database of purchases from an online grocery shop.
maintenance
Public sector data Data that is funded, created, maintained and held
by the public sector; Example: data on health

records of patients or individual tax records, or

data originating from the US GPS satellites 23.
Data with clearly defined ownership that is

protected by Intellectual Property Rights or any
Proprietary data other rights with a similar effect (OECD, 2019a);
this could include individual data as well as
organisational data 24.
Ownership or Data that is publicly available (as opposed to

right to use proprietary data), free to use by anyone for any
purpose without any legal restrictions. It is not
Open data protected by Intellectual Property Rights,
(or public domain copyrights or any other similar legal rights;
data) Example: data.gov.uk, which is a repository
maintained by the UK Government to make non-
personal UK government data available to the
public.
Personal data is any data that allows for the

identification of an individual data subject (OECD,
2013b). It can cover public and private sector data,
Personal data e.g. user-generated content (e.g. blogs, photos,
tweets) or geo-location data from mobiles as well
as public sector data (e.g. police records, social
security numbers).
Data subject
Organisational data describes data that allows for
the identification of organisations. This data is
Organisational usually controlled by organisations themselves,
data either legally or for contractual reasons. It can also
be held by public bodies such as tax authorities. It
is often commercially sensitive data.
User created data is data that has been made

available by an individual (e.g. telemetry tracking
data, consumer behaviour data collected through
User created data mobile apps or social media posts). This can be
volunteered data (i.e. “active”), observed data (i.e.
Data
“passive” or “implicit”), or derived data about a
generation
user (see also OECD, 2019c).
Machine generated data, e.g. machine-to-machine

Machine
communication (M2M); Internet of Things (IoT),
generated data
i.e. data collected from sensors.
Internal data is data that collected and
consolidated from different branches within a
business. For example, lists of purchase orders
Data source Internal data from the sales department, transactions from
accounting or any other internal source which is
responsible for recording information about a
business’ commercial interactions.

External data is data not collected internally, but

rather obtained from a source outside of company
External data - for instance, by purchasing access to a
proprietary database. This could be acquired data
as well.
In the previous sections of this paper, we have discussed the difficulties related to
measuring the value of data from the perspective of businesses. We have developed a
conceptual framework to approach the issue based on the business types, alongside
discussing some key data characteristics and types that are relevant for measuring the value
of data. It should be clear by now that this is no trivial exercise as it is related to the
longstanding question of the ‘value of information’. The following section adds another
dimension to this challenge, discussing the global nature of some data, including the use of
cross-border data flows that underpin many modern businesses operations.
While one might argue that these flows do not influence the value of data per se, we think
they are important as they critically enable the value creation processes highlighted in the
global data value cycle in Section 2. At the same time, the formation of barriers to cross-
border data flows can have a material impact on the ability of companies to monetise their
data. We first discuss the motivation of businesses to transfer data across borders, followed
by a brief discussion on both the scale and scope of data flows and the main barriers to
cross-border data flows.

4. Cross-border data flows
4.1. Creating value by transferring data across borders
The increasing digitalisation of the global economy is not only driving data flows within
countries but also across borders (European Commission, 2017). The main reason is that
digitalisation enables the physical detachment of data collection, aggregation or analysis,
storage, and use or monetisation, as outlined in Section 2. Each of these steps can take place
in multiple countries, raising various issues for the measurement and valuation of data and
data flows. This trend is likely going to accelerate due to the decreasing costs, as well as
increasing sophistication and ease of use of digital innovations such as cloud computing,
the Internet of Things (IoT), edge computing, artificial intelligence and machine learning
tools, robotics and automation, and digital distributed ledger-based transactions (i.e.
blockchains).
Cross-border data transfers enable businesses to build and maintain complex global value
chains. In other words, the creation of economic value is often very much dependent on the
ability to move and aggregate data across a number of locations scattered around the globe.
The ability to transfer data internationally enables firms to effectively coordinate their
research and development, supply chains, production, sales, and post-sales processes (US
Department of Commerce, 2016; OECD, 2018a). While the rationale behind introducing
barriers to cross-border data flows can be grounded in national security or data privacy
concerns, or the desire to protect domestic markets, it is clear that in many cases
impediments to international data transfers can have severe negative economic impacts on
businesses and ultimately on complex value chains and trade.
In terms of volume, global cross-border data flows were estimated to have exceeded 700
terabytes per second (Tbps) in 2017 (see Figure 4.1), which means that since 2007 they
increased by a factor of 64 (McKinsey Global Institute, 2019).25 Of course, volume (e.g.
measured in bytes) is not necessarily a good indicator of the value of data, i.e. the
information that it encodes, which is related to the potential for current or future
monetisation (see also Section 2.2). A previous study estimates that since 2015 the value
of cross-border data flows has exceeded the value of cross-border merchandise trade
(McKinsey Global Institute, 2016). However, the challenge remains in defining what
metrics are appropriate for assessing the actual ‘value-added’ of such flows (e.g. per GB of
data transferred) as this can differ vastly across firms and sectors depending on both the
informational content of the data itself and the analytics applied to it (among other factors).
The location of economic value creation is also the basis for national-level indicators such
as GDP, and estimates of productivity, meaning there is a clear need to analyse data value
chains in detail for policy-making processes.
It is estimated that in 2014 the international flow of data added USD 2.8 trillion to the
global economy and this is expected to grow to USD 11 trillion by 2025 (McKinsey Global
Institute, 2016). 26 The key question for measuring the value of cross-border data flows is
related to how businesses actually derive value from data. As discussed in Section 2, data
can either be used to enhance or augment existing business functions, or to enable entirely
new businesses models and revenue streams. This distinction also has a cross-border
dimension that can further complicate the way data flows can be measured as data can be
transferred across national jurisdictions.

In many cases it is possible that businesses aggregate data in locations (e.g. headquarters
or data analytics centre), which are not located in the same country where the original data
is collected (see Global Data Value Chain, Figure 2.1). For example, consider a set of data
points that are volunteered by, and collected from, the users of an online social network
free of charge and, hence, they do not generate any financial transactions in the country
where the user is based. However, once those data points are transferred and aggregated
with millions of other data items from across the globe, they become the basis for data
analytics and thus for value creation. Eventually, they are monetised by the provision of
data-based services (e.g. in the form of targeted advertising) or by licensing access to the
database. This potentially can haves large implications for taxation of multinational
enterprises. 27
Figure 4.1. Total used cross-border bandwidth, 2005-2017
Terabytes per second

Tbps
800
704
700
600
500 462
400
311
300
218
200 149
103
100 70
30 47
5 7 11 19
0
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Source: Telegeography; McKinsey Global Institute analysis, 2019
Nowadays, cross-border transfers of data underpin virtually all business relations in

international trade, international investment and global supply chains. Data transfers are
also fundamental in enabling business operations of multinational companies. This includes
coordinating human resource and R&D operations across foreign affiliates, often using
cloud-based solutions (Swedish National Board of Trade, 2014). In addition, the initiation
and completion of cross-border transactions between buyers and sellers typically involves
transferring some data, e.g. regarding contracts or specifications of orders, packaging and
delivery.
Similar to intermediate inputs in the ‘analogue’ value chain, the data value chain may
require data to be transferred across borders multiple times, or even on an on-going (real-
time) basis. For example, multinational enterprises owning plants in various countries are
often integrated in the same production process and need to constantly monitor and
communicate their production volumes.

4.2. The scale and scope of cross-border data flows
Data transferred across borders are not fundamentally different from data transferred within
countries. However, latency rates can be higher for data packages sent across longer
distances, which can affect their value by making data less accessible, less timely, less
frequent, less reliable, etc.28 Latency can also increase costs of data analytics if data have
to be locally stored in multiple locations. Data flows across borders can take place within
businesses, between businesses (B2B), between businesses and consumers (B2C), and
between machines (M2M). The latter is driving increasing amounts of global Internet
traffic based in the IoT. Cisco (2017) estimates that by 2021 there will be 11.6 billion
connected devices, including M2M devices, and that the traffic they generate will increase
by 70% annually between 2016 and 2021. In Box 2.1, several examples were given of
businesses making use of cross-border data flows in their day-to-day activities.
Many manufacturing companies monitor the status, performance and condition of their
machines in different locations. For example, Volkswagen and AWS (Amazon Web
Services) announced the co-development of the ‘industrial cloud’ in March 2019 with the
aim of connecting “data from all machines, plants and systems in all factories”
(Volkswagen, s.d.). This is enabled by sensors using cellular or satellite connectivity to
send signals. Real-time data is then aggregated at a global level and with the potential to
be monetised via a new service. In another example, reported in Pepper and Garrity (2016),
a General Electric jet plane turbine generates around 400 GB of data per day, that could
potentially be used to diagnose problems or optimise efficiency in real-time.
Similarly, many online platforms offer their services globally, relying on cross-border data
flows to deliver digital matching services (e.g. Uber, Airbnb, eBay). At the same time, they
also collect transaction and consumer behaviour data in various locations, which further
need to be transferred across borders in order to be stored, aggregated and analysed. Finally,
insights based on aggregated global data serve as the basis for commercial services that can
be delivered in multiple locations (e.g. targeted advertising, or demand forecasting, price
elasticities of consumers).
At least at an inter-continental level, the bulk of data is transferred via submarine cables,
making their use a useful indicator for the volume of cross-border data flows. A second
indicator of the scale of global data infrastructure are large data centres that enable the
storage of data, as well as remote computing via the Internet (cloud computing).
The installed capacity of submarine cables can provide a global view on which markets are
most integrated in terms of data connectivity. In 2017, there were around 428 submarine
cables in service, with a total length of around 1.1 million kilometres (TeleGeography Blog,
February 2017). Figure 4.2 reveals that some parts of the globe are much more connected
than others. Not surprisingly, the trans-Atlantic route between the East Coast of the US and
Europe, and trans-Pacific from the West Coast of the US to East Asia are especially well-
served with capacity. However, there is significant and increasing capacity in South-East
Asia as well. 29

Figure 4.2. Global Submarine Cable Map, 2018
Note: White dots represent cable landing points

Source: Submarine Cable Map, TeleGeography (TeleGeography), www.submarinecablemap.com, licensed
under CC BY-NC-SA 3.0 (creative commons).
Data centres are another important component of the global data infrastructure. They
consist of servers that can be used exclusively by a firm (often in the form of private cloud
or co-location services) or rented on demand from cloud service providers such as AWS or
Microsoft Azure (public cloud), or anything in-between (hybrid cloud). By 2020, global IT
infrastructure spending on traditional on-premises data centres is predicted to be surpassed
by spending on off-premises private and public cloud (IDC, s.d.). By some estimates there
were around 8.6 million data centres globally in 2017, though this number is likely going
to decline in the future due to ongoing centralisation in larger data centres (Cushman and
Wakefield, 2016).
International bandwidth usage is increasingly shifting towards content providers such as
Amazon, Google, Facebook, and Microsoft. In the past few years, their share of
international bandwidth usage has risen sharply to reach 54% in 2018, a similar share to
traditional Internet backbone providers (Mauldin, 2017). To ensure they can meet
increasing demand for their services, content providers themselves have become large
players in the development of global data infrastructure, including the construction of
submarine cables and data centres (Mauldin, 2017).

4.3. Barriers to cross-border data flows
Governments across the world face various challenges to their existing legal doctrines
brought about by digitalisation and have various reasons for imposing restrictions to cross-
border data flows. These are often related to data protection and privacy considerations (see
Box 4.1 for a more detailed overview) and their existence can offer ways of implicitly
valuing the targeted data and data flows. In simple terms, lower economic costs to
restricting them would suggest a lower value of the data flows. Note that while these costs
could be considerable, there could be political reasons that would justify incurring them.
Research shows that the number of regulations on data flows across national borders has
been increasing rapidly in recent years (Casalini and Lopez Gonzalez, 2019; Information
Technology and Innovation Foundation, 2017). Ferracane (2017) reports that in 2017 the
total number of restrictions reached 87 across 64 economies 30, of which half were put in
place in countries located in the Asia-Pacific region. Casalini and Lopez Gonzales (2019)
estimate that as of 2019 the total count of data regulations, including different types of
regulations relating to data transfers and local storage requirements, is above 200.
Studies have estimated that the economic costs associated with restricted international data
flows can be large. This suggests that they are highly valuable. 31 For example, the US
International Trade Commission (2014) estimates that US GDP would be 0.1 to 0.3%
higher if foreign digital trade barriers were removed. Similarly, for the EU, barriers to data
flows are estimated to reduce GDP by 0.4 to 1.1%, depending on the strength of data
localisation requirements (ECIPE, 2014). Another study finds that data regulations lead to
a 0.48% reduction of real GDP in the EU (Bauer, Ferracane, and van der Marel, 2016).
Box 4.1. Types of barriers to cross-border data flows
• Local storage and local processing regulations (i.e. the requirement to keep and/or
process data on servers located within a given country).
• Data protection regulation (i.e. laws governing the collection, use and transfer of
personal data. The most comprehensive example is GDPR in the European Union, which
has been in force since May 2018).
• Competition and antitrust law adapted to digital markets (i.e. a set of economic
policies that are designed to favour the exporting conditions faced by digital, data-enabled
enterprises of a particular nationality, e.g. EU Parliament voting for the legal breaking up
of Google operations in the EU).
• Cybersecurity (i.e. a collection of technologies, processes and controls designed to
protect systems, networks and data from an unauthorised exploitation, e.g. EU working
towards the introduction of a certification process for IoT devices to increase their
cybersecurity).
• Intellectual property rights (e.g. on digital content such as music, movies and
books).
• Restrictions on Internet use, censorship and blocks against data transfers.
Sources: ECIPE (2014), Swedish National Board of Trade (2014), US Department of Commerce (2016), Bauer,
Ferracane, and van der Marel (2016), ITIF (2017), Meltzer and Lovelock (2018).

Others study cloud computing services, as a key advantage of the cloud is that services can
be accessed remotely from anywhere as long as users can connect to the Internet. This
means restrictions to cross-border data flows potentially have severe negative
consequences for access to these services, which can be used to store or process data, access
software, or host platforms – usually at much cheaper costs than on-premises IT equipment.
Berry and Reisman (2012) conclude that in 2010 cloud services contributed around
USD 1.5 billion to US exports and USD 1.4 billion to the sales of US foreign affiliates. In
a study on the UK, Coyle and Nguyen (2018) calculate that cloud computing potentially
added USD 2.8 billion to UK exports in 2017.
4.4. Measurement challenges arising from cross-border data flows
The US Department of Commerce (2016) quotes a number of challenges that impede the
measurement of cross-border data flows. This includes the fact that there is limited
evidence on how firms actually use cross-border data flows to generate value, but also the
lack of a standard nomenclature as well as official data on cross-border data flows. The
reports stresses that existing studies are infrequent, not transparent enough on their methods
and often designed to capture only ‘tech’ sectors. It is clear that these fundamental
roadblocks need to be resolved as a matter of urgency to enable robust and coherent
measurement of the economic value of cross-border data flows.
Some aspects of cross-border data flows are already measured. For instance, Ferencz
(2019) introduces the OECD Digital Services Trade Restrictiveness Index (DSTRI), which
builds on the OECD Services Trade Restrictiveness Index and provides information on the
barriers companies face while supplying services using electronic networks. DSTRI indices
are available for 44 countries and quantify the impediments that affect any services traded
digitally in the main five categories: (1) Infrastructure and connectivity32, (2) Electronic
transactions, (3) Payment systems, (4) Intellectual property rights, and (5) Other barriers
affecting trade in digitally enabled services. In a similar vein, Ferracane et al. (2018) have
developed the Digital Trade Restrictiveness Index (DTRI) that measures the restrictiveness
of digital trade for 64 economies. A sub-category of DTRI captures the intensity of
restrictions to cross-border data transfers, which can be applied within standard gravity
models in order to quantify the impact on trade flows. This information is of high relevance
as international trade agreements are starting to include clauses governing data protection
and data flows (e.g. the EU – Japan Economic Partnership) 33.
The drawback of this type of study is that it is more difficult to account for heterogeneity
of various data types and data monetisation strategies that are needed to map value to
industry sectors.
It is clear that approximating the value of data flows between countries by the volume of
data has significant limitations. 34 This follows directly from our description of different
data monetisation strategies and business models (see Section 2). The value of data depends
on the information content it carries as well as on how it is used now or intended to be used
in the future. 35 Another reason is that the bulk of data traffic is expected to take place within
data centres (see Figure 4.3; Cisco, 2018), which is further complicated in the case of cross-
border flows. For example, two businesses might exchange data within a data centre in a
third country and hence no cross-border flow might be recorded, even though a database
might have moved across borders in the legal sense. Therefore, it seems that looking at
sheer data volumes or installed bandwidth (as shown in Figure 4.2) can be a misleading
when analysing the associated value of these flows. 36

Another approach would be to look in more detail into what types of data are being
transferred, and how their distribution differs across industries and sectors. For example,
Cisco (2018) report global data centre traffic broadly divided into ‘consumer’ and
‘business’, with further sub-divisions by type of application (see Figure 4.3). 37 Such
detailed data could be analysed based on the data taxonomy developed in Section 3, and
help to classify which types of data flows are high- or low value-added. If broken down by
sector and related to types of barriers this could prove to be highly valuable for
policymakers. For example, it could be that both automotive manufacturing and motion
picture production are characterised by large volumes of data traffic, though the value
added in films might be comparably low.
Figure 4.3. Global data centre traffic, by type and Consumer Internet Protocol (IP) traffic,
by sub-segment, 2015-22
Note: Zettabytes per year (left-hand panel) and Exabytes per month (right-hand panel). “To data centre” refers
to traffic flowing from one data centre to another, for example, moving data between clouds, or copying content
to multiple data centres as part of a content distribution network. “To user” refers to traffic that flows from the
data centre to end users through, for example, streaming video to a mobile device or PC. “Within data centres”
refers to traffic that remains within a data centre, for example, moving data from a development environment
to a production environment within a data centre, or writing data to a storage array.
Source: OECD (2019a), Measuring the Digital Transformation, calculations based on Cisco Global Cloud
Index 2016-2021 and Cisco Visual Networking Index 2017-2022, January 2019.
One important challenge is posed by the transfer of data that takes place within
multinational enterprises (van der Marel, 2015). In some sense, this is similar to
measurement issues related to firms’ internal transactions of intangible inputs such as R&D
(UNECE, 2015, Rugman & Eden, 1985). A review of that literature might offer additional
insights. In terms of measurement, it would be relevant to look into networks of affiliates
within the same business group, as cross-border coordination is often driven by offshoring
motives.
Lastly, researchers should make use of new and experimental data sources. The Internet
itself can provide new opportunities for measurement of some aspects of global data flows
that are beyond statistical surveys. This includes methods of web scraping, IoT search
engines, and sensor-based data generated by “smart-meters”.

5. Approaches to measuring the economic value of data
5.1. What aspects of data are already measured?
Although the measurement of the value of data still remains a challenge, we are able to
measure some aspects of data and data flows already. For example, some national statistical
organisations 38 conduct surveys covering the use of the Internet and digital technology (e.g.
the Community survey on ICT usage and e-commerce in enterprises conducted by EU
member states, or the Survey of digital technology and Internet use in Canada). These
surveys are designed to measure the use of digital technologies by businesses. However,
they do not directly measure the intensity of use of data or digital technologies by
businesses, which could be added to existing surveys relatively easily. For example,
regarding the use of cloud computing services, Eurostat currently only requires a “Yes/No”
question. It should be noted however that such data still provide valuable insights. The
OECD has coordinated experimental analysis using these data for factor and cluster
analysis on the complete set of the digital technology-related questions in ICT surveys from
four countries (including Italy, Poland, Sweden, and the United Kingdom) with the aim of
the measuring of the digital maturity of businesses (OECD, 2019a). Furthermore, this
analysis will be extended to look at the adoption and impact of data-intensive technologies
including cloud computing services and data analytics.
Regarding other statistical sources, Ker (forthcoming, [DSTI/CDEP/MADE(2019)6])
reviews what relevant breakdowns are prescribed in standard classifications of both
products and industries, finding various products related to data storage, management, and
processing as well as identifying several industries where firms primarily engaged in
compiling and monetising databases are classified. However, data availability for such
detailed product and industry classes is found to be very limited in practice.
Data centres, being an important component of the global data infrastructure, serve also as
a potential source for informative statistics. Cisco (2018) presents some forecasts on data
transfers by origin and destination, distinguishing the following flows:
• From data centre to user (i.e. traffic that flows from the data centre to end users
through the Internet or IP Wide Area Network);
• Flows of data from data centre to data centre;
• Flows within data centres (i.e. traffic that remains within the data centre, though
excluding traffic within the same rack of servers).
As shown in Figure 4.3, it is estimated that by 2021 over 70% data flows will take place
within individual data centres. Interestingly, the report states that the traffic between data
centres is growing faster than either traffic to end-users or traffic within data centres. The
reason given is “the increasing prevalence of content distribution networks, the
proliferation of cloud services and the need to transfer data between clouds, and the
growing volume of data that needs to be replicated across data centres” (Cisco 2018). We
think that working with the providers of data infrastructure represents a unique opportunity
to gain vital insights into the use of data by businesses.

5.2. Review of approaches to measuring the economic value of data
This section provides an overview of various approaches to measuring the value of data
with a focus on how businesses use data as an asset in their production process to generate
revenue. Most data are not transacted on traditional markets where we can observe, and
measure associated prices. As a consequence, the use of market (or: market-equivalent)
prices is often not possible and alternative approaches need to be found. In the this section
we review some of general existing valuation methods that were developed to e.g. measure
intangible assets but also propose more experimental methods with the aim to stimulate
further discussion. Where possible we comment on feasibility and strength or weaknesses
of approaches. We should also note that there are no off-the-shelf methods to measure the
value of data and much more research on this topic is needed.
5.2.1. Data valuation based on market prices

The value of a good or service is often identified with the market equilibrium price, i.e. the
meeting point between demand and supply. However, there are various challenges that
make this approach difficult to apply to data. First, we need the existence of a well-defined
market. While this might exist for some data (e.g. on marketing preferences of individuals),
it could prove to be difficult to find for many other types of data, often for good reasons.
In addition, as databases are traded relatively infrequently, transaction-based valuation may
often rely on obsolete information (similar to housing stock).
Secondly, since the value of data is highly context-dependent, the same dataset could be
valued differently across different data suppliers, users and regulators. As noted by OECD
(2013a): “[…] economic experiments and surveys in the United States indicate that
individuals are willing to reveal their social security numbers for USD 240 on average, the
same data sets can be obtained for less than USD 10 from U.S. data brokers such as
Pallorium and LexisNexis”. In addition, the ‘true’ value of data may not be known to the
seller and hence the market price could be misleading (i.e. due to asymmetric information).
Nevertheless, studying data brokers offers an interesting avenue for data valuation based
on market prices, as they typically value their own datasets by quoting detailed, itemised
price lists for compiled databases. Moreover, this approach is in line with 2008 System of
National Accounts as “databases for sale should be valued at their market price, which
includes the value of the information content” (OECD, 2009: p.120). However, the
identification of data brokering firms themselves may be problematic, as they are not
classified to a single class in standard industry classifications (OECD, 2013b).
Data brokerage companies typically collect information from publicly available personal
records and then aggregate, store and sell it to different customers (described in detail in
Gellman and Dixon, 2013). Their business models range from specialised business-to-
business services offering information such as background criminal record checks of
individuals or detailed information on potential business partners or competitors (e.g.
Experian, LexisNexis) to other business models such individual localisation services (e.g.
Locate Plus, LocatePeople.org).39
The insurance market could offer another potential avenue for deriving the value of data
held by businesses, since insurance companies need to assess the value of data held in-
house or in data centres. Insurance against data breaches is becoming more important given
the strong reliance on data by an increasing number of businesses. Therefore, the pricing
of (data) insurance as well as (possibly) negotiations around insurance claims in case of

data breaches could both potentially be informative (although the latter could also include
litigation payments or legal costs and hence go beyond the value of data per se).
Related to the insurance of data itself is the notion of insuring continuous service delivery
against failures of enabling IT infrastructure, including data centres. Interruptions of
service provision to the end-user can entail significant costs to businesses (loss of revenues
as well as reputation). Since under certain circumstances, the infrastructure providers might
be liable for such losses, they take out insurance to protect themselves against such
eventualities (Data Center Knowledge, July 2016). There are a number of factors shaping
these estimations, including third party liability and first part losses.40
In a recent study sponsored by IBM (IBM, 2019), the Ponemon Institute (2018) conducted
interviews with more than 2,200 IT, data protection, and compliance professionals from
477 companies that have experienced a data breach between July 2017 and July 2018. They
found that the average cost of data breaches in this period was USD 3.86 million. While
data breaches entails all kinds of costs this can at least serve as an indication. According to
another survey of 1,800 global business decision makers conducted by NTT Security (NTT,
2018), data breaches entail significant costs to businesses, including:
• Loss in customer confidence, damage to reputation and brand, financial losses;
• Expected revenue drops in excess of 10% on average;
• Costs of recovery, estimated to be around USD 1.5 million on average;
• Costs of cybersecurity insurance, which 38% of organisations have in place.
This section discussed a number of approaches that should be explored to measure the
business value of data at market prices. Specifically, we proposed to look at data brokers
and insurance for data breaches related to the data themselves or the data infrastructure.
The exact application of these experimental sources of information to the valuation of data
in production process would necessarily have to be further developed and its limitations
would need to be addressed (e.g. separation of ‘data value’ from the costs of legal process
or disentangling the contribution of data to output valuation in case of insurance
information). Nevertheless, we find these approaches as interesting options. The following
sections look at other approaches that may be useful when considering the value of data
and data flows.
5.2.2. Data as a knowledge-based asset

The intangible assets literature holds a number of insights when thinking about measuring
the value of data, as databases and software are an existing category of intangible assets.41
Corrado et al. (2005) outline a ‘best practice’ method for measuring intangible investment
in the National Accounts framework, distinguishing digitised information, innovative
property, and economic competences/organisational capital. This can be useful when
thinking about data as an knowledge-based intangible asset that contributes to the
production process of firms and previous reports have called for measuring the value of
digitised data as an intangible asset (OECD, 2014).
To measure intangible capital in the national accounts, surveys distributed by statistical
offices ask companies about their purchases, disposals and own-account investment in
databases and software. For example, the UK Office for National Statistics uses the
Quarterly Acquisitions and Disposals of Capital Assets Survey (QCAS) to collect
information on the value of capital assets bought and sold by businesses in the private
sector. However, the measurement of the value of data in these surveys suffers from a

number of problems. Firstly, response rates on databases tend to be relatively low. 42

Secondly, own-account databases are typically recorded as part of own-account software,
making it difficult to single out the value of data itself.43 Thirdly, at the industry-level it is
particularly difficult to gauge expenditures on the creation of databases, as data are to some
degree produced in virtually all sectors of the economy. 44
As a consequence, direct measures of investment in databases are often inadequate, and
therefore, statistical offices often rely on a cost-based approach. The cost-based method
uses labour market data, for the number of computer software developers (or other data-
related occupations such as economic researchers), their average hourly costs, combined
with assumptions on time-use, non-wage labour costs, and overheads, in order to estimate
the values of databases. 45 This approach has been explored recently by Statistics Canada
(2019) based on 8 occupational groups. Their calculations assume that some occupational
groups spend a higher share of their time producing data. For example, while ‘data entry
clerks’ are assumed to spend 100% of their time on producing data, it is only 20-30% for
‘economists and economic policy researchers and analysts’. Following this experimental
method, Statistics Canada estimates that the total investment in data amounted to CA$9.4
billion to CA$14.2 billion 2018.
It is also important to note that although databases qualify as an asset under the 2008
System of National Accounts (SNA 2008) 46, the data they hold are explicitly excluded. 47
Ahmad and van de Ven (2018: p.4) discuss in more detail how based on SNA 2008,
“databases should reflect only the value of the underlying database management systems
and the costs associated with the digitisation of data 48. This recommendation reflected the
view that the underlying value (information content) associated with the data itself was de
facto a non-produced asset, with outright purchases of databases that included the intrinsic
value of the underlying data recognised in the accounts as goodwill”. However, this
approach can pose challenges in practice. When databases (and the underlying value of
data) are acquired for example in taking over another company, the transfer is captured as
‘goodwill’, and, so in-line with the implicit treatment that data are non-produced. 49
However, in practice, when databases are acquired separately the transactions may be
recorded as Gross Fixed Capital Formation at the market price of the acquisition (including
the value of the underlying data). Discussions are on-going in the context of the update of
the 2008 SNA to assess whether this needs to be addressed and, in particular, how to make
data more visible in the accounts.
Under some conditions, data also qualify as an asset under the International Financial
Reporting Standards rules 50, yet it should be stressed that from an individual company
perspective intangible investment and data assets are likely to be largely underreported
(Tax Adviser, May 2018).
Box 5.1. Bankruptcy case studies highlighting the value of business data
Case 1: The bankruptcy of American Caesars Entertainment Operating Corp. Inc,

2015
During “chapter 11” bankruptcy proceedings, some creditors argued that Caesars’ customer
loyalty program “Total Rewards” held data worth USD 1 billion. In support of this
assessment, a report by a bankruptcy court examiner did note instances where sold-off
Caesars properties suffered a decline in earnings due to the loss of access to customer
analytics based on the Total Rewards database. However, it was also observed that it might
prove difficult to sell the Total Rewards data, e.g. for incorporation into another company’s

loyalty program. As Short and Todd (2017) note: “Although the Total Rewards system was
Caesars’ most valuable asset, its value to an outside party was an open question”.
Case 2: Bankruptcy of RadioShack, 2015
When the American company RadioShack filed for bankruptcy protection in 2015, the most
valuable asset that was put forward for bidding by the company’s creditors was a database
with records of purchases of roughly 67 million RadioShack customers, along with their
addresses.
Source: Short and Todd (2017).
A related approach based on the market valuation of data is to exploit information revealed
in mergers and acquisitions (M&A), which can potentially be motivated by access to data
(e.g. “data-seeking foreign direct investment”). This is feasible since the legal process
requires the acquired company to assign a value to its assets, which includes intangible
assets such as databases and software. Anecdotal evidence suggests that the number of
M&A transactions that are motivated by data access has been increasing (HuffPost,
February 2017). Some deals that were at least partially motivated by data access include
Microsoft’s acquisition of LinkedIn in 2016 and purchase of GitHub in 2018. Some Central
Banks or National Statistics Offices keep track of such activities and could add new
questions into their existing forms or surveys relatively easily.
Bankruptcy cases provide another opportunity to explore the value of business data as
negotiated in court. We provide some examples of high-profile court cases in Box 5.1.
Future research could look into what valuation methods are used by courts and whether
they could be used in other settings as well.
5.2.3. Revenue-based valuations and the ‘data value cycle’

In Section 2 we proposed a stylised taxonomy of different business models that are either
data-enhanced or data-enabled. Based on Figure 2.4 we suggested to explore the share of
revenue that is driven by data monetisation across different types of firms (e.g.
manufacturers, utility providers, banks, or online platforms). In practice this would involve
building up a repertoire of case studies across a number of businesses in different sectors
that in turn could be generalised to a sector. This should be supported by business surveys
and periodic revisions to keep shares up to date. In terms of theory this approach has could
be integrated with the notion of treating data as a knowledge-based asset as discussed in
Section 5.2.2. We believe this approach is feasible in practice.
Digital platforms are a key example of data-enabled businesses (see also Section 2.3) and
are disrupting a large number of industries, including transport, accommodation, retail, and
travel services. Li et al. (2019) propose a classification of online platforms 51 based on eight
major types, related to their underlying business models. For their empirical analysis, they
select key companies for each platform type to estimate a range for the economic value of
their data. Focusing on the ‘data value chain’, they show how online platforms create value
from data in stages, and how the value derived at each stage varies. Their approach
recognises two sources of data monetisation inferred from market transactions as well as
Selling, General and Administrative Expenses 52 as a proxy for a firm’s investment into a
data-driven business model. Then, an approach similar to R&D capitalisation is applied to
infer the value of the underlying data.

Li et al. (2019) find that online platforms can vary in their degree of vertical integration in
the data value chain, and that this variation can determine the means and effectiveness of
their data monetisation. For example, Amazon’s Marketplace value of data is estimated to
be USD 125 billion which accounts for 16% of the company’s current market value.
Overall, their study suggests large benefits from vertical integration within the data value
chain, and online platforms that are also data holders are best placed to exploit these.
One interesting example of a data-enabled business is Citymapper, which has been
experimenting with different ways to monetise the travel and movement data it has on
individuals using its app. While it discontinued a dedicated bus service in London, it started
offering the Citymapper Pass at prices which undercut the official Transport for London
prices for weekly travelcards by £4.10 in 2019. A crude calculation would suggest that the
business hence values annual travel and movement data for an individual at around
52*£4.10 = £213.20 (approximately 270 USD). It would be interesting and relatively
straightforward to seek similar examples for different types of data.
5.2.4. Other approaches to trace value of data flows

To better understand and trace the value associated with data flows it could be worth to
exploit variation found in input-output tables. As the OECD’s “Measuring the Digital
Transformation” measurement roadmap suggests, “Superimposing Input-Output tables
with data-flow tables [could be used] to assess whether flows of value added are
accompanied by flows of data.” (OECD, 2019a). Research by the European Commission
supports this idea as most data are “intermediary goods that are used in production
processes by other parties” (Duch-Brown et al., 2017: p. 28). It is fair to note that the
feasibility of this approach depends on solving some of the conceptual and measurement
challenges related to data as mentioned above. For example, one would need at least to
have a clear idea about the value-added by sector that is enabled by data assets in order to
track how this value is flowing across sectors. Hence, we see this as approach as less
feasible in the short-run.
This approach would capture the reliance of different industry sectors on inputs from data-
intensive industries; both domestic and foreign (see van der Marel, 2015). As the level of
reliance differs across sectors, input-output relations would provide a picture of the relative
importance of data inputs. In practical terms, the difficulty will be to actually assess which
sectors are providing data inputs to others, where, and to what degree. While it appears
easier to assign the output of some specific sectors to these activities (e.g. 63.1 – “Data
processing, hosting and related activities; web portals”), it would require considerable
research to figure out the appropriate coefficients other sectors. We note that this is similar
to some approaches and ideas discussed in the context of digital trade (e.g. López González
and Jouanjean, 2017).

6. Discussion and implications for measurement agenda
What is the purpose of measuring the value of data? In 1987 Robert Solow famously stated
that “You can see the computer age everywhere but in the productivity statistics” (Solow,
July 1987) – those words could be paraphrased today in the context of the seeing data
everywhere but in business balance sheets. It is clear that a better understanding of how
data contributes to adding value and raising productivity, in order to inform policymakers
on a wide range of areas where the impact of data-driven innovations is inevitable. But,
crucially, this requires consistent and reliable statistics that are able to capture the
complexity of data uses in the modern economy.
The development of methods for measuring the value of data is not an easy task and much
more research on the topic is vital. Also, both research and statistical communities would
benefit from more systematic thinking about this measurement challenge and available
options. As noted by ESCoE-OECD November 2018 workshop participants, some of the
issues raised in this paper might resemble the measurement challenges related to R&D
activities, software and intangible assets more generally. This also means we do not have
to start from scratch but can use methods and concepts tested and applied before. However,
it is important to stress that the challenges we are faced with when discussing the value of
data often go well beyond those surmounted when measuring R&D and software.
However, some characteristics of data could make it a special sort of intangible asset: some
data does not always become obsolete quickly53 and it can be continuously combined with
other data to create new value. Other types of data could be ‘worthless’ in a matter of
seconds. This is further complicated by the fact that the value of data is highly related to
context and dependent on its use within businesses. This paper discussed some of the data
characteristics that one would need to consider when looking into this issue further.
When it comes to valuing data and data flows, various avenues for further investigation
have been highlighted. We foresee the approach of valuing data based on a business model
perspective as the most comprehensive and therefore as the most promising (e.g. Li at al.,
2019). Nevertheless, more effort could be made towards a search for settings where market
valuations of data can be observed, with an aim to identify businesses of interest and
aggregate this information across industries. Here, information on data pricing from data
brokering agencies could prove helpful. Also, in the short term it seems feasible to derive
the value of data from data breaches, data insurance and the data-motivated Mergers &
Acquisitions – at least for some sectors.
It is certainly helpful (and feasible) to introduce new questions to existing business surveys,
which would allow both for a more precise measurement within currently adopted methods
in the national accounts and for a better understanding of different areas of data-driven
businesses activity across sectors. This could include the measurement of data licenses that
are in use across sectors, or how the demand for people doing data-related work within a
firm and sector changes over time. In the long term, an update of the Standard Occupational
Classification (SOC) to account for “new” data-related occupations should be explored.
More systematic work on supplementary National Accounts estimates (e.g. in the context
of Digital Supply-Use Tables [SDD/CSSP/WPNA(2019)1]) need to be explored further.
Lastly, also a more systematic approach to analyse specific business models and data
monetisation strategies will be unavoidable in the longer term.

This paper has highlighted a number of conceptual and empirical challenges regarding the
measurement of the value of data and data flows and proposed a number of approaches that
should be explored further to start accounting for this increasingly important resource in
modern economies. While the challenges seem daunting, we argue to put the business
model at the centre of any method of valuation.

References
Ahmad, N. and Schreyer, P. (2016), “Measuring GDP in a Digitalised Economy”. OECD Statistics
Working Paper 73, OECD Publishing: Paris.
Ahmad, N. and van de Ven, P. (2018), “Recording and measuring data in the System of National
Accounts”. Paper prepared for the Meeting of the Informal Advisory Group on measuring GDP in
a digitalised economy, 9 November 2018, OECD.
Bauer, M., Lee-Makiyama, H., van der Marel, E., and Verschelde, B. (2014), “The costs of data
localisation: Friendly fire on economic recovery”, ECIPE Occasional Paper No. 3/2014.
Bauer, M., Ferracane, M.F., van der Marel, E. (2016), “Tracing the Economic Impact of Regulations on
the Free Flow of Data and Data Localization”, Global Commission on Internet Governance
(GCIP) Series paper No. 30, May 2016.
Bloom, N., Schankerman, M., and Van Reenen, J. (2013), “Identifying Technology Spillovers and
Product Market Rivalry”, Econometrica, 81(4): 1347-1393.
Brynjolfsson, E., Collis, A. and Eggers, F. (2019), “Using massive online choice experiments to measure
changes in well-being”, Proceedings of the National Academy of Sciences, 116(15): 7250-7255.
Brynjolfsson, E. and K. McElheran (2016), "Data in Action: Data-Driven Decision Making in U.S.
Manufacturing", Working Paper 16-06, Center for Economic Studies, U.S. Census Bureau.
Business Roundtable (2015), “Putting Data to Work: Maximizing the Value of Information in an
Interconnected World”, Business Roundtable, January 2015.
Casalini, F. and J. López González (2019), “Trade and Cross-Border Data Flows”, OECD Trade Policy
Papers, No. 220, OECD Publishing, Paris.
CIGI (2018), “A National Data Strategy for Canada: Key Elements and Policy Considerations”, CIGI
Paper No. 160, Centre for International Governance Innovation, February 2018.
Cisco (2017), “Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2016–
2021”, White Paper, Cisco, February 2017.
Cisco (2018), “Cisco Global Cloud Index: Forecast and Methodology, 2016–2021.” White Paper, Cisco,
November 2018.
Cisco (2019), “Cisco Visual Networking Index: Forecast and Trends, 2017–2022”, White Paper, Cisco,
February 2019
Corrado, C., Hulten, C. and D. Sichel (2005): "Measuring Capital and Technology: An Expanded
Framework", NBER Chapters, in: Measuring Capital in the New Economy, pages 11-46, National
Bureau of Economic Research, Inc.
Coyle, D. (2017): “Do-it-yourself digital: the production boundary and the productivity puzzle”, ESCoE
Discussion Paper 2017-01, Economic Statistics Centre of Excellence, January 2017.

Coyle, D. and Nguyen, D. (2020), “Valuing Goods Online and Offline: The Impact of Covid-19”, CEPR
Covid Economics, Issue 33, 30 June 2020.
Coyle, D. and Nguyen, D. (2018), “Cloud Computing and National Accounting”, ESCoE Discussion
Paper 2018-19, Economic Statistics Centre of Excellence, December 2018.
Cushman & Wakefield (2016), “Data Centre Risk Index 2016: Informing Global Location Strategies in a
Digital World Expanding at a Phenomenal Pace”, Cushman & Wakefield, July 2016.
Cyphers, B. and Gebhart, G. (2019): " Behind the One-Way Mirror: A Deep Dive Into the Technology of
Corporate Surveillance”, Electronic Frontier Foundation, December 2019.
Deloitte, (2017): “Assessing the value of TfL's open data and digital partnerships”, Deloitte, July 2017.
Duch-Brown, N., Martins, B. and Mueller-Langer, F. (2017). “The economics of ownership, access and
trade in digital data”, JRC Technical Reports: JRC Digital Economy Working Paper 2017-01,
European Commission.
European Commission (2015). Public Consultation on the regulatory environment for platforms, online
intermediaries, data and cloud computing and the collaborative economy, Brussels.
European Commission (2017). Exchanging and Protecting Personal Data in a Globalised World.
Communication from the Commission to Parliament and the Council, Brussels.
Ferencz, J. (2019), "The OECD Digital Services Trade Restrictiveness Index", OECD Trade Policy
Ferracane, M.F. (2017). Restrictions to Cross-Border data flows: a Taxonomy, ECIPE Working Papers
No. 1/2017.
Ferracane, M.F., Lee-Makiyama, H. and van der Marel, E. (2018). Digital Trade Restrictiveness Index.
ECIPE Report April 2018.
Gartner (2011): “Gartner Says Solving “Big Data” Challenge Involves More than Just Managing
Volumes of Data”, Press release.
Gellman, R. and Dixon, G. (2013), “Data Brokers and the Federal Government: A New Front in the
Battle for Privacy Opens”, World Privacy Forum Report, October 2013.
Hartmann, P.M., M. Zaki, N. Feldmann, and A. Neely (2014): “Big Data for Big Businesses? A
Taxonomy of Data-driven business models used by start-up firms”, Cambridge Service Alliance
working paper.
Hofheinz, P. and Mandel, M. (2015). Uncovering the Hidden Value of Digital Trade: Towards a 21st
Century Agenda of Transatlantic Prosperity. Interactive Policy Brief Issue 19/2015. The Lisbon
Council and Progressive Policy Institute.
Hofheinz, P. and Osimo, D. (2017). Making Europe a Data Economy: A New Framework for Free
Movement of Data in the Digital Age. Policy Brief, the Lisbon Council.

Hughes-Cromwick, Ellen, and Julia Coronado (2019): "The Value of US Government Data to US
Business Decisions." Journal of Economic Perspectives, 33 (1): 131-46.
ITIF (2015). Cross-Border Data Flows Enable Growth in All Industries. Information Technology &
Innovation Foundation. February 2015.
ITIF (2017). Cross-Border Data Flows: Where Are the Barriers, and What Do They Cost?. May 2017.
Ker, D.(forthcoming). “Perspectives on the value of data and data flows”, DSTI/CDEP/MADE(2019)6,
OECD, Paris.
Kommerskollegium (2014). No transfer, No Trade: the Importance of Cross-Border Data Transfers for
Companies Based in Sweden. National Board of Trade, Sweden. January 2014
Laney D. (2011): “Infonomics: The Economics of Information and Principles of Information Asset
Management”, The Fifth MIT Information Quality Industry Symposium, July 13-15, 2011.
Li, W.C.Y., Nirei, M., & Yamana, K. (2019), „Value of Data: There’s No Such Thing As A Free Lunch
in the Digital Economy”, U.S. Bureau of Economic Analysis Working Paper, Washington, DC.
López González, J. and M. Jouanjean (2017), “Digital Trade: Developing a Framework for Analysis”,
OECD Trade Policy Papers, No. 205, OECD Publishing, Paris.
López González, J. and J. Ferencz (2018), “Digital Trade and Market Openness”, OECD Trade Policy
Magalhaes, G., and C. Roseira (2017): “Open Government Data and the Private Sector: An Empirical
View on Business Models and Value Creation.” Government Information Quarterly.
Mandel, M. (2017). “The Economic Impact of Data: Why Data is Not Like Oil.” Progressive Policy
Institute.
Mauldin, A. (2019). “A Complete List of Content Providers' Submarine Cable Holdings”.

(TeleGeography Blog, November 2017). Accessed February 2020.
McKinsey (2018): “Achieving business impact with data: A comprehensive perspective on the insights
value chain.” Digital/McKinsey, April 2018.
McKinsey Global Institute (2016a): “Digital Globalization: The New Era of Global Flows”. McKinsey &
Company, February 2016.
McKinsey Global Institute (2016b): “The Age of Analytics, competing in a data-driven world”.
McKinsey & Company, December 2016.
McKinsey Global Institute (2019): “Globalization in Transition: The Future of Trade and Value Chains”.
McKinsey & Company, January 2019.
Meltzer, J.P. and P. Lovelock (2018). Regulating for a global economy: Understanding the importance
of cross-border data flows in Asia. Brookings Global Economy & Development Working Paper
No. 113, March 2018.

Nakamura, Leonard, Jon Samuels, and Rachel Soloveichik (2018):“’Free’ Internet Content: Web 1.0,
Web 2.0 and the Sources of Economic Growth”; Paper prepared for the 35th IARIW General
Conference; Copenhagen, Denmark, August 20-25, 2018.
OECD (2009): “Handbook on Deriving Capital Measures of Intellectual Property Products”, OECD
Publishing, Paris.
OECD (2013a): “Introduction to Data and Analytics (Module 1): Taxonomy, Data Governance Issues,
and Implications for further Work”, Paper circulated for consultation; OECD,
DSTI/ICCP(2013)13.
OECD (2013b), “Exploring the Economics of Personal Data: A Survey of Methodologies for Measuring
Monetary Value”, OECD Digital Economy Papers, No. 220, OECD Publishing, Paris.
OECD (2014), “Measuring the Digital Economy: A New Perspective”, OECD Publishing, Paris.
OECD (2015), “Data Driven Innovation: Big Data for Growth and Well-being”, OECD Publishing,
Paris.
OECD (2018a), “Trade and cross-border data flows”, Preliminary Draft, TAD/TC, OECD Publishing,
Paris.
OECD (2018b). Working Party on International Trade in Goods and Trade in Services Statistics: Result
of the 2018 WPTGS Stocktaking Questionnaire. STD/CSSP/WPTGS(2018)3, 28th February 2018.
OECD (2019a), Measuring Digital Transformation: A Roadmap for the Future, OECD Publishing, Paris.
OECD (2019b), Digital Innovation: Seizing Policy Opportunities, OECD Publishing, Paris.
OECD (2019c), Understanding Enhanced Access to and Sharing of Data: Reconciling Risks and Benefits
for Data Re-use across Societies, OECD Publishing Paris.
Pepper, R. and Garrity, J. (2014). The Internet of everything: How the network unleashes the benefits of
big data. Global Information Technology Report 2014.
Rayport, J. F. and Sviokla, J.J. (1995), “Exploiting the Virtual Value Chain”, Harvard Business Review,
Vol. 73, No. 6, pp. 75–85.
Rugman, A.B., and L. Eden (1985). Multinationals and Transfer Pricing. London: Routledge.
Shapiro, R. and S. Aneja (2019) “Who Owns Americans’ Personal Information and What Is It Worth?”
Future Majority report.
Short, J.E. and S. Todd (2017): “What is your data worth?”, MIT Sloan Management Review, Vol. 58,
No.3.
Statistics Canada (2019): “The value of data in Canada: Experimental estimates”, Latest Developments
in the Canadian Economic Accounts, Statistics Canada, 10th July 2019.
System of National Accounts 2008. New York: United Nations, 2009.

Swedish National Board of Trade (2015): No Transfer, No Production: A Report on Cross-border Data
Transfers, Global Value Chains, and the Production of Goods”.
The Economist Intelligence Unit (2015): “The Business of Data”, The Economist Intelligence Unit
report.
United Nations Economic Commission for Europe (2015): “Guide on Measuring Global Production”,
December 2015.
US International Trade Commission (2014). Digital Trade in the U.S. and Global Economies, Part 2.
Publication Number 85, August 2014.
US Department of Commerce (2016). Measuring the Value of Cross-Border Data Flows. September
2016.
van der Marel, E. (2015). Disentangling the Flows of Data: Inside or Outside the Multinational
Company?, ECIPE Occasional Paper 07/2015.
World Bank (2016): “World Development Report 2016: Digital Dividends”.

Notes
1
However, we acknowledge that the literature on organisational change, innovation, and knowledge-based
capital holds relevant insights for studying the economic value of data as well.
2
This is related to the literature on R&D spillovers, where social returns to R&D investment are shown to
be higher than private returns, meaning that companies on their own are under-investing in R&D (Bloom et
al., 2013).
3 (The Economist, 2017) headlined that “The world’s most valuable resource is no longer oil, but data”. This is
underpinned by the fact that the most valuable companies nowadays include a number of technology giants that
critically rely on intangible assets, including vast databases, to generate large shares of their revenue (e.g. Apple,
Microsoft, Amazon, Alphabet, Facebook, and Alibaba).
4 Related to this, the OECD (2019b) distinguishes primarily physical (e.g. agriculture), mixed digital &
physical (e.g. retail), and primarily digital end products (e.g. media).
5
DDI is defined as “significant improvement of existing, or the development of new, products, processes,
organisational methods and markets” (OECD, 2015: 17).
6 Certainly, businesses might also collect some data aimed neither for sale, nor for use in a production
process, e.g. data could be gathered for the purpose of compliance with a government regulation. For
simplicity, we abstract from those considerations (they can be thought of as a form of taxation).
7
Amazon started operating as an online bookstore in 1994, and it now also sells consumer electronics,
produces movies, operates a supermarket chain, and provides cloud services, among others. We describe
Amazon as a data-enabled business since its core business model is based on offering ‘infinite shelf space’.
The company achieves this by providing large quantity of products at low prices delivered fast. It can do so
only with predictive data analytics and search algorithms at its heart. See also blog (Hunter, March 2019).
8 Freemium is a pricing strategy by which a product or service is provided free of charge, but a premium is charged
for additional features or services. Another example is Spotify, that charges a subscription fee for ad-free access to
its vast music library.
9
The business model may change over time as new opportunities to monetise data arise which were not
necessarily conceived initially.
10
Theoretically the nominal price (and hence value) of some data can be higher in the case of prohibitively
high barriers on cross-border flows. However, we are mainly concerned with data that only becomes valuable
when aggregated at a global level.
11
Also, as highlighted by participants from the business community at OECD-ESCoE workshop in
November 2018, a non-negligible share of value-added is generated by data analytics, which itself critically
depends on the quantity and quality of the underlying data.
12
Another important aspect to mention is that it matters whether data have been collected or acquired
legally and, hence, whether it can be legally commercialised. This is illustrated by the recent fine for the
parenting club Bounty which sold 34m personal data records to 39 different companies (incl. Axicom,
Equifax and Sky) without asking for the users’ permission. The British Information Commissioner’s
Office, which led the investigation case described that “data sharing was an integral part of their business
model at the time”; see (The Guardian, April 2019).

13
We are aware that not all companies (especially some start-ups) have positive revenue streams. This
taxonomy is highly stylised and revenue can be replaced by company ‘valuation’, which is naturally based
on future revenue streams.
14
High performers are defined as companies that had annual growth rate of 10% or more over the past 3
years.
15
Shapiro and Aneja (2019) provide estimates of the value of American personal data of based on the digital
advertising revenue of the major online platforms. They show, based on these companies’ financial
statements, that in 2018 they earned USD 111.1 billion from U.S. advertisers targeting American consumers.
Moreover, “Google and Facebook dominated this area in 2018, accounting respectively for 37.1 percent
($41.3 billion) and 20.6 percent ($22.9 billion) of total digital advertising revenues.” (p. 9)
16
According to TJI Research, Amazon sells products using 146 private label brands and 640 Amazon
exclusive brands (last updated November 2019), across different product categories including clothing,
electronics, food, furniture, household goods, and healthcare (TJI Research, s.d.).
17
Covered in Harvard Business School “Managing the Future of Work” podcast, 25th October 2018.
Konecranes reports that it already connected 18,000 cranes globally to TrueConnect and employs 600
software engineers to develop control systems (Harvard Business School, October 2018).
18
ISO (the International Organisation for Standardisation) and IEC (the International Electrotechnical
Commission) form the specialized system for worldwide standardization. National bodies that are members
of ISO or IEC participate in the development of International Standards through technical committees
established by these organisations to address particular fields of technical activity.
19
The is closely related to the concept of ‘data velocity’, which is used to refer to the speed of data flows. In
particular, it describes the efficiency of data analytics and network infrastructure enabling smooth processing
of data flows.
20
Data link-ability is directly linked to data variety. Data variety relates to “the capacity to analyse a variety
of mostly unstructured data sets from sources as diverse as web logs, social media, mobile communications,
sensors and financial transactions. This capacity is often associated with the capacity to link these diverse
data sets (linked data).” (OECD, 2013b)
21
Though beyond the scope of this paper, we also note the potential costs on the environment due to the vast
amounts of energy needed to power data centers.
22
We do not aim to provide an exhaustive list, but rather focus on the most frequently used data types that
are relevant in determining the economic value of data. For example, we do not discuss “big data” separately
as we do not see it as a data type. We would rather relate it to a set of data characteristics - such as volume -
discussed in the previous section.
23
In some instances, data can be generated by the private sector and held by the public sector (e.g. tax
records, employment records). Businesses spend considerable resources to file returns or fill-in business
surveys on an ongoing basis.
24
Data can also be ‘protected’ by being held as trade secrets, as is often the case for recipes (NOLO, s.d.)
25
However, as already mentioned, Cisco (2018) reports that as much as 75% of all IP traffic in 2017 was
video streaming. <This is the second time this is mentioned and I think you really need to make explicit what
you are suggesting their nature (video) means for their value>
26
The MGI model estimates the contribution of various flows - including data - to approximate their impact
on real GDP. Data flows are approximated by cross-border used bandwidth from TeleGeography (sum of
capacity for Internet backbones, private networks and switched voice networks). MGI run the model for 97
countries for the years 1995 - 2013 and find that a 10% increase in cross-border data flows raises GDP by
0.2.

27
For example, The Guardian (2018) reports that in 2017, Facebook only paid £15.8m in tax in the UK,
though it recorded GBP 1.3 billion in British sales. The article notes that taxable profits had been reduced by
‘administrative expenses’ of GBP 444m. In addition, the UK tax-to-sales ratios of Amazon, Google, and
Apple were similarly low. According to Facebook’s Annual Report 2017, the company recorded a profit
margin of 50% on its total global sales of USD 40 billion. This highlights the need for further research into
the potential role of data transfers that are internal to the firm and to what degree they are related to the
location of value creation.
28
Latency refers to the time delay in sending data packages across physical distance, e.g. via fibre optics.
29
Similarly, the Global Internet Map cam give some insights into the data volume in terms of installed
bandwidth (TeleGeography, 2018).
30
This includes all EU-28 and OECD member countries.
31
Various studies are summarised in ITIF (2017).
32
Among other categories, this area covers policies that affect connectivity such as measures on cross-border
data flows and data localisation.
33
The partnership was signed in July 2018 and is the largest trade agreement ever negotiated by the EU,
covering around a third of global GDP. It is complemented with reciprocal adequacy agreement on data
protection, meaning both countries will recognise respective data protection systems. Once ratified, this will
create the largest area of safe cross-border data flows. Source: European Commission, 17th July 2018
(European Commission, July 2018)
34
A study by McKinsey Global Institute (2016) relates data volume (or installed bandwidth) to its economic
value using regression analysis. This makes sense insofar as countries with more data flows (in terms of
volume) are more likely to have more data flows with higher values.
35
It also depends on whether data can be used legally, which might differ across countries.
36
This can be underpinned when considering that building and maintaining the physical infrastructure to
support cross-border (as well as within border) data flows is costly.
37
Consumer traffic is defined as traffic originating with or destined for consumer end users; business traffic
is traffic originating with or destined for business end users. Furthermore, consumer segment encompasses
Other Consumer Apps, Social Networking, Video Streaming and Search whereas business segment is
composed of Computation, Database/Analytics/IoT, Collaboration and Enterprise Resource Planning and
Other Business Applications (Cisco, 2018).
38
In some cases it could be other organisations such as universities or research institutes that conduct these
types of surveys.
39
It should be noted that data brokers often are not collecting their own datasets from scratch.
Instead, they might rely on various public and commercial datasets, to organise of aggregate the
data for easier use. Therefore, the market value of a dataset from the perspective of data brokers
could be much lower (or higher) than its original production cost.
40
This includes: Damage to the facility, premises and equipment; Employee health and safety; Business and
service interruption; Data security and privacy; Expenses resume operations after a loss event; Regulations
and compliance risks; and Human error.
41
It should be noted that the intangible assets literature offers many insights into the value of data, but its
focus is on long-lived assets. We acknowledge that many datasets are not capital assets in this sense and that
their rate of depreciation can be very high. However, in some sense we abstract from some of the implications
for measurement in this paper.
42
For example, for the ONS QCAS the response rates vary from 67% to 85%; see (Office for National
Statistics, 2014).

43
Moreover, the System of National Accounts 2008 guidelines suggest valuing own-account databases based
on the cost of organising and compiling data in an easy to use form, rather than the underlying data (Section
10.113). On the other hand, the value of the underlying data is included in purchased databases (10.114). It
is likely that most of the databases are however developed on an own-account basis and rarely transacted on
the market. Hence, surveys are likely to significantly underestimate of the total value of data.
44
Moreover, the Central Product Classification until recently has not provided an adequate set of categories
that cover databases without including too many other items. Central Product Classification Ver. 2 introduced
in 2008 new single category “Original compilations of facts/information” (83940) relates to databases.
45
It is also worth noting that surveys often use some sort of extrapolation from larger businesses to the
smaller ones. This is because large companies are typically asked to fill much more detailed forms, which
later serve as a basis for imputation of, for instance, the asset breakdown in a given industry. This implicitly
puts fairly strong assumptions on the asset mix distribution across the companies’ universe.
46
“All databases holding data with a useful life of more than one year should be recorded as fixed assets
providing they meet the general definition of an asset (i.e. are expected provide benefits to their owners and
over which ownership rights are exercised).” (OECD, 2009; p. 102)
47
For example, the ONS questionnaire for the Quarterly Acquisitions and Disposals of Capital Assets Survey
asks businesses to report databases that “are files of data that are organised to enable effective use of, and
access to, the data”, instead of the data itself.
48
In the National Accounts framework, both the database management software and on-going running costs
(e.g. for data storage services) are excluded from the capitalised value of databases. Further detail on this can
be found in Ker (2019) [DSTI/CDEP/MADE(2019)6].
49
The important point to note here is that the exclusion of data from the production boundary (and so do not
add to GDP when they are included in a database) is not the same thing as saying that data do not have value,
nor that they are not important sources of output and growth, as the SNA fully records revenue streams
generated from the use of data.
Data could be classified as an intangible asset under IFRS if it is “identifiable, non-monetary asset without
physical substance”; see (IAS Plus, s.d.).
51
The online platform is defined by the European Commission (2015) as “an undertaking operating in two
(or multi)-sided markets, which uses the Internet to enable interactions between two or more distinct but
interdependent groups of users so as to generate value for at least one of the groups” (p.5) (European
Commission, January 2016).
52
SG&A (alternatively SGA, SAG or SGNA) are Selling, General and Administrative Expenses which are
a major non-production cost reported in the balance sheets of companies.
53
A separate discussion on the depreciation rates appropriate for data and different data types is welcome,
however beyond the scope of this paper (see also note 22).

Measuring The Economic Value of Data and Cross-Border Data Flows

Uploaded by

Copyright:

Available Formats

Measuring The Economic Value of Data and Cross-Border Data Flows

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Measuring The Economic Value of Data and Cross-Border Data Flows

Uploaded by

Copyright:

Available Formats

MEASURING THE

OECD DIGITAL ECONOMY

OECD DIGITAL ECONOMY PAPERS

OECD DIGITAL ECONOMY PAPERS