The digital economy: A new set of data to assess the economic implications of digitalization

Abstract

The 2025 System of National Accounts (SNA) introduces significant advancements in the measurement of economic activities in the digital age. This article explores the methodologies developed to capture the impact of digitalization, including the recognition of data as an asset, the role of digital intermediation platforms, and the treatment of free digital products. These updates provide a more accurate and comprehensive understanding of the modern economy, reflecting the true nature of economic transactions today.

By enhancing the measurement of GDP and other key macroeconomic indicators, the 2025 SNA offers new insights into the digital economy's influence on production, consumption, and international trade. This article aims to provide a comprehensive understanding of these developments and their implications for economic measurement and analysis, highlighting the potential extension of analytical possibilities stemming from the innovations related to the digital economy in the 2025 SNA.

Keywords

digital economy data digital intermediation GDP measurement

1 Introduction

In today's rapidly evolving world, the digital economy has become a cornerstone of global development and innovation. As technology continues to advance at an unprecedented pace, the digital economy encompasses a wide range of activities, from e-commerce and digital services to online communication, data assets and analytics as well as artificial intelligence.

The rapid emergence and development of the digital economy have posed significant challenges to the measurement of economic activities, as well as to economic analysis and policy making. As technology advances, and business and consumer uptake increase, the need for changes to the System of National Accounts (SNA) has become evident. The 2025 SNA addresses these challenges by incorporating new elements that reflect the transformative impact of digitalization on production, consumption, and international trade.

Understanding and measuring the digital economy is crucial for several reasons. Firstly, it provides valuable insights into economic growth and productivity. By quantifying the contributions of digital activities, policymakers and businesses can make informed decisions that foster innovation and competitiveness. Secondly, measuring the digital economy helps identify emerging trends and opportunities, enabling countries to adapt and thrive in a digital-first world. Lastly, it ensures that the benefits of digital transformation are equitably distributed, addressing potential disparities and promoting inclusive growth.

In essence, measuring the digital economy is not just about numbers; it's about understanding the profound impact of digitalization on our lives and economies. It empowers us to harness the full potential of technology, driving sustainable development and improving the quality of life for people around the globe.

This article explores the methodologies and new datasets developed to measure digital products and services, including data on e-commerce, digital platforms, and intangible assets such as data and artificial intelligence. It highlights how these new data sets provide insights into the digital economy's impact on macroeconomic statistics. Additionally, the article distinguishes between elements that affect the measurement of GDP and other key macroeconomic measures, and those that required clarification in the accounts.

The introduction of data as an asset, the role of digital intermediation platforms, and the treatment of free digital products are among the key topics discussed. The article also touches upon the new issues addressed in the 2025 SNA, such as crypto assets and the digitalization of financial services, which link to other chapters of the publication. It also summarizes opportunities and challenges in the measurement of prices and volumes for the digital economy.

By examining these developments, the article aims to provide a comprehensive understanding of the digital economy's implications for economic measurement and analysis, offering a first impression of the potential extension of analytical possibilities stemming from the innovations related to the digital economy in the 2025 SNA.

2 Data as an asset

2.1 Why is data needed in the System of National Accounts?

The production of data is a fundamental process for most companies and organizations across the economy. Today, it is not hyperbole to state that for many enterprises, their very business model depends on the production of and use of data. Importantly, the use of data is not just reserved to “digital natives.”¹ The rapid digitalization of so many facets of the economy has created a substantial increase not just in information that can be collected but how insights from this data can be used in production. This trend, coupled with the long-term decrease in the cost of undertaking such work to businesses, has resulted in unprecedented amounts of data being produced and used in the modern economy.

For something so omnipresent, data as an input into (and/or output of) production is largely absent in official economic statistics, including previous iterations of the System of National Accounts (SNA). The lack of explicit inclusion of data in international classification and frameworks, such as the Central Product Classification (CPC) or the International Standard Industrial Classification (ISIC), combined with the general difficulty in conceptualizing exactly what data is as well as where the production of data starts and finishes, provides plenty of challenges in appropriately measuring it. The update to the SNA (as well as many other international statistical standards) provided the opportunity to work through these challenges and introduce data as an explicit output of production.

Appropriately recognizing data investment in the SNA production and asset boundary and then measuring this investment in a consistent way offers significant analytical insight to users of economic statistics. The recent accumulation and production of data have not only provided businesses the opportunity to use existing tools more productively, but also enabled the creation of brand new tools for production, such as generative AI. National Account users must be able to put a name (and more importantly, a value) to the product that is enhancing the capability of so many existing production process.

The addition of data as an explicit asset in the System of National Accounts (SNA) is not a brand-new asset category; rather, it builds on and expands the existing category of intellectual property products to better capture where economic value comes from. Previously, under the 2008 SNA, the asset category for ‘databases’ only included the “cost of preparing data in the appropriate format” but importantly not “the cost of acquiring or producing the data” (2008 SNA §10.113).¹ As a result, significant spending on collecting and creating data was treated as a regular operating expense, instead of as a capital investment in an asset that can be used repeatedly. The 2025 SNA corrects this by explicitly recognizing data—combined with databases—as a new asset category, ensuring these expenditures are properly measured as investments.

2.1 What data is and is not for the purpose of the national accounts

For something as multifaceted as data, it is necessary to have as specific definition as possible so that compilers and users are able to understand what is represented by the data estimates within the SNA. Within the 2025 SNA, data is considered “information content that is produced by accessing and observing phenomena, recording, and storing information elements from these phenomena in a digital format and that provides an economic benefit when used in productive activities” (2025 SNA §22.22).²

Data, as defined in the national accounts, is considered an output of production. If it is used repeatedly for more than a year, it qualifies as a produced asset. Often, data is collected and used quickly—sometimes only once—making it a short-lived output that is treated as intermediate consumption rather than capitalized. That said, on many other occasions the data will remain useful to the production process for well over a year (even if not used for its original use). On these occasions the expenditure on this data will be capitalized as an investment in a data asset.

To be considered an output of production or a produced asset within the SNA, data must fulfil several prerequisites, the most important of which is that it is providing an economic benefit to the owner. “Ancillary data,” which is information content created (often as an externality of production) by recording, and storing information elements but which is not providing an economic benefit, is outside the SNA production and asset boundary. This is a similar standard to other produced assets within the SNA, whereby to be considered a fixed asset, the capital formation must be providing an economic benefit to their owner. Additionally, within the SNA, data will refer only to digital data. Although numerous examples exist of collected and organised non-digital data being fundamental to economic production throughout history,² it is considered that in the modern economy, investment in non-digital data is fairly insignificant and its inclusion in the production boundary would only serve to unhelpfully broaden the conceptual description of data.

Finally, since headline measures within the SNA are often volume-based indicators, it is important to acknowledge that the quantity of “data” produced and sent around the internet is an extremely different concept to the volume of data assets produced in an economy. Modern lexicon such as asking about the “size” of a database or the amount of data required to steam a movie, do represent a quantity figure which could theoretically be considered the quantity (or volume) of data produced by the economy. As such there may be an expectation that the level of produced data within the national accounts should display an exponential growth over the recent past, similar to the growth in generated bytes which now sees over 400 million terabytes of data created each day.³

However, the bits and bytes that make up data flows are a different concept of data as defined for the SNA. A large amount of these bits and bytes includes digital activity such as photos, emails and other communications that fail the 2025 SNA data definition. These digital files are usually not produced by accessing and observing phenomena, are often not used in productive activities and more broadly simply reflect the nature of the digital service delivery used by business and consumers alike.

2.3 How data will be measured

Prior to its introduction into the 2025 SNA, there had been various attempts to put a value on data. Coyle and Manley⁴ provide a summary of the many different examples of trying to put a value on data. These methods include attempts to go beyond the standard approach of estimating the financial value to the owner, to also try and estimate the value of data, especially its interoperability, to society at large. While there are many approaches, these can usually be split up into three groups, market approach, cost approach and income approach.

Discussions on the incorporation of data into the SNA quickly focused on the cost approach (sum-of-cost approach in SNA lexicon). While there are some examples of data being sold,^5,6 much of this is not actually the data being sold but rather the insights of the data embedded into another good or service, such as advertising. Instead, most data used in production has usually been created on an own account basis.⁷ This fact, coupled with the homogenous nature of data, means that the calculation of representative market prices is both conceptually tricky and practically difficult. Fortunately, the sum-of-cost method is already well established in the compilation of national accounts, since it is the recommended approach when market prices are absent.² As such, it is often used for the estimation of non-market output and other own account assets such as research and development or computer software. This provided an important level of familiarity to compilers attempting to place a value on data for the first time.

The output of data measured via the sum of cost approach is undertaken by summing up the various costs of production. This process is similar to the calculation for other output and includes,

Source: 2025 SNA §4A.6.²

Countries in their initial attempts to measure data have all gravitated around this fundamental sum-of-cost formula. However, there are slight differences between the exact methodology in each country. Examples of some of these variations include

(a)
The choice of occupations and involvement rates used to derive the estimate of ROE (1).
(b)
The choice of mark-up applied to the ROE estimate, representing the other cost of productions (2–6).
(c)
Adjustments to represent the prevalence of short-lived data (data used for less than 1 year).

The handbook on measuring data in the system of national accounts⁸ provides an extensive compilation guide on producing estimates of data consistent with the SNA. This includes standard recommendations, which countries can look to implement immediately and will reduce the level of difference created by initial methodological variations. The handbook will also include advanced recommendations that can be further developed/refined as more countries undertake this work and new best practices are developed.
2.4 What will estimates of data look like in the national accounts

The approval of the 2025 SNA framework in March 2025 was an important milestone, as it acknowledged the role that data plays in the modern economy and the need for it to be explicitly measured and included in the production boundary. Encouragingly, even at this time of approval, there were already many estimates of the level of data investment measured in a way that is consistent with the SNA. These were compiled by national statistical offices but on occasions international organizations have also undertaken some work to further develop compilation methods. As displayed in Table 1, these estimates of the size of the investment have predominately been around 2–2.5% of GDP. There are instances of estimates being lower or higher than this, however since these were all done before the formalization of standard recommendation, these variances can be partially explained through slight conceptual and compilation differences.

Table 1.
Experimental estimates of data investment.

Country Year Value of data investment, % of GDP Value of data investment, % of total GFCF Compilation notes

Australia 2016 2.3% 9.7% Percentage is average of upper and lower range estimates.

Canada 2021 0.6% 2.3% Percentage is average of upper and lower range estimates.
Significant adjustment made to remove short lived data.

Germany 2018 2.2% 10.5%

India 2019 1.0% 3.2% Compiled by Asian Development Bank.

Japan 2020 2.4% 9.6% Percentage is average of upper and lower range estimates.

Netherlands 2017 2.4% 11.7% Private sector only.Percentage is average of upper and lower range estimates.

Pakistan 2022 0.8% 5.7%

United Kingdom 2020 4.9% 27.7% Compiled by OECD, using natural language processing.Includes production of databases and data science.Percentage is average of upper and lower range estimates.

USA 2021 0.8% 3.7% Private sector only.Significant adjustment made to remove short lived data.

Country	Year	Value of data investment, % of GDP	Value of data investment, % of total GFCF	Compilation notes
Australia	2016	2.3%	9.7%	Percentage is average of upper and lower range estimates.
Canada	2021	0.6%	2.3%	Percentage is average of upper and lower range estimates. Significant adjustment made to remove short lived data.
Germany	2018	2.2%	10.5%
India	2019	1.0%	3.2%	Compiled by Asian Development Bank.
Japan	2020	2.4%	9.6%	Percentage is average of upper and lower range estimates.
Netherlands	2017	2.4%	11.7%	Private sector only.Percentage is average of upper and lower range estimates.
Pakistan	2022	0.8%	5.7%
United Kingdom	2020	4.9%	27.7%	Compiled by OECD, using natural language processing.Includes production of databases and data science.Percentage is average of upper and lower range estimates.
USA	2021	0.8%	3.7%	Private sector only.Significant adjustment made to remove short lived data.

Source:^8–15

Like previous changes to the SNA, it is expected that compilation best practice will continue to improve as more countries undertake the work, new data sources become available and a greater understanding of how data is used in the economy helps guide assumptions and models. That's said, the initial work already undertaken by NSOs and IOs as well as the release of the handbook on measuring data in the system of national accounts provides a foundation of conceptual parameters and compilation guidance that countries can follow during their early compilation attempts.

3 Digital intermediation platforms and free digital products

The past years have seen a proliferation of online platforms. There are two broad groups of such platforms: digital intermediation platforms (DIPs), which intermediate online commerce between producers and consumers of goods and services, such as Airbnb or Uber; and platforms that facilitate non-economic interactions, e.g., social media and discussion sites like Instagram, YouTube or Facebook, providing free services to their users.

Business models for these two types of platforms substantially differ. Platforms that primarily facilitate non-economic interactions differ from DIPs, as they usually facilitate “interactions” between users, without facilitating a market transaction (nothing sold or bartered). While this characteristic is not absolute and there are examples of some economic transactions being derived from these types of platforms, these economic transactions are not the primary function of, or source of revenue for, the platform.

Importantly from a SNA perspective, on most occasions no explicit production occurs between the platform and either independent party involved in the “interaction”. This is in contrast with DIPs, which intermediate a transaction that not only creates an economic flow between two independent users, but also does so in exchange for an explicit fee (a market transaction) from either the producer, the consumer or both. Each of the two sub-sections below describe how each type operates.

3.1 Digital intermediation platforms

Digital Intermediation Platforms are a quintessential example of digitalization's dramatic effect on the economy. The digital transformation has allowed producers to interact with previously unreachable consumers (including those in other geographical locations) at relatively low costs, lowering the entry barrier and bringing in producers previously excluded from the market. At the same time, it has offered consumers greater knowledge on the different prices and quality of products on offer. Importantly, both producers and consumers are able to derive economic benefit from using a DIP even if it charges an explicit fee for the intermediation service.

This potential benefit to both has seen intermediation, which started with taxi services and holiday accommodation morph into a significant structural change to production lines where almost all services and or products accommodation

The success of intermediation platforms has seen significant structural change to certain sectors of the economy. The value chain for taxi services, accommodation, take-away food and a growing number of retail products have fundamentally been altered by intermediation platforms. As such, the SNA, as well as other statistical classifications, need to be better at identifying and measuring the respective output of the new actors. As while there are potential benefits to both producers and consumers, analysts and policy makers are keen to dissect the value chain to confirm if new players are adding value to the overall product consumed or rather if any new value added is simply being moved from one producer to another, with the possibility of leading to the potential rent seeking.

From the perspective of the SNA, a DIP is considered an enterprise that meets the two following criteria: it charges an explicit fee for digitally facilitating an economic transaction between two independent parties; it does not take economic ownership of the goods and services ultimately sold to the consumer.

Whether or not the DIP charges an explicit fee for facilitating a transaction is an important delineation point from a national accounts’ perspective, as the charging of a fee for an intermediation service creates value added. If a platform is not charging an explicit fee to either the consumer or producer, they are likely deriving their revenue from selling advertising space or information sourced from collected data, economic activities that differ from providing intermediation services.

3.1.1 Recording the flows of a DIP: Conceptual treatment

The 2025 SNA recommends following the “producer” approach to record the flows of a DIP. According to this approach, the buyer of the good or service pays directly the producer, for an amount that covers both the price of the good or service and the intermediation fee of the DIP, with the producer then paying the DIP. Following this approach, the output of the DIP is consumed by the producer as intermediate consumption. This results in no transaction being recorded between the consumer and the platform, even though in reality these flows are normally observed.

A “consumer” approach may be alternatively considered, where DIPs would operate in as similar way to traditional retailers: the consumer would pay the full amount to the DIP, which would then transfer the price of the good or service to the producer retaining the intermediation fee. However, this approach overlooks that DIPs neither assume ownership of the goods or services involved nor bear significant financial risk.

The producer approach is considered conceptually superior because it improves the interpretability of the value chain, better reflecting which units are adding value to the overall final consumption. In addition, considering that many DIPs that operate in a country are not resident in that economy, the producer approach prevents an artificial inflation of trade estimates.

3.1.2 Recording the flows of a DIP: Compilation challenges

There is no fundamental conceptual measurement issue in regard to DIPs, and the production and consumption associated with them should already be included in the national accounts. However, practical challenges to their measurement exist.

Since the intermediation service as provided by the DIP is often merged with the price of the service being sold, people may overlook that production is taking place and that value added is being created. In addition, DIPs pose several practical measurement challenges, stemming from the large amount of non-resident DIPs that may operate in a domestic economy. Another challenge, especially for comparing data across countries, is that digital intermediation platforms (DIPs) are not consistently classified within the ISIC system, nor are the services they provide consistently categorized in the CPA framework. This inconsistency makes international comparisons more difficult. Also, DIPs change the traditional consumer / producer paradigm, which may create instances where households play a different role in production than before. The number of informal workers who undertake economic activity from opportunities provided by DIPs may not be fully recorded, which possibly gives the impression of missing value added. Finally, DIPs also act as a vehicle to create data as an asset, which can form a significant element of gross value-added created by these businesses.

As the level of output they produce and their value-added grow, it will become even more important for statisticians to quantify the impact of DIPs on economic activity. This includes the level of inputs they use, the labour they employ, investment they make and the overall productivity gains/losses associated with these activities. The 2025 SNA and the revised ISIC and CPA classifications provide important guidance for the recording of DIPs, that should help overcoming these measurement issues. For example, to reflect the increasing importance of intermediation service activities in the economy, in the revised ISIC (ISIC Rev. 5), separate groups or classes have been created in the divisions where these goods and services are produced to better delineate the actual activity taking place.

The measurement of the activities of DIPs still poses significant practical challenges and data requirements. Reconciling the (total) amount paid by the consumer (usually derived from household surveys) with the amount received by the producer (normally derived from business surveys), at the product level without direct information from the DIPs involved will likely pose a large statistical challenge. Moreover, in many circumstances the DIP may take payment for a good or service which will not be provided for some time in the future; such circumstances can raise concerns about accurately recording the transaction on an accrual basis. Finally, if non-resident DIPs do not have any domestic affiliate and thus cannot be surveyed, it will be necessary to estimate the level of intermediation service being imported via household surveys or alternative data source. To overcome these measurement challenges, it is crucial that National Statistical Authorities establish contacts with the main domestic DIPs, to collect information directly from them, and cooperate with each other to exchange information on international flows.

3.2 Free digital products

Free media funded by advertising first emerged as an issue in the national accounts debate when television became a major source of entertainment in the 1950s. The issue then vanished from the debate, only to reemerge, arguably larger, when platforms offering free services became a major part of our digital lives. Unlike television, which generally offered only the service of entertainment, in the modern economy, due to the value that can be created harvesting information for data assets or selling advertising space, and abundance of services can now be obtained for free. Examples include communication services (free email accounts), information sharing (apps detailing sport results and current news) to navigational aids (free maps). By putting value estimates to these services (albeit outside the integrated framework of national accounts) greatly assist analysts and policy makers to better understand if these units are in fact reducing the level of income generated in the economy, due to reducing demand for paid versions of these services or simply changing the business model so that consumers pay for these services indirectly rather than directly.

Two main questions arise:

(1)
Are services freely available on the internet, which surely have a value for their users and producers, unmeasured in national accounts?

The 2025 SNA clarifies in this respect that they aren't. Therefore, the 2025 SNA did not change their treatment, but clarifies how these phenomena should be viewed.

There are two main types of assets that produce free digital services: open-source software (the services of free online platforms are primarily produced by software assets) and user-generated content available on platforms.

The value of open-source software produced by programmers employed by corporations, government, non-profit institutions providing services to households or produced by unincorporated business that is classified in the household institutional sector is part of the SNA asset boundary, and its production is already included in measures of own-account software investment. Open-source software produced by individual volunteers who are not remunerated in any way for their contribution is instead outside the SNA production boundary.

Another important category of intellectual property assets providing free services in the digital economy is entertainment, literary and artistic originals shared as user-generated content. To be considered as an asset of its creator, the user-generated content must fall within the production boundary of the SNA and be expected to yield economic benefits for its creator over a period of at least a year.

If a household receives remuneration for the content that it creates and uploads (e.g., ad revenue or subscription revenue), the own-account investment in artistic originals is within the SNA production boundary. For example, a social media influencer with earnings should be treated as a household unincorporated enterprise. On the other hand, creation of entertainment, literary and artistic originals for personal enjoyment – a common leisure activity – has always been treated as outside the production boundary. This longstanding treatment should also apply to user-generated digital content – for example, personal posts on Facebook – where no remuneration is received or expected. (2)
The second question concerns the business model of platforms offering free content.

Reinsdorf¹⁶ notes that outputs that are free, or at least priced below the cost of production, are common throughout the market economy. Market producers often include such items in the bundle of items that they supply. Items that are consumed together need not be individually profitable; making parts of the bundle free could avoid transactions costs, or subsidizing one item may help to sell another at a mark-up. In either scenario, operating surplus remains positive because the profits on the marked-up items fund the losses on the subsidized items. Taken as a whole, the bundle generates revenue that is commensurate with the amount of production taking place at the enterprise.

For platforms, bundling free and subsidized products are not merely common – they are the rule. Platform users consistently place different values on the opportunity to connect with those on the other side of the platform. The platform responds to these differences by subsidizing the users whose presence on the platform will raise the value of the platform to those with a high willingness-to-pay, while marking up the prices paid by those in the latter group. Platforms normally obtain enough revenue from their funder side to cover the cost of the subsidies to the other side, leaving them with a positive net operating surplus. The funding side (normally consisting of the producers and providers of a product attempting to reach a greater audience) in turn, recover the cost of the platform's services through mark-ups on the advertised products.

These platforms also collect and store information on their users and the user-generated content freely supplied by households. This own account gross fixed capital formation (GFCF) produced by the platform is then included as a capital input into the platform's output of advertising services. Targeted advertising is probably the main use of platform users’ information. Some digital platforms may also have another source of revenue by selling data assets derived from their users’ information.
3.2.1 Free digital products in an SNA extended account

The 2025 SNA clarifies how free digital products available on digital platforms are accounted for within the national accounts’ production and asset boundaries. However, it also includes a framework for an extended account on free services of digital platforms consumed by households. The 2025 SNA offers three options for such an extended account. The first option simply identifies and distinguishes the value of free digital products that are already included within the value of other products, as currently accounted for in the System of National Accounts (SNA). The second option builds on the first one by including costs associated with the production of a data asset, as suggested in OECD,¹⁷ specifically recording and processing costs plus the costs of collecting platform users’ information. The third option builds on the second one by including costs associated with the production of user-generated content. All options increase the visibility of the household's role as a final consumer of free digital products. The second and third options increase the visibility of the intersection of free digital products and data as an asset.

4 Cloud computing, artificial intelligence, non-fungible tokens

Digitalization has resulted in, and been accelerated by, the emergence of cloud computing as a new way of accessing ICT resources. It has also resulted in new types of assets, such as crypto assets including non-fungible tokens.

Cloud computing has become a global industry with extensive international transactions, and an estimated global output of several hundred billion US dollars, accounting for a sizeable share of IT investment. Importantly, the dynamics of the industry has meant that a huge majority of this expenditure taking place is concentrated with just a few major multinational providers. Accurate classification and measurement of cloud computing can assist users better understand the fundamental shift occurring in a wide range of industries from on sight capital expenditure to remote service delivery.¹⁸

At the same time, the marked technological progress in computer hardware and software as well as the storage and use of vast amounts of data over the last twenty years has led to the development of artificial intelligence (AI) systems. which find a. Business have invested considerable sums into AI with the expectation that they will not only have a broad range of applications in the economy, but that they may create a paradigm shift in the potential inputs (especially labour) required for production. By explicitly identifying this expenditure and classifying it separately to other software, analysts will have a better chance of identifying rates of return associated with the emergence of AI.

Crypto assets have also become prevalent in recent years. Crypto assets are of different nature and have different purposes. A main distinction is between fungible crypto assets and non-fungible tokens.³ This section focuses on the latter category, while fungible crypto assets are described in ref.¹⁹

All these digital products are relatively new and while they have not led to any conceptual change to the integrated framework of national accounts (except for the second type of non-fungible tokens described in section 3.1, that are recorded as a separate asset category), they were not separately described in the 2008 SNA.¹ The 2025 SNA clarifies the nature of these new digital products and their recording in the national accounts.²

4.1 Conceptual definitions

The 2025 SNA defines cloud computing services as consisting of computing, data storage, software, and related ICT services accessed remotely over a network, supplied on demand and with measured resource usage (2025 SNA §22.13).²

The main cloud computing products can be divided into three broad categories: (i) infrastructure-as-a-service (IaaS), which gives the user on-demand access to hardware such as a virtual server; (ii) platform-as-a-service (PaaS), which also includes access to a software platform; and (iii) software-as-a-service (SaaS), which includes access to the application software. Users of IaaS or PaaS provide their own software license, or software original.

The 2025 SNA treats payments for software subscriptions as purchases of a service, unless long-term licenses for software hosted in cloud computing data centres are more likely to represent software assets of the user. Furthermore, clarification is provided on the recording of gross fixed capital formation of data centres, including IT equipment, for own final use. As major suppliers of cloud computing have globe-spanning networks of computing facilities and clients, guidance is included on the measurement of the international transactions in cloud computing services across these networks.

While cloud computing services are not separately identifiable as an asset in the 2025 SNA asset classification, because of their importance cloud computing is one of the digital products shown separately in the digital SUTs described in section 5 of this chapter.

While several definitions of artificial intelligence have been given by different institutions over time, the 2025 SNA, for its purposes, states that “Artificial intelligence refers to capabilities of a computer program, or system controlled by a computer program, of recognition, reasoning, communication, and prediction emulating human recognition, reasoning, and communication.” (2025 SNA §22.33).²

Applications of AI systems include consumer electronics market, for instance in voice-controlled virtual assistants. AI tools are also used by Internet publishers, digital content subscription services, and social media networks. AI can identify and recommend content that an individual user is most likely to be interested in. Autonomous vehicles are another important application of AI. AI applications guide decision-making in a wide array of sectors, e.g., agriculture, manufacturing, health care or finance. The 2025 SNA gives explicit recognition to AI systems by including AI in the asset classification, as an “of which” category of computer software, including artificial intelligence systems.

Non-fungible tokens (NFTs) are digital records hosted on a blockchain that are associated with a digital or physical asset or product. An NFT is distinct from the associated asset or product and should not necessarily be recorded in the same way. There are many types of NFTs, and new types are being developed as the digital asset space evolves. A forward-looking way to categorize NFTs that should cover any new developments is according to the rights conferred upon the NFT owner: 1) NFTs that convey no ownership rights and only allow for personal use of another asset or product; 2) NFTs that convey limited ownership rights, beyond personal use for another asset or product; 3) NFTs that convey full ownership rights for another asset or product.

The 2025 SNA clarifies that in the first case NFTs should be recorded as consumption or, if over time an NFT gains more features of a valuable, it may transform into a valuable.² The second type of NFT should be recorded as contracts, licenses, or leases, distinct from the associated asset or product. The third type of NFTs are essentially inseparable from the asset or product and are simply a digital recording of ownership. So, if one is already recording the purchase of the associated asset or product, the NFT needs not be recorded as a separate purchase in the national accounts.

4.2 Measurement issues

For cloud computing services, dedicated access to remotely located IT assets as a substitute for owning an on-premises IT capital stock can raise questions for the measurement and interpretation of IT investment. The appropriate treatment in national accounts of a contract for dedicated access to a remotely located server or other IT assets is to be as a financial lease, making the user the economic owner of the IT asset. The SNA also clarifies that software licenses lasting more than a year are fixed assets. Having the software hosted in a cloud computing data center does not change the ownership of the license as a software asset.

A major issue for the recording of cloud computing services stems from the major suppliers of such services being multinational enterprises (MNEs), with customers and operations in many countries. Cloud computing enterprises supply services to this global marketplace both through cross-border data flows and by investing in local datacenters and other facilities linked by global cable networks that include subsea cables. Measuring cross-border flows of cloud computing services may be more of a challenge, as customers can place orders with a multinational cloud computing enterprise for computing services to be produced or used in a third country. Most cloud computing services are inputs into the production of something else. The resource pooling aspect of cloud computing technology could complicate the measurement of detailed trade flows when the supplier of the cloud computing services is an MNE with establishments in multiple countries and the purchaser of the computing service does not specify the location of the data center where the work must be done.

National accounts compilers’ expectation is that at a minimum cloud computing MNEs should be able to supply breakdowns of production and consumption of cloud computing services by country. If they can also provide data on the gross flows of cloud computing services, precise estimates of international transactions in those services can be developed. However, the payment flows on cloud computing should be more feasible to track. For an economy that is just an importer (and not a producer) of cloud computing services, this implies that the data on intermediate consumption of these services in the economy (assuming that the services are consumed by businesses) could provide a reliable estimate of imports. In any case, international collaboration between national accounts experts and experts on balance of payments statistics is needed to develop shared guidelines on international transactions involving remotely accessed computing services providers and the global cloud computing industry (see the UNECE Guide to Sharing Economic Data in Official Statistics for further discussion of international data sharing to improve measurement of MNEs and international transactions in general.²⁰).

In general, the valuation and recording of AI in the SNA can broadly follow that of software, as AI systems are produced, used in-house or exchanged following the business model of software. This regards for instance the case when AI is made available under a license. In such a case, it can be treated as a fixed asset if it is expected to be used in production for more than one year and the licensee assumes all the risks and rewards of ownership. There is however one particular aspect that should be considered for its recording: many applications of AI require either a “training data set” or continuous access to a large database that the AI uses to “interpret” and “reason.” The creation of these datasets may be regarded as investment in either AI systems or data assets. It is therefore of the utmost importance that they are counted only once. The 2025 SNA clarifies that “data assembled in a database solely as a step in the production of an artificial intelligence computer program and that cannot be re-used may be included in the costs of producing artificial intelligence programs, if the sum of costs method is used to value the relevant assets.” (2025 SNA §22.36).² However, if AI systems use data that are used also for other purposes, these should not be considered as part of investment in AI.

Regarding NFTs, detailed information would be needed on the rights that each NFT conveys, and the spending to purchase NFTs, separate from the purchases of the associated assets or products. These comprehensive data may be difficult to obtain. Fortunately, some useful information on NFTs is already available. Currently, only a few websites host the majority of the transactions and may be able to provide information on the range of rights conveyed by the NFTs.⁴

5 Digital supply and use tables

5.1 Introduction to digital SUTs

Digitalization impact on the economy is all pervasive. Consumers now have access to a larger variety of goods and services and the ability to buy through a variety of channels. Producers leverage digitalization in order to disrupt established markets, improve the efficiency of their production processes all while obtaining more information about their consumers (and potential consumers) than ever before. As such, many countries, international organizations and academics have attempted to measure digitalization impact on the economy in a more detailed manner to not only provide more material for policy decisions but also to reassure users that the value added stemming from digitalization is appropriately included in standard national account aggregates.

In order to increase the visibility of digitalization in the National Accounts in a more consistent manner, as well as foster international collaboration on this important topic, the OECD developed a statistical framework based on the concept of Digital Supply-use tables. Supply-use tables were an obvious choice for this work due to their comprehensiveness, availability across countries, as well as their multidimensionality. Additionally, producing outputs of digital activity that could easily be related back to the many already established outputs of the SNA improves the international comparability as well as the interpretability of the results.

A fundamental difference between the Digital SUTs and other attempts to measure the Digital Economy is that the digital SUT framework does not attempt to define the digital economy. This is not because one had not been proposed. In fact, Bukht and Heeks²¹ contains a list of not less than 20 definitions presented to the international community in the two decades proceeding 2017. Despite this plethora of options, the definitions often lack suitability for either users or compilers.²² For example, definitions that focus on specific products or industries are often considered too narrow because they miss the impact of digitalisation on the production of traditional goods and services. Conversely, definitions that focus on broader digital trends or the use of specific ICT products in production can become impractical for measurement purposes.

To overcome this issue, the Digital SUT framework also does not advocate a single measure or indicator as being representative of digitalisations impact. Rather the framework is designed to meet a multitude of needs and demands, by producing a suite of indicators on various aspects of digital activity in the economy. These indicators include the value of products digitally ordered or delivered digitally, the importance of certain digital products to firms during production, and the output and value added of specific types of firms that are significantly impacted by or completely reliant on digital technology. The framework itself is outlined in detail in the OECD Handbook on compiling digital supply and use tables.²³

5.2 Basic concepts of the digital SUTS

At its most basic, the digital supply and use table analyse the impact of digitalization along three dimensions.

The nature of the transaction (the “how”).

The goods and services produced (the “what”).

The specific value added coming from new digital industries, shown separately in the Digital SUTs (the “who”).

5.2.1 Nature of transaction

Conventional SUTs make no distinction on how a transaction is facilitated, focusing only on what product was produced and which industry produced it. However, the rise of online shopping combined with the movement to digital service delivery means that increasingly, economic transactions are undertaken entirely digitally. Digital ordering a product is defined as “The sale or purchase of a good or service, conducted over computer networks by methods specifically designed for the purpose of receiving or placing orders”.²³ Importantly, this split of products based on how the nature of the transaction allows for consistency with the framework outlining digital trade.²⁴ As such, estimates produced as part of the work on digital trade (which is a stand-alone statistical output) can be used for import and export estimates within the Digital SUTs.

5.2.2 Digital products

Four additional rows are added to the digital SUTs so that new totals and breakdowns may be compiled to provide visibility of the impact that digital products have in the economy. Two of these rows cover the goods and services outlined within the ICT product classification found in the Central Product Classification (CPC) Version 2.1. These rows aggregate estimates of output taken from numerous different product rows within the standard SUTs. This aggregation allows for easy interpretation of which industries use a large proportion of digital products as intermediate consumption and importantly how this has been evolving over time.

Two additional new rows will cover the output of cloud computing services (CCD) and digital intermediation service (DIS).⁵ The separate identification of these two products reflects the fundamental role the products play in the digitalized production and value chains of both traditional industries and the new digital industries. It should be noted that while the OECD Handbook on Compiling Digital Supply and Use Tables specifies only the two products, cloud services and digital intermediation services, a third, data, could likely be added, since it has now been officially included within the production and asset boundary of the SNA. Other products, such as Artificial Intelligence could also be separately identified if considered a useful addition.

It is worth highlighting here that digital SUTs can also be extended into a full digital economy thematic account by including products that go beyond the core principles of the SNA. For example, the OECD handbook notes that much work has been done to generate values associated with the consumption of certain free digital services.^25–27 The valuations described in these works are not always at market value (i.e., exchange value) and conceptually the interaction they describe is not considered an economic flow from the perspective of the 2025 SNA. However, due to their analytical value, and strong connection with the other aspects of the digital economy (e.g., advertising driven platforms) they are a logical extension beyond outputs produced as part of the sequence of economic accounts.

5.2.3 Digital industries

The Digital SUT framework includes seven new “digital industries” which delineate economic units in a way that is designed to highlight their role in digitalisation. This allows for traditional industry-based macroeconomic indicators such as output, gross value added (and its components such as gross operating surplus and compensation of employees) to be produced from a digitalisation perspective rather than with traditional economic activity breakdowns.

The seven digital industries are the digitally enabling industry, online platforms funded by advertising and data collection, digital intermediation platforms, producers that depend on digital intermediation platforms, e-tailers, financial service providers primarily operating digitally, and other producers operating only digitally. Detailed definitions of each industry are included in the OECD handbook.

It is important to reiterate that the economic units (along with their output and GVA) which are classified to these new industries are already represented in standard SUTs, classified within existing industry columns. In this regard the compilation of indicators associated with digital industries is not so much a new aggregation of outputs but a re-aggregation of existing estimates.

5.3 Examples of initial estimates consistent with digital SUTs

Several countries have produced estimates consistent with the Digital SUT framework. The specific outputs have varied between countries depending on source data availability or user demand. This has allowed for comparisons across countries, such as that shown in Table 2 which displays the proportion of total GVA associated with each new “digital Industry” as well as the aggregate.

Table 2.
Digital industry in selected countries as proportion of total GVA (%).

Sweden (2017) Netherlands (2018) Canada (2019) India (2022) USA (2021)

Digital enabling industries 5.23 5.25 4.83 7.83 7.40

DIPs charging a fee 0.05 0.80 0.15 0.41 --

Data- and advertising-driven digital platforms -- -- 0.05 0.05 --

Producers dependent on DIPs 0.06 0.10 -- 0.69 --

E-tailers 2.88 1.80 0.24 0.10 2.10

Financial service providers predominantly operating digitally -- 0.06 0.16 0.18 --

Other producers operating only digitally 0.88 -- 0.04 0.31 --

Total digital industries 9.10 7.90 5.46 11.74 9.50

	Sweden (2017)	Netherlands (2018)	Canada (2019)	India (2022)	USA (2021)
Digital enabling industries	5.23	5.25	4.83	7.83	7.40
DIPs charging a fee	0.05	0.80	0.15	0.41	--
Data- and advertising-driven digital platforms	--	--	0.05	0.05	--
Producers dependent on DIPs	0.06	0.10	--	0.69	--
E-tailers	2.88	1.80	0.24	0.10	2.10
Financial service providers predominantly operating digitally	--	0.06	0.16	0.18	--
Other producers operating only digitally	0.88	--	0.04	0.31	--
Total digital industries	9.10	7.90	5.46	11.74	9.50

Source: ^23,28–31

All the published work so far is still considered experimental in nature, as such, additional refinement of compilation methodology and source data is still ongoing. However, even at this initial stage, some clear take aways are noticeable. For example, the countries who have reported an estimate, the level of GVA for digital industries (5.5% – 12% of GVA) is considerably lower than the proportion of output considered to have been digitally ordered (8–25%).²³ This would suggest that the digital ordering extends well beyond the so call “digital native” type enterprises, with many traditional industries now levering digitalisation to interact with consumers. Which products and industries account for this higher digital ordering will become apparent as more columns and rows are completed within the Digital SUTs.

6 Price and volume measurement and implications for productivity estimates

Digitalization has introduced significant challenges in measuring prices and volumes for products, especially due to rapid quality changes, new digital products, and shifts in supply sources. Accurate deflators require frequent updates and adjustments to reflect these dynamics. Important issues that must be addressed in the measurement of the digital economy to accurately measure real growth and productivity include:

Quality adjustment for new models: Price indexes must adjust new models’ prices to reflect quality differences, often using hedonic regression or options pricing methods to capture improvements accurately.

Outlet substitution bias: When consumers switch to lower-priced sources enabled by digital platforms, unit value price indexes can capture price impacts, but assumptions about product homogeneity are critical.

ICT goods quality measurement: Frequent quality improvements in ICT and ICT-enabled goods require hedonic or options pricing methods, sometimes incorporating performance benchmarks, to adjust price indexes for quality changes.

Software and data valuation: Software and data assets pose measurement challenges due to their intangible nature; prices and volumes for own-account software are inferred from input costs, while data assets are valued based on production inputs and may require revaluation if sold.

Cloud computing price indexes: The variety and frequent updates of cloud services complicate price index construction, necessitating representative sampling and quality adjustments; cross-border transactions add further complexity requiring international cooperation.

Internet and telecom service volumes: Rapid growth in usage and quality improvements like data speed require up-to-date contract samples and quality-adjusted prices or direct volume estimation using detailed quantity indicators aggregated with expenditure weights.

E-commerce pricing representation: Price indexes must adequately include online sales and digital intermediation platforms, considering that online prices often differ from offline ones; frequent online price changes may be better captured by monthly unit values when expenditure and sales data are available.

Free digital products impact: The availability of free digital goods conceptually lowers consumption deflators by increasing volume and reducing price, but measuring their effects is complex and often beyond standard national accounts frameworks due to valuation challenges.

7 Conclusions

The 2025 SNA represents a significant advancement in the measurement of economic activities in the digital age. By incorporating new methodologies to capture the impact of digitalization, the 2025 SNA provides a more accurate and comprehensive understanding of the modern economy. The inclusion of data as an asset, the role of digital intermediation platforms, and the treatment of free digital products are pivotal in reflecting the true nature of economic transactions today. Table 3 provides an initial broad indication of how these changes as well as some others may impact various macroeconomic aggregates.

Table 3.

Impact of SNA changes on main macroeconomic aggregates.

Change to SNA	Expected impact on GDP	Impact on GFCF	Sectoral distribution effects	Stock/Balance sheet effects
Data as an asset	Increase level of GDP	Increase level of GFCF	—	Increase level of IPP capital stock
Digital Intermediation Platforms	—	—	Potential shift of income from corporations to households as individuals obtain access to markets	—
Cloud computing	—	Reallocation of IT GFCF towards cloud computing hyperscalers away from all other industries	__	Reallocation of IT Capital stock towards cloud computing hyperscalers away from all other industries
Crypto-assets	Potential effects	—	—	New financial/produced asset categories
Non Fungible Tokens	__	__	__	__
Free digital Services	__	__	Increased output and household consumption in supplementary tables/ thematic account	__
Artificial Intelligence (AI)	__	__	__	Potential splitting of computer software capital stock into specific AI category

Furthermore, the 2025 SNA addresses emerging issues such as crypto assets and the digitalization of financial services, ensuring that the accounts remain relevant and useful for policymakers and analysts. These updates not only enhance the measurement of GDP and other key macroeconomic indicators but also offer new insights into the digital economy's influence on production, consumption, and international trade.

The introduction of data as an asset is particularly noteworthy. In an era where data drives innovation and economic growth, recognizing data as an asset ensures that its value is accurately reflected in national accounts. This change acknowledges the importance of data in the digital economy and provides a framework for measuring its contribution to economic activities.

Digital intermediation platforms, such as online marketplaces and gig economy platforms, have transformed the way goods and services are exchanged. The 2025 SNA's inclusion of these platforms highlights their growing significance and provides a clearer picture of their economic impact. By capturing the value generated by these platforms, the SNA offers a more complete view of the digital economy.

The treatment of free digital products, such as social media services and search engines, is another critical update. These products, while free to consumers, generate significant value through advertising and data collection. The 2025 SNA's approach to measuring this value ensures that the economic contributions of these services are not overlooked.

The 2025 SNA marks a crucial step forward in economic measurement, providing the tools needed to analyze and understand the complexities of the digital economy. By incorporating new methodologies and addressing emerging issues, the 2025 SNA offers a comprehensive and accurate representation of modern economic activities. These advancements will be essential in guiding economic policy and fostering sustainable growth in an increasingly digital world.

Footnotes

ORCID iD

John Mitchell

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Notes

References

Eurostat, IMF, OECD, UNSD, World Bank. 2008 System of National Accounts, https://unstats.un.org/unsd/nationalaccount/docs/sna2008.pdf (2009).

Eurostat, IMF, OECD, UNSD, World Bank. 2025 System of National Accounts (white cover), https://unstats.un.org/UNSDWebsite/statcom/session_56/documents/BG-3a-ISWGNA-BG-2025-SNA-Combined-E.pdf (2025).

Digitalsilk. How Much Data Is Generated Per Day. 2025. [Online]. Available: https://www.digitalsilk.com/digital-trends/how-much-data-is-generated-per-day/#:∼:text=402.74%20million%20terabytes%20of%20data,to%20grow%20to%20394%20zettabytes.

Coyle

Manley

. What is the Value of Data? A review of empirical methods, https://www.bennettinstitute.cam.ac.uk/wp-content/uploads/2022/07/policy-brief_what-is-the-value-of-data.pdf (2022).

Ahmad

van de Ven

. Recording and measuring data in the System of National Accounts, https://unstats.un.org/unsd/nationalaccount/aeg/2018/M12_3c1_Data_SNA_asset_boundary.pdf (2018).

OECD. Measuring the value of data and data flows, https://www.oecd-ilibrary.org/docserver/923230a6-en.pdf?expires=1695240160&id=id&accname=guest&checksum=A23446F17CC73C2BAA1DEEAADB8AE8F3 (2022).

Eurostat. Digital Economy and Society (database). 2022. [Online], https://ec.europa.eu/eurostat/web/digital-economy-and-society/data/database.

IMF-Eurostat. Handbook on compilling estimates of Data in the System of National Accounts, https://unstats.un.org/UNSDWebsite/statcom/session_56/documents/BG-3a-ISWGNA-BG-Data-Handbook-E.pdf (forthcoming).

Smedes

Nguyen

Tenburren

. Valuing data as an asset, implications for economic measurement, https://www.abs.gov.au/system/files/documents/7bfccb4ddb08aded818330bebe6b76b14/Smedes%20-%20Valuing%20data%20as%20an%20asset.pdf (2022).

10.

Amegble

Bugge

Sinclair

. Estimation of Investment in and Stock of Data and Databases in the Canadian System of National Accounts, https://iariw.org/wp-content/uploads/2023/10/Amegble-Bugge-Sinclair.pdf (2023).

11.

Donado

Kuckelkorn

Weigelt

, et al. RECORDING DATA AS A SEPARATE ASSET IN NATIONAL ACCOUNTS - A WORKSHOP REPORT, Destatis, https://www.destatis.de/EN/Methods/WISTAScientificJournal/Downloads/recording-data-national-accounts-042024.pdf?__blob=publicationFile&v=2 (2024).

12.

Japanese Cabinet Office. Report of Research on the measurement of the Digital Economy toward 2025SNA : Recording Methods for Data as Capital, (in Japanese), https://www.esri.cao.go.jp/jp/esri/prj/hou/hou088/hou088.html (2024).

13.

De bondt

Mushkudiani . Estimating the Value of Data in the Netherlands, https://iariw.org/wp-content/uploads/2021/10/bondt_paper.pdf (2021).

14.

Schmidt

Pilgrim

Mourougane

. What is the role of data in jobs in the United Kingdom, Canada, and the United States?: A natural language processing approach, 10.1787/fa65d29e-en (2023).

15.

Calderón

Rassier

. Valuing the U.S. Data Economy Using Machine Learning and Online Job Postings, https://www.bea.gov/system/files/papers/BEA-WP2022-13.pdf (2022).

16.

Reinsdorf

. Measuring Economic Welfare: What and How?. IMF Policy Paper 20/028, Washington D.C, 2020.

17.

OECD. Measuring the value of data and data flows, Paris: OECD Digital Economy Papers No. 345, 10.1787/923230a6-en (2022).

18.

Baer

Lee

Tebrake

. Accounting for Cloud Computing in the National Accounts. IMF Working Paper, https://www.imf.org/en/publications/wp/issues/2020/07/17/accounting-for-cloud-computing-in-the-national-accounts-49564 (2020).

19.

Harutyunyan

. Capturing financial innovation: The 2025 SNA and BPM7 response to digital finance. Stat J IAOS 2026; 42.

20.

UNECE. Guide to sharing economic data in official statistics. Geneva: United Nations, 2020.

21.

Bukht

Heeks

. Defining, conceptualising and measuring the digital economy, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3431732 (2017).

22.

Mitchell

Ker

Lesher

. Measuring the economic value of data. Paris: OECD Publishing. https://www.oecd-ilibrary.org/science-and-technology/measuring-the-economic-value-of-data_f46b3691-en (2022).

23.

OECD. Handbook on compiling digital SUTs, https://www.oecd.org/content/dam/oecd/en/publications/reports/2023/11/oecd-handbook-on-compiling-digital-supply-and-use-tables_b127cb7a/11a0db02-en.pdf (2024).

24.

IMF, OECD, UNCTAD, WTO. Handbook on measuring digital trade - 2nd Edition, https://www.oecd.org/content/dam/oecd/en/publications/reports/2023/07/handbook-on-measuring-digital-trade-second-edition_099afd2f/ac99e6d3-en.pdf (2023).

25.

Coyle

Nguyen

. Free goods and economic welfare, https://escoe-website.s3.amazonaws.com/wp-content/uploads/2020/12/14161702/ESCoE-DP-2020-18.pdf (2020).

26.

Brynjolfsson

Eggers

Gannamaneni

. Using massive online choice experiments to measure changes in well being. NBER, Cambridge, MA 02138, 2018.

27.

Brynjolfsson

Collis

Diewert

, et al. GDP-B: Accounting for the value of new and free goods in the digital economy. NBER, p. https://www.nber.org/papers/w25695 (2019).

28.

I. Ministry of Electronics & Information Technology. Estimation and measurement of India’s digital economy, https://www.pib.gov.in/PressReleaseIframePage.aspx?PRID=2095260 (2025).

29.

Statistics Netherlands. Supply-Use tables for the digital economy, https://www.cbs.nl/-/media/_pdf/2021/49/nl_d3_1_finalreportdigitalsuts.pdf (2021).

30.

Statistics Canada. Digital supply and use tables, 2017 to 2019, https://www150.statcan.gc.ca/n1/daily-quotidien/210420/dq210420a-eng.htm (2021).

31.

Highfill

Surfield

. New and revised statistics of the U.S. Digital Economy, 2005–2020, https://www.bea.gov/sites/default/files/2022-05/New%20and%20Revised%20Statistics%20of%20the%20U.S.%20Digital%20Economy%202005-2020.pdf (2022).

32.

Ferguson

. The ascent of money : a financial history of the world. New York: Penguin Books, 2009.

33.

Supreme Court of the United States. Feist Publications, Inc. v. Rural Tel. Serv. Co., 499 U.S. 340, https://supreme.justia.com/cases/federal/us/499/340/ (1991).