## Chapter 22: Digitalisation ## Source details **Canonical URL:** [Chapter 22: Digitalisation](https://www.imf.org/-/media/files/data/statistics/bmp7/draft-chapters/sna-ch22-bpm7-ch16.pdf) ## Other formats - [Markdown version](/-/media/files/data/statistics/bmp7/draft-chapters/sna-ch22-bpm7-ch16.pdf.md) - [Structured JSON version](/-/media/files/data/statistics/bmp7/draft-chapters/sna-ch22-bpm7-ch16.pdf.json) --- ### A. Introduction and scope - SNA/BPM joint draft chapter frames digitalization as falling costs and rising capabilities to process, transmit, and store digitized data, enabled by semiconductor chips, computing and electronic communication equipment, software, the internet, and wireless digital telecommunication services (22.0). - Digital assets (crypto assets, data and software) assume roles as stores of wealth or inputs in production, creating needs for enhanced visibility of digital activities, products, and transactions in the macroeconomic accounts and guidance within the SNA/BPM conceptual framework (22.1–22.3). - Chapter organization: - Section B: digital products and assets; - Section C: two types of nonfinancial online platforms and free online products; - Section D: digitalization and the financial system; - Section E: price and volume measurement issues; - Section F: analytical tools including a thematic digital supply and use table (SUT) and an extended account for free digital platform services (22.3). ### B. Digital products and assets — cloud computing, data, AI, NFTs - Cloud computing (22.6–22.15) - Cloud computing shifts computing from users’ premises to remotely located data centers and substitutes purchased IT services for ownership of hardware and software; cloud services are primarily intermediate consumption (22.6). - Main product categories: Infrastructure-as-a-service (IaaS); Platform-as-a-service (PaaS); Software-as-a-service (SaaS); Function-as-a-service (FaaS); Business-process-as-a-service (BPaaS) (22.8). - Accounting and asset treatment for long-term cloud arrangements: - Users with long-term contracts for dedicated server access are economic owners and the contract is a financial lease (22.10). - Long-term software licenses accessed in the cloud (term more than a year, or one-year licenses that automatically renew) are software assets of the user; licenses of less than one year are intermediate consumption (22.10). - Software subscriptions with a long-term license are software assets; licenses under one year are treated as intermediate consumption (22.10). - Cross-border recording: consumption is recorded in the location of the production process that uses the service; exports/imports of cloud computing are recorded under computer and information services (22.14). - Data center construction is fixed capital formation; leases where the cloud enterprise bears operating risks should be treated as financial leases (22.12). - Shift from ICT fixed assets to cloud services complicates analysis of ICT fixed capital formation’s contribution to growth and TFP (22.13). - Data assets and databases (22.16–22.25, 22.23) - Definition: data asset = information content produced by accessing and observing phenomena and recording, organizing, and storing information elements in digital format that provides an economic benefit when used in production (22.17). - Own-account production: most data assets are produced internally and valued at cost of production, including expenses to develop strategy, collect/record information elements, and access observable phenomena (22.19). - Purchases: classification depends on duration and rights—purchases can be fixed capital formation or intermediate consumption; copies with general rights for more than a year are fixed assets (22.20). - Treatment of platform-collected data: payments to subjects for access to their observable phenomena are costs of producing the data asset and classified as distributions of income included in rents; agreeing to data collection alone does not constitute production (22.21). - Asset boundary/service-life: data used in production for more than a year conceptually meets the fixed asset boundary; many data types have short service lives and should be treated as intermediate consumption when appropriate (22.22). - Practical compilation choices: where separating long- and short-lived data costs is infeasible, either estimate combined costs and split by assumed ratio or assume a relatively short service life for data assets (22.23). - Artificial intelligence (AI) (22.26–22.30) - AI defined as capabilities of computer programs/systems for recognition, reasoning, communication, and prediction that emulate human abilities; includes machine learning and deep learning (22.26). - AI systems are classified as a special type of software within the IP product class “Computer Software, including Artificial Intelligence Systems,” with separate “of which” reporting encouraged (22.29). - Valuation guidance: value of data used to train or assist AI must be recorded separately from AI software; a database created solely as a step in AI production and not re-usable may be included in AI production costs (22.29). - Service life: AI software with learning capabilities may improve while in use and learning may extend service life; automatic internet updates can also improve performance (22.30). - Nonfungible tokens (NFTs) (22.31–22.35) - NFTs are digital records on a blockchain associated with an asset and certify ownership or authenticity; payments for NFTs usually occur in the blockchain’s native fungible crypto asset (22.31). - Three NFT classes by rights conveyed: 1. NFTs conveying no ownership rights—treated as consumption; creation = production of a service; rare high-price exceptions may qualify as valuables (22.31–22.32). 2. NFTs conveying limited ownership rights—fall in asset class containing contracts, leases and licenses if confer valuable realizable benefits (22.33). 3. NFTs conveying full ownership—recording of ownership similar to a property title; underlying asset already recorded in national accounts and NFT is not a separate asset (22.34–22.35). - Cross-border treatment follows underlying asset rules (22.35). ### C. Digital platforms, digital intermediation platforms (DIPs), and free online platforms - Definitions and platform taxonomy (22.36–22.44) - Digital platforms facilitate interactions via the internet between distinct sets of users and are subject to network effects (22.36). - Distinction: e-commerce firms take possession or produce the goods; digital intermediation matches parties and facilitates transactions without taking ownership (22.37–22.38). - Three platform types: a. Nonfinancial digital intermediation platforms (DIPs); b. Free online platforms (advertising/data-funded); c. Financial digital intermediation platforms (discussed in Section D) (22.36–22.38). - Nonfinancial DIPs — measurement and rerouting (22.39–22.44, 22.45–22.71) - Output of a DIP = digital intermediation services remunerated by fee/commission; does not include intermediated goods/services (22.39). - Producer approach: to capture economic substance when platforms pass payments to producers, reroute payments so that (i) direct sale recorded between producer and buyer, and (ii) producer purchases intermediation services from the platform (22.39–22.40). - Advantages of producer approach: production recorded in appropriate industry, consumption in appropriate commodity class, and intermediate consumption of DIP services is accounted for (22.40). - Measurement challenges: - Rerouting payments may require unavailable data, necessitating assumptions (22.45). - Distinguishing platform roles (intermediary vs. taking possession vs. employer) is difficult; cases where platform takes possession or acts as employer should be treated as e-commerce (22.45). - Growth of informal household enterprises and many small cross-border transactions complicate measurement (22.45–22.46). - Cross-border rules: digital intermediation services supplied by non-resident platforms are imports of intermediation for the compiling economy; goods/services produced domestically and sold via non-resident platform are recorded as produced/consumed domestically with the platform fee recorded as an import (22.43–22.44). - Free online platforms and free digital products (22.45–22.71, 22.97–22.99) - SNA valuation: market producers’ outputs are valued by price, which is zero for free products, so the SNA’s nonmarket cost-of-production valuation does not directly apply to most free digital products supplied by commercial enterprises (22.45). - Inclusion in GDP: free products are included in GDP insofar as they are part of bundles or help sell priced products; bundle revenues must cover operating costs of the free-product supplier (22.45–22.46). - Funding and data assets: - Commercial free platforms are generally funded by advertising and data collection; cost of free services is included in the price charged to advertisers (22.59–22.61). - Platforms produce own-account investment in data assets; short-lived data used for targeted advertising is intermediate consumption embedded in advertising services (22.63–22.64). - User-created content: - Unpaid user content produced as leisure lies outside the production boundary; users receiving monetary remuneration may be unincorporated household enterprises producing services (22.68–22.69). - Platforms derive economic benefits from user-generated content; extended accounts can optionally treat unpaid content as a service used by platforms, increasing measured own-account investment in data assets (22.72–22.74; 22.118). - Free software and open-source software (OSS): - OSS development funded by corporations or institutions is own-account software investment recorded by cost of production; volunteer OSS development is outside the production boundary (22.79–22.86). - App stores are DIPs where intermediated apps may be free and funded by advertising, data collection, or premium purchases (22.77–22.78). - Visibility: alternative measures and extended accounts are recommended to increase visibility of free platform services and their welfare impacts (22.97–22.99; 22.111–22.118). ### D. Digitalization and the financial system — fintech, e-money, crypto, and platforms - New digital financial services and reporting guidance (22.72–22.76, 22.82–22.86) - New digital financial services fit existing product categories but should be reported as “of which” items when important and separately identifiable (22.72). - Important suppliers include financial digital intermediation platforms, InsurTech, neobanks, e-money issuers, and online-only foreign exchange bureaus and money transfer operators (22.72). - BPM7 recommends “of which” categories for fintech companies within subsector classifications and for fintech-related instruments/services (22.72). - Financial digital intermediation platforms classification (22.72–22.71) - Financial digital intermediation platforms match suppliers and users of funds, do not take ownership of financial assets, receive fees/commissions, and are classified as financial auxiliaries (S126) (22.72). - Three types: 1. Peer-to-peer and other online lending platforms (household and SME lending) (22.72). 2. Equity-based crowdfunding platforms (facilitate equity stakes) (22.72). 3. Philanthropic crowdfunding platforms (mediated transfers, vetting services) (22.72). - Crypto asset exchanges, crypto asset typology, and recording (22.72–22.76) - Crypto asset exchanges and trading platforms facilitate buying, selling, and staking crypto assets for fees and are classified as financial auxiliaries (22.72). - Crypto assets with corresponding liabilities (asset-backed stablecoins, debt/equity security crypto assets) are financial assets; crypto assets without corresponding liabilities are non-produced, non-financial assets in a separate category (22.72). - Typology of crypto assets: - Fungible vs non-fungible (NFTs) (22.73). - Fungible crypto assets categories: (1) designed as general medium of exchange (subdivided by presence of corresponding liability); (2) designed as medium of exchange within a platform/network (subdivided by presence of liability); (3) security crypto assets (22.74). - Digital assets with corresponding liabilities include CBDCs, security crypto assets, payment tokens with corresponding liabilities, and most stablecoins; payment tokens entitle holder to future access to a good or service (22.74–22.75). - All crypto assets fall within the SNA asset boundary; those with corresponding liabilities are recorded as financial assets, those without are recorded as non-produced, non-financial assets (22.75). - Validation/mining: - Validation of crypto transactions is a service; mining (proof of work) rewards miners with newly released units plus explicit fees; validation rewarded with newly released units is assumed collectively consumed by existing holders; validation rewarded by explicit fees is consumed by the transactor paying the fee (22.76). ### E. Price and volume measurement issues — methods and challenges - Overarching issues (22.77–22.96) - Rapid changes in product characteristics, new digital products, quality improvements embedded in models or service contracts, growth in digital IP products with no physical units, and free products complicate price and volume measurement (22.77). - Regular refreshment of samples and matched models procedures are first steps to capture quality change; hedonic regression and machine learning-enhanced hedonic methods are recommended for large e-commerce datasets (22.82–22.84). - Outlet substitution bias from buyers shifting to new lower-price sources due to digitalization can be addressed via unit value price indices where homogeneous composition is met (22.85–22.86). - Software and data deflation: - Volume/price growth of own-account software and data often must be inferred from inputs’ prices and volumes, assuming no productivity change or using related activity productivity (22.92). - Continuous learning in AI software is not captured in volume of software investment (22.92). - Data heterogeneity and obsolescence affect valuation and volume measurement of gross investment in data (22.93). - Cloud computing deflation is challenged by product heterogeneity and frequent introductions; samples of stable product varieties can be used for deflators and international cooperation may be needed for cross-border consumption deflators (22.94). - Internet and telecommunications: CPI/PPI methods may miss volume growth from speed and coverage improvements; alternative is to define physical indicators (e.g., data usage) and aggregate with expenditure weights (22.95). - E-commerce and DIP deflators: deflators must represent e-commerce prices and online outlets; monthly unit values may be preferable for high-frequency online price changes (22.96). - Variety, customization, and free products: welfare gains from better matching and free products increase household consumption volume conceptually but are beyond standard production measurement; extended analytical measures can capture some of these effects (22.96–22.97). ### F. Analytical tools to increase visibility — thematic account, digital SUTs, and extended account on free services - Thematic account on the digital economy (22.100–22.104) - Purpose: increase visibility of digital segments of industries, products, and transactions by providing alternative aggregations and additional detail (22.100). - Priority content: digitally ordered and digitally delivered breakouts; consumption of cloud computing and digital intermediation services; output/value added of ICT and digital industries; supplier classification (e-tailers, data- and advertising-driven platforms, DIPs, digital financial providers, other digital-only producers) (22.101). - Additional content: ICT uses and investment, trade flows of ICT goods and digital services, decompositions by product category, and share of domestic transactions digitally ordered; free platform services can be included as addendum items (22.102–22.104). - Digital SUTs (22.105–22.110) - Purpose: analyze digitalization across transaction nature, products, and industries by adding detail to standard SUTs aligned with classifications (22.105). - High priority aggregates to reduce compilation burden: - decompositions by transaction nature; - output/intermediate consumption of digital intermediation services, cloud computing services, ICT goods and digital services; - output and gross value added of digital industries (22.106). - E‑commerce/digital ordering rows distinguish transactions ordered directly, via resident DIPs, and via non-resident DIPs (22.107). - Add rows for ICT goods, cloud computing, digital intermediation services, and other digital services; addendum for products beyond standard production boundary (free platform services, volunteer communities) (22.109). - New industry columns: one digitally enabling industry column and six digital-industry columns (advertising/data-funded platforms; DIPs; producers depending on DIPs; e-tailers; financial service providers primarily digital; other digital-only producers) (22.110). - Extended account to value free services consumed by households (22.111–22.118) - Rationale: households derive significant value from free advertiser- and data-funded platform services; estimating their value aids understanding of household consumption growth impacts (22.111). - Extended account flexibility: present expanded measures of household final consumption and platform output, accepting analytical double counting for usefulness (22.112–22.113). - Valuation approaches for imputed barter between households and platforms (four theoretical approaches) (22.115): 1. platform’s cost of producing free services; 2. value households place on free services; 3. value households place on privacy they give up; 4. economic benefits platforms derive from users’ attention and data. - Recommendation: use cost of production for the extended account for consistency with the system and practicality (22.115). - Classification of imputed license payments: the license to access households’ observable phenomena is not a service produced by households; the imputed payment is classified as a rent (22.116). - Accounting effects: - Platforms’ expanded output and value added increase by the imputed value of free services, balanced by imputed rent payments for data licenses; platforms’ balance of primary incomes remains unchanged and household saving is unchanged (22.117). - User-generated content: extended account can treat unpaid content as barter, increasing measured own-account investment in data assets (22.118). *Source: SNA2025/BPM7 chapter 22 content on digitalisation (SNA/BPM joint draft chapter).* ### Chapter 22: Digitalisation ### Chapter 22: Digitalisation ### A. Introduction - 22.0: Falling costs and rising capabilities to process, transmit, and store digitized data have resulted in extensive integration of digital technology into goods and services and the activities of production and consumption. This transformation is referred to as digitalization. Digitalization has been enabled by many types of information and communications technology (ICT) goods and services, including semiconductor chips, computing and electronic communication equipment, software, the internet, and wireless digital telecommunication services. - 22.1: Digital assets—assets that exist only in digital form (such as crypto assets, and data and software)—have assumed important roles as stores of wealth or inputs in production. The impact of digitalization on production, consumption, investment, prices, finance, and international trade calls for enhanced visibility of digital activities, products, and transactions in the macroeconomic accounts. Guidelines are needed on measuring the activities, products, and assets associated with digitalization in the conceptual framework of the SNA/BPM. - 22.2: Measurement issues related to digitalization are also discussed in other SNA/BPM chapters and manuals. Chapter 11 of the Balance of Payments and International Investment Position Manual recommends showing computer and information services as a first-level services category in the balance of payments current account. Non-financial digital assets discussed across capital account and goods and services chapters include ICT equipment, software, data and databases, crypto assets without a corresponding liability, and digital elements of research and development. Financial-account classification issues include crypto assets with a corresponding liability and electronic money (e-money). Price, volume, and productivity measurement challenges affect digital products and require quality-adjusted comparisons. - 22.3: The chapter consolidates conceptual and measurement guidance on digital products, activities, and assets—including associated cross-border transactions—and recommends tools to increase visibility of digitalization in national accounts and external sector statistics. The chapter is organized with Section B on digital products and assets, Section C on two types of nonfinancial online platforms and free online products, Section D on digitalization and the financial system, Section E on price and volume measurement issues, and Section F on analytical tools including a thematic digital supply and use table (SUT) and an extended account for free digital platform services. ### B. Digital Products — Overview - 22.4: Services supplied over a computer network are a defining feature of the digital economy. Digital economy services include wholesale and retail e-commerce distribution services, priced and free services of online platforms, audio and video streaming, and digital financial and payment services such as mobile money services. Production and consumption of these services are enabled by ICT equipment, software, and data and databases along with ICT consumer durable goods, and mobile and fixed line digital communication services. ICT components embedded in non-ICT equipment (such as semiconductor chips in motor vehicles) also facilitate consumption of digital services. - 22.5: Cloud computing has emerged as a new way of accessing IT resources and is associated with new types of assets. Cloud computing, data assets, AI systems, and non-fungible tokens are discussed in this subsection. Online platforms and digital financial services and fungible digital assets are addressed in other subsections. 1. Cloud computing - 22.6: Cloud computing shifts most computing from users’ premises to remotely located data centers accessed over a network (“the cloud”). Growing use of cloud computing has caused large-scale substitution of purchased IT services for ownership of hardware and software assets. Cloud computing services are primarily used as intermediate consumption (input into the production of other goods and services). - 22.7: Cloud computing services consist of computing, data storage, software, and related IT services accessed remotely, supplied on demand and with measured resource usage. Measured resource usage enables pay-per-use charging and efficient resource allocation via resource pooling. Rapid elasticity allows users to scale consumption up or down quickly. - 22.8: Main cloud computing product categories: - Infrastructure-as-a-service (IaaS): on-demand access to hardware such as a virtual server. - Platform-as-a-service (PaaS): access to a software platform. - Software-as-a-service (SaaS): access to application software. - Function-as-a-service (FaaS): simplified PaaS allowing execution of application functionalities in response to events. - Business-process-as-a-service (BPaaS): automation of business processes using cloud software and platforms (i.e., SaaS and PaaS). - 22.9: Cloud computing is part of a broader shift to remote computing including colocation and hosting services. IT users may lease colocation space, or lease servers from managed or unmanaged hosting suppliers, and often use combinations of colocation, hosting, and cloud services. - 22.10: Accounting and asset treatment for long-term cloud arrangements: - Users with long-term contracts for dedicated access to a server in a cloud datacenter are considered economic owners if operating risk is borne by the user, making the contract a financial lease. - Long-term software licenses accessed in the cloud (term more than a year, or one-year licenses that automatically renew) are software assets of the user rather than intermediate consumption. - Software subscriptions from publishers that come with a long-term license are software assets; licenses of less than one year are treated as intermediate consumption of software services. - Treatment of cross-border transactions in such licenses (more than one year/less than one year) is under discussion for inclusion in Chapter 11, Services Account, BPM7. - 22.11: Fixed capital formation of cloud computing enterprises may include own-account production of software and equipment or equipment designs; production of original equipment designs may be measured as part of own-account R&D capital formation. - 22.12: Data center construction is part of fixed capital formation associated with cloud computing. Real estate enterprises that construct and operate data center buildings often lease to cloud computing enterprises; if the cloud enterprise bears operating risks, the lease should be treated as a financial lease. - 22.13: The shift from purchasing ICT fixed assets to consuming cloud services complicates analysis of ICT fixed capital formation’s contribution to economic growth and total factor productivity (TFP) growth. Detailed data on consumption of cloud computing and hosting services may require adding product detail on cloud-related services to existing classifications, potentially via a digital supply and use table (SUT) discussed in Section F. - 22.14: Cross-border aspects: - Cloud computing and remote computing services are often supplied across borders, frequently by multinational enterprises with domestic and foreign computing establishments connected by cross-border networks. - Consumption occurs in the location of the production process that uses the service as an input. Example: a business in country A purchasing computing services from an establishment in country B implies an export for country B and an import for country A; such services are recorded under computer and information services in the balance of payments (refer to Chapter 11, Services Account, BPM7). - Resource pooling and workload shifts between servers/establishments can make physical production location hard to determine, but spending on exported and imported cloud computing should be trackable. - Ensuring consistency between estimates of exports/imports of cloud computing services and the net exports implied by an economy’s production and consumption of cloud computing services may improve estimates. For an economy that only imports cloud services (and does not produce them), data on intermediate consumption of these services by businesses could provide a reliable estimate of imports. - 22.15: Hosting and colocation services are exported when foreign-owned servers and software are hosted in a domestic data center, and imported when locally owned IT assets are hosted in a foreign data center. Investment in the IT assets should be recorded in the economy of their owner. 2. Data assets - 22.16: Data has emerged as an important type of intellectual property product. Many enterprises derive much of their value from holdings and uses of data, and many products and production processes depend on data for matching suppliers and buyers, product improvements, operational efficiency, planning, and marketing. - 22.17: Data as an asset is defined as information content produced by accessing and observing phenomena and recording, organizing, and storing information elements from these phenomena in digital format that provides an economic benefit when used in productive activities. Digitized information that does not provide a direct economic benefit to its owner, including ancillary by-product data, is excluded. Data assets are produced when information on observable phenomena (OP) such as facts, behaviors, and characteristics is recorded, organized, and stored in digital format. - 22.18: Producers derive economic benefits from data assets by creating and organizing databases that facilitate analysis and by extracting insights and knowledge; related intellectual property products include databases, software, research and development, and mineral exploration. The cost of acquiring data used to develop an intellectual property product is included in the value of that intellectual property product. - 22.19: Most data assets are produced internally. The value of own-account data assets is measured by their cost of production, including expenses to develop a data production strategy, collect and record information elements, and access the observable phenomena. - 22.20: Data can be acquired via purchase transactions. Classification depends on use and rights: - Purchasing data may represent fixed capital formation or intermediate consumption depending on duration of use in production and limitations on purchaser’s use. - Sale of the data original (including rights to sell copies or access and all ownership rights) transfers ownership of the data asset to the purchaser. - Copies of long-lived data with general rights to use in production for more than a year are classifiable as fixed assets of the purchaser. - A non-exclusive right to access a copy of data for a limited or specific purpose is generally treated as a service. - Right to use a purchased copy of data for less than a year is a service analogous to an operating lease. - Cross-border transactions in data assets are recorded in the services account (refer to BPM7 Chapter 11, Services Account). - 22.21: Digital platforms often collect user data and user-created content. If subjects of data collection receive payments for granting access to their observable phenomena, those payments are part of the cost of producing the data asset and are classified as distributions of income included in rents. Agreeing to data collection (e.g., by accepting a license agreement to use a free online platform) does not constitute production and is not a service. Platform users who receive payments for undertaking specific actions to assist data collection and recording supply a service, but such cases are likely rare. - 22.22: Asset boundary and service-life considerations: - The SNA asset boundary includes produced assets used in production for more than a year, so data used in production for more than a year is conceptually a fixed asset. - Many types of data (including behavioral data for targeted advertising) tend to have a relatively short service life. The value of goods and services produced with inputs of shorter-lived data conceptually includes the value of that data. - When feasible, shorter-lived data should be identified and treated as intermediate consumption rather than as a fixed asset. _Chapter 22: Digitalisation (SNA/BPM joint draft chapter)_ ### 22.23 In practice, the cost of producing long-lived data may be hard to estimate separately from the cos ### 22.23–22.45: Treatment of Data, AI, NFTs, and Digital Platforms ### Data and databases as assets (sections 22.23–22.25) - In practice, the cost of producing long-lived data may be hard to estimate separately from the cost of producing short-lived data, making it necessary to: - estimate the combined cost of producing data and split the combined estimate into short-lived and long-lived components based on an assumed ratio; or - assume a relatively short service life for data assets as estimated in practice to reflect the presence of types of data with a useful life of a year or less. - Including expenses that do not create economic benefits in future years in fixed capital formation overstates producers’ gross value added but can capture the value of stocks of data whose useful economic life is a year or less. - Service lives for data under a year are common; if the measure of production of data does not exclude all data with a service life of a year or less, a relatively short assumption for data service life is likely appropriate. - Expenditures to update or add to an existing own-account data asset are capitalized. - Suppliers of software and connected equipment with embedded software or AI systems that collect users’ data to update or add to their data assets: expenses to collect users’ data for these purposes are part of investment in data assets. - Data and databases are conceptually distinct, but: - they are produced with similar inputs and are often difficult to measure separately; - transactions in databases generally include the value of the data stored in the database. - Data and databases are combined into a single detailed class of intellectual property (IP) product known as “data and databases,” which is further combined with software (including AI) to form a higher-level class of IP product. - Distinction retained in labels: “data” = information content recorded in digital form; “databases” = design, structure, and organization of files to permit efficient access. - Costs of producing a database include planning and implementing structure/design and organizing data to facilitate use. *Key conceptual points about recording and valuation of data and databases drawn from the SNA/BPM treatment of intellectual property products.* ### Artificial intelligence (AI) (sections 22.26–22.30) - Definition and capabilities: - Artificial intelligence refers to capabilities of a computer program, or a system controlled by a computer program, of recognition, reasoning, communication, and prediction that emulate human recognition, reasoning, and communication. - Machine learning: data enables an AI software program to learn to predict or classify from experience. - Deep learning: a type of machine learning that enables some AI programs to improve from experience while being used in production. - AI often uses data beyond a scale that humans could analyze. - Examples of AI-enabled products/capabilities include: - text mining, computer vision/image recognition, speech recognition, natural language processing, personalized recommendations, content creation with generative AI; - translation, predictive modeling, risk assessment by lenders and insurers, data analytics, writing summaries of large textual databases, smart robots, autonomous drones and vehicles, face recognition, fraud detection, cybersecurity. - Economic implications: - AI has expanded the types of jobs or job elements potentially subject to automation and may affect patterns of international trade by facilitating digital ordering and changing relative costs of production. - The transformative impact of AI calls for granular data to analyze AI prevalence and effects on labor markets and international trade. - Classification and compilation guidance: - AI systems are distinguished as a special type of software within the IP product class “Computer Software, including Artificial Intelligence Systems,” with separate reporting of AI encouraged as an “of which” item. - AI is classified as a special type of software even when AI systems include data and hardware elements because the system is controlled by software; equipment that contains an embedded AI system (or other embedded software) is still classified as equipment. - General compilation guidelines for software, data and databases in chapter 11 of the 2025 SNA/chapter 11 of BPM7 apply to AI software, but AI raises special issues due to use of data and machine learning. - The value of data used to train AI or to help AI generate output should be recorded separately from the value of AI software, because the data could have multiple uses. - A database created solely as a step in production of an AI program and that cannot be re-used may be included in the cost of production used to value the AI program. - Service life and performance: - Unlike typical fixed assets that deteriorate over time, AI software with learning capabilities may improve as it is used, and learning from experience may extend the service life of many AI programs. - Automatic software updates delivered over the internet may also improve performance of software already in use. ### Nonfungible tokens (NFTs) (sections 22.31–22.35) - Definition and characteristics: - NFTs are digital records hosted on a blockchain associated with a digital or physical asset or product but distinct from that asset or product. - NFTs certify ownership of rights to use and benefit from the asset and may certify authenticity. - NFTs are nonfungible because the associated asset is unique and not interchangeable with other assets in the same class; payments for NFTs usually must be made in the fungible crypto asset native to the blockchain on which the NFT is hosted. - Three NFT classes (classification based on rights conveyed): 1. NFTs that convey no ownership rights and only allow personal use of another asset or product. - Purchase treated as consumption. - Not an investment item because it cannot be used in production and generally does not serve as a store of value. - Creation represents production of a service; cross-border transactions of this type are recorded under computer and information services. - Exception: in rare cases an NFT in this class that commands a high price when first auctioned and has enough exclusivity and appeal to have a lasting value may qualify as a valuable. 2. NFTs that convey limited ownership rights beyond personal use (include use for commercial purposes). - These NFTs fall in the SNA/BPM asset class containing contracts, leases and licenses if they confer valuable benefits that the holder can realize in practice. - Assets in this class are non-produced, nonfinancial assets; ownership rights conveyed may affect the value of the encumbered asset. 3. NFTs that convey full ownership rights. - NFTs conveying full ownership are a method of recording and verifying ownership of an underlying asset; the underlying asset should already be recorded in the national accounts. - Such an NFT is a digital recording of ownership similar to a property title, not a separate asset; purchasing the NFT is a way of purchasing the underlying asset. - For cross-border transactions: - If the underlying asset is digital, treatment follows principles for recording such assets (goods or computer services). - If underlying asset is physical (e.g., a house property), treatment follows existing principles for recording such assets. ### Digital platforms and digital intermediation (sections 22.36–22.44) - Digital (online) platforms: - Operators use digital technology to supply services that facilitate interactions via the internet between two or more distinct but interdependent sets of users (firms or individuals). - Platforms are subject to network effects: value to each user depends on number of other users. - Digital platforms (online platforms) deliver services via the internet. - Distinction from e-commerce firms: - E-commerce firms take possession of the goods they sell or directly produce the services they sell. - An establishment that sells its own merchandise and an establishment that facilitates selling by others can both be part of the same multi-establishment enterprise. - Both use data to produce matching services; digital intermediation uses digital technology and data to match parties that desire to transact and to facilitate transactions. - Three types of digital platforms: a. Nonfinancial digital intermediation platforms (DIPs): facilitate transactions between buyers and sellers for ordering and delivery of goods and services for a fee or commission without taking ownership of the intermediated goods or services. b. Free online platforms: facilitate non-commercial interactions between users or provide entertainment and information services, usually funded by advertising and user-data collection. c. Financial digital intermediation platforms: mediate funding or payment transactions (discussed separately in the section on digitalization and the financial system). - Nonfinancial DIPs (sections 22.39–22.44): - DIPs facilitate ordering, payment for, and delivery of goods and services supplied by producers to buyers and charge fees for digital intermediation services; they may provide quality assurance (e.g., vetting). - Output of a DIP is digital intermediation services, recompensed through a fee; it does not include the goods and services the DIP helps others sell. - Many DIPs accept customers’ payments for goods/services produced by platform users and deduct their fee before passing amounts to the producers. - To capture economic substance when platforms pass on payments after deducting fees, payments must be rerouted to reflect: - a direct sale of the intermediated product from producer to buyer, and - a purchase by the producer of intermediation services from the platform (the producer approach). - Advantages of the producer approach: - production of goods/services sold on DIPs is recorded in the appropriate industry; - consumption of goods/services is recorded in the appropriate commodity class; - the intermediate consumption of the DIP’s services on the producer’s value added is accounted for. - When the DIP invoices the buyer separately for its services and for the intermediated product, the buyer is recorded as purchasing intermediation services from the platform and the intermediated product from its producer; if the DIP charges fees to both buyer and seller, three transactions are recorded. - Cross-border aspects: - Digital intermediation services are frequently supplied across borders by non-resident platforms. - Goods and services produced by resident producers and consumed by resident buyers via a non-resident platform should be recorded as produced and consumed in the compiling economy; the platform’s fee/commission is recorded as an import of digital intermediation services of the compiling economy. - If a resident DIP collects cross-border payments for intermediated goods/services, rerouting is required so goods/services are not treated as being imported at the net price received by their producer and then exported at a marked-up price paid by the buyer. - Digital intermediation services can facilitate exports by resident suppliers or imports from foreign suppliers; digital intermediation services used by an exporter are included in the value of the exported good or service. - Example: If a DIP in country A facilitates supply of a service by a resident of country B to a resident of country C, the price paid by buyer in country C is the value of service exported by country B and imported by country C, and the platform fee is the value of digital intermediation services exported from country A and imported by country B. - Domestically produced digital intermediation services used by the producer of an imported good/service should conceptually be treated as an export of services and included in the value of the imported good/service; implementing this treatment may require rerouting the fee to pass through the foreign producer and may require assumptions if source data to compile rerouted flows are unavailable. *Source: SNA2025/BPM7 chapter 22 content on data, artificial intelligence, nonfungible tokens (NFTs), and digital platforms.* ### 22.45 In general, services sold for a fee or commission are straightforward to measure, but in the case of DIPs and ### 22.45–22.71 Digital intermediation, free online platforms, and digitalization of financial services ### Digital intermediation platforms (DIPs) — measurement challenges and effects - The chapter notes that services sold for a fee or commission are generally straightforward to measure, but DIPs and the goods and services they intermediate present common compilation challenges. - Key measurement difficulties: - Rerouting payments collected by a DIP to producers of intermediated goods and services may require unavailable data, making assumptions necessary. - Distinguishing transactions between platform users from transactions where the platform takes possession of the good, uses the service as an intermediate input to produce a different service, or acts as an employer is difficult. - Cases in which the platform takes possession or acts as employer should be treated as e‑commerce activity. - Economic effects of DIPs highlighted: - DIPs allow producers to interact with previously unreachable consumers, including those in other geographical locations, creating selling opportunities for previously excluded producers. - Growth in activity by informal household enterprises, which are likely missing from business registers and standard sources. - Rapid growth of small international transactions in goods and services that may be below de minimis thresholds for customs duties and documentation requirements. - Source data on DIPs with no local presence and on the activity they intermediate is often not easily available. ### Free online platforms and free digital products — valuation and inclusion in GDP - The chapter states that the SNA framework for valuing free outputs of nonmarket producers by cost of production does not apply to most free digital products because they are supplied by commercial enterprises; the outputs of commercial enterprises are valued by their price, which is zero for free products. - Inclusion of free products in GDP: - Free products supplied by market producers are included in GDP as part of the price of other products they help sell or with which they are bundled, directly or indirectly. - Items in the bundle must together generate at least enough revenue to cover the operating costs of the supplier of the free product, so the supplier’s overall output is not undermeasured. ### Free products supplied by non-platform firms — freemium and cross‑subsidy issues - Findings: - Free outputs can promote sales of priced outputs to the same set of customers; suppliers often use a “freemium” strategy where a free basic version promotes upgrades or a premium version. - Promotional outputs may have low prices subsidized by fully priced products (example: low-priced printers boosting sales of high-priced ink cartridges). - While zero or low prices do not undermeasure total producer output, they affect measurement of the composition of output. - Measurement implications: - Relative values within bundles matter for measurement of fixed capital formation when bundled items are used for fixed capital formation versus intermediate consumption (e.g., supplies, training, maintenance). - Relative values affect measurement of exports, imports, or trade patterns if parts of the bundle are sourced from different countries. - Effects on weights of price and volume indexes could matter if item prices in the bundle behave differently. ### Free online platforms — structure, funding, and data assets - Structure and funding: - Most free online platforms are commercial, often two‑sided (multi‑sided), charging one side and supplying free services to the other side to attract users. - Commercial free online platforms are generally funded by advertising and the collection of data on users; non-digital free radio and television broadcasters are funded by advertising. - Platforms funded by advertising assemble audiences via free services and include the cost of free services in the price charged to advertisers. - Measurement consequences: - The standard method of measuring a market producer’s output by the producer’s sales is applicable to both the platform and the funder-side users. - Households’ expenditures on products advertised on the platform include the cost of the platform’s services embedded in product prices. - Data assets and short‑lived data: - Digital platforms that collect and store data on users produce data assets as own-account investment (and may sell rights to use the data). - Value of own-account investment in data assets is usually measured by cost of production. - Platforms may collect short-lived data on recent browsing behavior used as an intermediate input for targeted advertising; value of short-lived data is part of the price of advertisement targeting services and can be treated as intermediate consumption embedded in the price of the products it helps produce. - Nonmarket free platforms: - A few free online platforms (e.g., public wikis maintained by volunteers) are owned by nonprofit institutions and operate as non-market producers; their output is valued by cost of production. - Production costs of nonprofit free platforms may be modest relative to physical indicators because volunteers do much of the work; volunteer work is outside the SNA production boundary. ### Content created by platform users — classification and measurement - The chapter states: - User-created content produced as leisure is outside the SNA production boundary; if the creator receives no remuneration, content is assumed leisure. - Households receiving monetary remuneration from advertisers or platforms for uploaded content may be unincorporated household enterprises supplying services to the advertiser or platform. - Where platforms take fees out of advertiser payments passed to content creators, content creators should be treated as purchasers of the platform’s services and as producers of services used by advertisers; if purchaser is a non-resident, the service should be included in exports of services. - Economic benefits and potential measurement options: - Economic benefits platforms receive from user-generated content include attracting users, selling advertising, and adding to the platform’s stock of data assets; these are positive externalities and do not imply the unpaid creator produced a service used by the platform. - Cost of supplying free services to user-creators may be included in the platform’s own-account investment in data assets. - An extended account can optionally treat user-generated content as a service produced with inputs of the platform’s free services and used by the platform as an input in its own-account production of data assets. ### Free software, open‑source software, and app stores — treatment as assets and measurement - Free software and distribution: - Free software is often free to download; hosting platform services may not be free. Free software may be used for final consumption or in production and is frequently supplied across borders. - App stores are a type of DIP where the intermediated service (the app) is often free and may be funded by advertising, data collection, premium purchases, or other services the app facilitates. - Open‑source software (OSS): - OSS has source code publicly available under a license to copy, use, inspect, modify, and share; it is used in production by corporations, governments, and nonprofit institutions. - OSS is usually developed, maintained, and supplied by a corporation, university, government research office, or nonprofit institution; the investment to develop the original is included in a general estimate of own-account software investment based on costs of production (e.g., compensation expense for software developers). - Complex OSS may contain components developed in multiple economies; allocation of investment among economies where development work takes place may be appropriate, or the entire investment may be attributed to the economy of residence of the owner of the software asset. - OSS can qualify as an asset of its developer if the producer bears the risks of production to claim benefits associated with the use of the asset; OSS developed by corporations is usually funded through sales of complementary services or other products it helps sell. - Unpaid OSS development by volunteers is outside the SNA production boundary; independent developers of OSS for commercial purposes are unincorporated household enterprises investing in own-account software. - Value of OSS copies supplied by enterprises may be embedded in the price of complementary outputs; substitution of OSS for purchased software affects measured software investment and uses of output. - Valuation principle reiterated: - The SNA values outputs of nonmarket producers by cost of production, not by inferred willingness-to-pay of users. ### Increasing the visibility of free online platforms and products - Analytical relevance: - The value of free services supplied by advertiser- and data-funded platforms to households is relevant for analytical purposes and understanding the broader impact on household consumption. - Alternative measures of household final consumption expenditures and of the output of free online platforms that include households’ direct consumption of advertiser- and data-funded platform services may be presented in an extended account on free online platforms (as discussed in Section F). - The effect of free platforms and products on volume growth of household consumption is conceptually relevant for understanding the impact of digitalization on prices and volumes (discussed in section E). ### Digitalization and the financial system — new services and means of payment - The chapter notes that digitalization has produced new financial service products and new digital assets designed as means of payment. - Classification and reporting guidance: - New digital financial services fall within existing product categories and new digital assets fall within existing asset categories, but they should be reported as “of which” items (or as part of product detail in a thematic account on the digital economy and associated digital supply and use tables) when important and separately identifiable. - Important suppliers of digital financial services include financial digital intermediation platforms, digital providers of insurance services (InsurTech), digital banking platforms operating solely online (neobanks), e-money issuers, and online-only foreign exchange bureaus and money transfer operators. - Important digital payment mechanisms include e-money (which includes mobile money), digital assets used as a means of payment including fungible crypto assets, and central bank digital currencies (CBDCs). - Note on reporting convention: - BPM7 recommends introducing “of which” categories for i) fintech companies within the subsector classification; and ii) instruments or services classifications where necessary to separate out fintech-related instruments and services. ### Financial digital intermediation platforms — types and classification - Classification: - Financial digital intermediation platforms provide matching services and facilitate financial transactions between suppliers of funds and users of funds; they do not take ownership of financial assets arising from claims on users of funds nor incur liabilities to suppliers of funds. - They receive fees or commissions and are classified as financial auxiliaries (S126). - They differ from conventional financial intermediaries that incur liabilities on their own account, acquire financial assets, and charge implicitly through interest rate margins. - Three types identified: 1. Peer-to-peer and other online lending platforms — peer-to-peer lending platforms facilitate loans between households; other lending platforms match households and small enterprises seeking funding to institutional investors seeking lending opportunities. 2. Equity-based crowdfunding platforms — facilitate financing transactions where funders receive equity stakes in enterprises or projects. 3. Philanthropic crowdfunding platforms — mediated funding transactions are current transfers or potentially capital transfers; an element of their services is vetting of causes seeking assistance. *Source: SNA/BPM7 draft chapter.* ### 22.72 Another type of financial digital intermediation platform consists of crypto asset exchanges and trad ### sna-ch22-bpm7-ch16 - 22.72 Another type of financial digital intermediation platform consists of crypto asset exchanges and trad ### Crypto asset exchanges and classification - Crypto asset exchanges and trading platforms allow users to buy, sell and stake (lend) crypto assets for a fee or commission. - Because financial assets are generally among the assets traded on a crypto asset exchange, crypto asset exchanges are classified as financial auxiliaries. - Crypto asset brokers act as intermediaries between buyers and sellers of crypto assets and embed their fee into their buying and selling prices. - Crypto assets with a corresponding liability (e.g., asset-backed stable coins, debt, and equity security crypto assets) are classified as financial assets. - Crypto assets without a corresponding liability are classified as non-produced, non-financial assets within a separate category. - If a crypto asset without a corresponding liability ever gains widespread acceptance as a general medium of exchange, the guidance on its classification may be re-considered. ### Digital representations of value and distinctions - Digital assets designed to act as a medium of exchange or financial instrument are digital representations of value recorded on a cryptographically secured distributed ledger or using a similar technology or issued by a central bank as a CBDC. - E-money differs from these digital assets: e-money is monetary value stored electronically on a physical device or stored remotely, represents a liability of the e-money issuer, is denominated in a fiat currency, and must represent general purchasing power. - Crypto assets use cryptography and distributed ledger technology (DLT) such as blockchains to enable direct transactions without a trusted intermediary; DLTs record, synchronize, and share transactions on multiple nodes; blockchains create cryptographic records of transactions and ownership that are impossible to alter. ### Typology of crypto assets - Crypto assets are classified as either fungible or non-fungible (NFTs are unique and non-divisible). - Fungible crypto assets are divisible and not unique (example: one bitcoin equals any other bitcoin). - Fungible crypto assets are classified into three broad categories: - (1) those designed to act as a general medium of exchange (subdivided into those with, and those without, a corresponding liability); - (2) those designed to act as a medium of exchange within a platform or network (subdivided into those with, and those without, a corresponding liability); - (3) security crypto assets. - Security crypto assets certify ownership of a financial instrument, always have a corresponding liability, and should be recorded as debt securities, equity securities, or financial derivatives depending on the nature of the claim on the issuer. - Digital assets with a corresponding liability include CBDCs, security crypto assets, payment tokens with a corresponding liability, and most stablecoins. - Payment tokens entitle the holder to future access to a good or service. - Stablecoins aim to maintain a stable value relative to a specified asset or basket of assets, usually by being backed by assets of the issuer; the stability mechanism might also be an algorithm that causes supply to respond automatically to demand changes. ### Asset boundary and recording guidance - All types of crypto assets are within the SNA asset boundary. - Crypto assets with a corresponding liability are recorded as financial assets; those without a corresponding liability are recorded as non-produced, non-financial assets in a separate category. - Additional details are provided in the SNA chapters on the capital account, financial account, and balance sheet (chapters 11, 12, and 14) and in BPM classification chapters. ### Validation (mining) as a service and its recording - Validation of crypto asset transactions is a service. - For crypto assets without a corresponding liability that rely on proof of work, the validation process is known as mining. - Mining includes the release of new units of the crypto asset as an implicit fee paid to the miner validating the transaction. - The miner also receives an explicit fee in crypto assets paid by the transactor (normally the sender/seller). - Validation services rewarded with newly released units are assumed to be collectively consumed by existing holders of the crypto asset units. - Validation services rewarded by explicit fees are consumed by the transactor paying the fee (normally the sender/seller). - Referenced recording guidance appears in Chapter 7, 2025 SNA and Chapter 11, BPM7 for output of mining and cross-border validation services. ### Measuring prices and volumes for products affected by digitalization — overarching issues - Price and volume measurement challenges often exceed those for output at current prices, because price change is straightforward to measure only when products and their characteristics remain static. - Digitalization causes rapid changes in product characteristics, frequent introduction and exit of digital products, quality improvements embedded in new models or service contracts, growth in digital intellectual property products and services with no physical units, and the frequent appearance or disappearance of free products. ### Sampling, matched models, and quality adjustment methods - Regular and timely refreshment of samples of models (and outlets) used to calculate price indexes is the first step to capture quality change for rapidly evolving digital products. - The “matched models” procedure considers only price changes of continuing models; models not present in both periods are excluded to avoid counting quality differences as inflation. - The matched models method implicitly assumes the quality-adjusted price of the new model equals the price of the model it replaces (after adjusting for general price change). - Hedonic regression models relating price to product characteristics are a recommended method for adjusting prices for quality change. - Hedonic models using machine learning to predict prior-period prices for new models and current-period prices for old models have been proposed for quality adjustment at scale when analyzing large e-commerce datasets. - Options pricing is another technique to adjust prices for quality changes, averaging observations on price differences caused by a characteristic offered as an option. - When a quality change is observed in price index data, prices of new models must be adjusted for the value of their quality difference from replaced models. ### Outlet substitution bias and unit value indices - Outlet substitution bias occurs when buyers obtain a substantially identical product at a lower price from a new source of supply (often enabled by digitalization). - The impact of households shifting to a new source of supply can be captured by compiling a unit value price index (total expenditure divided by total quantity). - A unit value index requires homogeneous composition (items included must have the same per-unit intrinsic value). - Under the weaker assumption that average quality is not falling, a unit value index can provide an upper bound measure of inflation for the product. ### Measuring quality change in ICT goods and goods with ICT components - Frequent quality changes in ICT equipment and durable goods enabled by semiconductor advances pose challenges; new models often offer substantial performance improvements at about the same price. - The matched models method assumes quality-adjusted prices are equal in overlap periods, which may not hold for rapid ICT improvements. - Hedonic regression methods or the option price method can estimate the value of quality changes. - If those methods are infeasible, information on the cost of production of a new product feature may be used to adjust the product’s price index (for example, producer’s cost plus distribution margins for an added vehicle capability). ### Software and data valuation and deflation - The growing importance of software increases uncertainty around deflators for investment in IT products because volume and quality of software packages are hard to assess and in-house production is common. - Price or volume growth of own-account software and data must generally be inferred from inputs’ prices and volumes, either assuming no productivity change or using the productivity change rate of a related activity. - Another option for deflating custom software is to use the price index of a related product, such as standardized software sold by publishers. - Continuous improvement in the performance of AI software that learns from experience is not included in the volume of software investment. - Data is heterogeneous; gross investment in data is measured by the volume and value of the inputs used, but holding gains/losses and normal obsolescence affect its value and volume. ### Cloud computing and deflation challenges - Cloud computing suppliers offer many detailed products within IaaS, PaaS, and SaaS, making a comprehensive price index difficult. - Frequent introductions of new or modified cloud products compound difficulty. - Price data from samples of major product varieties with relatively stable characteristics in each category can be used to estimate a deflator for cloud computing output. - Cross-border transactions complicate deflator estimation for consumption of cloud computing services; international collaboration may be required to obtain price data for cross-border cloud consumption deflators. ### Internet, telecommunications, and alternative volume measures - Household consumption of internet and telecommunications services has risen substantially with more online time via broadband and mobile connections and improved data transmission speeds. - Common CPI and PPI methods may fail to capture volume growth from improved speeds and coverage. - To measure change in cost of purchasing a given volume of internet and telecommunications services accurately, samples of contracts, products and carriers must be kept up to date and prices adjusted for quality changes like transmission speed or geographic coverage improvements. - If price indexes fail to capture appropriate quality improvements, an alternative is to estimate volume growth directly from physical indicators (e.g., data usage), defined at a detailed level and aggregated using expenditure or revenue weights to minimize composition distortion. ### E-commerce, digital intermediation platforms (DIPs), and consumer deflators - Deflators for household final consumption expenditures on items sold online must represent prices from e-commerce outlets and suppliers on digital platforms, because online prices may trend differently from offline prices. - High frequency of online price changes often makes a monthly unit value more suitable than a point-in-time supplier price. - If item-level expenditure data for online sellers is available, index weights should reflect that expenditure data and an index formula resistant to chain drift should be used. - Online prices from e-commerce retailers and platforms such as ridesharing and short-term rental platforms are often lower than offline comparable items. - The conceptual decline in the deflator when households substitute to lower-priced online sources is often not captured in practice due to practical difficulties and concerns about unobserved quality differences. ### Variety, customization, and free digital products — welfare and measurement considerations - Digital platforms and e-commerce expand households’ access to detailed varieties and enable better matching of product characteristics to tastes and needs, improving efficiency of use of produced goods and services. - Households’ material well-being (welfare) from consumption has grown faster than production due to improved matching and customization; these welfare gains could be relevant for extended analytical consumption volume concepts but are not part of production measured by the national accounts. - Increased availability of free digital products represents a decline in price and an increase in the volume of household final consumption. - When a free product directly replaces a priced product, measurement is straightforward—the same expenditure commands a greater volume and the deflator effect equals the decline in the cost of the bundle. - When free online platform services are bundled into prices of advertised products, the theoretical decline in bundle cost depends on the assumed value of the free online services. - An extended account that values free services of digital platforms by their cost of production is discussed under “Analytical Tools to Increase the Visibility of Digitalization.” *International Monetary Fund* ### 22.97 The free digital services and embedded product capabilities in devices such as the smartphone that ### 22.97 The free digital services and embedded product capabilities in devices such as the smartphone that have appeared since the start of digitalization ### Overview and scope (paragraphs 22.97–22.99) - Free digital services and embedded product capabilities in devices can enable households to achieve outcomes that previously required purchases of market goods and services, save time, or allow new activities (22.97). - Theoretical effects on household final consumption deflators and volume growth associated with the appearance or disappearance of free digital products that are not direct replacements for priced products require assumptions that would often be unacceptable for replicability; such effects are therefore beyond the scope of the national accounts measures (22.97). - Digitalization is multidimensional and requires multiple indicators and perspectives; standard national accounts aggregates provide limited information because digital products are often subsumed in broader aggregates and scattered across classifications (22.98). - Enhanced visibility into digital firms, products, and transactions is needed to understand digitalization’s effects on the economy and to reassure users that output and consumption of digital products are fully measured as part of aggregates (22.98). - Flexible, complementary analytical tools recommended: a thematic account on the digital economy, digital supply and use tables (SUTs), and an extended account on free services of online platforms (22.99). ### Thematic account on the digital economy (paragraphs 22.100–22.104) - Purpose: - Bring visibility to activities, products and transactions affected by digitalization that are subsumed in broader aggregates (22.100). - Provide alternative aggregations and additional detail that separately identify digital segments of industries, digital products, and digital transactions (22.100). - Priority items to include: - Breakouts of digitally ordered and digitally delivered transactions (22.101). - Production and intermediate consumption of cloud computing services and digital intermediation services (22.101). - Output and value added of industries supplying ICT goods, ICT services, and other digital services (22.101). - Classification of suppliers of other digital services into: e-tailers, data- and advertising-driven platforms, non-financial digital intermediation platforms, financial service providers primarily operating digitally, and other service providers operating only digitally (22.101). - Additional content: - Summaries of uses of ICT goods and digital services, including own-account and purchased investment, and fixed capital formation enabling digital activity (including ICT equipment, software, data, data centers and networks) (22.102). - Trade flows of ICT goods and digital services, decompositions of exports and imports by major product category, international transactions digitally ordered or digitally delivered, and the share of domestic transactions that are digitally ordered (22.103). - Indicators need not be limited to products within the SNA production boundary; free services of digital platforms valued by cost of production can be reported as an addendum item (22.104). ### Digital Supply and Use Tables (SUTs) (paragraphs 22.105–22.110) - Purpose and alignment: - Digital SUTs analyze digitalization along three dimensions — nature of the transaction, products, and industries — by adding detail and aggregations to standard SUTs while aligning with existing classifications (22.105). - Supply table classifications enable decomposition of domestic and imported sources of supply; use table classifications enable decomposition of uses for intermediate consumption, final consumption, investment, and exports (22.105). - Compilation practicalities: - Conceptually, digital SUTs have cells accommodating all potentially relevant transactions, yielding many possible entries; to reduce burden, a set of high priority aggregates is identified (22.106). - Recommended high priority items include: - (a) decompositions of expenditures by the nature of the transaction; - (b) output and intermediate consumption of digital intermediation services, cloud computing services, and ICT goods and digital services; - (c) the output and gross value added of digital industries (22.106). - E‑commerce and digital ordering: - E-commerce transactions are defined by digital ordering; digital SUTs include new rows that distinguish digitally ordered transactions and decompose them into: transactions ordered directly from a counterparty, transactions ordered through a resident digital intermediation platform, and transactions ordered through a non-resident digital intermediation platform (22.107). - Digital delivery: - Digitally delivered portions of domestic industry output and imports may be shown in “of which” columns in the digital supply table; digitally delivered portions of household final consumption and exports may be shown in “of which” columns in the digital use table (22.108). - New rows and addenda: - New rows are added for ICT goods, cloud computing services, digital intermediation services, and all other digital services; an addendum on products beyond the standard production boundary may include rows for free services of digital platforms and free services of online communities of volunteers (22.109). - Including imputed values of free services in the digital SUTs will facilitate compilation of an extended account on free digital services (22.109). - New industry columns: - The output and gross value added of digital industries are shown by incorporating seven new columns in the digital SUTs: - one column for the digitally enabling industry (producers of ICT goods and services); - six columns for: 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 (22.110). ### Extended account to increase visibility of free services consumed by households (paragraphs 22.111–22.118) - Rationale: - Households spend significant time using free services (social media, search, entertainment, information), indicating high household valuation of advertiser- and data-funded platform services; estimating their value is relevant to understanding impact on household consumption growth (22.111). - In the standard sequence of accounts, these free services’ costs are implicitly included in advertising services and data sold by the platform and the platform’s own-account investment in data assets; households’ benefits are treated as a positive externality (22.111). - Extended account as a tool: - Compilation of an extended account showing the value of free services is encouraged; extended accounts are flexible for presenting concepts that extend SNA boundaries, including expanded measures of economic activity and household final consumption expenditures (22.112). - Households’ consumption of advertiser-funded platform free services can be included in an extended account as part of expanded measures of household final consumption expenditures and output, accepting analytical double counting for usefulness (22.113). - Treatment of data-licensing and barter: - Households’ consumption of unpriced services produced by platforms in exchange for providing a license to collect data on their observable phenomena (OP) can be included; licenses to collect households’ data are viewed as a payment in kind for the platform’s unpriced services, and vice versa (22.114). - Imputed values of the bartered items (unpriced platform services consumed by households and opportunities for platforms to collect households’ data) must be equal (22.115). - Four theoretical valuation approaches (22.115): - (1) the platform’s cost of producing the free services; - (2) the value that the households place on the free services; - (3) the value that the households place on the privacy of their data that they give up; - (4) the economic benefits that the platforms derive from users’ attention and access to users’ data. - Recommended valuation and consistency: - Cost of production is identified as the most suitable valuation for the extended account because it ensures consistency within the system, aligns with measurement of own-account investment in data assets, and is more practical to estimate than the other three values (22.115). - Classification of the imputed payment for licenses: - The license to access households’ OP received by the platform is not a service produced by households; the imputed payment for that license is classified as a rent (22.116). - Licensing collection of data on one’s OP is not a production activity because households’ OP are not a produced asset; access to households’ OP cannot be considered consumption of services of a produced fixed asset. If households actively assist in data collection, the assisting steps can be considered production of services, though such cases are rare (22.116). - Accounting implications of the extended account: - Expanded measures of free platforms’ output and value added will include the imputed value of free services consumed by households, while the platforms’ balance of primary incomes will remain unchanged (22.117). - The platforms’ extra value added from imputed sales to households will equal platforms’ imputed payments of rent for licenses to collect households’ data (22.117). - Household saving will be unchanged because imputed household consumption expenditures on platform services will equal households’ imputed income from licensing collection of their data (22.117). - User-generated content: - Households provide economic benefits by supplying user-generated content without monetary compensation; in the standard accounts, unpaid content is beyond the production boundary (22.118). - The extended account can treat user-generated content uploaded without payment as a barter transaction in which platform users receive unpriced platform services in exchange for content; the platform’s unpriced services are used by creators as inputs to produce content, and content is used by the platform to produce data assets (22.118). - Net effect under this alternative: an increase in the measure of platforms’ own-account investment in data assets (22.118). --- _Source: https://www.imf.org/-/media/files/data/statistics/bmp7/draft-chapters/sna-ch22-bpm7-ch16.pdf_