Analytical Studies Branch Research Paper Series
Data, Intangible Capital and Economic Growth in Canada

Acknowledgements

The authors would like to thank Tim Sargent, Rachel Soloveichik, and participants in the International Association for Research in Income and Wealth and Centre for International Governance Innovation conference entitled “The Valuation of Data,” held in Waterloo, Canada, from November 2 to 3, 2023, for helpful discussions and comments. The authors would also like to thank Carol Corrado for providing the growth accounting results with intangibles for the United States.

Abstract

This paper constructed a new measure of data assets and updated the estimates of other intangible capital for the Canadian business sector. The methodology used by Corrado et al. (2024) for estimating data assets and other intangibles for the United States and European countries is applied in this paper to compare Canada with the United States and European countries. The paper finds that about 30% of all intangibles were data assets in 2019. Intangible capital is found to make a significant contribution to labour productivity growth, especially in service-producing industries. A comparison with the labour productivity growth results for the United States from Corrado et al. (2024) shows that the relatively lower labour productivity growth in Canada for the 2000-to-2019 period was attributable to the relatively lower contribution of intangibles in data and non-data assets and lower multifactor productivity (MFP) growth in Canada during that period. The contribution of tangible capital to labour productivity growth was similar in the two countries. A comparison with the results for European countries shows that Canada and European countries had similar labour productivity growth for the 2000-to-2019 period, with similar capital deepening effects, similar MFP growth and similar labour compositional change. But the relative importance and mix of the capital deepening effect between tangibles and intangibles are different. European countries had a relatively larger contribution from intangible assets, while Canada had a relatively larger contribution from tangible assets.

1 Introduction

Intangible assets constitute a major source of capital in modern economies that is hard to quantify given the ephemeral nature of intangible investments. Ideas, stories, models, data, designs and embodied knowledge of how systems work are integral to modern production processes and economic progress. They lack physical form and market valuations, making accounting exercises designed to value and summarize these types of assets difficult. Nevertheless, a rapid increase in these assets—which coincides with the rise of digital and knowledge-based production systems—is expected to affect all aspects of the economy, including product provision; product development; the dynamics of firm competition and concentration; the growth of firms, industries and countries; and capital accumulation, productivity growth and overall economic growth.

Given the importance of intangible assets for modern economies, it is crucial to apply the best measurement practices available to provide estimates of the magnitude and influence of intangible capital. This paper adds to the information on intangible capital in Canada by updating previous estimates (Gu and Macdonald, 2020), expanding the number of data-related assets under consideration and providing—for the first time for Canada—a set of estimates for artistic originals that capitalizes outputs of creative industries.^Note Artistic originals include books, TV programs, music and movies.

The integration of artistic originals reconciles the full set of Canadian intangible capital estimates with American intangibles for the first time. The 2008 System of National Accounts (SNA) recommends including artistic originals, and this is done by the U.S. Bureau of Economic Analysis (BEA). However, estimates for Canada have not previously been presented, and this work offers a first comparison for the stock of these assets between the two countries.

Whether an item is treated as an asset in the national accounts is not an esoteric question (Corrado et al., 2022). The definition of what constitutes a capital asset affects the understanding of the growth process for an economy and thus has real implications for economic policy. A broader definition for capital assets increases the scope of capital investment in the economy and affects estimates of gross fixed capital formation (GFCF) and the capital stock. These estimates are key to understanding and sustaining economic growth. The level and growth path of gross domestic product (GDP) are affected, as is the division of GDP between capital and labour income, estimates for consumption, and estimates for saving. All of these are key concepts in the consideration of macroeconomic policies.

Appropriate definitions and measurement strategies for assets are also critical for accurately representing the ways in which capital and labour are combined to produce outputs. In industries, such as emerging industries for artificial intelligence, having assets like datasets and trained models recognized is vitally important for understanding the nature of the production process.

Currently, a suite of intangible categories is treated as GFCF in the Canadian System of Macroeconomic Accounts (CSMA). These are software, mineral exploration, and research and development (R&D). Intangible categories related to data assets, artistic originals, brand equity, firm organization, knowledge embedded in workers and financial innovation are not capitalized. While these assets are acknowledged by researchers as capital assets, the measurement and data sources for them have not yet been sufficiently developed to be included in the 2008 SNA. As a result, it is necessary to adjust existing national accounts values to include the missing intangible categories and develop satisfactory methods and data sources for estimating their capital stocks.

Significant progress has been made in this area in the last 15 years. Researchers and national statistical agencies have explored data sources and methods for estimating the value of intangibles and examined the effect of including them in the national accounts on macro aggregates such as GDP, investment and productivity. Corrado, Hulten and Sichel (2005, 2009) undertook seminal work in defining and estimating a range of intangible assets. These types of intangible capital estimates were subsequently explored in studies in other countries. Statistics Canada (2019a, 2019b) and the BEA (Rassier, Kornfeld and Strassner, 2019) further explored the addition of data as an asset, and the latter presented a framework for measuring data as an investment, along with a set of preliminary estimates. The 2025 SNA is expected to expand the types of intangible investment in GDP to include data assets.

Because intangible assets are often not sold on the market, they typically lack market valuations. In the absence of a market valuation, the current practice in national accounting is to estimate the value of the assets applying the cost of the inputs used to produce them. The paper adopts this sum of costs approach for estimating the value of intangibles to adhere as closely as possible to standard recommendations. In the process, it highlights methodological and data challenges that must be overcome for the approach to work.

The rest of the paper is organized as follows. In Section 2, the framework for integrating intangibles and data assets in the national accounts and for examining the contribution to labour productivity growth is presented. In Section 3, the data sources and methods for estimating intangible assets (including data and artistic originals) are presented. In Section 4, the estimates of intangible investment are examined. In Section 5, the contribution of intangible capital to labour productivity growth in Canada is illustrated. The estimates for Canada are also compared with those for the United States and European countries reported by Corrado et al. (2024) using the industry productivity database EUKLEMS & INTANProd (LLEE, 2023). That study included nine European countries: Denmark, Germany, Finland, France, Italy, the Netherlands, Spain, Sweden and the United Kingdom. However, it is important to note at the outset that there are methodological differences that arise from differences in the national statistical systems used as the basis for estimation.^Note Consequently, estimates of intangibles for Canada in this paper may not be entirely comparable with the estimates for the United States and European countries, and comparative analysis between Canada and other countries on intangibles should be best viewed as suggestive. Section 6 concludes and summarizes the data challenges for measuring intangible capital.

2 Intangibles as assets in the national accounts

Many intangible assets are associated with knowledge that is owned by economic agents and is non-rivalrous, but not entirely excludable. They yield economic benefits to the economic agents over time. Counting these assets, and valuing them, presents important challenges for the accounting community, given the lack of physical representation and market valuation.

The seminal work of Corrado, Hulten and Sichel (2005, 2009) categorized intangible assets into three broad categories, and these categories have been generally adopted by the international community (see, for example, Corrado et al., 2024). The three broad categories of intangible assets are digitized information, innovative property and economic competencies. Within these categories is a combination of assets currently treated as such in the 2008 SNA and assets that are still not incorporated into official recommendations (Table 1).

The first category covers digitized information, which includes software, data and tools derived from data. The expansion of software during the 1990s and the subsequent rise of data, tools and models with the digital economy of the 21st century make this one of the fastest-growing and most influential forms of intangible capital in the economy. While software is included in current CSMA estimates, only a small fraction of data assets is considered in this paper; databases, embedded in software, are included.

Data assets are created from economic activities referred to as the data value chain creation or data stack. The data value chain creation includes data collection and data storage, data transformation, and data science; most values along this value chain are derived from data science and data intelligence (Corrado et al., 2024). According to the data value chain creation, a database is more closely related to data storage. But a significant part of the value created from data collection and use—including data transformation and data science—is not capitalized. Therefore, an important aspect of the economic activities associated with expenditures for producing data assets is currently expensed rather than being capitalized. In this paper, these values are treated as capital investments, and a broader measure of the value of databases and database-derived models is presented. The 2025 SNA is expected to capitalize data assets. However, only activities related to data collection and data storage will be capitalized. Data science activities that produce new insights and intelligence are not capitalized. Statistics Canada (2019a, 2019b) experimented with this approach and estimated that the value of data assets was about 1.8% of GDP in Canada in 2018.^Note

The second category corresponds to innovative property or activities that improve the understanding of the natural world. It covers assets included in the CSMA (R&D and mineral exploration) and assets not currently included (financial innovation, architectural and engineering designs, and artistic originals). These assets are directly related to the creation of new ideas, knowledge, products or processes. They are often represented by a legal construct that gives rise to a physical asset that protects the use of the idea in Western legal systems. For example, R&D often leads to patents that protect the intangible asset, while artistic originals lead to copyrights that protect the property rights of creators.

Within this category, artistic originals merit particular attention, because they have not been previously included with intangible capital estimates for Canada. The 2008 SNA recommends that entertainment, literary and artistic originals be treated as capital outputs. Artistic originals are not capitalized in the CSMA. Artistic originals include original movies, sound recordings, manuscripts, tapes, models, etc., on which drama performances, radio and TV programming, musical performances, sporting events, literary and artistic output, etc., are recorded or embodied.^Note Such works are frequently developed on an own-account basis. Subsequently, they may be sold outright or by means of licences.

The Organisation for Economic Co-operation and Development (OECD, 2010) outlined practical guidelines on the treatment of artistic originals and other intellectual property products as assets in the 2008 SNA. For an item to be considered an entertainment, literary or artistic original, it should satisfy the following criteria: the item must be covered by copyright, the work should have primary artistic intent and the item must satisfy the capitalization criteria to be included as GFCF. That is, the 2008 SNA requires that a capital asset must be intended for use in the process of production repeatedly or continuously for more than one year. Finally, the item must not be covered elsewhere in the national accounts. According to these criteria, entertainment, literary and artistic originals should be defined to include at a minimum (1) films; (2) TV and radio stock programs that exclude radio and TV news, as well as sports events; (3) literary works that exclude magazines and newspapers; and (4) musical works.

The third category relates to economic competencies. These assets relate to brand equity, firm organizational structures, operating models, and knowledge or human capital embedded in workers. These assets are among the most difficult to measure but are also important in understanding the growth process, because they relate directly to a firm’s ability to organize itself and compete in markets. They are critical capacities for any business that will succeed over time.

2.1 Data and intangibles in expanded national accounts

In the 2008 SNA and the CSMA, three approaches (the value-added, final expenditure and income approaches) to estimating GDP are integrated. This means changing the treatment of an intangible asset to become GFCF, requiring a series of adjustments across the three approaches to measuring GDP. It also means that the saving and capital accumulation accounts and the national balance sheet must be adjusted to reflect the higher saving and GFCF that occurs, as well as the increase in national net worth from the new assets. There are three distinct cases in which adjustments are made when capitalizing intangibles.

Case 1—Intangible assets currently classified as final consumption

In cases where an intangible asset is produced and sold on markets but is classified as final consumption, the adjustment through the sequence of accounts is the most straightforward. It begins with GDP at the total economy level. An example of such an asset is software purchased by households or governments as final consumption. The value of the asset in consumption is reclassified as GFCF. This leaves total GDP unaffected, but the composition of GDP changes. Since consumption falls, saving rises, doing so by the same amount as the increase in GFCF, which maintains the equality between saving and investment identity. The value of the capital asset is added to the national balance sheet, and the value of its depreciation is added to consumption of fixed capital. There is no change to the income-based estimate of GDP or to value-added calculations.

 

Case 2—Intangible assets currently classified as intermediate inputs

When an intangible asset is classified as an intermediate input, reclassifying it as GFCF raises GDP, saving and national net worth. Examples of such assets are software and R&D purchased by firms as inputs in their production of goods and services that are ultimately sold on the markets. This is reflected in higher GFCF in the final demand GDP estimate, as well as an increase in gross operating surplus of income-based GDP estimates. Value added rises, because the value of gross output is unchanged but the value of intermediate inputs falls. The rise in value added and gross operating surplus leads to an increase in saving that is equal to the growth in GFCF. When the intangible category is capitalized, the asset side of the national balance sheet rises, as does the estimate for the consumption of fixed capital.

Case 3—Intangible assets not currently included in national accounts

When an asset is not currently included in the CSMA, the value of the estimated investment stream of the asset is added to the gross output of the industries producing it. Examples of such assets are software and R&D produced in house by firms and used as inputs in their production of goods and services (own-account software and own-account R&D). Since, by assumption, these assets are being created using current inputs but are not recognized, this change increases value added because gross output rises but intermediate inputs do not. Gross operating surplus rises in the industries with the new outputs, as does GFCF. Saving rises by the same amount as gross operating surplus and GFCF. The value of assets on the national balance sheets increases to reflect the new assets, and consumption of fixed capital rises to reflect their depreciation.

In general, treating intangibles as capital tends to induce similar changes to aggregate GDP and its composition on the income side (the labour and capital income mix) and on the expenditure side (the consumption and investment mix) regardless of the case. First, the level of GDP would be higher because own-account and purchased intangibles are added to GDP when capitalizing intangibles; the growth of output is thus expected to be higher as investment in intangibles often increases at a higher rate than consumption and investment in tangible capital. Second, the share of labour income in GDP is expected to decline while the share of capital income is expected to increase because of the expanded capital base. Third, the saving rate and investment rate (share of investment in GDP) for the economy will be higher, while the share of consumption in GDP will be lower.

2.2 Data and intangibles in industry production accounts

Capitalizing intangible assets will introduce changes to the industry production accounts that show the production of gross output from the use of capital, labour and intermediate inputs. As a result of changes to the industry production account, the capitalization of intangibles will affect the analysis of the sources of real output growth accounted for by growth in capital, labour and intermediate inputs, and multifactor productivity (MFP). It will also affect the source of growth in labour productivity accounted for by input deepening from capital, labour and intermediate inputs, and MFP growth.

In general, intangible assets can be produced in house for own use and for sale to other industries as intermediate consumption and to governments and households as final consumption. The first case corresponds to the in-house production of intangible assets for own use, while the last two cases correspond to the production of intangibles for sale.

When intangibles are produced in house and used in own production, capitalization will increase gross output, which now includes the value of intangibles. The capital input used in the production is expanded to include intangibles, while labour and intermediate inputs remain unchanged. Value-added output, which is the difference between gross output and intermediate inputs, will also increase as a result of a rise in gross output and no change to intermediate inputs.

When intangibles are produced for sale, capitalization will impact output and inputs of downstream industries that purchase the intangibles and will have no effect on industry production accounts for the industries that produce the intangibles for sale. For downstream industries, expenditures on intangibles will be recorded as investment instead of intermediate inputs. Value added will increase, while intermediate input will decline by the same amount, leaving gross output unchanged. Investment in intangibles is added to capital input, and this capital stock then returns a flow of services, while labour input does not change. For the industries that produce intangibles, there are no changes to gross output, capital, labour and intermediate inputs, except the portion of gross output is reclassified as intangible assets from goods and services used as intermediate inputs.

The analysis of the source of growth, expanded to include intangibles, starts with the output equation, where the nominal value of output is equal to the nominal value of inputs for an industry. The presentation of the growth accounting framework below will focus on the source of growth in value-added output. A presentation of the source of growth in gross output is similar.

When intangible expenditures are classified as investments, the value-added output identity is expanded to include investment in intangible capital on the output side and the flow of services from intangible capital on the income side:

$$P^Y\left(t\right)Y\left(t\right)+P^N\left(t\right)N\left(t\right)=P^L\left(t\right)L\left(t\right)+P^K\left(t\right)K\left(t\right)+P^R\left(t\right)R\left(t\right). (1)$$

Y is real value-added output in an industry (or GDP in constant dollars) before the capitalization of intangibles. N is real investment in an industry, which includes intangibles produced in house for own use and purchased intangibles. Their corresponding prices are denoted by $P^Y$ and $P^N$.

The sum of the two terms on the left-hand side of equation (1) is adjusted nominal value-added output, which now includes the value of intangibles.

Value-added output is produced using three inputs: labour input (L), tangible capital stock (K) and intangible capital stock (N). It is equal to the sum of the flow of services from the inputs on the income side. $P^L$ is the price of labour input, $P^K$ is the user cost of tangible capital and $P^N$ is the user cost of intangible capital.

The extended growth accounting equation when intangibles are capitalized as investment can be written as

$$\Delta Ln{Y}^{adj}\left(={\overline{s}}_Y\Delta \ln Y+{\overline{s}}_N\Delta \ln N\right)={\overline{s}}_L\Delta \ln L+{\overline{s}}_K\Delta \ln K+{\overline{s}}_R\Delta \ln R+\Delta \ln A, (2)$$

where $\overline{s}$ denotes the average shares of inputs and outputs in nominal GDP and $\Delta \ln$ denotes the log difference between two periods. The growth rate of the value-added output adjusted for intangibles $Y^{adj}$ on the left-hand side of the equation is equal to the share-weighted growth rates of inputs plus MFP growth. $\Delta \ln$A is the growth in MFP. This equation shows that the growth in adjusted GDP can be decomposed into contributions from labour input, contributions from tangible capital input, intangible capital input and MFP growth.^Note

The growth accounting equation can be written to examine the sources of labour productivity growth, defined as growth in output per hour worked. The sources of labour productivity growth can be expressed as

$$\Delta \ln (Y^{adj}/H)={\overline{s}}_K\Delta \ln (K/H)+{\overline{s}}_R\Delta \ln (R/H)+{\overline{s}}_L\Delta \ln (L/H)+\Delta \ln A, (3)$$

where H denotes hours worked. This equation shows that the growth in labour productivity can be broken into four components. The first two components are the contributions from capital deepening attributable to tangibles and intangibles. The third component is the contribution of shifts in labour composition (e.g., toward more educated and more experienced workers). The fourth component is MFP growth, which is often associated with technological change, organizational change or economies of scale.

When the expanded set of intangibles is capitalized, the effect of capital deepening on labour productivity growth is expected to increase. This is because the growth in intangible capital is often faster than the growth in tangible capital. Capital, expanded to include the full set of intangibles, will increase at a faster rate, and the contribution of capital deepening to labour productivity growth will therefore rise.

The estimated growth in MFP will also change. The change in estimated MFP growth depends on two effects: the increase in measured labour productivity growth from including fast-growing intangibles as investment output and the rise in the component of labour productivity growth allocated to capital deepening effects.

The rest of this section provides details on how the growth accounting framework expanded to include intangibles can be implemented. A detailed discussion of that implementation can be found in the work of Corrado, Hulten and Sichel (2005, 2009) and Baldwin, Gu and Macdonald (2012).

To implement growth accounting with intangibles, nominal investment in intangibles is first estimated. It is then deflated to obtain investment in intangibles in constant prices. Investment in intangibles in constant prices is then accumulated to arrive at an estimate for the capital stock of intangibles using the perpetual inventory method (PIM). The PIM is a model that accounts for the depreciation and decline in the efficiency of intangibles in the production process over time.

Finally, the nominal value and volume of the flow of capital services from intangibles, together with those from tangibles, are to be estimated. The nominal value of capital service from capital assets is estimated using the user cost of capital formula, where the flow of service from capital, or the user cost of capital, is equal to the sum of the real rate of return plus depreciation minus the capital gains of the asset, with adjustments for the effects of corporate taxes and capital consumption.

A main parameter for estimating the capital service of intangible capital and implementing growth accounting with intangibles is the rate of return to an asset. Two main alternatives have been used for estimating the rate of return on capital and the user cost of capital for intangible assets: endogenous rates of return that are calculated from capital income or exogenous rates of return that are chosen from observed market rates, such as a government bond rate, a corporate debt rate, or a weighted average of corporate debt and corporate equity rates. The endogenous rate of return is solved using the equation that the sum of capital costs across all capital assets is equal to total capital income. For example, Corrado, Hulten and Sichel (2009) used endogenous rates of return to calculate the user cost of intangible capital for the United States and the United Kingdom. Van Rooijen-Horsten et al. (2008) used exogenous rates of return for the Netherlands. For practical purposes, the choice of those alternative methods has little effect on estimated MFP growth at the broad sector level. However, it may influence the estimates at the detailed industry level. For this paper, the endogenous rate of return method is used to estimate the user cost of capital for intangibles and tangibles to be consistent with the estimation of capital input and MFP in the official estimates.

3 Methods and data sources for valuing data and other intangible assets

This section summarizes the methods for estimating investment in intangible capital and the data sources used to estimate intangibles. In general, there are three methods for valuing assets, including both tangible and intangible assets. When there are markets for an asset, the market price will be used. This approach is used to value most, if not all, tangible assets for machinery and equipment and buildings and structures. The second approach is the income-based approach, which estimates the value of an asset as the present discounted value of income from the use of the asset in production. But for intangible assets, there are limited market transactions, and the market price of most intangibles does not exist. In addition, the future stream of earnings from the assets is difficult to project. As a result, the third approach—the sum of costs approach—is used.

The sum of costs approach is used for estimating most types of intangible assets, such as software, R&D and mineral exploration, that are already capitalized in the CSMA. It is also used to estimate intangibles that have not been capitalized in the national accounts in most previous research. It typically starts with estimating the payroll of workers engaged in producing intangible assets. The costs are then scaled up to include non-labour costs, which are the sum of capital and intermediate input costs in the production of intangible assets when non-labour costs are available.

Here, the sum of costs approach is also used for estimating data assets not currently included in the CSMA and for estimating artistic originals. For data assets, the occupations involved in the three stages of the data value chain creation are identified. The share of time spent in producing an asset by workers in those occupations is assumed. The hours and labour compensation of workers in those occupations are estimated, to derive the labour cost component for the production of data assets. The labour costs are then scaled up to account for non-labour cost components (intermediate consumption and capital costs) and to derive the total costs of production of data assets that will be used to value investment in data assets.

The occupations identified as producing data assets vary across studies. For example, Corrado et al. (2024) took a broader approach that covers most, if not all, forms of data intelligence and data science in the generation of virtual intangibles and knowledge assets. Statistics Canada (2019a, 2019b) took a narrow view and identified occupations engaged in data science in financial and marketing activities but did not include data-driven industrial and computing engineering design as in the work of Corrado et al. (2022). The broader approach in Corrado et al. (2024) will be used in this paper to enable a comparison of the results for Canada with the results for the United States of Corrado et al. (2024).

Like most other intangible assets, there are few market transactions for artistic originals, leading to the use of the sum of costs approach for estimating artistic originals. The OECD (2010) recommends that the value of film and TV and radio program originals should be measured by the sum of costs approach. Production costs should include royalty payments made for the use of other originals in the production of films and TV and radio programs.

By contrast, the value of books and music should be measured using the income approach and by modelling royalty flows over the lifetime of the assets. The authors’ and musicians’ royalties are the income the authors and musicians derive for the assets. The value of book and music assets is estimated as the present discount value of the royalty payments. That approach has been experimented with for estimating artistic originals in the United Kingdom (Goodridge and Haskel, 2011). But the data required for implementing the income approach to estimating artistic originals are often incomplete. As a result, the sum of costs approach is adopted to estimate the value of book and music assets by Goodridge and Haskel (2011).

To implement the sum of costs approach, the production costs of artistic originals must be collected. Industry accounts and Statistics Canada’s supply and use tables have information about the total sales of artistic originals that are produced domestically by Canadian industries. The total sales values need three adjustments to arrive at investment estimates of artistic originals. First, the value of the sales includes advertising costs and other non-art costs for selling artistic originals. They must be removed as they lead either to double counting (advertising) or to including expenditures not used for producing the intangible asset. Second, total sales represent the sales of artistic originals that were created over time, and only a portion of the receipts from the sales is from the production of artistic originals in the current year. Third, TV and radio programs can be characterized as either stock or flow. Stock programs include documentary, drama, music, arts, history and education programs. Flow programs include news, sports and game show episodes. Stock programs have a longer life because they are suitable for repeat performances or replicated in different countries, and they are included as artistic original assets. Flow programs have a shorter life and are unlikely to be repeated, and they should not be included as artistic original assets.

To adjust sales, the ratio of investment to sales from Soloveichik and Wasshausen (2013) is used. They estimated the ratio for the United States using the industry data. The ratios are 0.4, 0.5, 0.1 and 0.5 for books, music, TV programs and movies, respectively. The low ratio of investment to sales for TV programs reflects the fact that flow programs such as news and sports programs are not counted as investment assets.

4 Investment in data assets and other intangible assets

The time series estimates of intangible investment by asset type at the M level of industry aggregation (North American Industry Classification System [NAICS] two- to three-digit industries) have been developed for the total economy for the 2000-to-2019 period. The discussion will focus mainly on the trends in the business sector.

4.1 Nominal investment

Tables 3 and 4 present nominal investment in tangible and intangible assets in the business sector for the 2000-to-2019 period. Table 3 divides assets into two main categories: tangible (which is further disaggregated into information and communications technology [ICT] and non-ICT tangible assets) and intangible assets from 2000 to 2019. Table 4 presents the estimates by detailed intangible asset category for selected years.

Intangible investment was lower than investment in tangibles for the 2000-to-2019 period, but intangible investment rose faster than tangible investment (Table 3). As a result, the ratio of nominal investment in intangibles to tangibles increased from 0.73 in 2000 to 0.83 in 2019.

In 2019, nominal investment in intangibles was valued at about $213 billion, and nominal investment in tangible assets was valued at about $256 billion.

The largest component of intangibles was data assets, at about $57 billion in 2019 ($54 billion for own-account data assets and $3 billion for purchased data assets). The size of data assets relative to other intangible asset categories underscores the importance of data and data-derived tools for modern production processes.

The industries with the largest investment in data assets include professional, scientific and technical services; finance and insurance; and information and cultural industries (Appendix Table A.1). Professional, scientific and technical services accounted for 41.5% of total investment in data assets in the business sector in 2019. Finance and insurance accounted for 13.9% of total investment in data assets, and information and cultural industries accounted for 7.4% of investment in data assets in that year.

Investment in artistic originals, which has not previously been capitalized in the CSMA, is also a significant asset.^Note Investment in artistic originals was about $6 billion in 2019 (Table 4). The largest component of artistic originals is movie assets ($4.5 billion), followed by TV and radio programs ($0.9 billion), books ($0.6 billion), and music assets ($0.1 billion).

The intangible categories of R&D, software investment and mineral exploration are currently capitalized in the CSMA. The investment in these intangibles is only a small fraction of the total intangible investment measured in this paper. In 2019, the intangibles that have been capitalized in the CSMA accounted for 20% of total intangible investment (Table 4).

4.2 Nominal investment as share of gross domestic product

Chart 1 presents nominal investment in tangible and intangible assets as the share of GDP in the Canadian business sector. The ratio of tangible investment in nominal GDP was volatile from 2000 to 2019 and averaged about 16% over the period, while the ratio of intangibles to GDP increased steadily from 11.7% in 2000 to 13.5% in 2019.

Data table for Chart 1

Data table for Chart 1 Table summary
The information is grouped by Year (appearing as row headers), All tangibles and All intangibles, calculated using percent units of measure (appearing as column headers).

Year	All tangibles	All intangibles
	percent
Source: Statistics Canada, author's tabulations.
2000	16.2	11.7
2001	15.9	12.2
2002	15.3	11.9
2003	15.0	11.8
2004	15.4	11.8
2005	16.2	12.1
2006	17.3	12.1
2007	17.0	12.4
2008	17.2	12.3
2009	15.6	12.9
2010	16.3	12.8
2011	17.1	12.5
2012	18.4	12.5
2013	18.8	12.4
2014	19.2	12.6
2015	18.2	12.9
2016	16.5	13.0
2017	15.9	12.8
2018	16.0	12.9
2019	16.2	13.5

Previous studies in Canada have shown that investment in Canada was weak after the financial crisis in 2007 and 2008, and especially after 2014, with the collapse of commodity prices compared with the 1990s and the early 2000s (Gu, 2024). Chart 1 shows that while the share of investment in tangibles in value added declined after 2014, the share of intangibles continued to increase after 2014. However, the rise in the intangible share did not fully offset the decline in the tangible share. The sum of intangible and tangible shares declined from 32% in 2014 to 30% in 2019.

Table 5 shows the share of intangible assets in GDP by detailed asset type. The share of intangibles in GDP increased for almost all types of intangibles. The biggest increase was for software, data assets, and new architecture and engineering design. For example, the share of investment in data assets in GDP increased from 2.6% in 2000 to 3.6% in 2019, also shown in Chart 2.

Data table for Chart 2

Data table for Chart 2 Table summary
The information is grouped by Year (appearing as row headers), , calculated using (appearing as column headers).

Year	Percent
Source: Statistics Canada, author's tabulations.
2000	2.6
2001	2.7
2002	2.7
2003	2.7
2004	2.6
2005	2.8
2006	2.7
2007	2.9
2008	2.8
2009	3.2
2010	3.1
2011	3.1
2012	3.1
2013	3.1
2014	3.3
2015	3.4
2016	3.5
2017	3.3
2018	3.3
2019	3.6

In terms of artistic originals, the ratio of investment to GDP increased for movie assets, declined for books and music assets, and had no change for TV and radio program assets (Chart 3). The share of artistic originals in GDP declined from 0.37% in 2000 to 0.30% in 2014 and then increased after 2014 to about 0.39% in 2019.

Data table for Chart 3

Data table for Chart 3 Table summary
The information is grouped by Year (appearing as row headers), Books, Music assets, TV programs, Movies and All artistic originals, calculated using percent units of measure (appearing as column headers).

Year	Books	Music assets	TV programs	Movies	All artistic originals
	percent
Source: Statistics Canada, author's tabulations.
2000	0.11	0.03	0.06	0.17	0.37
2001	0.12	0.03	0.06	0.16	0.37
2002	0.13	0.03	0.06	0.16	0.38
2003	0.12	0.03	0.06	0.16	0.37
2004	0.12	0.03	0.06	0.16	0.37
2005	0.10	0.02	0.06	0.17	0.35
2006	0.09	0.02	0.06	0.17	0.35
2007	0.09	0.02	0.06	0.17	0.34
2008	0.08	0.02	0.06	0.15	0.32
2009	0.10	0.02	0.07	0.18	0.36
2010	0.09	0.02	0.07	0.17	0.35
2011	0.07	0.02	0.07	0.17	0.33
2012	0.07	0.02	0.07	0.17	0.32
2013	0.06	0.01	0.07	0.17	0.31
2014	0.04	0.01	0.07	0.18	0.30
2015	0.04	0.01	0.07	0.22	0.34
2016	0.04	0.02	0.07	0.24	0.37
2017	0.04	0.01	0.06	0.27	0.38
2018	0.04	0.01	0.06	0.27	0.38
2019	0.04	0.01	0.06	0.29	0.39

The share of artistic originals in GDP in Canada was lower than the estimate for the United States. The share of artistic originals in nominal GDP for the total economy was about 0.23% in Canada in 2007, while that share was 0.35% in the United States (Soloveichik, 2011).

Chart 4 presents the average annual growth in real investment in tangible and intangible assets by asset type for the 2000-to-2019 period. Real investment increased for all asset types except for mineral exploration, books and music assets. The largest increases in real investment were for software, data assets and organizational capital. Movie assets also had a large increase over this period.

Data table for Chart 4

Data table for Chart 4 Table summary
This table displays the results of Data table for Chart 4 , calculated using (appearing as column headers).

	Percent per year
Note: ICT = information and communications technology. Source: Statistics Canada, author's tabulations.
ICT tangibles	3.9
Non-ICT tangibles	1.8
Research and development	0.8
Own-account research and development (except software development)	0.5
General-purpose software	7.0
Custom software design and development	4.5
Own-account software design and development	4.8
Mineral and oil and gas exploration	-4.2
Development costs in financial industry	1.2
New architecture and engineering design	3.4
Own-account data	3.5
Purchased data	4.6
Advertising	0.7
Firm-specific human capital	1.9
Purchased organizational capital	4.0
Own-account organizational capital	1.2
Books	-4.1
Music assets	-5.4
TV programs	2.1
Movies	4.5

5 Data, intangible capital and labour productivity growth

Intangibles can contribute to growth in labour productivity through the capital deepening effect. They can also contribute to growth in labour productivity through spillover effects and complementary changes in organizations and business processes, which are captured by the effect of intangibles on MFP growth. In this section, the growth accounting framework is used to decompose labour productivity growth into contributions from the capital deepening effect of tangibles and intangibles for the total business sector and for detailed industries in the business sector. Regression analysis is then used to examine the relationship between intangibles and MFP growth across industries. The hypothesis is that MFP growth will be positively correlated with investment in intangibles across industries if there are spillover effects from intangibles such as data assets. Corrado et al. (2024) conjecture that the spillover effects from data assets are limited, as the data assets are appropriable and excludable (if not entirely).

5.1 The contribution of intangibles to labour productivity growth in the business sector

Chart 5 presents the decomposition results when additional intangibles are capitalized. For comparison, the decomposition results using the national accounts data that include only a small part of intangibles (R&D, software and mineral exploration) as assets are also provided.

Data table for Chart 5

Data table for Chart 5 Table summary
This table displays the results of Data table for Chart 5 SNA and SNA with additional intangibles, calculated using percent per year units of measure (appearing as column headers).

	SNA	SNA with additional intangibles
	percent per year
Note: SNA = System of National Accounts; MFP = multifactor productivity. Source: Statistics Canada, author's tabulations.
Labour productivity growth	0.96	1.00
Capital deepening	0.79	0.81
Labour composition	0.26	0.23
MFP growth	-0.09	-0.05

Consistent with previous studies for Canada, the capitalization of additional intangibles increased the growth of real GDP and labour productivity in the Canadian business sector by about 0.04% per year from 2000 to 2019. This occurs because real intangible investment rose faster than real GDP during this period. The capitalization of additional intangibles increased the level of nominal GDP by 8% to 9% for the 2000-to-2019 period (not shown in the chart).

The capitalization of intangibles is also found to increase the estimated contribution of capital deepening to labour productivity growth. The contribution of capital deepening to labour productivity growth increased from 0.79 percentage points to 0.81 percentage points per year from 2000 to 2019. The contribution of labour composition to labour productivity growth declined slightly from 0.26 percentage points to 0.23 percentage points during this period. This is a result of a decline in the estimated share of labour income in the expanded output base used to weight the labour compositional changes to derive the productivity contribution of labour composition. Finally, the estimated MFP growth declined slightly over the 2000-to-2019 period when all intangible capital is included.

Table 6 presents the results from the decomposition of labour productivity growth in the Canadian business sector for the 2000-to-2019 period and two subperiods (2000 to 2015 and 2015 to 2019) where the contribution of capital deepening is presented separately for each asset type. A summary of the decomposition of labour productivity growth is presented in Chart 6.

Data table for Chart 6

Data table for Chart 6 Table summary
This table displays the results of Data table for Chart 6 2000 to 2015 and 2015 to 2019, calculated using percent per year units of measure (appearing as column headers).

	2000 to 2015	2015 to 2019
	percent per year
Note: MFP = multifactor productivity. Source: Statistics Canada, author's tabulations.
Labour productivity growth	1.03	0.91
Tangible capital deepening	0.69	0.22
Intangible capital deepening	0.25	0.14
Labour composition	0.24	0.21
MFP growth	-0.15	0.34

Intangible capital is found to make a significant contribution to labour productivity growth, accounting for about one-quarter of the total capital deepening effects in the 2000-to-2019 period. Of all the categories of intangible capital, data assets contributed the most to the increase in labour productivity growth.

Previous studies have shown that investment in Canada was weak after the financial crisis of 2007 and 2008, and especially after 2015, with the collapse of commodity prices that started in 2014 (Gu, 2024). The capital deepening effects from investment in tangible capital and from investment in the intangible capital that is capitalized in the national accounts (which represents 20% to 30% of all intangibles estimated in this paper) declined significantly after 2015. Chart 6 and Table 6 show that much of the decline in the capital deepening effect was attributable to the decline in the capital deepening effect from tangible assets after 2015. The decrease in intangible capital deepening was small after 2015. The capital deepening effect from tangibles declined from 0.64 percentage points in the 2000-to-2015 period to 0.21 percentage points in the 2015-to-2019 period—down 0.43 percentage points. By contrast, the capital deepening effect from intangibles declined from 0.22 percentage points to 0.14 percentage points between those two periods—down 0.08 percentage points. The intangible capital deepening effect increased for software and data assets after 2015 (Table 6).

While not entirely comparable, the contribution of capital to labour productivity growth for Canada can be compared with that of the United States and European countries from Corrado et al. (2024). These comparisons are presented in Chart 7.

Data table for Chart 7

Data table for Chart 7 Table summary
This table displays the results of Data table for Chart 7 Canada, Unites States and Europe, calculated using percent per year units of measure (appearing as column headers).

	Canada	Unites States	Europe
	percent per year
Note: MFP = multifactor productivity. Source: Statistics Canada, author's tabulations and and LLEE (2023).
Labour productivity growth	1.00	1.90	1.05
Tangible capital deepening	0.59	0.50	0.42
Intangible capital deepening—non-data	0.17	0.42	0.25
Intangible capital deepening—data	0.06	0.35	0.22
Labour composition	0.23	0.27	0.26
MFP growth	-0.05	0.35	-0.10

For the United States, the contribution of tangible capital to labour productivity growth was 0.59 percentage points per year from 2000 to 2019, while the contribution in Canada was 0.50 percentage points per year during the same period. The contribution of intangibles to labour productivity growth in the United States was about 0.77 percentage points during that period, while intangibles contributed about 0.22 percentage points in Canada. MFP growth in the United States was 0.35% per year during that period, while MFP growth was -0.10% per year in Canada. The effect of labour compositional changes was similar in the two countries. These findings suggest that capital deepening for physical assets progressed at roughly the same rate for both countries. However, the 0.55 percentage point difference for intangibles suggests that the United States developed significantly more intangible capital over the 2000-to-2019 period than Canada. Overall, the relatively lower labour productivity growth in Canada after 2000 was attributable to the relatively lower contribution of intangibles and lower MFP growth in Canada. The relatively lower contribution of intangibles in Canada lies in both data and non-data assets. The contribution of tangible capital to labour productivity growth was similar in the two countries.

Canada and nine European countries had similar labour productivity growth in the 2000-to-2019 period. The nine European countries are Denmark, Germany, Finland, France, Italy, the Netherlands, Spain, Sweden and the United Kingdom. During this period, Canada and these nine European countries had similar productivity growth with similar capital deepening effects, similar MFP growth and similar labour compositional change. But the relative importance and mix of capital deepening between tangibles and intangibles were different. European countries had a relatively larger contribution from intangible assets, while Canada had a relatively larger contribution from tangible assets.

5.2 The contribution of intangibles to labour productivity growth at the industry level

Table 7 presents the growth accounting results for the sources of labour productivity growth by industry for the period from 2000 to 2019.

The contribution of intangible capital to labour productivity growth was much higher in the service-producing industries than in the goods-producing industries. For the 2000-to-2019 period, intangible capital contributed 0.21 percentage points per year to the growth in labour productivity in average service-producing industries. Its contribution to the growth in labour productivity was 0.06 percentage points per year in non-manufacturing goods-producing industries and -0.04 percentage points per year for manufacturing industries.

For service-producing industries, the contribution of intangibles was as important as the contribution of tangibles. By contrast, for goods-producing industries, the predominant form of capital was investment in tangible assets, such as machinery and equipment and building structures. The productivity contribution of intangibles was small for goods-producing industries.

The largest contribution of intangibles to labour productivity growth was in service-producing industries such as wholesale trade; information and cultural industries; and professional, scientific and technical services. Intangible capital also made large contributions in several goods-producing industries, such as construction; apparel manufacturing and leather and allied product manufacturing; and electrical equipment, appliance and component manufacturing.

5.3 The relationship between intangibles and multifactor productivity growth

While intangibles are found to contribute to labour productivity growth through capital deepening, they may also contribute through their spillover effect on MFP growth. For the ICT changes and rapid labour productivity growth of the 1990s and early 2000s in Canada and the United States, the acceleration of labour productivity growth during this period is related to both the capital deepening effect of ICT capital and the spillover effect of ICT capital on MFP growth. The industries with high ICT capital intensity experienced high MFP growth in both countries (see Stiroh [2002] for the United States and Gu and Wang [2004] for Canada).

To examine the effect of intangibles and data assets on MFP growth, the regression equation that relates MFP growth to the average intangible investment/GDP ratio or the average data asset investment/GDP ratio is estimated using industry data for the 2000-to-2019 period at the two- to three-digit level of NAICS industry aggregation for a total of 40 industries in the business sector. The coefficient estimates on intangible intensity and data intensity are not statistically significant, with robust t statistics around 1 for both equations. This suggests that intangibles and data assets are not related to MFP growth across industries for the 2000-to-2019 period.

There are two potential explanations for the lack of correlation between intangibles and data assets and MFP growth. The first explanation relates to the hypothesis that intangibles and data assets can be largely appropriated by firms that invest in these assets. Unlike ICT capital, there are limited spillover effects from data and other intangible capital. The second explanation relates to the lag hypothesis of intangibles and data assets. Intangibles and data assets are expected to have an effect on MFP growth when complementary changes related to those in organization and business practices are introduced. The effect of intangibles on MFP growth is yet to come.

6 Conclusion

This paper constructed a new measure of data assets and artistic originals and updated the estimates of other intangible capital in Canada to provide a more comprehensive measure of investment and to examine the contribution of these intangibles to labour productivity growth in Canadian business sector industries. The results were then compared with the results for the United States and European countries.

The largest component of intangibles was data assets, which reached about $57 billion in 2019 ($54 billion for own-account data assets and $3 billion for purchased data assets). The share of investment in data assets in GDP increased from 2.6% in 2000 to 3.6% in 2019.

Investment in artistic originals, which has yet to be capitalized in the Canadian national accounts, is also a significant asset. Investment in artistic originals reached about $6 billion in 2019. The largest component of artistic originals was movie assets ($4.5 billion), followed by TV and radio programs ($0.9 billion), books ($0.6 billion), and music assets ($0.1 billion).

The capitalization of intangibles was found to affect growth in GDP and labour productivity growth, as well as affect the source of growth in labour productivity. When intangibles and data assets are not included in the national accounts but are included as investment, GDP growth and labour productivity growth increased slightly from 2000 to 2019 (+0.04% per year).

The effect of capital deepening on labour productivity growth for the 2000-to-2019 period increased from 0.79% per year to 0.81% per year when those intangibles and data were included.

Overall, intangible capital made a significant contribution to labour productivity growth, especially in service-producing industries. Intangible capital accounted for about one-quarter of the total capital deepening effect in the 2000-to-2019 period. Of all the categories of intangible capital, data assets contributed the most to labour productivity growth.

Much of the decline in the capital deepening effect in Canada after 2015 was attributable to the decrease in the capital deepening effect from tangible assets. The decline in intangible capital deepening was small after 2015.

The largest contribution of intangibles to labour productivity growth was in service-producing industries such as wholesale trade; information and cultural industries; and professional, scientific and technical services. Intangible capital also made large contributions in several goods-producing industries, such as construction; apparel manufacturing and leather and allied product manufacturing; and electrical equipment, appliance and component manufacturing. Overall, the contribution of intangible capital to labour productivity growth was much higher in service-producing industries than in goods-producing industries.

Data assets and intangible intensities are not correlated with MFP growth across industries for the 2000-to-2019 period. This lack of cross-industry correlation between intangibles and MFP growth may suggest that intangibles and data assets can be largely appropriated by businesses (Corrado et al., 2024). Unlike ICT capital, there are limited spillover effects from data and other intangible capital. Alternatively, this lack of correlation between intangibles and MFP growth may suggest that intangibles and data assets are expected to have an effect on MFP growth only when complementary changes related to those in organization and business practices are introduced. The effect of intangibles on MFP growth is yet to come.

A comparison with results for the United States shows that the contribution of tangible capital to labour productivity growth was similar in Canada and the United States. However, the contribution of intangible capital in both data and non-data assets was much lower in Canada. The lower labour productivity growth in Canada compared with the United States after 2000 was attributable to the lower contribution of intangibles in both data and non-data assets to labour productivity growth and lower MFP growth in Canada.

A comparison with the results for nine European countries shows that Canada and these countries had similar labour productivity growth for the 2000-to-2019 period, with similar capital deepening effects, similar MFP growth and similar labour compositional change. But the relative importance and mix of capital deepening between tangibles and intangibles were different. These European countries had a relatively larger contribution from intangible assets, while Canada had a relatively larger contribution from tangible assets.

The results in this paper provide empirical evidence in support of the longstanding concern that Canada is not taking advantage of opportunities to increase output and labour productivity through investment, especially in knowledge intangibles, and innovation in products and processes (OECD, 2023).

Appendix

References

Baldwin, J. R., W. Gu, and R. Macdonald (2012), “Intangible Capital and Productivity Growth in Canada,” The Canadian Productivity Review, Catalogue no. 15-206-X — No. 029, Statistics Canada.

Corrado, C., J. Haskel, M. Iommi, C. Jona-Lasinio, and F. Bontadini (2024), “Data, Intangible Capital, and Productivity,” revised paper presented to NBER/CRIW Conference on Technology, Productivity and Economic Growth, 2022.

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

Corrado, C., C. Hulten, and D. Sichel (2005), “Measuring Capital and Technology: An Expanded Framework,” in C. Corrado, J. Haltiwanger, and D. Sichel (eds), Measuring Capital in the New Economy, Studies in Income and Wealth No. 65, University of Chicago Press, Chicago, IL, 11–46.

Corrado, C., C. Hulten, and D. Sichel (2009), “Intangible Capital and U.S. Economic Growth,” Review of Income and Wealth, 55, 661–85.

Goodridge, P. and J. Haskel (2011), “Film, Television & Radio, Books, Music and Art: UK Investment in Artistic Originals” Available at SSRN: https://ssrn.com/abstract=2707171 or http://dx.doi.org/10.2139/ssrn.2707171

Gu, W. (2024), “Investment Slowdown in Canada After the Mid-2000s: The Role of Competition and Intangibles,” Analytical Studies Branch Research Paper Series.

Gu, W. and R. Macdonald (2020), “Business Sector Intangible Capital and Sources of Labour Productivity Growth in Canada,” Analytical Studies Branch Research Paper Series, no. 442, Statistics Canada.

Gu, W. and W. Wang (2004), “Information Technology and Productivity Growth,” in Economic Growth in Canada and the United States in the Information Age, edited by Dale W. Jorgenson. Industry Canada monograph, Ottawa, Canada.

Jackson, C. (2001), “Capitalization of Software in the National Accounts,” Income and Expenditure Accounts Technical Series, Statistics Canada.

LLEE (Luiss Lab of European Economics) (2023), “EUKLEMS & INTANProd database.” LUISS.

OECD (Organisation for Economic Co-operation and Development) (2023), OECD Economic Surveys: Canada 2023, OECD Publishing, Paris, https://doi.org/10.1787/7eb16f83-en.

OECD (Organisation for Economic Co-operation and Development) (2010), Handbook on Deriving Capital Measures of Intellectual Property Products, Paris, OECD.

Rassier, D. G., R. J. Kornfeld, and E. H. Strassner (2019), “Treatment of Data in National Accounts,” Technical report, Paper prepared for the U.S. Bureau of Economic Analysis Advisory Committee.

Soloveichik, R. C. (2011), “Research spotlight: Artistic originals as capital assets,” Survey of Current Business, June.

Soloveichik, R. C. and D. Wasshausen (2013), “Copyright-Protected Assets in the National Accounts,” Bureau of Economic Analysis.

Statistics Canada (2019a), “Measuring Investment in Data, Databases and Data Science: Conceptual Framework,” Technical report, Statistics Canada.

Statistics Canada (2019b), “The Value of Data in Canada: Experimental Estimates,” Technical report, Statistics Canada.

Stiroh, K. J. (2002), “Information Technology and the U.S. Productivity Revival: What Do the Industry Data Say?” American Economic Review 92, 5: 1559-1576.

Van Rooijen-Horsten, M., D. van den Bergen, M. de Haan, A. Klinkers, and M. Tanriseven (2008), Intangible Capital in the Netherlands: Measurement and Contribution to Economic Growth, Discussion Paper No. 08016, Statistics Netherlands, The Hague.