Main article

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

1. Introduction
2. Background
3. The aggregate Social Accounting Matrix for Canada (Tables 1 and 2)
4. A disaggregate Social Accounting Matrix
5. Three Social Accounting Matrix applications
6. First application: Expanding the household sector of the Social Accounting Matrix
7. Integration of the Survey of Household Spending and System of National Accounts household accounts
8. A transaction-based integrated household account
9. Second application: Breakdown of employee compensation by social attributes
10. Third application: Expanding the government account (taxes on products by type and by origin)
11. Conclusion
12. Appendix

1   Introduction

A Social Accounting Matrix (SAM) is a framework for analytical presentation of economic data jointly with other relevant data—such as data on social conditions or the environment—as an integrated whole. Such presentations go beyond what is available today from published national accounting statistics. Being an extension of existing national accounts, SAMs empower users of the accounts to more easily analyze socioeconomic questions, to supply the information needed for policy development, and to build general equilibrium models.

The System of National Accounts 1993 (SNA 1993) describes SAM in its broadest form, namely as a means of presenting national accounting data in the form of a matrix:

"A SAM is defined here as the presentation of SNA accounts in a matrix which elaborates the linkages between a supply and use table and institutional sector accounts". ¹  (SNA 1993, paragraph 20.4)

A SAM depicts the entire circular flow of income for an economy in a (square) matrix format. It shows production leading to the generation of incomes, which, in turn, are allocated to institutional sectors. ²  In addition, it presents the redistribution of income leading to disposable income of institutional sectors. These incomes are either spent on products or saved. Expenditures by institutional sectors lead to production by domestic industries as well as supply from imports.

One decisive advantage of SAMs is their flexibility, permitting meso-level classification of social and economic statistics in a way that suits varied analytical objectives. For instance, the cells in Table 5, which show the compensation of employees (intersection of row 3a and columns 2a to 2c) in a macro SAM, can be expanded into an applied SAM to reveal the breakdown of this total by industry and by age group; this allows for an analysis of how different industries serve as income sources for different generations of workers. SNA 1993 describes such an applied SAM as follows:

"In many instances SAMs have been applied to an analysis of interrelationships between structural features of an economy and the distribution of income and expenditure among household groups. Evidently, SAMs are closely related to national accounts whereby their typical focus on the role of people in the economy may be reflected by, among other things, extra breakdowns of the household sector and a disaggregated representation of labour markets (i.e., distinguishing various categories of employed persons)." (SNA 1993, paragraph 20.4)

A key advantage of a SAM over the existing supply and use tables lies in the fact that it can be applied to a disaggregated household sector for the purpose of modeling the impact of exogenous changes on the system. Unlike a closed input-output model that uses a simple household sector, a SAM is capable of modeling inter-sectoral impacts by incorporating a complex household sector (e.g., households with different income levels and different induced expenditure patterns). Roland-Holst (1990, p. 125) suggests that inter-industry analysis that omits these considerations "can be seriously misleading."

This paper presents two versions of SAM for Canada as described in the SNA 1993: an aggregate SAM and a disaggregate SAM. The latter is a compact disaggregation of SAM designed to illustrate SAM's potential to expand along any dimension. In addition, the paper presents three applications of SAM. The aggregate and disaggregate SAM use mainly data from the integrated economic accounts of the Canadian System of National Accounts (CSNA) recently completed at Statistics Canada (see Siddiqi [2004]).

The first of the three applications presented here shows the income, expenditures, and savings of the household sector by income quintile and other household attributes for the year 2000. This involves an integration of macro-level national accounts data by household sector with those of Statistics Canada's Survey of Household Spending (SHS). The second application expands aggregate employment income into industries and attributes, ³  such as gender, age, educational attainment, and hours worked. It permits a range of analyses, such as how average hourly earnings vary according to age, gender, and education. The third application is constructed around indirect taxes, showing how the aggregate amount of commodity taxes varies by type of tax (and by jurisdiction), and provides the tax base to which tax rates are applied. These applications of SAM are built through integration of SNA data sources (e.g., as input-output tables) and other Statistics Canada sources (such as the SHS, the Survey of Employment, Payrolls and Hours, and the Labour Force Survey) as well as external data sources, such as income tax data from the Canada Revenue Agency (CRA).

2   Background

Historically, SAMs have been produced by developing countries more than for developed countries. Given the greater prevalence of activism and centrally driven economic development policies in developing countries, it is not surprising to find that many developing countries have used SAMs for economic planning (see Pyatt and Round 1977). Resosudarmo and Thorbecke (1996) discuss the analysis of the impact of environmental pollution abatement policies on household incomes for different socioeconomic classes in Indonesia. More recent years have seen a resurgence of interest in applications of SAM to socioeconomic analysis. ⁴  In 1996, the U.K. Office of National Statistics developed a pilot SAM for the United Kingdom for the year 1993 (Stuttard and Frogner 2003a). Reinert and Roland-Holst (1992) built a detailed SAM on trade policy analysis for the United States with respect to reference year 1988; they suggest that SAMs are rapidly becoming the standard data construct for general equilibrium models of trade policy. SAMs are now annually compiled by Statistics Netherlands for the domestic economy, and the Italian and British statistical offices are working on regular production of SAMs (Timmerman and Vande Ven 1994). ⁵  In addition, regional SAMs have been built, in both developing and developed countries, for purposes of interregional and regional economic analysis (Thorbecke 1998, p. 317).

3   The aggregate Social Accounting Matrix for Canada (Tables 1 and 2)

In a SAM, rows record receipts (incomings) by origin, and columns record outlays (outgoings) by destination. Total receipts (row sums) equal total outlays (column sums). Accordingly, each account in a SAM is represented by a row and column pair, identically named. The SAM framework of the SNA 1993 distinguishes eleven accounts; these are listed in Tables 1 and 2 and Tables 5, 6 and 7, along both rows and columns. Tables 3 and 4 present the same information as that set out in Tables 1 and 2 while separating the government tax account.

In this section of the paper, the authors describe the aggregate SAM for Canada with respect to reference year 2000. The aggregate SAM provides coherent economic aggregates without sector or institutional detail. Each entry is in fact the grand total of a sub-matrix. For example, the entry described as "Output" at the intersection of the "Production (industries)" row and the "Goods and services (Products)" column of Table 1 is the sum total of all goods and services domestically produced in Canada for reference year 2000 contained in the make matrix of the Canadian input-output tables. That matrix sets out domestic production for 727 goods and services produced by 300 industries.

The entry "Intermediate consumption," at the intersection of the "Goods and services (Products)" row and "Production (industries)" column of Table 1, represents the sum total of the intermediate use matrix of the Canadian input-output tables. The entry "Gross value added" represents primary inputs cross-classified by industry.

Imports of goods and services are shown at the intersection of the "Rest of the word – current" row and the "Goods and Services (products)" column of Table 1. In the input-output tables, imports are potentially classified by 727 commodities. The import duties included in the value of imports in the input-output tables are removed here and shown with commodity taxes, at the intersection of the "Allocation of primary income" row and "Goods and services (products)" column of Table 1.

The "Goods and services (products)" row, shows the use of goods and services at purchasers' prices ($2,479 billion). The breakdown of this figure in this row is the following: intermediate consumption (in the "Production [industries]" column of Table 1); final consumption expenditure (in the "Use of income [institutional sectors]" column of Table 1); changes in inventories (in the "Capital [institutional sectors]" column of Table 2); gross fixed capital formation (in the "Fixed capital formation [industries]" column of Table 2); and exports (in the "Rest of the word – current" column of Table 2). Column 1 of Table 1 shows how the supply of these goods and services is made up of domestic production (in the "Production [industries]" row), imports (in the "Rest of the world – current" row), and taxes on products (in the "Allocation of primary income [institutional sectors]" row).

In Table 1, the "Production (industries)" row presents the total domestic production of goods and services ($1,965 billion) (in the "Goods and services [products]" column), while the "Production (industries)" column shows the breakdown of this figure into intermediate consumption (in the "Goods and services [products]" row) and gross value added (in the "Generation of income [value added categories]" row). The accounts represented by these two rows and columns are aggregate versions of the supply and use (input-output) tables of the CSNA, which are linked with the other accounts of the system.

The "Generation of income (value added categories)" row shows the categories of value added (in the "Production [industries]" column in Table 1). The corresponding column shows the payment of these incomes to institutional sectors.

The "Allocation of primary income (institutional sectors)" row records how primary incomes are allocated to institutional sectors: gross generated incomes (in the "Generation of income [value added categories]" column), taxes on products (in the "Goods and services [products]" column), property income received from other sectors (in the "Allocation of primary income [institutional sectors]" column), ⁶  all in Table 1, and property income received from the rest of the world (in the "Rest of the world – current" column of Table 2). The "Allocation of primary income (institutional sectors)" column shows the property income paid to other sectors and to the rest of the world. The balancing item of this account (at the intersection of the "Secondary distribution of income [institutional sectors]" row and the "Allocation of primary income [institutional sectors]" column in Table 1) is national income.

The "Secondary distribution of income (institutional sectors)" account shows the relationship between the national income and disposable income. In the row are recorded the national income as well as inter-sectoral transfers, which include current transfers to and from the rest of the world. The balancing item (at the intersection of the "Use of income [institutional sectors]" row and the "Secondary distribution of income (institutional sectors)" column of Table 1), which equates the column and row totals, is disposable income.

The "Use of income (institutional sectors)" row and column in Table 1 describe, the use of disposable income. The column shows the amount of income used as final consumption expenditure (in the "Goods and services [products]" row) and gross saving (in the "Capital [institutional sectors]" row). This saving is carried forward into the capital account.

The "Capital (institutional sectors)" row ⁷  shows the availability of funds coming from saving (in the "Use of income [institutional sectors]" column in Table 1), borrowing (in the "Financial [financial assets]" column in Table 2), inter-sectoral capital transfers (in the "Capital [institutional sectors]" column in Table 2), and capital transfers from the rest of the world (in the "Rest of the world – capital" column). The "Capital (institutional sectors)" column (in Table 2) records the allocation of these funds, namely, changes in inventories (in the "Goods and services [products]" row), inter-sectoral capital transfers (in the "Capital [institutional sectors]" row), gross fixed capital formation (in the "Fixed capital formation [industries]" row), lending (in the "Financial [financial assets]" row), and transfers payable to the rest of the world (in the "Rest of world – capital" row ). The balancing item—the net lending of the nation—can also be derived from the difference between borrowing and lending.

The "Fixed capital formation (industries)" row and column (Table 2) show gross capital formation ($207 billion). This account can be expanded to show the composition of investment by sectors (household, business, and government) cross-classified by industry and type of capital good, etc.

The "Financial (financial assets)" row and column (Table 2) summarize the financial account, showing lending in the row and borrowing in the column. The balancing item is shown in the row because it equals net lending to the rest of the world.

The current transactions and capital transactions for the rest of the world are shown in the "Rest of the world – current" and "Rest of the world – capital" rows and columns, respectively (Table 2). The balancing item of the current account is viewed from the perspective of the rest of the world.

4   A disaggregate Social Accounting Matrix

As mentioned earlier, a SAM can be estimated and presented for any level and type of aggregation that analysis demands, provided that data sources permit. Following the SNA 1993, the authors present a disaggregate version of the SAM in Tables 5, 6 and 7. Since a fully disaggregate SAM would have unmanageable dimensions, they present a compact disaggregation consisting of three industry/commodity groups. This version of SAM articulates a sub-account for each account presented in Tables 1 and 2. As with the SNA 1993, the first number of each sub-account refers to the summary Tables 1 and 2, while the second label refers to the details shown for that sub-account.

The domestic output shown at the intersection of the "Production (industries)" row and "Goods and services (products)" column in Table 1 is disaggregated in Table 5 by industry and commodity (rows coded 2a to 2c, columns coded 1a to 1c). This block shows the commodities produced by each industry. At the limit, this account can be disaggregated to 300 industries and 727 goods and services on the basis of the most detailed Canadian input-output tables. Columns 2a, 2b, and 2c represent the industry use matrix. The block (rows coded 1a, 1b, 1c, and columns coded 2a, 2b, 2c) is an intermediate input commodity-by-industry matrix, while the block (rows 3a, 3b, 3c, 3d, and columns 2a, 2b, 2c) shows primary inputs cross-classified by industry.

Taxes on products and services are shown in the "Allocation of primary income" row 4c of Tables 5, 6 and 7. Canadian accounts distinguish fourteen types of product taxes. These taxes are allocated to each purchaser of a commodity (on the basis of taxability) whether industries or final demand. The sub-matrix on commodity taxes is expanded in Section 10. The Final Demand Table comprises 48 categories of personal expenditure, 39 industry groups purchasing machinery and equipment, and 40 industry groups purchasing construction, 6 functional classes of government expenditure, 2 categories of inventories, and 2 categories of exports. Each of these classes is cross-classified by commodity. Thus, there are fourteen tax matrices (corresponding to types of taxes) relating to industries and final demand categories. This matrix corresponds to the cell at the intersection of the "Allocation of primary income (institutional sectors)" row and the "Goods and services (products)" column in Table 1 in the aggregate matrix.

The block (row 10 and columns 1a, 1b, and 1c) (Table 5) shows imports by commodities. From Canadian input-output tables, imports data can be obtained for potentially 727 goods and services. The import duties normally included in the value of imports are shown separately here with commodity taxes at row 4c.

The totals of columns 1a, 1b, and 1c (Table 5) represent the supply of each commodity. The supply of each commodity (for example, the commodity found in row 1c) shown in the column total is equal to domestic production plus taxes on products plus imports. The corresponding row total, for example, the total for row 1c (Table 7), is equal to the sum of intermediate use (for example: row 1c and columns 2a, 2b, and 2c in Table 5) and final use, namely, consumer expenditure (row 1c and column 6a in Table 6), government expenditure (row 1c and column 6c in Table 6), fixed capital formation used by various industries (row 1c and columns 8a, 8b, 8c in Table 7), and exports (row 1c and column 10 in Table 7). The explanation for the rest of the accounts is the same as the aggregate SAM except for the sector or transaction details.

5   Three Social Accounting Matrix applications

This paper presents three applications of SAM with respect to the Canadian economy for reference year 2000. Each application entails expansion of the aggregate SAM along an economic or social dimension in a way that is both conceptually coherent and statistically integrated.

The first application of SAM presents incomes, outlays, and savings of the household sector by income quintile and other household attributes. This study would affect four accounts (see Tables 5, 6 and 7): "Allocation of primary income" (code 4a), "Secondary distribution of income" (code 5a), "Use of disposable income" (code 6a), and the "Capital" account (code 7a). The household sector in each of these accounts will be disaggregated by income quintile with SHS data for reference year 2000. The "Allocation of primary income" account would record primary income accruing to various income brackets (quintiles); the secondary distribution of income would include the transfers received and paid by each income quintile; and the use of disposal income would record the expenditures and savings by each income group. Expanding along the income dimension requires that the macro household estimates be integrated with data from household expenditure surveys and from other data sources on income and spending behaviour of households with different income profiles. The result is a fully integrated micro-macro framework that reveals not only how savings differ by income group, but also how different types of income (e.g., employment income versus social transfers) and different types of expenses (e.g., transfers to governments versus consumer expenditure) vary by income group.

The second application of SAM expands aggregate employment income (wages, salaries, and supplementary labour income). This estimate is expanded along the industry dimension, showing the origin of labour income of households, and along four socioeconomic dimensions, namely, age, education, gender of the worker, and hours worked. The first expansion merely presents a CSNA industry breakdown of employment income, whereas the second integrates these statistics with data on age, education, gender, and hours worked obtained from other surveys and administrative sources. In this second application, the household sector in the "Generation of income" account and industries in the "Production" account will be expanded. The result is a SAM that permits analyses such as with respect to how average hourly earnings of workers vary according to their age, their years of schooling, and their gender in a statistically integrated framework.

In the third application of SAM, taxes on products are expanded to show their makeup in terms of tax bases (e.g., final consumption, fixed capital formation), in terms of the jurisdiction of the tax (federal, provincial), and in terms of the type of tax applied. This expansion of SAM permits an analysis of the origins of tax revenues and the relative contributions of tax jurisdictions. For this purpose, the authors have removed taxes on products from the allocation of primary income and created a product tax account (Tables 3 and 4, "Taxes on products" row and column). The tax account row depicts taxes paid on intermediate consumption and on final consumption, and the column shows the total tax on products at the intersection of the tax account column and the allocation of primary income.

The principal contribution of these applications of SAM is in making available a coherent framework for analysis that integrates macro statistics from the national accounts with micro statistics on agents from social or economic surveys. In the first two applications presented here, the constructs involve an integration of two or more statistical sources, typically one at the macro level and one or more at the micro level. At the micro level, the focus is on the decision-making of individual units such as households or persons. Decisions made by these units are a function of certain observable attributes, such as the individuals' ages, number of children, source of employment, level of income, and single versus multi-parent family type. While statistics on these attributes are available from surveys (e.g., household surveys or labour force surveys), these bodies of information are usually separate from those that make up national accounts aggregates. National accounting data are generally concerned with aggregates that describe macroeconomic variables such as a sector's disposable income, its savings, or its expenditures. So long as these statistics in this domain remain separate from statistics on microeconomic agents, one cannot explain macroeconomic phenomena described in national accounts in terms of changes in behaviour that occur at the micro level, such as that of the household, the individual, or the firm.

Using statistics at the micro level is often problematic because the concepts and definitions relating to these statistics are heterogeneous; consequently, this makes linkages among them, and their integration with macro statistics, difficult. Over the last decade, Statistics Canada has streamlined the concepts and definitions relating to its survey and administrative data collection vehicles in order to maximize the consistency of its statistical outputs. Nevertheless, significant consistency issues in both concepts and methods still remain because data collection vehicles serve fundamentally different needs. For instance, variables used in household spending surveys are defined for capturing attributes that relate to consumption habits and that meet specific data collection objectives; these objectives may differ from those of labour force surveys, which are intended to gather a coherent picture of labour market participation.

6   First application: Expanding the household sector of the Social Accounting Matrix

This section expands the household sector of the disaggregated SAM presented in Tables 5, 6 and 7 so that incomes, outlays, and savings can be shown by socioeconomic attributes, such as income bracket or household type. After removing non-profit institutions from the sector, the household sector account proper can be integrated with socioeconomic data from the SHS and personal income tax data from the CRA. The literature shows two clearly different approaches to such integration. The first approach—used by den Bakker et al. (1994) in building a historical SAM for the Netherlands—expands the household sector within the SNA framework. Adapting the (micro level) household statistics to the national accounting control totals results in national accounting concepts and imputations being preserved while the data are disaggregated by household attributes.

The second approach, advocated by Ruggles and Ruggles (1986) and followed in this section, integrates national accounts aggregates with household level data on the basis of a market transaction view of household incomes and outlays. This involves recording transactions that constitute incomes and outlays of the households as they occur rather than following some SNA conventions. While the SNA uses the transaction as the basis for measuring the flow of economic activity between institutional units, the system provides for three explicit exceptions, whereby transactions are not recorded as observed (SNA 1993, paragraphs 3.12 to 3.15). In each of these "rearrangements", transactions are rearranged so as to "bring out the underlying economic relationships more clearly" (SNA 1993, paragraph 3.23). In compiling the incomes and outlays of the household sector according to the SNA 1993, some transactions are re-routed and others are partitioned in order to portray the economic significance of the activity between the parties. These rearrangements result in income, outlay, and savings for the household sector that differ from what this sector actually experiences. In the market-transaction approach followed in this paper, these rearrangements are reversed in order to conceptually integrate macro and micro statistics and to arrive at savings consistent with households' actual experience. The rearrangements addressed in this paper relate to the treatment of owner-occupied dwellings, the treatment of investment incomes of insurance carriers and pension funds, the contribution of employers to pension and social security schemes, the treatment of insurance other than life insurance, and the treatment of imputed financial services. Although using the transaction basis alters national accounts aggregates such as household income, outlay, and savings, this method introduces a new perspective on the household sector by showing disaggregate socioeconomic data that are especially suited for analyzing present and future household behaviour in areas such as savings, investment, and consumption. The integrated household account can also be used to address questions such as the adequacy of actual savings in providing households with income security, the role of government transfers in fiscal redistribution among different income levels, family structures, and employment profiles. SAMs that serve other analytical needs can be constructed from the methodology and the integrated database used in this section. Readers should note that changes to national accounts aggregates resulting from the application of the transaction criterion are confined only to the household account shown in this section of the paper. Aggregates shown in Tables 1 and 2 and Tables 5, 6, and 7, and those shown in subsequent applications of SAM are not modified on the basis of the work done in this section of the paper.

In addition, using the market transaction criterion entails including realized capital gains as household income. This is consistent with the disposable income recorded in the accounts, which nets out income taxes on all forms of income, including taxable capital gains.

7   Integration of the Survey of Household Spending and System of National Accounts household accounts

The data sources used for this study are the reference year 2000 SHS and personal income tax data from the Canada Revenue Agency (CRA) for the same year. The survey collected data from a sample of about 21,000 households on their income, expenses, dwelling characteristics, household equipment, family structure or household type, household balance sheet, and a number of other analytically useful characteristics. In order to construct a coherent SAM for the household sector, it is necessary to harmonize the concepts underlying the SNA household sector account with those of the SHS and to achieve an acceptable measure of statistical integration by making appropriate modifications to one or both sides. However, integration of the two sets of accounts is problematic because the system of national accounts follow definitions and standards designed to facilitate macroeconomic analysis; they adhere to concepts such as neoclassical theories of consumption and production and to international standards and conventions for compilation of national economic accounts. These concepts are often at variance with individuals' or households' perception of what they consume, what they earn, and what they save because individuals and households are concerned solely with the micro unit, namely, the household. For instance, a household would consider the funds it receives from a pension plan, from disability insurance, from interest on bank deposits, or from gains from sales of its assets to be its income for household management purposes. By contrast, none of the above are considered income for the household sector of the economy since they do not originate in current economic production. Similarly, households do not consider as income the contributions of an employer to their pension plans or to disability insurance, since they cannot use these sums to defray household expenses (until they receive the associated benefits). However, all these items enter into income of the household sector for national accounts purposes. In this study, the authors follow a market-transaction approach, as proposed by Ruggles and Ruggles (1992), for harmonizing the concepts of income and expenses between the micro units (households) and the national accounts aggregates. As elaborated below, aggregates (such as income and savings) constructed under this approach differ substantially from those that follow SNA 1993 definitions. Table 8 sets out the CSNA household income and outlay for the year 2000 as published by the CSNA. It shows savings of $29.9 billion as the balancing item of income ($840.4 billion) over outlays ($810.5 billion) of the sector.

Steps that are taken to harmonize and integrate the two sets of accounts are discussed below under separate headings.

7.1  Delineation of the household sector

The CSNA presently consolidates non-profit institutions serving households with the household sector into "Persons, Unincorporated Businesses, and Non-profit Institutions Serving Households." The operating expenses of non-profit institutions are financed by contributions from governments, businesses, and households. ⁸  The presence of non-profit institutions in the sector makes it difficult to relate the aggregate sector with the behaviour of the households that make up most of the sector, since the economic behaviour of these organizations differs significantly from that of persons. Sectoring out these institutions as recommended by the SNA 1993 (SNA 1993, paragraph 4.10) would constitute an appropriate treatment for improving macro household statistics. This will reduce household revenues and expenses by $9.8 billion (Table 9). While this will have no effect on gross saving, it will reduce the level of income shown for the sector and harmonize income and expense aggregates with the SHS.

7.2  Employment benefits

Following SNA conventions (SNA 1993, paragraphs 7.43 and 7.44), national accounts aggregates show all contributions made by employers toward social insurance, private insurance, and pension plans of employees as labour compensation even though these are paid directly into public or private plans. ⁹  The economic logic underlying this convention is that these expenditures benefit employees and represent a cost to the employer. The convention facilitates macro economic analysis and to account for production and consumption as aggregate concepts. However, it is inherently problematic for microeconomic analysis of the household. While it is true that household decisions are made in full cognizance of the benefits receivable from these plans, such benefits cannot be considered income of the household for microeconomic purposes. Employer contributions provide income in the future (in the case of pensions and life insurance), which will be accounted for in those periods. In the current period, households have no discretion in deciding whether to spend these sums on goods and services or to save them.

However, two classes of employment benefits should be included as income for the current year because these constitute the consumption of real services in the current period: retirement allowances ($4.8 billion), which individuals receive in cash upon leaving the workforce; and employer contributions to employee welfare benefits ($13.9 billion), which include dental insurance plans, drug plans, and disability insurance (Table 10).

In the CSNA, contributions to social insurance, such as the Canada Pension Plan (CPP) and the Quebec Pension Plan (QPP) ($12.0 billion), Employment insurance ($11.0 billion), and workers' compensation ($6.5 billion), are shown as transfers back to government. In order to reconcile the macro concept of household income with the household budget, these contributions are removed from both the income and the outlay sides of the account. This would affect only net saving. For the same reasons, employers' contributions to pensions are also removed from income. Benefits received under the above programs constitute current income and are included in the income and outlay account of households.

7.3  Owner-occupied dwellings

One of the national accounting conventions that clearly impede micro-macro integration relates to the treatment of owner-occupied dwellings (Ruggles and Ruggles 1986, p. 251-252). The SNA shows the ownership of owner-occupied homes as an industry. Under this convention, owner-occupied homes are treated as if they were rented to their owners at a competitive rental rate by this fictive industry, which receives the rental incomes, pays mortgage interest, operation, and maintenance expenses and taxes, sets aside funds for depreciation, and pays the rest to households as net rental income. While this treatment may be expedient for many purposes, it has significant drawbacks when national accounts data are used to analyze household behaviour and household savings (see Ruggles and Ruggles 1992 and Webb 1980). Essentially, this is a consequence of removing housing from the domain of the household and classifying it as a business industry. Unless a number of complex adjustments are made, this distorts the analysis of homeowner decisions with respect to borrowing, investment, and saving when aggregative household account data are used. In addition, the existing treatment includes the depreciation of homeowners' property as part of the imputed rent of owner-occupied dwellings, which is considered a consumption expenditure for the current period. Families typically pay for repair and maintenance of their dwellings as they arise. The treatment understates the amount that households actually put aside for future consumption by classifying it as an amount consumed in the current period. Gross saving shown in the SNA sector accounts (which include capital cost allowance) shows the appropriate amount of savings available to households.

An appropriate reconciliation between the macro and micro sides is to dissolve the housing industry and to allocate the current costs of home ownership (excluding depreciation) to households as suggested by Ruggles and Ruggles (1986). This method consists of moving the expenses of the housing industry, such as repairs, mortgage interest, insurance, maintenance, and property taxes, to the household sector, where they are shown as household expenses. Capital consumption allowance of housing, a major expense at $14 billion in 2000, is no longer charged as a current household expense. As Table 11 shows, household incomes will be reduced by the amount of the imputed rental income of owner-occupied dwellings (net rental income of unincorporated business) by about $24 billion in 2000. Household outlays will be reduced by $38.4 billion as depreciation and mixed income are no longer considered costs. This treatment will not affect gross saving but will affect net saving.

7.4  Insurance income and pension income

Another national accounting treatment that differentiates micro and macro concepts relates to life insurance and pensions. National accounting conventions have always treated the reserves of these institutions as if they were assets of persons who ultimately receive benefit payments. It follows from this principle that any returns earned on invested assets of these funds ¹⁰  are recorded in the national accounts as income of households in the current period. It also follows that benefits (e.g., annuity payments, policy dividends) paid to households are not recorded as households' income: they appear only as expenses of life insurers and pension funds. In addition, lump-sum life insurance settlements paid to households do not explicitly appear in the accounts because they are treated as intra-sectoral transfers, whereby insurance assets are consolidated with all other assets within the household sector.

By contrast, data compilation at the micro level follows business accounting conventions. Households, which are unaware of investment incomes of insurance and pension funds, simply report their receipts from pension plans, annuities, life insurance claims, and dividends as their current income. They similarly consider gross premium payments as the cost of acquiring the security of insurance and report these payments as their outlays. Households plan their spending, investment, and saving decisions in light of these incomes and outlays.

In order to make the macro accounts consistent with the household budget reflected in the micro data, investment returns originating from insurance reserves and pension assets are removed from the income side of the household sector; they are replaced by dividend income, annuity receipts, pension benefits, and life insurance claims received (Table 12).

In order to account for household expenditure on insurance, payments of life insurance premiums should be shown as an outlay. However, this should include only premium payments for life insurance products, such as term life insurance actually consumed by the household, and should exclude premiums for insurance plans with a saving feature such as endowment insurance. ¹¹  While premiums paid for term life insurance are unambiguously household expenses at par with premiums for fire and casualty insurance that purchase risk coverage, certain other types of life insurance (endowment insurance and some forms of universal insurance) contain an investment component that is managed as an investment portfolio by insurance companies. These investment components have attributes that make them close substitutes of tax-sheltered savings plans and savings accounts. Consequently, it is necessary to separate the data on life insurance premiums into data on pure premiums and data on investment contributions, and to treat these two types of data accordingly. This separate treatment has not been implemented in this paper; this has resulted in a slight understatement in household savings.

In keeping with the same rationale, household contributions toward pensions (paid to pension managers, life insurers, etc.) are not shown as an outlay of the household because they are a form of saving that increase household assets.

7.5  Insurance other than life insurance

The treatment of insurance other than life insurance (property insurance and casualty insurance) in the accounts is another area where existing conventions necessitate an adjustment. In accordance with conventions recommended in SNA 1993, households' expenditure on insurance other than life insurance (e.g., automobile, property, accident, and sickness) is measured by means of a cost-of-service concept. Premiums and claims appear on neither the income side nor the expense side of the accounts. Instead, a cost-of-service appears as the outlay of consumers of insurance, which includes premiums paid less adjustment expenses less claims received by households plus the investment income (premium supplement) of policy reserves. Consequently, household expenses shown in the SNA sector accounts are net amounts that are smaller than the actual (gross) outlays on premiums. The investment income associated with this cost-of-service is also entered on the income side of the SNA sector accounts. Although insurance settlements (claims) received by households finance their expenditures on goods and services, they are added only to income, but are shown as part of depreciation in the sector accounts.

At the micro level, data from SHS household expenses exclude both income from settlement of insurance claims and the expenses of goods and services financed by them. The survey records only the deductible portion of expenditures covered or financed by insurance as a household expense. To reconcile the household accounts with the actual budget of the household, the imputation for cost-of-service is removed from personal expenditure, and the full amount of premiums paid by the sector is shown as an outlay. On the income side, insurance settlements received by households are added to income while the same amount is removed from the sector's capital cost allowance. ¹² 

7.6  The treatment of interest

The treatment of interest income and interest expense in the national accounts differs significantly from the way in which households perceive and report on these items. The CSNA income and outlay account of the households sector shows total interest and other property incomes of households. In addition to total interest received, the CSNA includes an imputation for FISIM (financial intermediation services indirectly measured) on deposits of households held by financial institutions, as recommended by SNA 1993. This imputation, $6.7 billion in 2000, is the difference between what the household sector actually received and the opportunity cost of those funds to deposit-taking institutions. An identical amount is shown as personal expenditure of households on financial services. On the borrowing side of financial transactions, the interest paid by households is divided into two parts: imputed loan interest and pure interest. The imputed loan interest is shown as consisting of two components: an expenditure by households ($7.5 billion in 2000) and a pure interest payment ($11.7 billion in 2000) to the financial sector.

In contrast to the transfers and imputations provided for under the SNA, the micro data on household spending show gross interest (and dividends) income from domestic and foreign sources and gross principal-and-interest payments on consumer loans and mortgages (including principal-residence mortgages). Interest income is also available from income tax data. In making the SNA and the micro data sources consistent, the authors' objective is not only to accurately reflect the household budget by showing the actual net interest income (expense), but also to show gross interest (and dividend) income on the income side and gross interest expenses on the cost side, so that each side can be matched with the desired data on social attributes of households. To make these modifications in the household accounts data, the imputation for financial services (FISIM) on deposits previously shown as part of income is removed. As well, all FISIM imputations are eliminated from the expenditure side of the account and replaced with gross payments of interest on consumer credit (i.e., all interest payments of the sector except those in respect of mortgages on owner-occupied housing). As mentioned in Subsection 7.3, homeowners' mortgage interest is added to the expenses side of households and the CSNA industry known as "Owner-Occupied Dwellings" is dissolved. These adjustments reconcile both the concepts and the values in the household sector accounts with those of micro data from the expenditure survey.

7.7  The treatment of employees' contributions to pensions

Households' contributions to pensions sponsored by employers or managed by insurance companies are (implicitly) included in savings in this study, rather than being treated as an outlay of households. The reason for this treatment is that pension assets belong to households, that they are generally portable as employees change jobs, support households' borrowing power, and relieve them of the necessity of voluntarily saving for retirement through other vehicles such as tax shelters or other forms of retirement savings plans. However, this may not be the most appropriate treatment, although the appropriateness of such treatment depends on the objectives of the analysis. If not appropriate, one would simply adjust the expense and saving figures accordingly.

The case of contributions to social insurance schemes such as the CPP and the QPP is more straightforward. Household contributions to these schemes are included as outlays (as is current practice under the CSNA) rather than as savings under present regimes. Contributions to these social insurance schemes have many of the attributes of taxes that support social safety nets: namely, they are (mostly) unfunded, offer terms and conditions that can be changed unilaterally by governments as policies change, and involve no legal or contractual obligations. As some of these conditions change, it may be more appropriate to include these payments in household savings. For instance, an increasing proportion of CPP assets are funded and invested in marketable securities. As Canadians begin to view these investments as their retirement savings, household contributions shown as outlays should be reduced in order to allow for an increase in corresponding savings.

7.8  The treatment of holding gains

The SNA 1993 defines income as "[…] the maximum amount that a household, or other unit, can consume without reducing its real net worth. However, the real net worth of a unit may be changed as a result of the receipt or payment of capital transfers and as a result of real holding gains or losses that accrue on its assets or liabilities" (SNA 1993, paragraph 8.15). Holding gains that result from changes in the prices of assets (fixed and financial) are excluded from income under the SNA because income is measured on the same basis as production. Following this definition, household receipts that originate from capital gains when they dispose of financial or non-financial assets are not reflected on the income side of the SNA sector account, even though the income tax paid on these gains is shown as an expense in calculating disposable income. In addition, goods and services purchased by households with capital gains proceeds appear on the expense side as personal expenditures. The actual saving by the sector, which is a residual of income over outlay, is thus understated by the amount by which consumption is financed through this source. ¹³  The Expert Group on Household Income Statistics (the Canberra Group), which deliberated revisions to the SNA 1993, acknowledged in its 2001 report that households' use of these proceeds to finance consumption argues in favour of their inclusion in income. The Canberra Group recognized also that "there are good reasons in some areas for departing from the recommendations embodied in SNA93, reflecting the different purposes of the statistics to be compiled" (Canberra Group, 2001, p. 16) and recommended that capital gains and losses should be treated as a memorandum item which may be added to income for purposes of certain analyses (Canberra Group, p. 17). The report of the Canberra Group also recognized that certain features of macro statistics are not suitable for microeconomic purposes, such as for analysis of income distribution, because households with higher income levels receive a disproportional share of capital gains income. ¹⁴ 

The household account presented in this paper uses the transaction as the basis for recognition of incomes and outlays of the household sector. As such, it includes proceeds from all sources, including realized capital gains. The SNA 1993 recognizes that the use of SAM for general equilibrium modeling and policy analysis requires that it adapt to the specific needs of these applications. The SNA 1993 affirms that "if one thinks that […] capital gains are directly and to a large extent reflected in final consumption expenditure, these values could be booked as an incoming on the (secondary) distribution of income account" (SNA 1993, paragraph 20.130). This is the approach that the authors take in this paper. Unfortunately, data are available only on taxable capital gains at the present time. It would be appropriate to also include here the proceeds from other realized capital gains, such as the sale of principal residences (a non-taxable income in Canada).

8   A transaction-based integrated household account

Tables 13 and 14 show at an aggregative level how the existing CSNA account for the household sector is modified in order to be integrated with the survey data on household spending. The household account shows $814.2 billion in income in 2000 (Table 14), rather than $840.4 billion as shown in the National Economic Accounts (Table 13). The household income now excludes $42.7 billion in employers' contributions to employee benefits (see Table 10) and excludes $24.0 billion in a national accounting imputation of net income for owner-occupied dwellings (see Table 11). Removing the incomes of non-profit organizations (contributions of government and business) provides an additional reduction of $8.0 billion. However, the new estimate of income includes $1.7 billion in life insurance dividends, $31.0 in capital gains income (as reported by the CRA), $16.7 billion in insurance claim settlements, and larger estimates for life insurance and annuity benefits ($18.5 billion) and for pension benefits ($39.0 billion).

On the expense side, the sector's outlays are only $740.8 billion, rather than $810.5 billion as shown in CSNA accounts. Firstly, a key difference with the CSNA sector account is that outlays on housing are limited to actual expenses of maintaining owner-occupied dwellings, including taxes and mortgage interest, which add up to $42.9 billion, compared to imputed rent at $82.6 billion. Secondly, the imputation for financial services (FISIM) is removed, but outlays now include gross interest payment, compared with the existing sector accounts, which include only the transfer portion of interest. Thirdly, outlays that represent contributions to social insurance are significantly smaller, at $20.7 billion, because only the portion contributed by employees is shown as an expense of the household sector; this leaves out the portion paid out by employers ($29.0 billion). Finally, outlays on life insurance and on insurance other than life insurance are larger ($32.5 billion versus $17.0 billion) under the transaction approach; this accounts for the full value of policy premiums (excluding annuities).

The amount of savings generated by the sector is redefined in that it includes capital gains reported to the Canada Revenue Agency as taxable income. In accordance with this approach, income should also include non-taxable capital gains, such as gains in tax-sheltered assets (e.g., RRSPs). The present study overlooks these considerations, since no consistent estimates are available. Nonetheless, net savings (excluding depreciation) are $73.4 billion, compared to $29.9 billion; this result is consistent with national accounting conventions.

Disposable income, the residual of incomes over transfers to governments, is $642.5 billion, yielding a savings rate of 11.4%, compared to the rate of 4.7% consistent with SNA conventions.

Table 15 presents a summary of incomes and expenses by income quintile as well as social statistics relevant to each quintile. The data by quintile are based on the SHS, whereas the totals are derived from national accounts data modified according to concepts and procedures proposed in this paper, in order to ensure their integration with micro data on households. ¹⁵ 

To better facilitate the analysis of transfers by income level, transfers to governments and transfers from governments are shown by level of government for each quintile. The net amounts of these transfers are also shown in a subsequent line of the table. While the three levels of government received transfers of $171.8 billion from households, they transferred $102.5 billion to households; this led to a net transfer to governments of $69.2 billion. The table also shows net savings of $73.4 billion, or 11.4 % of households' disposable income.

Table 16 analyzes the results of Table 15, focusing on the relationship between income quintiles, transfers, and savings. SHS data for the first quintile show that about 91% of these households have no full-time wage earner, that such households consist mostly (62.3%) of single individuals. The average size of households in the first quintile is 1.5, compared to 2.5 for all quintiles. While households in the first quintile make up 20% of Canadian households, they earn only 2.7% of non-transfer or earned income. When one takes into account government transfers from all sources and all levels of government (these households receive more than 30% of all government transfers), the combined incomes of these households rise to only 6.1% of the household sector. Since their outlays account for 6.8% of those of the sector, their saving is negative (-$0.4 billion) (see Table 15). This result is not surprising given that these households account for 20% of households but receive only 7.4% of the sector's disposable income.

The second quintile has a slightly larger household size and includes a greater number of full-time earners; half of households in this quintile are made up of husband-wife family units. The non-transfer income of households in this quintile is somewhat larger, at 8.4% of that of the sector, but these households require more than 28% of all government transfers in order to raise their incomes to 10.9% of those of all households. Personal disposable income for this quintile was 12.1% of that of all households: the savings of households in this quintile totaled $2.0 billion in 2000 (Table 15).

The third quintile has very different attributes from those of the first two. The family size of households in this quintile is slightly larger than the average; they have more full-time earners than the average household, consist mostly (64%) of husband-wife family units, and paid $5.6 billion more in government transfers than they received in 2000 (see Table 15). However, this made them net borrowers of $2.3 billion, or about $1,000 per household, in 2000. This finding is discussed in greater detail in the Appendix (Subsection 12.1).

In the fourth quintile, families are larger (3.05 persons), and about one-third have two or more full-time wage earners. This quintile comprises predominantly (78.2%) husband-wife families with average incomes of $80,900, or 14% higher than the national average. These households receive a larger share of non-transfer income (23.9%) than their sheer number accounts for (20%). While they are recipients of 13.6% of all government transfers, they pay 24.4% of all transfers to governments. They earned about 22% of all personal disposable income in 2000. Nevertheless, they accounted for only 6.6% of all household savings. On average, these households saved $2,121, or 3.4% of their disposable income, during 2000 (see Table 15).

Finally, the fifth quintile, whose average income is $159,900 (123% higher than the average), has more units with two or more full-time earners (55.0%) than units with a single full-time earner (37.2%); this quintile had a family size much larger than the average for all quintiles (3.47 compared to 2.57). These households accounted for almost half of all non-transfer income. This group pays more than half of all government transfers (53.6%) and receives only 9.7% of transfers paid to the sector. In 2000, this quintile's net transfer to governments was $92.1 billion, a figure larger than the net transfer from all other households combined (see Table 15). These households saved more than 25% of their disposable income, or $30,481 per household, in 2000. In fact, households in this income bracket saved more than the rest of households in Canada in 2000.

The household income includes declared capital gains incomes of approximately $31 billion for 2000, a year in which households' gains from investments in equities were probably at their peak. It is important to know to what extent these incomes affect the savings picture described above because extraordinary gains in that year were much larger than in years before or since reference year 2000. A comparison of savings with and without declared realized capital gains is shown in Table 17. Excluding declared taxable income from capital gains (and gains within tax shelters, plus unrealized gains), total savings of about $42 billion out of (a smaller) disposable income indicate an average savings rate of 6.9%. Savings rates for all quintiles are lower than in the case of capital gains income, with the first and the third quintile showing larger rates of dissaving. For the highest income quintile, the savings rate out of disposable income would be 17.8%, versus 25.5% if capital gains were taken into account. The finding that savings are made predominantly by the 2.3 million households in the highest income quintile remains unchanged.

9   Second application: Breakdown of employee compensation by social attributes

This part of the paper expands on the labour compensation of employees with respect to the "Generation of income" account (row 3a in Tables 5, 6, and 7). The labour compensation data are shown by industry and cross-classified by social characteristics—age, gender, and educational attainment—in order to explain differences in employee compensation. Employee compensation in the national accounts consists of wages and salaries of employed persons plus supplementary income, which includes employment benefits paid by the employer.

Tables 24, 25, 26 and 27 in the Appendix (Subsection 12.2) present labour compensation for 20 broad industry groups in the business sector of the Canadian economy. The business sector excludes general government and non-profit institutions serving households (NPISHs). ¹⁶  Using data on hours worked corresponding to the employment income, Tables 28, 29, 30 and 31 in the Appendix (Subsection 12.3) present the hourly compensation of employees in each of these industries for reference year 2000. The labour compensation data represent the reconciliation of data collected from payroll surveys, labour force surveys, and administrative data. These sources are integrated at Statistics Canada ¹⁷  with data on employee attributes by establishment collected primarily from the Survey of Employment, Payrolls and Hours.

Labour force attributes for which labour compensation is shown are gender, seven worker-age brackets, and six successive levels of education. When one combines the three groups of attributes, labour compensation is shown for 84 gender/age/education combinations; this allows for the analysis of how labour compensation varies among attributes and among industries.

Tables 18 and 19, below, are an excerpt of the complete results presented in Subsection 12.2 of the Appendix. They reveal that about one-quarter of paid labour compensation in Canadian industries (excluding government and NPISHs) is received by women. They show, however, that this figure varies from as little as 7% in "Forestry and logging" to about 50% in the "Finance, insurance and real estate" group, and reaches almost 80% in "Health care and social assistance." It is easily evident from the tables that most of the compensation received by women in Canada goes to the 35-to-44 age bracket and that almost 60% being received by 25-to-44-year-old paid workers. These tables also show that a similar pattern prevails among men, although the age brackets that account for 60% of labour compensation are those from 35 to 54 years of age; this is somewhat older than the age group of women that come under this category.

While these results answer a wide range of queries pertaining to wages, salaries, and benefits, a more powerful and revealing picture of earnings differentials emerges from analysis of average labour compensation per hour, obtained by dividing annual compensation (Subsection 12.2), by similar tables of data on hours worked by the same employees. Complete results for average labour compensation per hour are presented in Subsection 12.3. Tables 18 and 19 below are an excerpt of Tables 28, 29, 30 and 31 (Subsection 12.3); they show that, in 2000, employees received an average of $21.50 per hour for work outside the general government and the non-profit sector of the Canadian economy. Workers in the "Utilities" industries, owing to their higher technical qualifications, received more than $35 an hour, a 65% advantage; workers in "Finance" saw a wage of more than $32 an hour, a 51% advantage.

Tables 20 and 21 show that, while women's average hourly compensation is about 82% of that of men in Canadian industries, these figures vary from lows of about 78% in the "Construction" and "Manufacturing" industries to highs of 96% in "Educational services" and to 98% in "Health care and social assistance." Importantly, the latter two industries employ a very high proportion of women.

Analysis of the tables in Subsection 12.3 of the Appendix also reveals that women's highest earning years are 35 to 44, when their earnings per hour are 12% higher than their lifetime average (see Tables 20 and 21). For men, this peak occurs in the 45-to-54 bracket, when they earn about 18% more than the lifetime average for their gender. At 65 years and over, women earn about 68% of their lifetime average hourly compensation. Men in this age bracket earn about 59% of the lifetime average for their gender. Furthermore, while men in the 55-to-64 bracket still earn 4% more than their life-time average hourly compensation, women's compensation in that age group stands at 4% below their lifetime average.

Tables 20and 21 also shed light on how educational attainment may affect employees' hourly compensation. For example, the first column of these tables illustrates the benefit of a postgraduate degree with respect to an employee's earnings, compared to the average for his or her age group for all industries. The premium for women varies from 8% for the 65-and-over group to 49% for the 18-to-24 age group. These figures show that those with a postgraduate degree earn up to 49% more in hourly compensation than their peer group. In the highest-earning bracket, namely the 35-to-44 age group, a postgraduate degree confers a 45% earning advantage whereas a bachelor's degree provides a 37% premium over the average. For men, a postgraduate degree boosts hourly earnings by only 31% over the average during their highest income earning years of 45 to 54 years of age, whereas the advantage is as high as 36% for the younger groups, 25-to-34 and 35-to-44 years of age, which have slightly lower earnings as a rule; this shows the diminishing returns to education in higher income brackets. The advantage conferred by a bachelor's degree is highest for the 18-to-24 age group and tends to decline as a worker ages. Within every age group, having both a bachelor's degree and a postgraduate degree tends to raise the earnings of women by a greater amount above the average wage than the earnings of men.

10   Third application: Expanding the government account (taxes on products by type and by origin)

This section provides a breakdown of taxes on products. The goal of this study is to present an appropriate breakdown of tax types, along with the associated tax bases (types of products), showing the origin of government revenues from indirect taxes. These taxes do not include taxes levied on incomes of persons or on incomes of incorporated and unincorporated enterprises, or taxes on production, such as property taxes, school taxes, and capital taxes. The latter taxes could also be expanded through a study similar to what appears in this paper.

At the most disaggregate level, the CSNA compiles 14 types of product taxes in its national and provincial input-output tables. ¹⁸  For the purposes of this exercise, these product taxes have been grouped into six broad tax groups.

Federal taxes: These consist of federal trading profits, the federal gasoline sales tax, the federal air (transportation) tax, the excise tax, and the excise duty.

Provincial taxes: These are taxes imposed by various provincial governments in Canada. They consist of the provincial gallon tax (on alcoholic beverages), the tax on trading profits, and the provincial gasoline sales tax.

Municipal sales taxes: these consist only of amusement taxes levied by certain municipalities.

Goods and Services Tax (GST): The GST, imposed nation-wide by the federal government, accounts for the largest proportion of indirect tax revenue. With relatively few exceptions, such as some types of store-purchased foods and financial services, all goods and services transacted in Canada are subject to the GST. Businesses receive refunds for GST paid on intermediate inputs. The Harmonized Sales Tax (HST) has recently replaced the GST in some provinces, where, subject to a federal-provincial tax accord on sales tax harmonization, the federal government collects the provincial sales tax and the GST in a single collection at the point of sale.

Provincial Sales Tax (PST): This is the provincial counterpart of the GST. It is imposed by most provinces on most types of transactions involving goods.

Custom Duties: These taxes are imposed by the federal government on imports of goods. Data on import duties were estimated from the input-output tables on the basis of the import-share assumption conventionally used for these calculations.

The disaggregate SAM presented in Table 22 shows the values of taxes collected for each of the above six types and lists the types of products affected. This is a disaggregation of the "Taxes less subsidies on products," which has a value of $86.2 billion, as shown in the cell at the intersection of the "Allocation of primary income" row and the "Goods and services (products)" column of Table 1 or, alternatively, in Table 5 as the sum of (4c, 1a), (4c, 1b), and (4c, 1c). Table 23 expresses the effective tax rates that underlie Table 22, showing those taxes as a percentage of the value of goods and services associated with them.

Table 22 reveals the preponderance of product tax revenue in final expenditure, which accounts for about 80% of the total taxes on product; about 85% of this total comes from personal consumption expenditure. Slightly more than 12% of taxes on final expenditure were collected through levies on capital formation (such as construction, and machinery and equipment). Exports accounted for a negligible share of taxes. Taxes on intermediate consumption of goods and services by domestic industry accounted for the remaining 20% of taxes collected, with some 87% of such taxes coming from industries other than construction. Table 22 also shows that provincial sales taxes account for more than 80% of the taxes imposed on the construction industry's intermediate expenses, with another 10% taking the form of other provincial taxes. Import duties and other federal taxes comprise less than 10% of the taxes paid by the construction industry.

Table 23 contains effective tax rates that illustrate the relative incidence of taxes on products shown in Table 22. ¹⁹  By far, the highest incidence occurs on personal consumption expenditure, whereby consumers pay about 11%. While the federal GST applies at the rate of 7% (in 2000), tax exemptions for consumer expenditures such as food and rent and for financial services, such as banking, reduce the effective rate to 3.7%. The effective tax rate on capital formation was 4.3% in 2000; this tax rate comes mostly through federal and provincial sales taxes paid on investment goods and services that are capitalized. Canadian industry paid about 2.0% on its intermediate expenses, about the same as governments paid on their expenditures (1.9%). However, the construction industry paid a substantially higher tax rate (3.4%) on the goods and services that it bought in order to produce its output. This industry is defined in the national accounts as including not only the builders of residential and non-residential structures, but also all other construction activities conducted on own-account elsewhere in the economy. The tax on exports was less than one-tenth of one percent, whereas imports were subject to an effective rate of 0.6% in custom duties. The highest rate of import duties is found in goods that enter into personal consumption expenditure.

11   Conclusion

This study has demonstrated the feasibility of constructing a Social Accounting Matrix (SAM) for the Canadian economy with existing Statistics Canada economic and social statistics. At the macro level, the data sources are the income and outlay accounts and the national input-output tables. Micro variables are obtained from a household spending survey, payroll and labour force surveys, and income tax files. The study has shown that SAM is a useful construct in its own right as an integrated system of socioeconomic statistics.

This paper shows three applications of SAM integrating macro and micro concepts and data sources. The most interesting case is the socioeconomic disaggregation of the household sector (the first application), which is possible only in a SAM framework. The paper shows that a market transaction approach to measuring the household sector works well for integrating incomes, outlays, and savings of the sector with socioeconomic data on household attributes. The authors suggest that a supplementary income and outlay account for the household sector based on the transaction approach would be a useful addition to existing statistical measures. While the paper presents a complete integration of macro and micro concepts that employs the market transaction approach, statistical integration of household spending data, tax data, and national accounts aggregates is a broader project than the objective of this paper. Work is under way by the authors and others at Statistics Canada to enhance the statistical reconciliation of household income data though the integration of household surveys with income tax data and other sources.

12   Appendix

12.1  Results for households in the third income quintile

The third income quintile shows a net borrowing of $2.3 billion, or $995 per household, in 2000 (Table 15). By contrast, the second quintile has a positive saving of $881 per household, and higher quintiles also show positive savings that rise with households' average income levels. This finding challenges the expectation that saving levels would monotonically rise with income levels for each quintile. The purpose of this subsection is to explore possible explanations for the saving pattern observed across quintiles, by means of the limited data available at the present time. A definitive explanation would require data on all elements of income and outlays over a period of several years. However, some tentative explanations can be gleaned from available data on household attributes.

Comparing the households in this quintile with others in the SHS suggests explanations in terms of the average age of household members and in terms of their employment status. Households in the third quintile have an average income of $56,500, about 44% higher than the $39,000 earned by those in the second quintile. However, it could be argued that these households expect a still higher long-term income, because 69% of them are holders of full-time jobs, whereas the rate is only 40% for the second quintile (the rate is 85% for the fourth quintile). Full-time employment brings with it not only a higher expected lifetime income to support consumption, but also the ability to finance consumptions in excess of current income on a temporary basis. This explanation is consistent with predictions of the permanent income hypothesis of consumption.

A second and complementary explanation relates to the average age of wage-earners who head households. A proxy for this available from the SHS is the average age of the reference person—the individual who is normally responsible for financial maintenance of the household, such as paying bills. The third quintile's reference person is on average 47 years old, and nearly a half (46%) of all reference persons are between 25 and 44 years of age. The reference person in the second quintile is considerably older, at 53 years of age. In the second quintile, references persons are fairly evenly spread: about one-third are in the 25-to-44 age bracket; one-third are in the 45-to-64 age bracket; and about 31% are 65 years of age or over. In the fourth and fifth quintiles, the average age declines only slightly, to 45 years and 46 years, respectively. However, these households have substantially higher levels of income on both a per-adult basis and a per-person basis, compared with households in the third quintile. This explanation is consistent with the life-cycle theory of consumption.

In sum, the likely explanation of the observed saving pattern is that households in the third income quintile have a disproportionate number of people who are economically secure since they hold full-time jobs, yet are within an age bracket when major expenditures are undertaken that are associated with early stages of life in anticipation of higher incomes in the future.

12.2  Labour compensation in 2000 — Tables

12.3  Compensation per hour in 2000 — Tables