Appendix — Technical and Methodological Issues
Consumer Price Index
Statistics Canada and its predecessors have been publishing consumer price indices since early in the 20th century. The current Consumer Price Index (CPI) is obtained by comparing through time the cost of a fixed basket of commodities purchased by consumers. The weight given to the data for each commodity included in the Index is principally derived from various periodic family expenditure surveys.
The CPI is often used to eliminate the impact of inflation in time-series data (or, in other words, to convert time-series data from current to constant dollars). It is worth noting that the CPI is not a cost of living index, but rather is based on average price changes for a fixed basket of goods over a given period, in a given geographical area.
Some caution must be exercised with respect to the use of CPI information. A change in the quality of a service, for instance, may affect its price and hence its contribution to movement in the CPI. Statistics Canada has made great strides overcoming this limitation, but researchers must remain aware of the constraints, given that the problems encountered in adjusting prices for quality changes are complex and sometimes impossible to solve in a fully satisfactory manner.1
A further limitation of the CPI is apparent in Chapter 9 on consumer spending, that is, the inability to use the CPI components to directly deflate the detailed spending categories in the Family Expenditure Survey (FAMEX) (now called the Survey of Household Spending and the Food Expenditure Survey). This problem arises because the goods and services used to determine the CPI are not always the same as those covered by the Family Expenditure Survey. Statistics Canada provides this caution:
The commodities (goods and services) represented by the CPI are the commodities that can be associated with a retail price, i.e., with the amount of money that a consumer must pay to purchase a specific quantity and quality of a good or service. Because of this restriction, it is possible to compute the CPI as a measure of price movement related to commodities of unchanging or equivalent quantity and quality. Family expenditure surveys, from which the CPI basket is derived, do not follow the restriction; hence their commodity coverage is broader than that of the CPI (Statistics Canada 1995, 5).
Consumer income: analyzing after-tax family income using quintiles
As demonstrated in Chapter 5, quintile analysis is often used to draw conclusions about the relative situation of people at either end, or in the middle, of the income scale. The quintile analysis in this report is based solely on families using after-tax income. To determine these groups, all economic families are ranked from highest to lowest after-tax income. These families are then divided into five equally sized groups. Hence, the same quintile definition is applied to the population for each year studied, which keeps the size of each quintile group constant, although the economic families per se will not be the same ones in a given quintile across all years (see Figure A.1).
Plotting the upper limit of each quintile gives the range of after-tax family incomes that make up each group. For example, the middle (family) quintile had after-tax incomes ranging from $43 642 to $58 567 in 2001. The highest (family) quintile, on the other hand, had after-tax incomes of $79 041 and above. Confidentiality restrictions prevent the publication of the upper limit for the highest quintile, which represents the economic family in Canada with the highest after-tax income.
Statistics Canada's low income cut-off (LICO)
The low income cut-off (LICO) is a measure developed by Statistics Canada to convey an income level at which a family may be in difficult economic circumstances because it has to spend a larger proportion of its income on necessities than does the average family of a similar size.
For example, the most recent base for the LICO is the 1992 FAMEX, which found that the average spent on food, shelter and clothing by all households was 43.6 percent of after-tax income.2 In this instance, the income at which a typical family spends 20 percentage points more — 63.6 percent — of its after-tax income on the three necessities is set as the LICO. In 1992, the income level at which a family of four living in a city of 30 000-99 999 people spent 63.6 percent of its after-tax income on food, shelter and clothing was approximately $21 300. This figure is set as the after-tax LICO for 1992.
Several LICOs are calculated in order to factor in differences in family sizes (as larger families need more income to meet their needs) and community sizes (to account for differences in the cost of living). The LICOs are updated annually using the CPI. It is worth noting that the relationship between spending and income is only used to produce the LICO threshold; low-income status is then determined by comparing the family's income to this threshold.
For example, using the above 1992 threshold for a family of four living in a city of 30 000-99 999, a family spending 90 percent of a $60 000 income on food, shelter and clothing would not be counted as low income (i.e. since the LICO is set at $21 300), whereas a family that spends 50 percent of a $20 000 income would be (Statistics Canada 2001, 9).
Sources of net worth (assets and debt) data
Survey of Financial Security
Chapters 6 through 8 present information on the assets, debts and net worth of Canadians.
The source of most of this information is the Survey of Financial Security (SFS). The SFS was conducted between May and July, 1999, and collected information on the assets and debts of families and unattached individuals.
The SFS was conducted in the 10 provinces, with those living in the territories excluded. Furthermore, other segments of the population (e.g. those living on reserves, inmates of penal institutions, etc.) were excluded. Statistics Canada states that the survey covered 98 percent of Canadians living in the 10 provinces.
The results in this report on assets, debt and net worth that are attributed to Statistics Canada are primarily based on Cat. No. 13-595 (for net worth) and custom tabulations (for assets and debts). Note that the value of pensions (for those who belong to, or once belonged to, a pension plan provided by their employer) is excluded from the majority of the results. Some information on pensions is provided in Chapter 6 and Chapter 8, and is based entirely on Statistics Canada Cat. No. 13-596.
Unless otherwise stated, all time-series comparisons are based on data for 1999 that have been adjusted to make them comparable to data for 1984 (the last time Statistics Canada conducted a wealth survey). The adjustment — proposed by Statistics Canada — is necessary because certain assets (i.e. contents of home, collectibles and valuables, and annuities and registered retirement income funds) included in the 1999 survey were not included in the 1984 survey (and hence, these assets have to be removed to make the two series comparable). Furthermore, all net worth, asset and debt information is expressed in constant 1999 dollars.
The SFS collected information for the family unit and not for each individual in the family. Hence, the term family unit includes both economic families (which accounted for approximately 68 percent of family units in 1999) and unattached individuals (which accounted for approximately 32 percent of family units in 1999). The definition of economic family is consistent with that provided in Chapter 5 on consumer income.
Finally, the chapters on assets, debt and net worth feature data based on median values (i.e. as opposed to average values). This is recommended by Statistics Canada, since those in the highest net worth decile hold a relatively large proportion of net wealth. This has a substantial impact on the calculations: in 1999, average net worth in Canada is approximately 2.5 times higher than median net worth. Median values better reflect the net worth of Canadians in the middle of the wealth distribution (as they are less affected by extremely high or low values).
Variables received from Equifax Canada
OCA obtained data from Equifax on four key variables: credit files, tradelines, credit balance and credit limit.
This information was disaggregated by loan categories (defined below), age groups and credit ratings.
A tradeline represents one financial obligation held by a consumer. A consumer's credit file contains all tradelines for that particular consumer. The credit balance represents the total amount owing on a particular tradeline. The credit limit represents one of two things: for an instalment loan, the credit limit is equal to the original amount of the loan (i.e. for a car loan, the credit limit equals the face value of the car loan until the balance is paid off), while for a revolving loan, the credit limit is equal to the maximum amount of money that can be drawn against it (i.e. for a credit card, the credit limit is equal to what is traditionally viewed as a credit limit — the maximum amount that a consumer can borrow off the card).
Equifax's database only provides details on consumers' short-term credit (i.e. it excludes data on mortgages).
More specifically, the database captures instalment loans (i.e. bank loans, finance company loans, automotive sales finance loans) and revolving loans (i.e. national credit cards, department store cards, other retail cards and other revolving credit).
Each financial obligation incurred by a consumer (e.g. a bank loan, a credit card transaction) is recorded as a separate tradeline. Each bit of credit history in an Equifax credit file is assigned a credit rating by the credit grantor. OCA obtained tradeline data broken down by North American Standard Account Ratings:
'Good' credit ratings
- R0: Too new to rate; approved but not used.
- R1: Pays (or paid) within 30 days of payment due date or not over one payment past due.
'Minor Delinquent' credit ratings
- R2: Pays (or paid) in more than 30 days from payment due date, but not more than 60 days, or not more than two payments past due.
- R3: Pays (or paid) in more than 60 days from payment due date, but not more than 90 days, or not more than three payments past due.
'Serious Delinquent' credit ratings
- R4: Pays (or paid) in more than 90 days from payment due date but not more than 120 days, or four payments past due.
- R5: Account is at least 120 days overdue, but is not yet rated '9.'
'Serious Delinquent, Action Taken' credit ratings
- R7: Making regular payments through a special arrangement to settle debts.
- R8: Repossession (voluntary or involuntary return of merchandise).
- R9: Bad debt; placed for collection; moved without giving a new address.
In the above credit ratings, the concept of 'pays (or paid)' is for the agreed amount that must be paid. For example, if a consumer pays their 'minimum payment' amount on a credit card (within 30 days of the payment due date), this tradeline will be assigned an R1 rating, even though there remains an outstanding balance on the credit card.
The data received by OCA were based on a 5.2 percent3 random sample of active tradelines (i.e. activity within the last 12 months) in Equifax's database. A sample was drawn from July 1992, 1997 and 2002.
Equifax's database includes the major financial players within each loan category as described above, with the exception of Desjardins. Prior to 2000, Desjardins only reported 'bad' tradelines (R2 to R9) to Equifax. Since R0 and R1 information is not available for 1992 and 1997, Desjardins was removed from the sample to generate a more consistent time-series.
Limitations of Equifax data
Ideally, if one wishes to specifically analyze the financial situation of a consumer, the most appropriate indicator is the 'credit file' variable, which contains the complete assortment of a consumer's outstanding obligations.
To obtain information by credit rating, the Equifax data were drawn at the individual tradeline level. Equifax was able to provide OCA with the corresponding number of credit files that these tradelines were based on. However, due to a number of concerns (outlined below), OCA determined that the 'credit file' variable does not provide an accurate representation of the adult Canadian population.
For example, one notable concern of OCA is with respect to the fragmentation of credit files within Equifax's database. Fragmentation occurs when tradelines from one specific consumer get mapped to more than one credit file. This may occur for a variety of reasons.
For example, when people use multiple addresses (such as for work and home) for different credit accounts, it is possible that a credit file may be set up under both. A similar situation may occur when it is not easy to distinguish between a first name and a surname. Furthermore, a recently married person who changes his or her name could potentially have a credit file set up under both the current and former surname.
It is worth noting that fragmentation of tradelines between multiple credit files has no impact on the aggregate credit balance or aggregate credit limit for various financial products. However, because fragmentation overstates the 'true' number of credit files, any average created using the credit file variable (in the denominator) will be understated.
In this report, aggregate Equifax data (i.e. total outstanding balances and total credit limits) have frequently been adjusted by relevant population data (from Statistics Canada) to generate average values per adult Canadian. Recall that file fragmentation biases the 'credit file' variable but has no impact on the outstanding balances and credit limits. Also note that the all-items consumer price index was used to present all monetary figures in 2002 dollars.
Sources of household spending data
Chapter 9 on consumer spending is primarily based on data from Statistics Canada's Family Expenditure Survey (FAMEX 1982, 1986, 1992) and Survey of Household Spending (SHS 1997-2002).4 Statistics Canada's Spending Patterns in Canada (Catalogue No. 62-202) has also been used. Recall that the family expenditure surveys cover a broader basket of commodities (goods and services) than does the CPI.
As shown in Figure A.3 of this appendix, household spending data are broken down according to 17 main spending categories. Statistics Canada only produces detailed FAMEX and SHS spending tables in current dollars.5 Therefore, unless otherwise noted, figures presented in Chapter 9 are in current dollars.6
Due to significant differences between these two surveys, interpreting data on shelter from a time-series perspective is complex.
On advice from Statistics Canada, the Office of Consumer Affairs (OCA) did attempt to reconcile the shelter and other data (i.e. to adjust them to take into account differences in the definitions of certain variables between the FAMEX and SHS). Despite these efforts, however, a clear break remained in the data, making interpretation of this revised shelter series unclear. Information on OCA's attempt to reconcile the FAMEX and SHS data, presenting some broad comparisons using adjusted 1982 FAMEX data, is provided later in this appendix. The interpretation of the adjusted shelter variable is judged to be somewhat less difficult for 1982 (see below).
The sectoral case studies, however, offer more detailed historical comparisons given that, for spending categories other than shelter, FAMEX and SHS data require fewer adjustments and the time-series are easier to interpret. As a consequence of being unable to create a consistent time-series for shelter (and, consequently, total expenditures), the first two overview sections of Chapter 9 are based exclusively on SHS data (i.e. a time-series restricted to 1997-2002).
The data used in Chapter 9 are based on information collected from private households in Canada's 10 provinces. Detailed spending tables were obtained per income quintile and household type. Figure A.2 presents the main SHS household types7 used in this report's analysis, structured under two main groups.
|Total One-Person Households||Total Husband-Wife Households|
|One-person households (65 years of age and older)8|| |
Husband-wife households (both 65 years of age and older)
Husband-wife households without children (and without additional persons)
Husband-wife households with children (and without additional persons)9
Two other SHS household types are not analyzed, that is, 'husband-wife households with additional persons' and 'other households.' There is a relatively lower degree of homogeneity in the composition of these two groups. Data on 'husband-wife households with additional persons' combine information for both those with and without children. 'Other households' may include a number of situations.10 Given the greater lack of homogeneity in the household circumstances captured by these two groups, the related data were considered to be too difficult to interpret.
Using the FAMEX database to analyze consumer spending
Significant differences exist between Statistics Canada's FAMEX and SHS.11 Adjustments were made by OCA to the 1982 FAMEX data, in an attempt to produce a longer time-series from the two data sets. An overview of the changes is presented below. The limitations of the reconciliation, as well as other sources of discrepancies inherent to the time-series, are also identified. Finally, a general comparison is provided for the composition of the average household expenditure, using the adjusted FAMEX data and the 2002 SHS data.
OCA's adjustments to 1982 FAMEX data
The most significant difference between the FAMEX and SHS affects the important category of shelter. The FAMEX includes only the interest portion of a mortgage payment as an expenditure, while the principal portion is attributed to a 'net change in assets' variable. The SHS includes both the interest and principal component of regular mortgage payments under shelter expenditures. The FAMEX shelter definition was therefore more precise in differentiating between 'pure' household expenditures and spending that resembles an investment. Given Chapter 9's primary concern about where Canadians' money goes, we have used the SHS definition for an overview (1997-2002) of spending patterns.
OCA therefore attempted to adjust the FAMEX shelter data to more closely match it to the SHS definition. The net 'change in principal of mortgage' was removed from 'net change in assets' in FAMEX and transferred into shelter expenditures. This remains an imperfect reconciliation, however, as the net change accounts for more than the principal portion of regular payments. It adds irregular and lump sum payments, net of any money borrowed on the mortgage during the year (the latter, however, was likely small given the high interest rates in 1982). Information using the adjusted data is presented below.
Other adjustments to the FAMEX data involved relatively less significant changes and categories of expenditures. More specifically:
- Complete re-roofing of a house is deleted from shelter expenditures in the FAMEX, as the SHS now includes it under money flows — investments in the home.12
- A gross games of chance category is created in the FAMEX with the lotteries and tickets information available under miscellaneous expenditures.13
- Interest paid on personal loans and mortgages for other property is deleted from the miscellaneous category in the FAMEX, as it is now part of money flows in the SHS.
- Other (non-monetary) gifts are deleted from the FAMEX gifts and contributions, given that only money gifts are included in the SHS.14
Other limitations of comparisons over time
Independent of the above adjustments, a number of differences remain between the surveys over time. First, a reduction in the number of survey questions limits analysis of spending to the more summary-level data found in the SHS detailed tables. While containing fewer survey questions, however, the SHS uses a larger sample size (for example, the 1997 SHS is about 50 percent larger than the 1996 FAMEX), which can affect the survey's degree of accuracy. While similar questions were used throughout the FAMEX and SHS, variations in actual wording may, however, still influence the results. Furthermore, new products and services have been introduced at various points in time, such as cellular services in 1996. Finally, products may be qualitatively very different today, such as in the case of computer hardware and home entertainment equipment.
A broad comparison of adjusted 1982 FAMEX and 2002 SHS data
At present, an adjusted FAMEX/SHS time-series is not available from Statistics Canada. OCA's adjustments to 1982 FAMEX data represent an alternative to address the main differences between the two surveys, with the variables that are currently available. This information is presented as a way to fill the existing data gap and to provide general indications of how today's household spending varies from 20 years ago. As outlined above, however, the reconciliation between OCA's adjusted 1982 FAMEX data and the unadjusted 2002 data is imperfect.
Based on the adjusted FAMEX data, the average Canadian household spent $27 402 in 1982, compared to $60 090 in 2002. This corresponds to an average annual growth rate of 4.0 percent, slightly higher than the 3.8 percent annual growth from 1997 to 2002.
In terms of the composition of the average household expenditure, one change compared to 1982 is that basic expenditures — food, shelter and clothing — now represent a smaller proportion of the average household's spending (see Figure A.3). Price is one factor behind these trends as, for example, the food and clothing price indices rose relatively slowly over the period. In contrast, personal taxes represent a higher proportion of average household expenditures, and have now surpassed shelter. Transportation expenditures have similarly increased in importance.
|Sources: Statistics Canada, FAMEX (1982, adjusted) and SHS (2002).|
|Personal Insurance Payments and Pension Contributions||4.2||5.7|
|Household Furnishings and Equipment||3.5||3.0|
|Tobacco Products and Alcoholic Beverages||3.3||2.5|
|Gifts of Money and Contributions||2.1||2.4|
|Games of Chance||0.4||0.5|
|Reading Materials and Other Printed Matter||0.6||0.5|
1 'For example, it is fairly easy to monitor changes in bus ticket prices. But, how would you attach a dollar value to the changes in the frequency or punctuality of the bus service? A change in the quality of that service may well have contributed towards a change in the bus ticket prices.' See Statistics Canada 1996, 7. Back to text
2 Statistics Canada publishes both before– and after-tax LICOs. When selecting a measure, Statistics Canada notes that one consideration may be that goods and services are purchased with after-tax dollars. After-tax LICOs were used in this study. Back to text
5 At the time of publication, Statistics Canada was in the process of preparing a research paper dealing with the issue of using CPI information as a deflator for household spending data. Any subsequent proposal from Statistics Canada will need to be taken into consideration in future consumer research that uses the spending data. Back to text
6 Complete conversion of all data into constant dollars was not attempted in light of various differences between the CPI components and the SHS spending categories. In certain instances, however, Chapter 9 provides CPI information as a general indication of price trends. Back to text
8 Statistics Canada's standard spending table by household type does not provide data for non-elderly one-person households. Information on this non-elderly group can therefore only be inferred by comparing spending trends for all one-person households to that of senior one-person households. Back to text
9 ''Additional persons' include sons, daughters and foster children whose marital status is other than 'single, never-married,' other relatives by birth or marriage, and unrelated persons.' See Statistics Canada 2003, 24. Back to text
10 'Other households may be broken down into households composed of relatives only and households having at least one unrelated person. Relatives may include: sons, daughters and foster children of the reference person whose marital status is other than 'single, never-married'; relatives of the reference person by birth or marriage (not spouse, son, daughter, or foster child); spouse of the reference person who was not present in the household on December 31, 2002; other households having at least one household member who is unrelated to the reference person (e.g. lodger, roommate, employee).' See Statistics Canada 2003, 24. Back to text
13 Differences remain, however, as the reconstructed FAMEX variable still does not account for bingos and casinos (found in the SHS). In addition, the FAMEX reports gross expenses while the SHS reports net expenses (i.e. reduces it by the amount of money won). Back to text
14 This is an imperfect reconciliation, however, given that the non-monetary gifts are still included in the SHS, under their respective categories (e.g. clothing gifts are part of clothing expenditures). Back to text