1. The personal saving rate in Canada, as measured in the national accounts, generally increased during the 1960s, 1970s, and early 1980s, reaching a peak of close to 19 percent of disposable personal income in 1982 (Chart 1). Subsequently, it has fallen sharply, declining to 2.3 percent in the first three quarters of 1997. This paper examines some of the factors that help to explain the decline in the personal saving rate and tries to assess the significance of this decline. In particular, it notes the shortcomings in the national income accounts measure of personal saving and looks at alternative measures that can be derived from the national balance sheet accounts. The paper concludes that the sharp downward trend in the national accounts measure of the personal savings rate since 1982 is largely related to definitional problems in measuring personal savings (in particular, reflecting the treatment of consumer durables and the exclusion of changes in the value of financial assets in the national accounts measure) and the effect on nominal saving of the decline in inflation (nominal personal income and saving tend to be positively related to inflation). Adjusting for definitional differences, the personal savings rate in real terms has been largely trendless.

Abstract

1. The personal saving rate in Canada, as measured in the national accounts, generally increased during the 1960s, 1970s, and early 1980s, reaching a peak of close to 19 percent of disposable personal income in 1982 (Chart 1). Subsequently, it has fallen sharply, declining to 2.3 percent in the first three quarters of 1997. This paper examines some of the factors that help to explain the decline in the personal saving rate and tries to assess the significance of this decline. In particular, it notes the shortcomings in the national income accounts measure of personal saving and looks at alternative measures that can be derived from the national balance sheet accounts. The paper concludes that the sharp downward trend in the national accounts measure of the personal savings rate since 1982 is largely related to definitional problems in measuring personal savings (in particular, reflecting the treatment of consumer durables and the exclusion of changes in the value of financial assets in the national accounts measure) and the effect on nominal saving of the decline in inflation (nominal personal income and saving tend to be positively related to inflation). Adjusting for definitional differences, the personal savings rate in real terms has been largely trendless.

II. Recent Trends in Personal Saving in Canada1

1. The personal saving rate in Canada, as measured in the national accounts, generally increased during the 1960s, 1970s, and early 1980s, reaching a peak of close to 19 percent of disposable personal income in 1982 (Chart 1). Subsequently, it has fallen sharply, declining to 2.3 percent in the first three quarters of 1997. This paper examines some of the factors that help to explain the decline in the personal saving rate and tries to assess the significance of this decline. In particular, it notes the shortcomings in the national income accounts measure of personal saving and looks at alternative measures that can be derived from the national balance sheet accounts. The paper concludes that the sharp downward trend in the national accounts measure of the personal savings rate since 1982 is largely related to definitional problems in measuring personal savings (in particular, reflecting the treatment of consumer durables and the exclusion of changes in the value of financial assets in the national accounts measure) and the effect on nominal saving of the decline in inflation (nominal personal income and saving tend to be positively related to inflation). Adjusting for definitional differences, the personal savings rate in real terms has been largely trendless.

CHART 1
CHART 1

CANADA: NATIONAL ACCOUNTS PERSONAL SAVING RATE

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A002

Source: Statistics Canada.

A. Measuring the Personal Saving Rate

2. Theoretically, a saving rate should measure the fraction of income not consumed and thus available for future consumption. For the personal sector, disposable income should include all earnings from labor (including wages, salaries, and employer’s social insurance contributions), returns on prior saving (rental income, net interest, dividends, and capital gains), and government transfers less taxes. In addition, income should include real returns on the personal sector’s assets and not nominal returns on these assets, since the difference between real and nominal asset yields represents an inflation premium to compensate households for losses in the value of their assets owing to inflation. This premium does not provide additional resources for future real consumption.

3. In Canada’s national income accounts, personal saving is measured by subtracting personal consumption and current transfers from persons to corporations and to nonresidents from personal disposable income. The personal saving rate is then derived as the ratio of personal saving to personal disposable income.2

4. Measured in this way, the personal saving rate inadequately accounts for inflation in part because nominal, rather than real, asset returns are included as part of income. In addition, inflation could influence the behavior of personal saving. Current inflation could induce households to consume less and save more because of uncertainty about relative prices,3 and uncertainty about future inflation could induce households to save more in order to hedge against the possible future erosion of real wealth.4 In fact, the national income accounts measure of the personal saving rate is positively correlated with consumer price inflation (Chart 2),5 reflecting the inclusion of the inflation premium in returns to personal sector assets and the uncertainty about current and future inflation.

CHART 2
CHART 2

CANADA: NATIONAL ACCOUNTS PERSONAL SAVING RATE AND INFLATION

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A002

Source: Statistics Canada.

5. The national accounts measure of personal saving also does not include both realized and unrealized capital gains on the assets of the household sector. To the extent that capital gains are due to fundamental changes in the real value of assets (as opposed to speculative bubbles), they should be included in the measure of saving. The exclusion of capital gains may be a significant factor in explaining the decline in the national accounts measure of the personal saving rate because households have significantly increased their ownership of stocks and mutual funds in recent years and these assets have increased sharply in value.

6. Errors in the national accounts measure of personal saving may also result from the fact that important data for the household sector tend to be estimated as residuals from the activities of other sectors that are easier to measure. In addition, the unincorporated business sector—which includes private nonprofit institutions, trusteed pension plans, and the investment activities of life insurance companies—are included in the personal sector.6

7. Other shortcomings of the national accounts measure of personal saving are that it excludes saving in the form of consumer durables and investment in human capital, and does not factor in the depreciation of physical assets.7 Consumer durables are goods, such as motor vehicles or major appliances, which have expected lifetimes of more than one year, and provide consumption services beyond the year in which they are acquired. The national accounts measure treats these durable goods as consumption in the year that they are purchased, and thereby, would tend to overstate consumption in that year and understate a “true” measure of saving. Investments in human capital (for example, outlays for education) may increase the future earning potential of households, and including these expenditures as consumption in the national accounts saving measure also may understate “true” saving. In contrast, the depreciation of physical assets reduces the future returns from these assets, and the exclusion of this depreciation from estimates of personal consumption tends to overstate saving.

8. An alternative measure of personal saving, which addresses most of the shortcomings in the national accounts measure, can be derived from the national balance sheet accounts data, essentially by defining saving as the change in the real net worth of the personal sector.8 This measure of personal saving includes capital gains and purchases of durable goods less the depreciation of the stock of consumer durables and excludes the inflation premium from asset returns. The personal saving rate derived on this basis (referred to below as the real net worth personal saving rate) is calculated by dividing this measure of saving by a disposable real income term defined as the real net worth measure of personal saving plus real personal consumption of nondurable goods and services plus the real depreciation of durable goods.

9. Chart 3 shows the trends in the real net worth personal saving rate and the national accounts personal saving rate. The real net worth personal saving rate is relatively more volatile than the national accounts personal saving rate, with a peak of 30 percent in 1973 and a trough of below -10 percent in 1982; however, this measure displays little or no trend. In 1996, this saving rate was about 12 percent, compared with its average of about 15 percent over the period since 1961. It is interesting to note that this saving rate reached its low point in the same year that the national accounts measure peaked. In other words, analysis based on the real net worth saving rate presents a very different picture than one based on the national accounts measure. The two measures are, in fact, negatively correlated (Table 1).

CHART 3
CHART 3

CANADA: NATIONAL ACCOUNTS PERSONAL SAVING RATE VS. REAL NET WORTH SAVING RATE

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A002

Sources: Statistics Canada and Fund staff estimates.
Table 1.

Correlation Matrix of Different Measures of Personal Saving Rates

article image
Source: Fund staff estimates.

Net investment in consumer durables is defined as expenditures less the depreciation on the stock of consumer durables.

Derived from the national balance sheet accounts, expressed in constant 1992 dollars. Differences from national accounts measure include net investment in consumer durables and changes in the value of financial assets.

10. To examine the differences between the two measures, the national accounts personal saving rate was adjusted to reflect definitional differences, and nominal and real national accounts personal saving rates including net investment in durable goods (consumer durable purchases less depreciation on the stock of consumer durables) were derived (Chart 4). With these adjustments, the real national accounts personal saving rate is positively correlated with the real net worth saving rate (see Table 1).9 However, if only the adjustment is made for net investment in consumer durables, then the national accounts saving rate is still negatively correlated with the real net worth saving rate, suggesting that a major difference in the national accounts personal saving rate and the real net worth saving rate appears to reflect the failure to adjust for inflation.

CHART 4
CHART 4

CANADA: DIFFERENT MEASURES OF PERSONAL SAVING RATES

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A002

Sources: Statistics Canada and Fund staff estimates.

B. Factors Affecting the Personal Saving Rate

11. The permanent income hypothesis suggests that households will consume constant fractions of their permanent income, and fluctuations in savings should reflect transitory fluctuations in income. The effect of interest rate changes on saving is ambiguous because of offsetting income and substitution effects. To test these hypotheses, an equation relating the real net worth saving rate (Δnw) to the percentage change in real disposable income per capita (%ΔY) and the real three-month interest rate (r) was estimated for the period 1963 to 1996.10 Lagged values of the dependent variable also were included on the assumption that households adjust their savings only gradually to shocks. The results are summarized below and indicate that changes in real disposable income per capita almost completely explain the behavior of the real net worth saving rate, with the interest rate being insignificant :11

Δnw=0.922(0.99)+0.834(18.39)Δnw(1)0.019r(0.19)+0.82%(23.67)ΔY
article image

Other explanatory variables, in particular the rate of inflation and the dependency ratio (defined as working-age population divided by nonworking-age population), are insignificant when %ΔY and lagged values of the dependent variable are included in the various regressions; the lagged value of %ΔY also was insignificant. The Chow test shows evidence of parameter instability, as it does not reject a structural break in the data around 1981. When added to the equation, a trend term is significant and positive, suggesting a slightly rising real net worth saving rate over time, in contrast to the downward trend in the national accounts measure of the personal saving rate.

12. A similar equation was estimated for the real national accounts personal saving rate including net investment in consumer durables:

ΔsRDur=2.599(1.97)+0.641Δ(6.98)sRDur(1)+0.115r(0.75)+0.541%ΔY(9.90)
article image

As with the real net worth saving rate equation, when %ΔY and lagged values of the dependent variable are included, other variables (including inflation, the dependency ratio, and lagged %ΔY) are insignificant. In contrast to the real net worth saving rate equation, the trend term is insignificant, although a structural break around 1981 cannot be rejected.

13. In sum, these relationships suggest that when saving rate measures are adjusted to remove the inflation premium on asset returns, the movements in saving are well explained by changes in real disposable income per capita (Chart 5). Inflation has no separate explanatory power, and behavioral changes related to the effects of current and future inflation may not be significant. The interest rate is found to be insignificant which may imply that income effects offset substitution effects. Furthermore, when changes in real per capita disposable income are taken into account, the real saving rate measures have either no trend or a slight positive trend.

CHART 5
CHART 5

CANADA: REAL NET WORTH SAVING RATE AND PER CAPITA DISPOSABLE INCOME

(In percent)

Citation: IMF Staff Country Reports 1998, 055; 10.5089/9781451806915.002.A002

Sources: Statistics Canada and Fund staff estimates.
Table 2.

Canada: Stationarity Tests 1/ 2/

article image
Source: Fund staff estimates.

Number of lags in parenthesis. Optimal lag length is determined by the Akaike information criterion.

Augmented Dicky-Fuller tests were also run including a trend with similar results (not reported).

* Hypothesis of a unit root is rejected at a 5 percent level of significance

List of References

  • Clift, Barbara, 1988, “Components of Personal Saving,” Canadian Economic Observer.

  • Dagenais, Macel G., 1992, “Measuring Personal Savings, Consumption, and Disposable Income in Canada,” Canadian Journal of Economics, Vol. 25, pp. 681707.

    • Search Google Scholar
    • Export Citation
  • Davidson, Russell and James G. MacKinnon, 1983, “Inflation and the Savings Rate,” Applied Economics, Vol. 15, pp. 731743.

  • Deaton, Angus S., 1977, “Involuntary Saving through Unanticipated Inflation,” The American Economic Review, Vol. 67, pp. 899910.

  • Howard, David H., 1978, “Personal Saving Behavior and the Rate of Inflation,” The Review of Economics and Statistics, Vol. 6, pp. 547554.

    • Search Google Scholar
    • Export Citation
  • Jump, G.V., 1980, “Interest Rates, Inflation Expectations, and Spurious Elements in Measured Real Income and Savings,” The American Economic Review, Vol. 70, pp. 9901004.

    • Search Google Scholar
    • Export Citation
  • Lipsey, Robert E. and Irving B. Kravis, 1987, “Is the U.S. a Spendthrift Nation?” NBER Working Paper No. 2274 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Randall, John D., 1989, A User Guide to the Canadian System of National Accounts, (Ottawa: Statistics Canada).

  • Siegal, J.J, 1979, “Inflation-induced Distortions in Government and Private Savings Statistics,” The Review of Economics and Statistics, Vol. 61, pp. 1831.

    • Search Google Scholar
    • Export Citation
1

Prepared by Ranil Salgado and Yutong Li.

2

The guidelines for the System of National Accounts (United Nations, 1993) suggest that personal saving be calculated as personal disposable income less personal consumption. In contrast to the Canadian national accounts, the guidelines suggest that current transfers from persons to corporations and to nonresidents should not be included in the measure of personal disposable income. Adjusting the Canadian personal saving rate to the one suggested by the 1993 System of National Accounts has only a minor effect (less than a tenth of a percentage point in 1996, for example) on the personal saving rate, and the adjusted measure exhibits a decline similar to the measure derived from Canada’s national income accounts. Current transfers from persons to corporations are defined as the interest paid on consumer debt less the administrative expense of lending to consumers, and current transfers to nonresidents include private remittances and withholding taxes paid abroad.

4

Empirically, the level and variance of inflation are positively related.

5

See, for example, Howard (1978), Siegal (1979), Jump (1980), and Davidson and MacKinnon (1983) for more discussion.

6

Clift (1988) suggests, however, that saving from these institutions is relatively small and has little impact on overall personal saving trends.

7

See Lipsey and Kravis (1987) for more discussion.

8

As with the national accounts, personal sector data in the national balance sheet accounts are estimated as a residual from the activities of other sectors. In addition, investment in human capital is not included in this measure of saving.

9

Other differences between the national accounts saving rate and the real net worth saving rate include a different treatment of capital gains and of depreciation of real assets (as noted above).

10

Table 2 shows the results of unit root tests for the stationarity of various measures of saving and potential explanatory variables. The real measures of the saving rate were found to be stationary (unit roots tests are rejected), while the nominal measures are not (unit roots tests are not rejected). The stationarity of the real measures suggests that ordinary least squares can be used to estimate equations for these variables.

11

The absolute values of the t-statistics are shown in the parentheses. The LM(p) statistic tests for autocorrelation and is asymptotically distributed χ2(p). The ARCH(p) statistic tests for autoregressive conditional heteroskedasticity and is asymptotically distributed χ2(p). The Chow statistic tests for a structural break in 1981 and is distributed F(4,26); an “*” signifies that the statistic is significant at 5 percent level.

  • Collapse
  • Expand
Canada: Selected Issues
Author:
International Monetary Fund
  • CHART 1

    CANADA: NATIONAL ACCOUNTS PERSONAL SAVING RATE

    (In percent)

  • CHART 2

    CANADA: NATIONAL ACCOUNTS PERSONAL SAVING RATE AND INFLATION

    (In percent)

  • CHART 3

    CANADA: NATIONAL ACCOUNTS PERSONAL SAVING RATE VS. REAL NET WORTH SAVING RATE

    (In percent)

  • CHART 4

    CANADA: DIFFERENT MEASURES OF PERSONAL SAVING RATES

    (In percent)

  • CHART 5

    CANADA: REAL NET WORTH SAVING RATE AND PER CAPITA DISPOSABLE INCOME

    (In percent)