ERI Discussion Paper Series No.53
The Long-Run Demand for Money
Canada Savings Bonds and Treasury Bills in Canada

May 1991
Kim McPhail
(Senior Analyst, Department of Monetary and Financial Analysis, Bank of Canada)

(Introduction)

This paper describes the result of research assessing the long-term relationship between various financial aggregates and the major economic variables believed to be relevant to their demand. Three assets are considered:(1)a broad monetary aggregate (M2P)1, (2) the outstanding stock of Canada Savings Bonds (CSB's)2 and (3) holdings of Treasury Bills in the hands of the public (T-Bills)3.

Our major interest is in the behaviour of the monetary aggregate, M2P.Previous work4 has shown, however, that Canada Savings Bonds are strongly substitutable with M2P.Hence,in modeling M2P,this substitutability needs to be taken into account. In addition, in recent years T-Bills have grown to represent a sizeable component of individuals' portfolios. These too may show significant substitutability with M2P.

It has been difficult to demonstrate that there is a stable long-run relationship between the levels of a broad monetary aggregate, real income, prices, and interest rates -- in the sense that money demand is cointegrated -- unless a dummy variable allowing for some shift in the level or trend of the aggregate is included for the early 1980's. It has been difficult to find an economic explanation for this break that is statistically supported by the data.

Taking account of shifts between money and CSB's by including CSB's as a regressor in the money demand equation does not alter these findings as, in this augmented specification, there is also little evidence in favour of cointegration5.

This study extends previous research by considering whether the combined aggregates M2P+CSB and M2P+CSB+T-Bills exhibit cointegration. If we find that M2P+CSB's is cointegrated (with prices, real income, and interest rates), for example, then the lack of evidence in the previous studies may be due to the different long-run functional relationship between M2P and CSB's in the two cases or to the higher power of the test in the present case (because of the smaller dimension of the system of variables). If, instead, we find that M2P+CSB+T-Bills is cointegrated but M2P+CSB's is not, this would suggest that portfolio shifts between T-Bills and the other two assets may be an important factor in accounting for long-term trends in M2P.

We examine this issue with the residual-based tests for cointegration that have been used in previous Canadian studies. One pervasive finding is that evidence in favour of cointegration varies according to the measure of prices used -- the GDP deflator provides much less evidence for cointegration than the CPI.

The econometric tests used in this study do not appear to have sufficient power to distinguish clearly between the hypotheses that M2P, M2P+CSB, and M2P+CSB+T-Bills, in constant dollars, have long-run trends accounted for by movements in real income, interest rates, and possibly inflation. There is some evidence that each of these three aggregates (deflated by the CPI) is part of a cointegration vector. The strongest and most robust evidence of such a stable relationship is for the combined aggregate M2P+CSB+T-Bills.


1.M2P includes currency, demand, savings, and notice deposits at banks and near-banks and term deposits held by the household sector.

2.Canada Savings Bonds are non-transferable long-term bonds issued by the Government of Canada to individuals. Their characteristics have changed considerably over the years with respect to how and when interest on the bonds is paid, the maximum limit that any individual is allowed to hold, and how many bonds are issued each year. The government in general announces a "sales period" which, recently has extended from October to early November (the government reserves the right to cut short the length of the sales period however). During much of the remainder of the year, CSB's can be redeemed with interest.

3.This measure of Treasury Bills represents total outstanding issues of Government of Canada Treasury Bills excluding those held by the Bank of Canada, any Government of Canada accounts, chartered banks (including their mortgage loan and investment dealer subsidiaries), and trust and other mortgage loan companies.

4.See McPhail and Caramazza (MC 1989), Caramazza, Hostland, and Poloz (CHP 1989), and Caramazza, Hostland and McPhail (CHM 1990).

5.CHP (1990)


Structure of the whole text

  1. full textI will open in a new window.(PDF-Format 980 KB)
  2. page1
    I. Introduction and Summary of Conclusions
  3. page2
    II. The Policy Enviroment and Recent Estimates of Broad Money Demand Function in Canada
  4. page3
    III. The Econometric Issue : Cointegration
  5. page3
    IV. Description of the Data
  6. page5
    V. Unit Root Tests
  7. page5
    VI. Tests for Cointegration
  8. page8
    VII. Sub-samole Regressions
  9. page8
    VIII.Conclusions and Implications for Modelling the Dynamics of M2P Demand
  10. page10
    Bibliography
  11. page11
    Appendix
  • 1-6-1 Nagata-cho, Chiyoda-ku, Tokyo 100-8914, Japan.
    Tel: +81-3-5253-2111