Publications and Research

Summer 2005

Changes in the Indicator Properties of Narrow Monetary Aggregates

Tracy Chan, Ramdane Djoudad, and Jackson Loi, Department of Monetary and Financial Analysis

• Past research has shown that, compared with other monetary aggregates and expressed in real terms, net M1 and gross M1 have traditionally provided superior leading information for output growth.
• Financial innovations and the removal of reserve requirements have made it increasingly difficult to differentiate between demand and notice deposits. This suggests the need to re-examine the information content of narrow monetary aggregates (such as net M1 and gross M1) that depend on this distinction.
• Evidence examined in this article shows that, since 1993, real M1+^* has become a better indicator of future output growth than real gross and net M1.

While many countries have abandoned monetary targeting¹ over the past two decades, monetary aggregates are still useful indicators of future economic activity. This is true even though growth in these aggregates has at times been affected by shifts in the demand for money. As suggested in Longworth (2003), there are several reasons to believe money can provide leading information for output growth, including its role in the transmission of monetary policy. In Canada, the relationship with output growth is shown in the literature to be the strongest for narrow monetary aggregates (Hostland, Poloz, and Storer 1987; Muller 1992; Maclean 2001; Siklos and Burton 2001; Hassapis 2003). However, some authors have found that the link between real economic activity and monetary aggregates has weakened over the past two decades (Siklos and Burton 2001).

At the Bank of Canada, narrow monetary aggregates expressed in real terms (i.e., deflated by a price index) continue to be monitored to assess their information content for real output. Past studies have found that narrow monetary aggregates, particularly real net M1 and gross M1, contain explanatory power for real output growth one to two quarters ahead. But no study compares how the leading-indicator properties of various narrow aggregates (net M1, gross M1, M1+, and M1++) for output growth have evolved over the recent period. (See Box 1 for definitions of narrow monetary aggregates.)

Financial innovations in banking products over the years have made it increasingly difficult to differentiate between demand and notice deposit accounts. For example, both types of account now offer similar interest rates and comparable accessibility to funds. The elimination, between 1992 and 1994, of reserve requirements on all bank accounts in Canada has removed the need for banks to discriminate between demand and notice deposit accounts (Aubry and Nott 2000).² As a result, the classification of accounts by financial institutions between demand or notice deposits has become increasingly arbitrary.

The blurred distinction between the two types of deposit raises questions about the value of those monetary aggregates whose very definition is based on such a distinction. Specifically, M1 and GM1, which include currency and demand deposit accounts, are directly affected by this classification issue. The broader measures of narrow money, namely M1+ and M1++, capture both demand and notice deposits and, hence, should not be affected. Since this classification has become somewhat artificial, it is possible that the narrower aggregates (GM1 and M1) no longer contain superior information to that of M1+ and M1++. It is therefore interesting to compare the various narrow monetary aggregates with respect to their properties as leading indicators for output growth.

Creation of the Narrow Monetary Aggregates in Canada

There are many ways to aggregate various financial assets and money stocks to represent the supply of money. Economists generally aggregate money using two approaches (Laidler 1969). The first approach is to group those monetary assets that most closely represent some underlying definition of money, such as a medium of exchange or a store of value. The second approach is to define money as an aggregation of financial assets that have the most significant empirical relationship with certain macroeconomic variables, such as real output and inflation. However, no single method of monetary aggregation has been universally accepted, because there is no simple "one size fits all" approach to deal with the numerous economic concepts of money (Laidler 1999). As White (1976, 49) remarked, "the answer to . . . the related choice between alternative money definitions [is] based on the usefulness of the various aggregates for policy purposes."

The Bank of Canada began publishing monthly data for monetary components well before 1970. It was not until the 1970s, however, that the monetary aggregate M1 was reported. During the 1980s, the Bank also began reporting M1A, which is defined as the sum of M1 plus daily-interest chequing accounts and nonpersonal notice deposits. This aggregation comprised the most liquid monetary accounts and was intended to represent money for transactions purposes and purchasing power.

Financial Innovations and Money Distortions

In the past 20 years, financial innovations have played a significant role in the way economic agents have managed their money and financial assets. These innovations have caused important shifts among the monetary accounts, ultimately blurring the distinction between the various narrow monetary aggregates. The first wave of innovations in banking products, which took place from 1978 to 1986, significantly reduced the demand for M1 in both the corporate and household sectors in Canada (Aubry and Nott 2000). On the corporate side, a number of new cash-management packages allowed businesses to consolidate several accounts into one centralized account. As a result, firms were able to reduce their total working cash balances. For households, the introduction of daily-interest savings accounts (chequable and non-chequable) boosted incentives to deposit and transfer money into these accounts, which were not included in the measurement of M1 because they were unlikely to have been used for transactions purposes before the adoption of such financial innovations. Throughout this period, new financial products introduced by deposit-taking institutions continued to offer households and firms increasing flexibility in the type of account in which to hold deposits.

The second major wave of financial innovations began around 1993. Mutual fund products gained popularity relative to notice deposits as a saving vehicle, and free credit balances (cash or margin accounts intended for trading financial assets) grew rapidly. More importantly, as mentioned earlier, the removal of reserve requirements in the mid-1990s eliminated the need for banks to differentiate between demand (transactions) and notice (savings) deposits for reserve purposes. Indeed, many banks can no longer distinguish "demand" deposits from some types of notice deposit. As well, interest payments on some types of demand deposit became more common. In addition, the innovations in business accounts also made a significant contribution to the boost in the growth of GM1. A sizable share of GM1 was thus allotted to the sale and purchase of financial assets rather than to transactions for purchasing goods and services (Aubry and Nott 2000). Lastly, the development of Internet banking during the late 1990s enabled bank clients to easily transfer money between non-savings and savings accounts. This allows bank clients to deposit money in accounts that yield higher interest, while still being able to transfer money for transactions purposes without first having to give notice to the bank.

Towards M1+ and M1++

Thus, over the years, it has become increasingly difficult to differentiate between money held for transactions purposes and money held as savings. This has ultimately led to concerns about whether M1 and GM1 are adequate measures of transactions balances. Financial institutions are also experiencing difficulties in classifying and reporting their deposit accounts as either demand or notice, raising concerns about the quality of M1 and GM1 data. In an effort to capture a broader notion of transactions money and to internalize the shifts occurring in some of the components, two alternative measures of narrow money, M1+ and M1++, have been published and monitored by the Bank since 1999. M1+ and M1++ are not affected by the distinction between demand and notice deposits because they incorporate both account categories. As such, they capture the components related to transactions purposes, as well as to savings purposes. For all of these reasons, the Bank of Canada has been motivated to explore new ways to define measures of transactions money (Gilbert and Pichette 2003).

Evolution of the Information Content of Narrow Monetary Aggregates

It has been generally determined that the growth of narrow money tends to precede growth in real output. Early research has verified the significance of this relationship over long historical samples (Hostland, Poloz, and Storer 1987; Muller 1992). Given the changes in the financial and regulatory environment over the 1990s, it is essential to examine how this relationship between narrow money and output has evolved over time.

Charts 1 and 2, which are similar to a chart published in the Bank of Canada's semi-annual Monetary Policy Report,³ plot the quarterly growth of real gross domestic product (GDP) and the two-quarter moving average⁴ of the growth of various real narrow monetary aggregates (lagged one quarter). The charts suggest that movements in the real monetary aggregates have usually preceded movements in real output growth, indicating that the movements in money growth have some leading information for future output growth. In the literature, this lag effect is traditionally shown to be the strongest between output growth and the growth of GM1 and M1.

To quantitatively assess how this lead-lag relationship has evolved over time, a simple empirical exercise is performed to calculate the rolling correlations between the lagged two-quarter moving average of real narrow money growth and real output growth. The total sample is derived from the period 1975Q1 to 2005Q1. A 10-year correlation for the period 1975Q4 to 1985Q3⁵ is calculated for each of the combinations considered (GM1, GDP), (M1+, GDP), and (M1++, GDP). The start and end dates are then rolled forward (1976Q1 to 1985Q4), and the 10-year correlations are calculated again. The start and end dates continue to be rolled forward, and the same exercise is performed until 2005Q1. For simplicity, the results using real M1 are not presented, since they are broadly consistent with those using real GM1.

Chart 3 shows the results of the rolling exercise for the 10-year correlations between output growth and the lagged two-quarter moving average of real money growth.⁶ The following conclusions can be drawn from this chart:

1. Over the period 1985 to 1996, the correlations using real GM1 were generally higher than those using real M1+ and real M1++. Over the period 2000 to 2005, however, there has been a clear deterioration in the correlations using real GM1. In the more recent period, the correlation using real GM1 has fallen to about 0.30, which is close to the lowest value over the entire sample.
2. Since 2000, the correlations using real M1+ outperformed the ones using real GM1.
3. The correlations pertaining to real M1+ have been fairly stable over the whole sample and have generally been around 0.45, on average.

These results suggest that a shift has likely occurred in the information content of real narrow monetary aggregates for output growth. While GM1 had higher correlations over the first part of our sample period, M1+ had stronger correlations in more recent years. Thus, the information content of real GM1 has deteriorated over time, while the information coming from real M1+ has been stable.

On a more formal basis, the results described in Box 2 support this view and determine that 1992 was the year when a shift occurred.⁷ During the period from 1975 to 1991, real GM1 had better leading information for output growth. But real M1+ has become the more relevant indicator since 1993. This new regime is likely to persist, since the developments that made it difficult to distinguish between demand and notice deposits are permanent. This finding is consistent with the existence of a shift in the estimated parameters of the money demand equation that occurred over that period (Hendry 1995; Maclean 2001).

Conclusion

Financial innovations and the removal of reserve requirements have made the distinction between demand and notice deposits artificial. As a result, financial institutions are finding it increasingly difficult to allocate new accounts between these two categories. Thus, there are growing concerns that this change may have eroded the leading information of M1 and GM1 for future GDP growth. Consequently, M1 and GM1 may no longer provide more information than M1+ and M1++.

Our findings suggest that the leading-indicator properties of M1, GM1, M1+, and M1++ for GDP growth have shifted over time. Previous empirical results had suggested that real M1 and real GM1 were traditionally better indicators for future output growth. More recently, however, real M1+ has become more informative. Thus, we find evidence in favour of the existence of a regime shift in the indicator properties of narrow money for output growth. This regime change occurred in 1992 and is likely to persist.

When constructing the narrow monetary aggregates, the primary goal was to capture the supply of transactions money. Given institutional changes and financial innovations, the concept of transactions money is no longer likely to be adequately captured by GM1 or M1. We argue that the broader measure M1+ now better defines transactions money. Indeed, today there is less need for agents to carefully consider their holding of cash, since many non-term assets are easily converted into cash. This renders the distinction between demand and notice deposits less relevant for money demand.

---

Literature Cited

Aubry, J.-P. and L. Nott. 2000. "Measuring Transactions Money in a World of Financial Innovation." In Money, Monetary Policy, and Transmission Mechanisms. Proceedings of a conference held by the Bank of Canada, 3–4 November 1999. Ottawa: Bank of Canada.

Chan, T., R. Djoudad, and J. Loi. 2005. "Changes in the Indicator Properties of Narrow Monetary Aggregates." Bank of Canada Working Paper (forthcoming).

Cozier, B. 1993. "Recent Developments in Canadian Monetary Aggregates." Bank of Canada Review (Spring): 31–42.

Gilbert, P. and L. Pichette. 2003. "Dynamic Factor Analysis for Measuring Money." Bank of Canada Working Paper No. 2003–21.

Hassapis, C. 2003. "Financial Variables and Real Activity in Canada." Canadian Journal of Economics 36 (2): 421–42.

Hendry, S. 1995. "Long-Run Demand for M1." Bank of Canada Working Paper No. 95–11.

Hostland, D., S. Poloz, and P. Storer. 1987. An Analysis of the Information Content of Alternative Monetary Aggregates. Technical Report No. 48. Ottawa: Bank of Canada.

Kottaras, J. 2003. "The Construction of Continuity-Adjusted Monetary Aggregate Components." Bank of Canada Working Paper No. 2003–22.

Laidler, D. 1969. "The Definition of Money: Theoretical and Empirical Problems." Journal of Money, Credit, and Banking 1 (3): 508–25.

———. 1999. "The Quantity of Money and Monetary Policy." Bank of Canada Working Paper No. 99–5.

Longworth, D. 2003. Money in the Bank (of Canada). Bank of Canada Technical Report No. 93. Ottawa: Bank of Canada.

Maclean, D. 2001. "Analyzing the Monetary Aggregates." Bank of Canada Review (Summer): 31–43.

Muller, P. 1992. "The Information Content of Financial Aggregates during the 1980s." In Monetary Seminar. Proceedings of a conference held by the Bank of Canada, 7–9 May 1990. Ottawa: Bank of Canada.

Siklos, P. and A. Burton. 2001. "Monetary Aggregates as Indicators of Economic Activity in Canada: Empirical Evidence." Canadian Journal of Economics 34 (1): 1–17.

White, W. 1976. "The Demand for Money in Canada and the Control of Monetary Aggregates: Evidence from the Monthly Data." Bank of Canada Research Studies No. 12.