Abstract
In this study, we utilise a time-varying parameter vector autoregressive model to investigate the relationship between central bank liabilities and the overnight policy rate across three countries with distinct monetary policy frameworks. Our findings reveal that a consistent negative correlation between these variables is present only in the case of the conventional reserve regime. For the other regimes, an initial significant negative relationship between reserves and interest rates is observed, but this effect diminishes with the adoption of new frameworks. These results suggest that the emergence of novel operational structures for central banks has disrupted the traditional approach to monetary policy implementation. One practical consequence of this” decoupling” between interest rates and reserves is that central banks in the United States, Canada, and similar systems can effectively employ their balance sheets in conjunction with standard interest rate policies. However, this does not apply to reserve regimes that continue to depend on open market operations for steering interest rates
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Which is determined by the Monetary Policy Committee (MPC).
2 Factors that central banks retain no control over.
3 Factors that central banks actively manage.
4 These deposits are semi-autonomous.
5 Non-borrowed reserves generally are calculated as total reserves minus borrowed reserves. Borrowed reserves are found on the asset side of the balance sheet, normally referred to as loans, borrowing, liquidity provided, etc. In the case of South Africa borrowed reserves were defined as the total liquidity provided (i.e., money market shortage).
6 Especially after 1951, when the Fed gained its independence from the government with the Federal Reserve Accord.
7 Movements from this band were corrected through market intervention.
8 Also known as the Great Inflation.
9 In this study Canada was chosen as there is an existing body of literature on the liquidity effect. In addition, it is one of few countries that has employed a corridor regime for an extended time period and therefore concerns over data availability are reduced
10 Countries with a corridor system and reserve requirement are also referred to as hybrid regimes
11 It was set at 25 basis points above the average yield on three-month treasury bills at the Federal government’s weekly auction.
12 The Bank rate forms the ceiling in this corridor system, 25 basis points above the target overnight rate.
13 The growth rate (first difference) of the liquidity measures was not always used, but was used in the cases where the inclusion of the variable in level terms resulted in an unstable VAR.
14 Data for the United States and South Africa on NBR were unfortunately discontinued and further extension was not possible. Any attempt at reconstruction with new data would deliver results that are not directly comparable to the original data.
15 As presented in Nakajima (Citation2011).
16 Several plausible alternative orderings were tried with no reversal of sign or significance encountered.
17 For a detailed discussion on prior selection in TVP-VAR models, refer to the monograph of Koop and Korobilis (Citation2009).
18 It should be noted that confidence intervals for the IRFs are not included in the analysis. The primary reason for this exclusion is that the software implementation utilised for the analysis does not allow the construction of intervals. An attempt at engineering these intervals in this software implementation might deliver erroneous results and is therefore not presented. To gain insight into the significance of the IRFs, one could look at the results from the SVARs explored in the previous section. In addition, IRFs for TVP-VAR analysis are often excluded to improve the legibility of the graphs. Examples of this includes Galí and Gambetti (Citation2009) and Nakajima (Citation2011).