RDP 9504: The Link Between the Cash Rate and Market Interest Rates 3. The Pass-through of Changes in the Cash Rate

Changes in the cash rate affect the entire structure of interest rates in the economy. However, not all interest rates move one for one with the cash rate. Some interest rates do not move at all, while others, such as long-term bond rates, may actually move in the opposite direction. Some measure of the average degree of “pass-through” of cash rate changes to various other interest rates can be obtained by estimating the following equation:

If pass-through is instantaneous and complete, β should be insignificantly different from one. In practice, the pass-through of cash rate changes may not be instantaneous. In particular, rates widely advertised by the banks may take longer to adjust than rates determined in continuously operating auction markets. To allow for this possibility, we also estimate the following model:

The long-run response of the relevant interest rate to a change in the cash rate is given by the parameter δ.

Equation (1) is estimated by ordinary least squares and equation (2) by instrumental variables using monthly data over the period from January 1986 to October 1994.^[6] The sample period is constrained by the fact that prior to April 1985 most lending rates were the subject of regulation.^[7] We chose January 1986 as the starting point to allow for a period of adjustment to the deregulated environment. Equation (1) is estimated in both levels and differences. The results are reported in Table 1. For the autoregressive distributed lag model we simply report the estimate of the long-run response.^[8]

In discussing the extent of “pass-through” it is useful to distinguish among four classes of interest rates:

• short-term money-market rates;
• rates on long-term securities;
• deposit rates;
• lending rates.

We discuss each of these in turn. Figure 2 shows the various interest rates.

3.1 Short-Term Money-Market Rates

In general, the pass-through of the cash rate to short-term money-market interest rates is rapid and complete. Using the level of the interest rates, we typically find that the coefficient on the contemporaneous cash rate is quite close to one in both an economic and statistical sense and that the intercept is not significantly different from zero. In contrast, in the regressions using changes, the coefficients on the change in the cash rate are all significantly less than one, ranging between 0.67 and 0.88. In part, the lower coefficients in the “change” regressions reflect the fact that short rates often move prior to changes in the cash rate. When movements in the cash rate are expected, the short-term money-market rates reflect these expectations. This can be seen in Figure 3 which shows the level of the cash rate and the spread between the 90-day bank bill and the cash rate. At least since 1987, during periods when the cash rate was falling, bill rates tended to lie below the cash rate, while during periods when the cash rate was rising, bill rates tended to be higher than the cash rate.

3.2 Rates on Long-Term Securities

The relationship between the cash rate and the longer end of the yield curve is more complex. Changes in the cash rate affect the expected future level of short rates, and thus the current level of long rates, in two ways. The first channel is often referred to as the liquidity effect – lower short rates today imply lower short rates for some period into the future. As a result, the lower short rates today put downward pressure on long-term bond rates. The second channel works primarily through changing expectations of future inflation. If a lowering of the cash rate generates expectations of higher future inflation, short-term rates will be expected to increase at some point in the future. This will put upward pressure on nominal long-term bond rates that may offset the impact of lower short rates in the near term.

The second block of Table 1 reports the interest rate pass-through regressions for two, five and ten-year government bonds. The long-run pass-through coefficient and the coefficient on the cash rate in the “levels” regressions are positive for all three rates, but the size of the coefficients declines the longer is the maturity of the security. On average, over the sample period, higher cash rates have tended to be associated with higher rates on government bonds. Given that the liquidity effect is relatively more important the shorter is the maturity, it is not surprising that the rates on shorter maturities are more responsive to changes in the cash rate than is the 10-year rate.

The results using the changes in interest rates are less clear cut, particularly for the ten-year bond. The two-year bond is the only one to have a significant positive coefficient on changes in the current cash rate. The contrasting results from the levels and difference equations suggest that while long-term bonds eventually move in the same direction as the cash rate, the timing of movements in the cash rate and bond rates often differs. Indeed, the relationship between bond rates and the cash rate changes through time. During some periods, bond rates tend to move in the same direction as the cash rate, while during others, they move in the opposite direction.

Examining the fifteen reductions in interest rates that took place between January 1990 and July 1993, we see that on every occasion, except one, the ten-year bond rate on the day following the easing was lower than on the day prior to the easing. However, we also see that on a number of occasions, the ten-year bond rate was higher on the day following the easing, than it was on the day following the previous easing (see Figure 4). In particular, the reductions in the first three quarters of 1990 saw some net increase in long-term bond rates. In contrast, from the last quarter of 1990, bond rates tended to move in the same downward direction as the cash rate. To a large extent, this change in behaviour can be attributed to a progressive reduction in inflation expectations. The easings in the first three quarters of 1990 were made prior to progress on inflation being widely apparent. However, from late 1990 onwards, it became clearer that the economy was weaker than had been expected and that there had been a structural break in inflation. Accordingly, reductions in the cash rate did not lead to expectations of higher future inflation, and bond rates fell.

More recently, long-term bond rates increased by about 3 percentage points over the first half of 1994, despite no change in the cash rate. This increase reflected higher expected future short rates, as expectations of future real rates and future inflation were revised upwards. When the cash rate was increased by 2.75 percentage points between August and December 1994, long bond rates rose by a further 1 percentage point. At the end of 1994, however, long bond rates were below their levels in early October 1994, even though cash rates had risen by 2 percentage points over that period. Clearly, while cash rates and bond rates tend to move together over the longer term, expectations about future inflation can lead to very weak short-run linkages between changes in the cash rate and bond yields.

An additional factor influencing the relationship between the short and long ends of the yield curve is the behaviour of world real long-term interest rates. Australia cannot isolate itself from the effects of a change in world real long interest rates, particularly due to changes in world inflation and savings-investment patterns. Higher world real long interest rates mean higher Australian real long rates. While such a change might be expected to eventually flow through to short rates (reversing the direction of “causation” discussed above), this process may be drawn out over a considerable period of time.

3.3 Deposit Rates

We now turn to an examination of the relationship between the cash rate and deposit rates. We examine three different classes of deposits: cash management trusts, transaction – investment accounts and fixed deposits. The estimation results are presented in the third block of Table 1. The cash management trusts have the most responsive interest rates. This is hardly surprising as they invest almost entirely in short-term money-market instruments. The interest rates on these accounts behave similarly to the money-market rates, with the exception that, on average, the rates paid by cash management trusts tend to be slightly lower than the cash rate (the intercept is significantly less than zero). This reflects the management expenses of the trust.

The picture is somewhat different for the interest rates paid on the transaction – investment accounts. These rates adjust much less to changes in the cash rate than do the money-market interest rates. The interest rates paid on the low-balance accounts are particularly sticky. Competition for these accounts is relatively limited. They tend to have high operating costs and the accounts are held by depositors with relatively low interest-rate sensitivity. Competition for accounts with larger balances is more aggressive, and hence deposit rates on these accounts tend to move more closely in line with the rates on alternative investments such as cash management trusts. Nevertheless, the adjustment in rates appears to be slower than that for money-market rates. The pass-through of cash rate changes to one-month and one-year fixed deposits is similar to that for the high-balance transaction accounts, although the long-run pass-through appears smaller. The estimates of α also indicate that, on average, these fixed deposits pay a higher rate of interest than that paid on transaction accounts.

3.4 Lending Rates

Finally, we turn to the lending rates. Changes in the cash rate are not always passed through completely and immediately into lending rates. This can be seen in the final section of Table 1. It shows quite different degrees of pass-through for the four loan rates examined. The stickiest interest rate is the rate on credit cards. Between January 1986 and October 1994, the credit card rate ranged between 14.4 per cent and 24.7 per cent whereas the cash rate varied in a range between 4.75 per cent and 18.9 per cent. The rates on personal instalment finance and housing loans also tend not to move one for one with the cash rate. This is the case in both the short run and the long run. Finally, while the business indicator rate is the least sticky loan rate, changes in the cash rate do not appear to be always completely passed through.

Figure 5 shows the spreads between the business indicator rate and the cash rate and between the owner-occupied housing rate and the cash rate. It also shows the level of the cash rate. Given the lower pass-through of changes in the cash rate to the housing rate, it is hardly surprising that the housing – cash rate spread has been more volatile than the business indicator – cash rate spread. When cash rates reached their 1982, 1985 and 1989 peaks, the housing rate actually lay below the cash rate. In the earlier episodes, the large negative spread was in part a direct result of the regulation of housing interest rates and in 1989, the fact that mortgage rates did not move up one for one with the cash rate, partly reflected changes in the arrangements for banks' Non-callable Deposits with the Reserve Bank.

More generally, while the same factors making for incomplete pass-through into the business indicator rate (see below) are probably at work for the housing rate, two factors may also have played a role. The first possibility is that banks smooth lending rates over the cycle to reduce the variability in borrowers' repayment burdens. The second possibility is that the costs incurred by a borrower when switching banks gives each bank some market power over their existing customers and this is reflected in interest rates that move by less than the change in the banks' cost of funds.^[9] These switching costs include both loan establishment fees and taxes (stamp duties). Notwithstanding these explanations, business rates and mortgage rates have moved similarly over recent years.

The results in Table 1 suggest that, on average, while changes in the business indicator rate have matched changes in the cash rate reasonably closely, the pass-through has not always been complete. Figure 5 indicates that this incomplete pass-through is most apparent in 1982–83 and 1990–91. 1994 also saw the business indicator rate move by less than the cash rate, unwinding much of the increase in the spread in 1990–91. Outside these periods, changes in the indicator rate appear to have matched changes in the cash rate quite closely. Overall, relative to the size of the changes in the cash rate, changes in the margin have been relatively small.

In contrast to the volatility in the spread between lending rates and the marginal cost of funds, the spread between the average interest rate received by banks on their loans and the average interest rate paid on their deposits has remained relatively constant.^[10] As Figure 6 shows, this spread has been about 4.5 per cent over recent years. This is similar to the average spread throughout much of the 1980s.

Combining Figures 5 and 6, it appears that changes in the average cost of deposits have been translated into similar changes in the average lending rate, but changes in the cash rate have not always been fully translated into changes in the business indicator rate. We now explore possible reasons why the degree of pass-through of cash rate changes might vary through time.

Footnotes

In estimating (2), the lagged interest rate is used as the instrument for the current change in the interest rate. [6]

While the regulation of most lending rates was lifted in April 1985, the ceiling on new owner-occupied housing loans was not removed until April 1986. The interest rate on overdrafts greater than $50,000 was lifted in February 1972. In February 1976, the threshold level was increased to $100,000. Interest rate ceilings on all trading and savings bank deposits were removed in December 1980. [7]

We take an agnostic view with respect to the order of integration of the various interest rates. Appendix 1 reports the results of unit root tests using the Augmented Dickey-Fuller test. For all interest rates it is not possible to reject the unit root null hypothesis. The Appendix also reports the results of co-integration tests (using the Unrestricted Error-Correction Model) between the cash rate and various interest rates. For some interest rates it is possible to reject the null hypothesis that the rate is co-integrated with the cash rate while for others it is not possible to do so. Given the relatively low power of the tests, especially when the sample period is relatively short, the results should be interpreted with caution. If interest rates are indeed non-stationary, it would be surprising if some rates were co-integrated with the cash rate, while others were not. It would also be surprising if the true long-run response was substantially different from one. [8]

Lowe and Rohling (1992) discuss these explanations in more detail. They also canvass other possibilities for the stickiness of loan rates. [9]

To obtain this figure, interest foregone on non-accrual and restructured loans has been added to the actual interest received. [10]