RDP 2022-01: MARTIN Gets a Bank Account: Adding a Banking Sector to the RBA's Macroeconometric Model 5. How Does the Pass-through of Monetary Policy Change with the State of the Economy?

In this section, we detail how the deposit rate lower bound dampens the pass-through of cash rate changes to other interest rates in the economy once the level of the cash rate moves below 1.5 per cent. We also explain how a deterioration in banks' capital adequacy may also change the pass-through of monetary policy. We then construct scenarios to show how the responses of banks' lending rates and macroeconomic variables can depend on the level of the cash rate and banks' capital positions (i.e. state-dependence). To provide context for the results, we compare the BA-MARTIN responses to those from the standard MARTIN model.

5.1 The mechanics of state-dependent pass-through

The pass-through of changes in the cash rate to lending rates in MARTIN has historically been one-for-one – that is, a 25 basis point decline in the cash rate reduces lending rates by 25 basis points – irrespective of the level of the cash rate. While this is a simplifying assumption incorporated within many macro models (see Appendix A), there is strong evidence that pass-through may deviate from one-for-one in low-rate and/or stressed environments.^[25] The banking sector in BA-MARTIN facilitates a state-dependent transmission mechanism that more accurately reflects the true pass-through of policy rates to other interest rates in the economy.

There are several features of the banking system that can cause the pass-through of monetary policy to differ from one-for-one in some states (all of which are incorporated within BA-MARTIN).

5.1.1 Deposit rate lower bound

Most deposit accounts have an interest rate that typically moves with either the cash rate or longer term rates related to cash rate expectations. If banks are not willing to reduce retail deposit rates below zero because they fear a run to physical currency, then this typical co-movement with the cash rate will cease for these interest rates once they reach near zero. In other words, there may be a deposit rate lower bound that reduces the pass-through from cash rate changes to banks' funding costs, thereby reducing the pass-through to lending rates.

5.1.2 Banks' capital positions

As discussed in previous sections, in stressed states (i.e. when banks suffer large loan losses), banks may find it difficult to raise equity externally and may therefore rely on retained earnings and/or reduced lending to restore their capital ratios. A reduction in NIMs or increase in credit growth would slow this recapitalisation, while any reduction in losses would reduce the amount of recapitalisation that is required.

Cash rate reductions potentially lead to all three of these recapitalisation effects. So if banks desire/require a particular speed of recapitalisation, they may change lending spreads in response to the cash rate change, thereby changing pass-through. The NIM and credit growth effects reduce pass-through – banks would increase spreads in response to a cash rate reduction in order to increase NIMs and reduce credit growth – whereas the losses effect increases pass-through. Therefore, even the direction of the change in pass-through will depend on the state of the economy.^[26]

In addition to this pass-through change that results from banks endogenously responding to the effect the cash rate change has on their capital ratios, banks' creditors might also respond to lower capital ratios by increasing the risk compensation they require when lending to banks. So any change in banks' capital ratios that results from a change in the cash rate (via the three recapitalisation effects) may also lead to an change in debt funding costs, thereby changing the pass-through of the cash rate to lending rates.^[27]

5.2 Pass-through to banks' lending rates

The combination of all the above channels leads to a complex relationship between the state of the economy and interest rate pass-through – where even the direction of the pass-through (from the usual one-for-one) depends upon the state of the economy. To illustrate these relationships, we explore the state-dependence of pass-through using four scenarios:

1. Illustrating the deposit lower bound mechanism with capital ratios remaining at target.
2. Elevated losses cause a capital shortfall but deposit rates remain above their lower bound.
3. Elevated losses cause a capital shortfall and some deposit rates reach their lower bound.
4. Our alternative assumption in which banks respond to a capital shortfall by restricting the supply of new loans only (so pass-through differs between new and outstanding loans).

In Scenarios 2 to 4, the economy suffers an equivalently sized downturn (27 per cent fall in housing prices and unemployment rising 3 percentage points). This ensures that in each scenario the economy is in the same ‘state’ before monetary policy is changed. Importantly, changing the initial state will change the pass-through of monetary policy, so this section should be interpreted as illustrating how pass-through can change in a small number of possible scenarios. Moreover, all scenarios start from a benign economic position before the downturn occurs. If the economy is still recovering from a previous downturn, pass-through could deviate even further from the scenarios we explore.

5.2.1 Scenario 1: the deposit lower bound mechanism

Based on our estimates of the increasing share of deposits reaching their lower bound as the cash rate falls (Figure 3), Figure 10 shows how pass-through to debt funding costs is expected to fall as the cash rate falls below 1.5 per cent. Assuming banks' capital ratios remain at their desired level, changes to debt funding costs pass through one-for-one to household and business lending rates.^[28] So as the cash rate declines below 1.5 per cent, this muted pass-through to debt funding costs leads to an equivalently muted pass-through to lending rates.

5.2.2 Scenario 2: capital shortfall when deposit rates remain above their lower bound

In stressed scenarios (i.e. when banks cannot raise capital externally), cash rate reductions inhibit recapitalisation by reducing NIMs and increasing credit growth (henceforth, the direct channel). However, these cash rate reductions also reduce unemployment and interest payments, and increase housing prices, thereby reducing bank losses and increasing banks' capital (henceforth, the indirect channel).

When the economy is in a benign economic position before the downturn occurs, we find that the indirect channel is the more powerful channel. Cash rate cuts therefore reduce banks' recapitalisation needs, allowing for competitive pressures to reduce lending spreads.^[29] As a result, the pass-through of monetary policy exceeds 100 per cent (Figure 11).

For the direct channel to dominate, banks' capital shortfalls would need to be large relative to any newly provisioned losses. This would require a scenario in which banks previously provisioned heavily, causing a capital shortfall, but retained both the capital shortfall and the inability to raise equity externally even after provisioning returned to normal – in other words, the economy would not be starting from a benign economic position. The ‘excluding losses effect’ in Figure 11 illustrates the less than 100 per cent pass-through that would result in this situation.

Figure 11 also shows how pass-through evolves over time. The point at which pass-through reaches its peak depends on the profile for losses (which depends on the shocks that hit the economy) and the point at which monetary policy changes. In our example, the cash rate is cut when the downturn hits the economy. Peak pass-through is not reached for another seven quarters; this is due to the delayed effect cash rate changes have on unemployment and housing prices in MARTIN.

Over time, as the economy begins to improve and losses return to normal, the beneficial effect of lower losses wanes and pass-through moves back towards 100 per cent. In our example, capital returns to target around the same time as losses normalise, such that pass-through does not move below 100 per cent. Situations in which capital shortfalls and an inability to obtain external equity persist longer than the point at which losses normalise would see pass-through temporarily move below 100 per cent before returning to normal.

5.2.3 Scenario 3: capital shortfall with some deposit rates reaching their lower bound

The existence of a lower bound for deposit rates means interest rate pass-through falls as deposit rates progressively reach their lower bound (as shown in Scenario 1). When the cash rate gets very low, the deposit lower bound outweighs the indirect channel described in Scenario 2. So at very low rates, even the peak pass-through is far less than 100 per cent (Figure 12), while the lowest pass-through will closely follow the pass-through in Scenario 1.

In this scenario, we do not consider the effect of policies designed to directly influence longer-term interest rates (such as bond purchases, yield curve targeting or forward guidance). Moreover, any policies designed to directly reduce banks' non-deposit debt funding costs (e.g. the Term Funding Facility) would also have a large effect on lending rates, potentially offsetting the effect of the deposit rate lower bound. These policies can be accounted for in BA-MARTIN by exogenously lowering debt funding costs.

5.2.4 Scenario 4: banks restrict supply of new loans only

As discussed in previous sections, since restricting the supply of new loans does not immediately benefit banks' NIMs, achieving the same speed of capital ratio restoration as when all lending rates are increased requires a larger credit supply response. Therefore, the causal effect that cash rate changes have on banks' capital ratios (via the direct NIM and credit growth channel, and the indirect losses channel) will have a much larger effect on the pass-through to new loans than was the case when rates on all loans were adjusted (i.e. relative to Scenarios 2 and 3), while the pass-through to outstanding loans will be lower than was seen in Scenarios 2 and 3 (Figure 13).^[30]

5.3 Pass-through to the economy

Since the addition of the banking sector to MARTIN changes how lending rates and credit growth respond to changes in the cash rate (in some states of the economy), it will also change how other macroeconomic variables respond to monetary policy. This section analyses the macroeconomic implications of the state-dependent pass-through scenarios from Section 5.2.

Figure 14 shows the responses of GDP, consumption and dwelling investment to an exogenous 25 basis point cash rate reduction that persists for the entire period of analysis; the responses we show are consistent with Scenarios 2 to 4 in Section 5.2. To show the effect of having a banking sector, these responses are shown for both BA-MARTIN and MARTIN.

The overall differences between the MARTIN and BA-MARTIN responses are a straightforward mapping from the discussion in Section 5.2. When deposit rates are away from their lower bound (left-hand panels), the responses in BA-MARTIN are larger than in MARTIN due to the beneficial indirect channel dominating the deleterious direct channel. This is amplified when banks restore their capital ratios by restricting the supply of new loans only.

However, while the overall differences are a straightforward mapping, the response profiles are not. This is because the profiles depend not only on how the pass-through to interest rates evolves over the cycle, but how persistent the effect of changes in interest rates are on different parts of the economy. This is why, even after accounting for the differences in scale, the response profile of dwelling investment is noticeably different to consumption and GDP.

Once a sufficient proportion of deposit rates have reached their lower bound (right-hand panels), the reduced pass-through leads to smaller responses in BA-MARTIN than in MARTIN – consistent with Figure 12 showing that even the peak pass-through is less than 100 per cent when reducing the cash rate from 0 to –0.25 per cent. In the top right-hand panel, the response of GDP to the cash rate reduction is around 12 per cent smaller in BA-MARTIN than in MARTIN (when credit supply changes are implemented via all loans).

Figure 15 shows the responses of unemployment and inflation to the same exogenous changes in the cash rate as above. The high degree of estimated persistence embedded in these variables mutes the effect of the state-dependent pass-through. When the cash rate is above the point at which deposits increasingly reach their lower bound (left-hand panels), the inflation and unemployment responses are 7–9 per cent larger than in MARTIN (when credit supply changes are implemented via all loans). The effect of the deposit lower bound has a more noticeable effect on pass-through (right-hand panels); the inflation and unemployment responses are 10–12 per cent smaller than in MARTIN.

Our analysis in this section produces two broad implications for monetary policy. First, the persistence of macroeconomic variables mutes the effects the temporary lending rate pass-through amplifications have on economic activity, the labour market and consumer prices. Second, the deposit lower bound causes an economically significant and persistent reduction in expansionary policy pass-through when the level of interest rates is sufficiently low. The welfare implications of this reduced pass-through are not simply the differences between the MARTIN and BA-MARTIN curves at each point in time, but the cumulative difference. So it is important for policymakers to take this reduced pass-through into account when formulating policy.

In this section we have only considered the effectiveness of policy designed to expand economic activity. But an important implication of the deposit lower bound is that it has the opposite effect on the pass-through of policy tightening. As the cash rate is increased from these low levels, the interest paid on deposit accounts is expected to move away from the lower bound, causing a stronger tightening than if these deposit rates had remained at the lower bound.

Footnotes

See CGFS (2019), Altavilla et al (2020) and Hack and Nicholls (2021) for international evidence. Our focus is on the cash rate pass-through excluding unconventional policies that work through other interest rates. [25]

It is theoretically possible for the NIM and credit growth effects to be so strong that lending spreads increase by more than the cash rate reduction. The point at which this occurs is known as the ‘reversal rate’ (Brunnermeier and Koby 2018). We leave investigation of the reversal rate in Australia for future research (see Section 6). [26]

The cost of deposits is also affected as banks compete for deposit funding when faced with more expensive wholesale funding. [27]

MARTIN assumes all loans are variable-rate loans, so this pass-through immediately affects all outstanding loans. [28]

This does not imply that banks could improve their capital positions by reducing lending spreads. For a given reduction in lending rates, the difference between this reduction occurring via the cash rate and via lending spreads is that a cash rate reduction has a much smaller effect on NIMs than a reduction in lending spreads. [29]

How to quantify pass-through when credit supply is restricted differently across borrowers (e.g. a tightening of lending standards) is not obvious. We measure pass-through via the implied effect on $z_t^{*}$ explained in Section 3.8.3. The pass-through to outstanding loans may still differ from 100 per cent due to the deposit lower bound and banks' creditors changing their required risk compensation. [30]