RDP 9504: The Link Between the Cash Rate and Market Interest Rates 4. What Determines the Degree of Pass-through?

What factors determine the degree to which a change in the cash rate is passed through to banks' indicator lending rates? Below we focus on the following three (not mutually exclusive) explanations:

1. the structure of banks' deposits and interest rates;
2. the riskiness of bank lending;
3. the degree of competition in banking.

We use the following simple framework to examine how these factors might affect the relationship between indicator lending rates and the cash rate. By assuming that the loan rate is set to equalise the expected return from the loan with the cost of the providing the loan (including the return on capital) we have:

where: r_l = loan rate;

p = probability that interest is paid;

q = probability that the loan principal is repaid;

α = capital requirement;

r_c = rate of return to capital;

r_d = interest rate paid on deposits;

c = administrative costs of providing a loan (less fee income).

The expected gross return from lending one dollar (the left hand side of (3)) consists of two components: the expected interest payments (r_lp) and the expected principal repayment (q). If there is no default risk (p = q = 1), the gross return from lending one dollar will be r_l+1.

The “cost” of providing the loan is given by the right hand side of (3). Each dollar of loans is funded using a fraction α of capital and 1-α of deposits. Thus, the cost of providing the loan consists of three components: the amount that must be paid to the owners of the capital (α(1 + r_c)), the amount that must be paid to depositors ((1 − α)(1 + r_d)) and operating or administrative costs (c).

Re-arranging (3), we obtain an expression for the lending rate:

So far we have treated all deposits as attracting the same interest rate (r_d). As discussed above, the costs of various types of deposits do not always move together, with short-term money-market interest rates moving much more closely with the cash rate than do rates on many other classes of deposits. For the sake of simplicity, we assume that there are two types of deposits; deposits that attract an interest rate equal to the cash rate and deposits that attract lower and less variable interest rates than the cash rate. Thus, the average cost of deposits is given by:

Substituting (5) into (4) and rearranging we can obtain an expression for the difference between the lending rate and the cash rate. This is given by the following:

From this equation it can be seen that the spread between the lending rate and the cash rate will increase if:

1. the required return to capital increases relative to the average rate on bank deposits (i.e., r_c−r_d increases);
2. the share of loans funded by capital increases (i.e., α increases);
3. administrative costs increase (i.e., c increases);
4. the probability of either principal or interest repayment falls (i.e., either p or q decreases);
5. the fraction of deposits attracting “low” interest rates falls (i.e., βincreases);
6. the cost of “cheap” deposits relative to the cash rate increases (i.e., r_cash − r_low falls);
7. the general structure of interest rate increases (i.e., r_cash increases).

The intuition behind each of these factors is relatively straightforward. If bank shareholders can successfully demand higher rates of return, the loan rate – cash rate spread must increase to generate those higher returns. Also, the larger is the share of loans that are funded by the relatively expensive capital, the higher must be the loan rate to achieve a given return to capital. If administrative costs increase, or the probability that the loan will default increases, the bank must charge a higher interest rate to make any given rate of return.^[11] Similarly, if the bank loses access to its cheap deposit base, or the cheap deposits become relatively more expensive, lending rates must increase to generate the same expected rate of return. Finally, an increase in the general level of interest rates makes an interest payment default by a customer more costly to the bank. To cover this decline in expected return, the bank must charge a higher loan rate.

Of these various factors, a number can be eliminated as important determinants of the observed behaviour of the spread through time. First, bank operating costs have been declining gradually through time, and, if anything, should be contributing to a trend decline in the spread between the lending rate and the cash rate. Since the early 1980s, the ratio of bank operating expenses (less non-interest income) to total assets has fallen by about one percentage point (see Figure 7). Against this downward trend, the last couple of years have seen a slight increase in operating expenses as a number of banks have incurred substantial rationalisation and restructuring costs (which tend to be concentrated in the early years). In aggregate, however, the increase has been relatively small.

Second, while the capital ratios of the banks did increase over the 1980s, the above model suggests that these movements should have had only a small effect on lending spreads. If the premium over the deposit rate that is paid to equity is 5 per cent, then a two percentage point increase in the capital ratio should add just over 0.1 of a percentage point to the loan rate. Clearly, movements in capital ratios do not account for the observed behaviour of the lending spreads. Third, changes in the general level of interest rates affecting expected losses from a given loan default are unable to explain changes in the margin. Not only is any effect quite small theoretically, but just at the time when interest rates were at their lowest, margins were at their highest.

Changes in Deposit Structure

Each of the three remaining factors – changes in deposit structure, default probabilities and the degree of competition – probably play some role in influencing the degree of pass-through. The change in the banks' deposit and interest rate structure has two key elements. The first is the drift of deposits out of low-interest-rate accounts and into deposits that pay interest rates at, or near, the cash rate (i.e., an increase in β) and the second is the rise, in relative terms, in the rates paid on the cheap deposits that remain (a decline in r_cash − r_low).

This shift in the structure of the major banks' liabilities can be seen in the bottom panel of Figure 8 which shows the share of total deposits that attract “low” interest rates.^[12] In the early 1980s, these deposits accounted for around 50 per cent of total deposits. By 1990, this share had fallen to around one quarter and by 1994 it had fallen further to around 15 per cent. This decline has been driven by the increasing range of unit trusts and deposit products that followed deregulation and by the fact that high inflation induced depositors to switch out of low-interest-rate accounts.

While high inflation encouraged depositors to switch into accounts that paid market interest rates, it significantly increased the value to the banking sector of the low-interest-rate accounts that the banks managed to maintain. As discussed above, an increase in the cash rate is more fully passed on to money-market rates than to rates on transaction accounts. Thus, with relatively high inflation and high money-market rates in the second half of the 1980s, the spread between the cash rate and the average rate paid on accounts attracting low-interest rates was quite wide. This access to deposits that attracted interest rates considerably below money-market rates was an important source of bank profitability during this period. The fall in interest rates in the early 1990s meant that the differential between money-market rates and the average rate paid on low-interest-rate accounts declined significantly (see Figure 8).^[13] This reduced the benefit to banks of the cheap deposits that remained.

A final issue concerning the changes in deposit and interest rate structure is whether the changes have affected banks' average or marginal costs. If the banking industry is perfectly competitive and the marginal cost of funds is given by short-term money-market interest rates, then the changes discussed above should have affected the intra-marginal, but not the marginal cost of funds. As a result, they should not have affected lending rates. Instead, these changes should have reduced the profitability of bank lending. On the other hand, if the average cost of deposits is important in bank pricing and/or the marginal cost of funds is a weighted average of all funding sources, changes in the deposit structure should have an impact on lending spreads.

Certainly Figure 6 suggests that the average cost of bank deposits has been an important factor in the pricing of bank loans. The reduction in the cash rate between 1990 and 1993 was larger than the reduction in the average cost of deposits due to the stickiness of many deposit rates. Because average deposit rates adjust less than the cash rate, loan rates also adjust less than the cash rate. As a result, the spread between the cash rate and the loan rate widened as the cash rate fell. This explanation leaves unresolved the important question of bank competition. To some extent, the combination of lower inflation and increased competition for deposits should have led to a reduction in the difference between the average interest rate received and paid, and in so doing, reduced the profitability of core lending business. Working in the other direction, strong demand for housing loans may have meant that banks could maintain these margins. These issues are discussed in more detail below.

The Riskiness of Lending

The second factor that probably plays some role in influencing the degree of pass-through is the perceived riskiness of bank lending. In Figure 5, the lending spread appears to behave counter-cyclically – in particular, the periods of slow economic activity are associated with the largest lending spreads. If periods of recessed activity are associated with a higher probability of loan default (that is, a lower p or q) then it is natural for lending spreads to widen in recessions.

Prior to deregulation of the financial markets, banks had to ration credit. Those customers with the highest credit rating stood at the front of the queue. As a result, loan defaults were relatively rare (see bottom panel of Figure 9). Following deregulation, the need for regulation-induced credit rationing no longer existed and banks engaged in aggressive competition for business. On the demand side, increasing asset prices and the weakening of financial constraints saw significant increases in demand for intermediated finance from the corporate sector. These increased demands were willingly met by the newly deregulated banking sector. The result of this process was a significant increase in the gearing of the corporate sector and an increase in the ratio of business credit outstanding to GDP (see top panel of Figure 9).

With increases in corporate indebtedness and lower average credit standards, it was not surprising that when the early-1990s recession arrived, record corporate liquidations and loan defaults occurred (see middle panel of Figure 9). The model presented above suggests that for every one percentage decline in the probability of principal repayment, the loan rate should increase by slightly over one percentage point. Certainly, the size of the increase in defaults in the early 1990s is not inconsistent with a widening of the business loan spread at around that time. Further, the subsequent recent narrowing of spreads follows a significant reduction in banks' charges for bad and doubtful debts. However, the exact timing of changes in the spread is a little surprising. During the late 1980s, when corporate lending was increasing rapidly, and the probability of a significant slow-down in growth was rising, there was no widening of the spread. Certainly with hindsight it appears that during the final stages of the boom, lending was becoming more and more risky as the likelihood of a fall in asset prices and a recession increased. However, it was only after the recession hit, and the banks experienced record bad debts, that spreads widened. Similarly, margins have narrowed only after bad debts expenses have declined. This suggests that the banks may have been under-pricing risk during the late 1980s and over-pricing risk in the first half of the 1990s.^[14]

If changes in the riskiness of bank lending are responsible for change in margins, it suggests that banks have not been undertaking “dynamic provisioning”; that is, at the top of a boom, they have not increased loan margins and provisions in expectation of higher future loan defaults and they have not reduced margins and provisions at the bottom of the cycle. Instead it appears that pricing and provisioning is static, with margins only being changed once a change in default experience actually occurs.

Competition

A third possible explanation for the widening of the spread between the business indicator rate and the cash rate in the early 1990s is that banks were attempting to earn higher real rates of return on new and existing loans. With a number of banks suffering considerable losses as a result of their lending decisions in the second half of the 1980s, there may have been pressure to partially offset these losses by earning higher rates of return on loans that were actually being serviced (although the data on average spreads do not lend support to this view). In addition, while deposit rates declined as inflation fell, it is sometimes argued that there is considerable inertia in firms' (including banks') expected nominal return on equity, so that expected returns on bank lending did not fall as quickly as deposit rates. To the extent that banks are able to increase r_c − r_d (the premium that equity earns), lending margins must increase.

This explanation of widening spreads, as does the explanation based on changes in the structure of banks deposits, rests on a departure from the textbook model of perfect competition. In this model, price (in this case, lending rates) should equal the marginal cost of the extra loan. It should not be possible for banks to widen spreads either to increase their rate of return, or to protect a given rate of return from the adverse effects associated with a change in deposit structure.

As is the case in goods markets there is no consensus as to whether the degree of competition in loan markets is pro or counter-cyclical. If the degree of competition is procyclical (that is, competition for a loan of a given quality is more intense in the upswing) then lending margins should narrow in booms and widen in recessions. In the recession of the early 1990s, lending margins stayed high for a longer period than was previously the case – the business indicator spread and the owner-occupied housing spread first reached 4 per cent in early 1992 and were still around that level 2½ years later. It was only in the last quarter of 1994 that both spreads came back down under 4 per cent.

Certainly, a spread of 4 per cent between the variable-rate mortgage interest rate and the cash rate is not consistent with the simple perfect competition framework. In that framework, a spread of 4 per cent between the business indicator rate and the cash rate also appears high, although a proper assessment in this case is complicated by the difficulty in determining risks and the margins over/under the indicator rate that the banks' charge.

Suppose that the cash rate is the marginal costs of funds and that the rate is 4.75 per cent (as it was for much of 1994). Further, suppose that the market return on equity is 15 per cent and that each dollar of housing loan is funded by 92 cents of deposits and 8 cents of capital, and that administrative costs (less fees) on a loan amount to 1¾ per cent. Further, assume that the probability of default is 0.5 per cent (p = q = 0.995). Using equation (4), the equilibrium loan rate is 7.86 per cent. For most of the time that the cash rate was 4.75 per cent, the standard mortgage interest rate was around 8.75 per cent. While altering the assumptions changes the calculated lending rates, quite large and unrealistic changes to the assumptions must be made to generate a spread of 4 per cent and, at the same time, maintain moderate returns on equity. This suggests that lending for housing was particularly profitable at these spreads. This is evidenced by a number of regional banks specialising in housing lending earning high after-tax rates of return in 1993/94.

In textbook markets of perfect competition, high rates of return get bid away very quickly, either by existing firms lowering their prices, or by new firms entering the market. In banking, as in many other real world markets, these forces do not always operate so quickly. Three characteristics of banking contracts are particularly important in understanding the dynamics of competition in response to profitable opportunities at the margin. The first is that banking contracts tend to be long-term contracts, which rely heavily on information. The second is that, historically, intermediation and transactions services have not been priced separately by banks. The third is that changing an indicator rate changes the interest rate paid, not just by new customers, but also by all previous customers who still have loans outstanding.

The textbook world of price equals marginal cost is one of auction markets with no long-term contracts, no customer loyalty, perfect information and no joint provision of products. In contrast, loan contracts are characterised by imperfect information, typically last for a number of years and involve the provision of some payment services. Further, largely as a legacy of financial regulation, customer loyalty to a particular bank has been relatively high in Australia. Of these departures from the textbook world, perhaps the most important is the long-term nature of loan contacts. This departure allows lenders to smooth the loan interest rate over the course of the interest-rate cycle to reduce the variability in repayment burdens. This means that margins widen at the trough of the cash rate cycle, and narrow at the peak. This outcome need not be inconsistent with perfect competition. It may be optimal for both sides of the contract, particularly if borrowers dislike cash-flow volatility more than lenders, and the writing of more explicit contracts is difficult and costly.

In the medium term, however, it may be difficult to sustain this form of interest rate smoothing. The smoothing means that at the trough of an interest rate cycle, margins are large. This allows other institutions to offer lending rates below the smoothed lending rates. If customers can switch lenders for only a small cost, it pays to switch to an institution offering an interest rate closer to marginal cost, but then to switch back to the original lender towards the peak of the interest rate cycle. This possibility may make it unprofitable for lenders to smooth lending rates over the interest rate cycle, unless substantial switching costs are imposed. In the absence of these costs, it is likely that, eventually, price will follow marginal cost more closely than has been the case in the past. This is particularly the case for housing loans, as information about borrowers is less costly to obtain and assess than is information about business borrowers. This makes it relatively easy for customers with housing loans to switch banks. For small business, switching is more difficult because of information asymmetries. At the trough of the cycle it may be difficult for the new bank to assess whether a borrower is leaving the existing bank to take advantage of a lower interest rate, or because the existing bank is withdrawing credit following a deterioration in the business's creditworthiness.

Another important banking-industry deviation from the textbook world is the joint provision and pricing of distinctly different services. In particular, for many years, banks have jointly provided both intermediation and payment services. While payment services have historically been seen as “free”, they have not been without cost. In its 1994 Annual Report, the Reserve Bank of Australia (p. 32) noted that the high “interest margin is being used to subsidise other types of activity, particularly those concerned with payments transactions.” This subsidisation has opened the window for new entrants to split the provision of loans and payments services. Unencumbered by the need to offer transaction accounts to their borrowers, new entrants can offer loans at rates lower than those of the banks. In the longer term this represents a considerable problem for the banks, for if the subsidisation continues, the banks will end up with a disproportionate share of customers that are heavy uses of subsidised payment services. The resulting pressure on bank profitability is likely to see banks make greater use of fees, resulting in a reduction in average margins, and loan rates that move more closely with banks' marginal costs of funds.

A third factor influencing the dynamics of competition is the fact that a change in the indicator rate affects not just new borrowers but also existing borrowers. Thus, in reducing an indicator rate in response to profitable lending opportunities, banks must weigh the benefit of increased demand for new loans, against the loss of revenue on existing loans. As the profitability of new loans increased in the early 1990s, banks attempted to differentiate between existing and new customers by introducing “honeymoon” rates or “special-deals” for a limited period of time (most often one year). These special rates offered new borrowers a temporary discount on the standard indicator rate, after which the loan rate reverted to the indicator rate. By 1993, the competition between banks on these special rates had become intense, with some banks offering discounts of up to 3.25 percentage points off the standard rate for housing loans. The bulk of new housing loans was written at discounted rates, while existing customers continued to pay the higher variable loan rate.

This type of price discrimination is only sustainable if the banks' existing customers do not switch lenders to take advantage of another bank's special offer for new customers. While loan establishment fees and mortgage stamp duty introduce costs to switching, the special offers were often sufficiently attractive to make it worthwhile to do so. It is likely that the willingness of customers to switch banks played some role in the narrowing of the spread between the variable loan rate and the cash rate in the last quarter of 1994.

The introduction of new providers of finance, particularly housing finance, has also influenced the dynamics of competition. Two types of non-bank lenders have entered the market for mortgages – mortgage originators and mortgage lenders. Mortgage originators arrange loans but then pass the loans on for securitisation, while mortgage lenders keep the loans on their own balance sheets. These new providers of finance have typically offered interest rates 1 per cent below the standard variable loan rate. They tend to have lower overhead costs than the banks, often relying on mobile loan managers rather than operating an extensive branch network. Over recent years, the lower operating costs, together with the high margin between the banks' variable loan rate and marginal cost of funds, meant that these new providers could undercut the banks and still return satisfactory rates of return.

The undercutting of the existing institutions by new entrants has also been helped by the fact that the new entrants do not have an established book of loans whose interest rates are reduced when the standard variable rate falls. Over recent years, established institutions with lower than average operating costs could have reduced their standard variable lending rates (relative to their competitors) and gained market share. However, this would have meant earning a lower rate of return on the whole stock of existing loans, rather than just on new loans. New entrants were not constrained in this way and have been a principal catalyst for greater competition in the market.

From the perspective of the textbook, it took a relatively long time for these new providers of loans to enter the market. This reflects a number of factors. First, much of the household sector is quite conservative and is reluctant to take out a housing loan from an institution with which it is not familiar. The loyalty to a particular bank that was a legacy of credit rationing in earlier decades, also reduced the willingness of customers to borrow from new entrants. Second, the start-up costs for new entrants are relatively high. These include marketing and advertising costs to gain name recognition, information technology costs and systems and staffing costs. Before incurring these costs, those providing the financing for the new entrants require that the profit opportunities that they are to exploit, will not disappear quickly. Third, new institutions are typically at a disadvantage regarding information collection and assessment. The existing institutions have knowledge about the credit worthiness of their customers that is not available to the new entrants. For housing lending, this is not a major problem as credit assessment can be undertaken using a relatively small number of easily observed and assessable pieces of information. For lending to small business, the asymmetry of information between existing lenders and potential lenders is more problematic and makes successful entry more difficult.

Footnotes

Here we are ruling out the possibility that as the interest rate rises the probability of default also rises. In certain cases, it is impossible to make the interest rate sufficiently high to give the bank its required rate of return. In such cases, equilibrium credit rationing occurs (see Stiglitz and Weiss (1981)). [11]

“Low-interest-rate” deposits are defined as those in non-interest and interest-bearing cheque accounts, passbook savings accounts and statement savings accounts. [12]

Bordes, Goyeau, Melitz and Sauviat (1991) present evidence for OECD countries that anticipated inflation increases bank profitability by increasing the interest differential on assets relative to liabilities. [13]

One piece of evidence that, in the past, banks have not accurately assessed risk is that when banks moved from “rationers” of credit under the regulated system to credit “marketers” under the deregulated system, they did not substantially increase general provisions for bad debts, despite the move further out the risk spectrum. [14]