Speech Some Recent Banking Developments
Talk to CS First Boston, Australasian Banking Conference
When I was invited to this banking conference, I agreed to talk about bank interest margins – in particular, the international comparison of these margins. As it turns out, we at the Reserve Bank have said about as much as can fruitfully be said on this subject in our recent Annual Report. Therefore, I will quickly summarise this subject in the first third of this talk, then move on to something which is both more interesting and more topical – namely, banks' housing lending.
Aggregate Bank Interest Margins
This is a subject which has raised its head quite a few times since banking was deregulated. It is usually raised by people who think that banks are charging too much on their loans and not paying enough on their deposits and thereby making higher profits than they should. Another string to their bow is usually a desire to show that deregulation and competition have not delivered any benefits to bank customers.
To date, virtually all the discussion has been based on very aggregate concepts such as the net interest margin, which is usually taken to mean banks' net interest income divided by their interest-earning assets. The first episode in this continuing discussion involved calculating the net interest margin for the major Australian banks and seeing how it has moved since deregulation. The results show that it has been relatively flat with, if anything, a slight tendency to decline. There is not really much of a story to tell: banks have not taken advantage of deregulation to widen their margins, but, on the other hand, competition did not bring forth a noticeable reduction as a number of people probably expected it would.
Over the past year or so, a number of people put forward the view that it was not whether bank interest margins were going up or down that mattered, but whether they were high or low compared with interest margins in other countries. This set people looking for international comparisons of bank interest margins and a couple came quickly to hand – one produced by the OECD and one by Salomon Brothers. On the first ranking Australia was a bit above the middle, and on the second it was near the top. The House of Representatives Standing Committee on Banking, Finance and Public Administration, which had been looking at these types of issues for a few years, asked the Reserve Bank to see whether they could do some research which would help adjudicate on this matter. I will now briefly summarise what we concluded.
- The existing comparisons of interest rate margins did not provide a good enough basis from which to draw meaningful conclusions. Salomon Brothers have in fact stopped publishing their international ranking, presumably because they were reaching the same conclusion. The OECD ranking was so surrounded by disclaimers that it did not inspire much confidence.
- There were two very large problems with the international comparisons which made it difficult to compare like with like. First, a wholesale operation like a US money centre bank cannot meaningfully be compared with a full-service retail bank. The former can operate on a margin of less than 2 per cent, while the latter typically needs more than 4 per cent. The sample for each country should be one or the other, not a mixture of both with different proportions in different countries. Second, domestic business is nearly always done at much higher margins than offshore business, because the latter is more akin to the business done by money centre banks. To make meaningful comparisons, each country's sample should be confined to the domestic business of full-service retail banks. We made a very extensive investigation of all the data available from the usual sources and on a bank-by-bank basis and came to the conclusion that a reasonable degree of comparability could only be achieved for Australia, the United States, the United Kingdom, Canada and New Zealand. All attempts to include continental European banks in the sample failed.
- On this basis, there was some support for the view that Australian banks' margins tend to be on the highish side. The average margin in Australia was second highest of the five countries, with the United States being the highest.
- When we turned to the ratio of total income to assets, the Australian figure was in about the middle of the field. This implies that non-interest income to assets was a bit lower in Australia than the average for the other countries. Two possible explanations for this were that trading income was lower in Australia or that fees and charges were lower. There was no reason to believe that trading income would be any lower than in these other countries and so we looked at fees and charges.
- Although we have not yet done an exhaustive comparison in this area, the available evidence tends to support the view that fees and charges as a proportion of assets are a little lower in Australia. International data on account-keeping fees, transactions fees and loan establishment fees show them as lower in Australia than in the other countries in the sample.
- Various measures of underlying profit to assets or to equity also had Australian banks in the middle of the field. By implication from this and from the figures on total income to assets, the ratio of costs to assets for Australian banks was not markedly different to that for full-service retail banks in other countries.
We put a lot of work into this exercise and collected a huge amount of data, but in the end had to discard most of it because it was not good enough to make publishable comparisons. Some would say that in the end our conclusion was that there was no story to tell. This is probably a bit of an over-simplification because I think we did find something – namely that margins were a bit on the high side and that fees and charges were a bit on the low side.
Nearly all of the discussion in Australia to date on margins has focused on the aggregate picture. By now we have extracted all the useful information we are ever going to extract from aggregate measures of margins, and the story that it tells is not a particularly exciting one. Where I think the discussions will soon move is to the subject of margins on particular products. It seems clear that some of these margins look to be quite fat, others are extremely fine, and some services are probably being provided at a loss. There is still a large amount of cross-subsidisation going on in the Australian banking sector. If people are worried about the slowness of competition to bring about the textbook results, this is the area they should be looking at rather than aggregate measures of margins.
The variable rate housing mortgage is widely regarded as being a product with a relatively fat margin. Banks earn a high rate of return on this product, some of which goes towards defraying the losses on transactions-related activities. Competition from within and without is beginning to whittle away this margin, but until recently it has had little effect. This brings me to my second subject, which is lending for housing.
Lending for Housing
What is happening?
It is worth making a few points about what is actually happening in the area of housing lending before moving on to the question of why it is happening and whether there are any risks involved.
The first point I want to make is that the rise in housing lending that has occurred over the past three years has been extremely large by historical standards. In an economy where nominal magnitudes (such as GDP) have grown at an average rate of about 5 per cent per annum over the past three years, the rises in housing loans outstanding of 13 per cent in 1991/92, 19 per cent in 1992/93 and 23 per cent in 1993/94 stand out. When put in a long-run historical context, by comparing the stock of housing loans outstanding to nominal GDP (Graph 1), it shows that the recent lift in housing lending dwarfs any of its predecessors. It is not something that we at the Reserve Bank believe we can take lightly.
It is also worth mentioning that approvals for loans for investment housing at their recent peak in April were running at twice the level that they were in the 1988 peak.
My second point is that it is not only housing lending that has been strong by historical standards, it is also true of dwelling construction. Some would go even further and make the claim that there are too many dwellings being built relative to underlying demand. The Indicative Planning Council for the Housing Industry (IPC) has for many years estimated the underlying need for dwellings on the basis of demographic and income trends. The present level of dwelling construction exceeds the medium-term requirement by a bigger margin than in previous cycles (Graph 2). If the IPC's estimate is anywhere near the mark (and I concede that it often gets revised after the event), there is a distinct possibility that we may be close to a situation where there is an excess supply of dwellings.
I am not aware that there is a big stock of unsold new houses and units, so someone must be buying them. Perhaps the purchasers are mainly owner-occupiers, but there is also a chance that there is a disproportionately large number of new investors being pulled into the housing market with promises of high returns through such things as rental guarantees and negative gearing.
My third point is that recent reports of a fall in housing lending, like Mark Twain's death, are greatly exaggerated. It is true that the monthly rate of housing approvals is down from its peak, but this does not mean much (Graph 3). They have still been running at over $5 billion per month. No doubt they will come off further, but they have to do so by a very large amount before we reach the point where the rate of growth of loans outstanding is back to a more reasonable rate. Even if they fell to $4 billion a month, outstandings would still be growing at over 15 per cent in a year's time. It is this latter concept – the rate of growth of housing loans outstanding – which is the best measure of what is happening. It is the one that I quoted at the beginning which showed 13 per cent, 19 per cent and 23 per cent over the past three financial years. This is also the one that matters for banks. It tells them how fast they are growing their housing assets.
The other thing that stands out from the top two lines in Graph 3 is how much higher the monthly flow of approvals and new credit is now compared with 1988 and 1989. At that time, we had a huge surge in house prices which was widely attributed to excessive credit being pumped into the housing sector. With the benefit of hindsight, it now looks like a small hill compared to the reasonably large mountain of the past three years.
Whenever an economic variable, whether it is a price or a quantity, is growing at an unsustainable rate, there is a tendency for a lot of people to say that we should not worry because it will correct itself. This is true in that eventually nearly everything does correct itself; the issue is how long does the excess have to continue and how large will the eventual correction have to be? There is a legitimate role for policy in bringing forward the correction and thus helping to reduce its severity. This is one of the reasons for the tightening of monetary policy last month, and it is a consideration we have to bear in mind in carrying out our prudential responsibilities.
Causes of the strong growth in housing lending
As usual, there is both a demand and a supply explanation for the strong growth in housing lending, but there is general agreement that the major cause is the very strong demand by borrowers.
The Australian public has retained a strong attachment to property, both for owner-occupation and investment purposes. With mortgage rates in single digits for the first time since the late 1970s, the attraction of purchasing a house has become very powerful. Whether an 8¾ or 9½ per cent mortgage rate is a real bargain in a country with an underlying rate of inflation of 2 per cent is an open question, but the public seem to think it is.
From a central banking point of view, this is a bit disappointing and seems to suggest that inflationary expectations have not come down as much as we would have liked. On the other hand, I do not think people who are buying property now are doing so because they think that inflation is going to take off again and they are going to make a big profit. They do it for a variety of reasons, some of which are not necessarily economically motivated. To the extent that pecuniary considerations are important, I think there is a feeling that ‘you can't go wrong with housing – you can't lose in the long run’. This is probably correct, but there can be transitional difficulties if borrowers have not factored in rising interest rates, have assumed that the debt servicing burden will be quickly eroded by rising incomes or, in the case of investment property, that it can be sold quickly without loss if a need arises.
Banks have also been keen to supply housing loans and have put a lot of resources into advertising, product design and delivery systems. An important reason for this is that the interest margin on variable rate mortgages has been high. The desire by banks to expand their housing lending has been keen, but until about a month ago it did not involve price competition for established borrowers. No doubt a marketing expert could give a lot of logical reasons why this would be the last aspect of mortgage lending to be affected by price competition among existing players.
What is more surprising is that competition from outside the banking/building society industries took so long to have an impact. One NSW insurance company entered the market in 1990 offering a lower interest margin for housing loans, but others were slow to follow. By our reckoning, there were only two other entrants until this year, when five insurance companies either entered the market or announced plans to do so. There has also been some competition from mortgage companies, but overall the share of housing finance approvals accounted for by banks and building societies remains at about 97 per cent.
Consequences of the surge in housing lending
Our main concern has been on the monetary policy front, but there are also subsidiary issues which have prudential implications.
Lending for housing over the past three years has clearly been exceptionally high by historical standards. A continuation of these rates of lending would result in excessive overall credit expansion once business lending returns to more normal growth as it is now in the process of doing. It would also normally be expected to produce sharply rising house prices, which would have unwelcome consequences for asset price inflation, price expectations and inflation generally. To date, the evidence of rising house prices is mixed (Table 1). The three indices of house prices show rises in excess of the general rate of inflation, but vary between 3½ and 7½ per cent over the past twelve months. Incidentally, the one that usually shows the smallest rise is the one produced by the ABS, but it excludes the top 10 per cent of the housing market where most of the action usually takes place (at least in the first instance).
|REIA||6.6||3 months to July|
Reasonable people could differ on how to interpret these statistics and the anecdotal evidence from the real estate industry. Some would say that prices are clearly on the move, the top end of the market has shown some very large rises, and it is just a matter of time before it feeds through more widely. Others would emphasise that most of the movement in the aggregate indices is due to Sydney prices (and to a lesser extent Perth). They would also point out that a 6 or 7 per cent aggregate increase in prices in the past year is not very high given the exceptionally strong growth in housing lending for the past three years. In most capitals, it has only returned house prices to their 1989 peaks.
If this latter interpretation is anywhere near the mark, prudential issues start to raise their head. If rates of growth of housing lending in excess of 20 per cent have done no more than keep the majority of house prices stable, what will happen when housing lending returns to more normal rates? Will some house prices fall?
It does not have to be an either/or situation; it is quite possible that the two scenarios outlined above could co-exist. A pronounced slowing in housing lending might have little effect in the areas now experiencing rising prices, but lead to falls in prices in areas which are less favoured or where supply has increased most rapidly. We may have to rethink our assumptions about the housing market; perhaps there may be two totally different sub-markets within the Australian housing market.
The main thing to point out about housing lending is just how big the current expansion has been compared with earlier phases of growth. A lot of people have taken advantage of the circumstances of the past few years to buy homes for the first time or to upgrade. The house-building industry has had a good run, which has contributed to output and employment at a good time for the economy, that is, it has been counter-cyclical. Banks have done well out of lending for housing. It helped them grow their assets at a time when business demand for funds was low, and it was also very profitable for them, as shown most clearly by the high rates of return on equity earned by the specialist housing lenders such as the regional banks.
All good things must come to an end, and we are starting to see the early signs of a slowing in housing lending. But as I have pointed out, finance approvals will have to fall by a lot more than they have to date before the rate of growth of housing loans outstanding gets anywhere near back to a ‘normal’ rate. We are still going to be surprised by high figures for the rate of growth of housing loans outstanding for quite a while to come. During this phase, when housing lending is still high in absolute terms, but on the way down, there will be conflicting pressures at work in the housing market. Among housing lenders, there is also a similar conflict which we are seeing played out at the moment. Cyclical pressures related to monetary policy will be pushing up mortgage rates while structural pressures associated with the intensification of competition will be squeezing margins.
Graph 3 sets out the steps involved in moving from finance approvals data to the rate of growth of loans outstanding. The first line is monthly housing finance approvals in billions of dollars – this is the gross flow of new approvals (for owner-occupiers and investors). Not all of this will flow through to the stock of housing credit outstanding. For a start, some people who obtain an approval will not take it up. There is also a continual reduction in outstandings as people pay back part of their principal through their regular monthly repayments or through using the proceeds of the sale of their existing house to pay off an old mortgage. In addition, there is housing finance for alterations and additions that needs to be added because it is not in approvals. When these four effects are taken into account, the resulting figure is the monthly change in housing credit outstanding (the second line). This is about half as big as the gross approvals figure, but it follows much the same pattern. If the monthly increments in credit are accumulated, the resulting series is the level of housing credit outstanding (the third line). Finally, the fourth line shows the annual rate of change of housing credit outstanding. For most purposes, this is the most useful measure of the growth of housing lending. 
The Australian Bureau of Statistics divides new loan approvals into those made by banks, building societies and other lenders. In July 1994, the first two accounted for about 97 per cent of total approvals. In these figures, ‘other lenders’ include credit unions, life or general insurance companies, and superannuation funds but, where loan funds are raised directly in the secondary mortgage market through securitisation, only those which are associated with State Government housing schemes are included. Some of the new lenders do raise funds in this way and thus will be excluded from the figures. This is not a serious omission at present, as the volumes of loans made in this way to date are small.