Building society margins
UK MORTGAGE MARGINS We begin by examining a measure
of the mortgage margin for the
By Niall Gallagher and Alistair Milne, the Bank of England building society sector as a whole,
computed using published interest
Over the past two years there has been intense competition for mortgage rate and balance sheet statistics.
business, with offers of substantial interest rate discounts or cash This appears, together with the
payments to borrowers and the emergence of an active re-mortgage underlying retail and wholesale
market. What impact have these developments had on the margins of spreads, as a solid line in Chart 1.
mortgage lenders, and to what extent do they represent a general pruden- The box on the next page gives defi-
tial concern? nitions of these three measures of
the interest margin.
There is a close relationship
Over the past two years lenders between retail spreads and the
have fought hard for share of a rela- building society mortgage margin,
tively static mortgage market, by due to the dominance of retail
offering either interest rate dis- deposits, which accounted for 73%
counts or cash payments to eligible of total liabilities at end 1995.
new borrowers (‘cashbacks’). Spreads and the mortgage
These incentives have fuelled margin widened during the early
an active re-mortgage market, with 1990s, as building societies altered
many borrowers switching from administered deposit and lending
one lender to another so as to take rates in response to increasing prob-
advantage of the best deals on offer. lems of arrears and loan losses.
Mortgage lenders themselves now Wholesale spreads subsequently
talk about intense competition and narrowed by around 200 basis
both the Bank of England and the points, while retail spreads fell by
Building Societies Commission around 30 basis points, partly
have reminded mortgage lenders because of the deliberate policy of
that they need to take full account of ‘committed’ mutuals to pass on the
the risks involved when competing benefits of mutuality to their depos-
for this business. itors and borrowers. The overall
To assess the prudential impli- mortgage margin for building soci-
cations of these developments, we eties has fallen by around 60 basis
have examined data on United points since early 1994. But by end
Kingdom mortgage margins and 1996 it was still close to its average
considered the circumstances which during the second half of the 1980s
might generate widespread and and, according to the less compre-
substantial losses for mortgage hensive data available prior to 1985,
lenders.1 Our findings are supported wider than at any previous period
by a technical paper which explains back to the early 1960s.
Over the past two years borrowers
have taken advantage of attractive
the methodology and data sources. The statistics in Chart 1 largely
cashback and discount deals This is available on request.2 exclude cashback offers. Lack of
38
UK MORTGAGE MARGINS
data makes it impossible to provide effects of conversion could reduce
an accurate figure for the impact of the retail spreads by 10 to 30 basis THE MORTGAGE MARGIN
these offers. We have made some points. This would leave margins
illustrative calculations and exam- slightly below their average level Mortgage margin is sometimes used
ined accounting data (see box on during the second half of the 1980s. to describe the difference between
p41 for details). This suggests that mortgage interest and deposit rates.
in 1995 cashbacks would have Bank margins This is the retail mortgage spread —
reduced spreads and the mortgage We have also calculated a measure the difference between mortgage rates
margin by between 9 and 13 basis of bank mortgage margins. This and the cost of wholesale funds is the
points. Thus we find that the decline allows broad comparisons to be wholesale mortgage spread.
of building society margins, while made, although the series is not The mortgage margin is an
slightly greater than shown in Chart directly comparable with the average of these spreads adjusted for
1, has still been modest and margins building society calculations. The the endowment effect — the degree to
remain at a similar level to the end reason for this is that interest rates which mortgages are financed by non-
of the 1980s. for the banking sector are not interest bearing liabilities. It is defined
There is likely to be some further published in sufficient detail to as the average yield on mortgage assets,
decline of mortgage margins, as construct an entirely reliable weig- minus the average cost of interest
‘locked-in’ deposits are released hted average of funding costs. bearing liabilities, plus the proportion
following the demutualisation of According to our measure, of liabilities that are non interest paying,
several major building societies later bank mortgage margins have multiplied by the average cost of
this year. It is difficult to quantify recently been around 1.0-1.5% interest bearing liabilities. This means
the magnitude of this effect. We higher than those of the building the mortgage margin is always wider
believe that the unwinding of the societies, despite the greater reli- than the average mortgage spread.
Spreads, the endowment and the
mortgage margin cannot be calculated
Chart 1: The building society mortgage margin
from accounting data alone; they must
4.00 also use average interest rates.
The net interest margin is a
3.00
broader accounting-based measure
defined for all interest earning assets.
It also takes account both of spreads
2.00
between interest rates and of the
endowment effect.
1.00
Unlike the mortgage margin it can
be computed from annual accounts as
0.00 the ratio of net interest income to
85M1
86M1
87M1
88M1
89M1
90M1
91M1
92M1
93M1
94M1
95M1
96M1
interest earning assets. Both measure
-1.00 the average yield on assets less the
average cost of total funding. As such
Wholesale spread
-2.00
margins will differ for each lender,
Mortgage margin
Retail spread depending on the individual lender’s
mix of assets and liabilities.
-3.00
39
UK MORTGAGE MARGINS
Chart 2: Bank and building society mortgage margins ance of the banks on wholesale
funding (illustrated in Chart 2).
5
Lower average retail deposit
4.5 rates reflect banks’ traditional role
Banks
Building Societies
in providing liquidity and transac-
4
tions services. In order to make a
3.5
fair comparison with building
society mortgage margins, the oper-
3 ating expenses associated with
% (end year)
providing these services should be
2.5
offset against the gross mortgage
2 margin. In practice, we cannot do
this because published statistics on
1.5
costs are not sufficiently detailed. In
1 1995, bank and building society
operating expenses were 2.7% and
0.5
1.4% of total liabilities respectively,
0 but this difference reflects greater
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
relative costs of managing non-
mortgage assets as well as costs of
providing deposit services. Overall,
we judge that bank mortgage
margins are broadly comparable to
those of the building societies.
Chart 3: Building society sector income and provisions Bank mortgage margins have
fallen over the past decade because
1.60
of the decline in the proportion of
Income net of costs non-interest bearing accounts from
1.40 Provisions
around 15% to 5% of total liabili-
ties. Our weightings do not capture
1.20
the corresponding increase in low
interest chequeable accounts, which
1.00
means that we may have overstated
% mean assets
0.80
the decline in bank mortgage
margins.
0.60
Bank mortgage margins are
more sensitive than building society
0.40 mortgage margins to the fluctua-
tions in the wholesale mortgage
0.20 spread, as wholesale funding
accounts for more than 50% of bank
0.00 liabilities compared to less than
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
20% for building societies. This is
40
UK MORTGAGE MARGINS
THE IMPACT OF CASHBACKS ON MARGINS
Because they affect the interest charged to the borrower, mort- ‘average’ cashback equal to 3% of principal then it will be worth
gage discounts are usually taken into account in the published between 9 and 13 basis points of outstanding mortgage balances.
data on average mortgage interest rates, and hence are already An alternative calculation of the impact of cashbacks on
included in the retail and wholesale spreads shown in Chart 1. margins can also be made using annual accounts for individual
Cashbacks do not involve a reduction in the average mortgage lenders. The table below shows figures for lenders which amor-
interest rate and are thus not reflected in these spreads. In this box tise cashbacks over a period of years and report the unamortised
we consider how much building society spreads and margins balance in their accounts. By deducting this unamortised balance
should be reduced in order to take account of the impact of cash- from the published net interest margin we obtain an adjusted
backs. figure for the net interest margin which takes full account of the
We have no direct measure of this impact. Nevertheless a impact of cashbacks. This adjustment is a measure of the impact
rough estimate can be made using a combination of official statis- of cashbacks on mortgage margins and on mortgage spreads. This
tics and a number of ‘assumptions’. Anecdotal evidence suggests is not a precise measure as unamortised balances include cash-
that around 20-30% of gross mortgage lending has been trans- backs offered prior to 1995 and exclude that part of the cashback
acted on cashback terms over the past two years, whilst official treated as a first year expense. Nonetheless these figures suggest
data show that the annual level of gross lending has been equal to that an estimate for the effect of including cashbacks in published
just under 15% of the average outstanding stock. If we assume an statistics of 9-13 basis points is plausible.
Net income as a Based on published accounts Adjusted Impact of cashback adjustment
% of mean assets (1) (2) (1)-(2)
Abbey National 1.76 1.64 0.12
Woolwich 2.08 1.98 0.10
Northern Rock 1.97 1.83 0.14
Chelsea 2.07 1.95 0.12
West Bromwich 2.17 1.96 0.21
Source: 1995 annual reports. Adjustment described in text.
the principal reason for the larger growth of assets. We now examine Chart 3 compares the net inc-
year-to-year variations in bank the data on net income, loan loss ome of this sector (total income net
mortgage margins, compared to provisions and capitalisation of of costs) with provisions for loan
those of building societies. retail funded mortgage lenders, as losses. Net income rose from
a yardstick for assessing future around 1.2% of mean assets at the
Net income and provisions prudential risks. Again, we have a end of the 1980s to over 1.4% of
Whether the current level of mort- problem with data for banks, which mean assets in the early 1990s. This
gage margins is adequate for prevents us from distinguishing that rise was, in part, a widening of
supporting lending risks depends part of total income which can be administered spreads in response to
upon the potential scale of loan attributed to their mortgage business, the high level of loan losses. Net
losses, the level of capitalisation of and are forced to rely primarily on income has since fallen back slightly
lenders, and their expected rate of data for building societies. as loan losses have been reduced.
41
UK MORTGAGE MARGINS
Loan losses themselves peaked at a capital. Building societies, like the their mortgage income comfortably
little under 0.8% of mean assets in banks, made particular efforts over exceeded their levels of mortgage
1992; with cumulative loan losses, these years to raise their risk asset loan loss provisions in the early
between 1990 and 1994, of around ratios in order to comply with 1990s.
2.5% of mean assets. capital based regulatory regimes. The lenders who got into most
A striking feature of Chart 3 is But unlike the banks, they were serious difficulties in the early
that income net of costs has always under no pressure from share- 1990s were wholesale funded
been comfortably greater than holders to economise on their use of centralised lenders. These lenders
provisions, leaving the sector in financial capital and many societies never held more than a small
surplus throughout a period of raised capital levels comfortably proportion of the market, but
unprecedentedly severe difficulties above their regulatory requirements. accounted for all the insolvencies
with mortgage lending. These amongst mortgage lenders of the
surpluses amounted to around 0.8% early 1990s. This is unsurprising
of mean assets in the late 1980s and given that they lacked a retail
were still 0.54% of mean assets in Centralised funding base, that their mortgage
1992, the year of peak provisioning. assets were often of below average
This continuing surplus of lenders never quality and that they entered the
income over provisions was large recession with a relatively smaller
enough to increase reserves from h e l d m o re t h a n a proportion of mature low risk mort-
4.4% of mean assets in 1985 to gages on their books.
5.8% of mean assets in 1995,
s m a l l p ro p o rt i o n Qualifications must be made
despite relatively rapid balance about the use of this data as a guide
sheet expansion.
of the market, to the security of mortgage lenders
A further factor increasing but accounted in the face of future loan difficul-
building society capitalisation was ties. The loan loss provisions shown
the issue of interest bearing capital, for all the in Chart 2 were reduced by mort-
which was possible from 1988 gage indemnity guarantees, which
onwards. By 1995, issued capital insolvencies are now provided on much less
(permanent interest bearing shares generous terms than in the 1980s.
and subordinated debt) amounted to amongst The risks associated with mortgage
1.2% of mean assets, increasing lending have also increased because
total capital (reserves plus issued m o rt g a g e l e n d e r s of a recent tightening of social secu-
capital) to 7% of mean assets. rity rules, restricting the ability of
Computed on a risk weighted borrowers to claim mortgage
basis, the 1995 total risk weighted Examining the experience of interest payments on loans taken out
capital ratio for the building society other lenders would not lead to very after October 1995.
sector was 14.1% and tier-1 capital different conclusions. The major Nonetheless, this data still de-
ratio 12.7%.3 These risk asset ratios banks also benefited from access to livers a clear message: the profits
compare with average 1995 ratios low cost deposits and their mort- from retail financed mortgage len-
for the major United Kingdom gage loan loss provisions were no ding have comfortably exceeded
banks, computed using the Basle greater, in relation to their stock of loan losses even in exceptionally
1988 risk weightings, of 10.8% for lending, than those of building soci- difficult economic conditions. This
total capital and 6.6% for tier 1 eties. Like the building societies is a clear indication of the important
42
UK MORTGAGE MARGINS
role played by the ‘retail franchise’, in costs arising from, for example, the increasing interest rates from 6% to
ie access to lower cost retail introduction of a single European 12% over the period from mid-
deposits, in cushioning lenders from currency. 1997 to end-1999, with a
the problem of loan losses during (ii) Mortgage incentives spreading consequent deterioration in loan
the early 1990s. Provided the retail to 75% of the mortgage stock and performance.
franchise is not significantly eroded, eventually reducing the mortgage Net interest income before
prudential risk for retail-funded margin by 45 basis points a year. provisions rises substantially (acco-
mortgage lenders remains low. (iii) One-third of any post-tax rding to our calculations by around
surplus paid out, either in the form 45 basis points per annum as a share
Worst case scenarios of dividends (for a converted of mean assets). This is because of
In order to assess the potential scale the increased value of the endowment
of any future losses on mortgage of non-interest bearing liabilities
lending, we have analysed the when interest rates rise.
impact of some worst case scenarios In the context of such an
on lender income. These scenarios interest rate shock it seems reason-
all involve a major deterioration in P ro v i d e d t h e able to assume that loan losses are
asset quality for a ‘typical’ retail on about the same scale as experi-
funded lender, which we take to be
retail franchise enced by the average building
a lender with a portfolio composition society lender in the early 1990s.
and cost structure corresponding to
i s n o t e ro d e d , The increase in net interest income
the average of the present building p ru d e n t i a l r i s k then exceeds the rise in provisions
society sector. in all but the peak year of provi-
We first developed a baseline f o r re t a i l f u n d e d sioning. This, combined with slower
scenario for the years 1996-2004, growth in the stock of mortgages,
which assumes market interest rates m o rt g a g e l e n d e r s increases the risk asset ratio to 20%
of 6%, loan loss provisions of 0.2% in 2004.
of mean assets per year and growth remains low The second scenario we have
in the stock of mortgages and retail considered is a housing market
deposits of 6% per year. In this boom and bust, repeating the expe-
baseline, net income before provi- rience of the late 1980s and early
sions initially declines and then 1990s.
settles down at around 0.9% of lender) or as ‘quasi-dividend’ paid During the housing boom there
mean assets, while the risk asset in the form of bonuses or beneficial is a period of 15% pa growth of the
ratio of our typical lender rises to interest rates to members. mortgage stock. This reduces the
around 17.0% in 1997 and changes (iv) The retail spread declines by 25 risk asset ratio to less than 12%, as
little thereafter. basis points between end 1996 and the stock of assets outstrips capital
There are a number of specific 1998, to allow for the unwinding of reserves; and also increases average
assumptions which underlie this the ‘lock-in’ of deposits with conve- funding costs, as greater reliance is
baseline: rting societies, and remains constant placed on wholesale funding.
(i) No change in management costs thereafter. The boom sows the seeds for
as a proportion of total assets. This Against this background, we further large scale loan loss provi-
implies that we have taken no have considered the impact of a sions which, as a proportion of
account of potential one-off increases short term interest rate shock, mean assets, are nearly twice as
43
UK MORTGAGE MARGINS
STATISTICS ON INDIVIDUAL LENDERS
What do statistics for individual lenders add to our analysis of and the Abbey National, whose lending is dominated by mort-
mortgage margins? The table shows 1995 accounting ratios for the gages, and the other three banks which conduct substantial
eight leading United Kingdom mortgage lenders; these comprise non-mortgage lending business (column 4 of the table). Another
four mutuals, of whom all but the Nationwide are converting later contrast is between those institutions which provide substantial
this year, and four banks. Together these institutions hold 66% of money transmission services (Lloyds TSB, Barclays, NatWest,
the stock of UK mortgages (the first column of the table records and the Alliance and Leicester, the latter having acquired this
their individual market shares).The remaining shares of the mort- business through its purchase of Girobank in 1992) and those
gage stock are accounted for by smaller building societies (16%); which do not. The former group have higher interest and non-
other major banks (8%); specialised mortgage lenders, some of interest income, in relation to the size of their balance sheets, but
which are licensed as banks (8%); and other lenders (1%). also higher costs.
The second column of the table shows a measure of whole- For all these lenders their retail deposit franchise allows
sale funds as a proportion of total liabilities. The institutions with them to earn a healthy level of net income (column five of the
the highest proportions of wholesale funding, Abbey National, table). In the case of the four mutuals net income is close to the
Barclays and NatWest, are also the institutions with the most average for the present building society sector. There is greater
significant involvement in treasury and investment banking activ- variation amongst the banks, reflecting their different asset mixes.
ities. This is confirmed by the lower proportion of loans and The final column of the table shows the total risk-weighted
advances to customers in their balance sheet (column three of the capital ratio computed using the standard Basle 1988 weightings.
table). On this measure it is apparent that the mutual institutions are
There are several significant differences between these particularly well capitalised, but the banks also all comfortably
lenders. There is a contrast between the four mutual institutions exceed the 8% international minimum capital standard.
% Share Wholesale Loans Mortgages/ Income Risk
of UK funding/ and advances/ loans net of costs/ weighted
mortgage total total and total total capital
stock liabilities assets+ advances+ assets ratio
Halifax 19.8 16 81 97 1.4 15.0
Abbey National 12.3 36* 52 93 1.3 11.7
Lloyds TSB 9.6 21* 54 48 2.4 9.6
Nationwide 7.1 20 81 91 1.6 13.7
Woolwich 5.6 19 80 97 1.4 14.8
Alliance &Leicester 4.1 22 76 93 1.6 15.9
Barclays 3.8 31 49 19 1.5 10.9
NatWest 3.8 28* 53 20 1.7 10.7
Source: computed from IBCA database. All accounts are year ending December 1995, except Nationwide (March 1996).
* For these three banks, figure shown is other deposits/total liabilities and thus excludes some wholesale time-deposits.
+ For the societies, loans and advances are the total of class 1, 2 and 3 commercial assets.
44
UK MORTGAGE MARGINS
great again as those experienced in threatened to fall below the required
the early 1990s. Loan loss provi- level, regulators would be likely to
sions exceed net income for three insist on the lender putting in place
years in succession, but even so the management plans to increase net
risk weighted total capital ratio of interest income and restore capital-
our typical lender still remains just isation.
over 9% in 2004. Nevertheless, a comparison of
Even with such a housing mar- these last two scenarios supports
ket boom and bust, the risk asset our main finding: provided the retail
ratio of our typical lender remains franchise is not significantly eroded,
above the Basle international min- the possibility of loan losses trig-
imum of 8%. Nevertheless it is still gering widespread and substantial
worth asking the question: what P ru d e n t i a l deterioration in capitalisation of
extreme circumstances, in the abse- retail-funded mortgage lenders
nce of any response by lenders or c o n c e rn c o u l d seems remote.
regulators, would reduce capitalisa- We should, of course, point out
tion to well below required minimum still arise over that this reassuring conclusion does
levels? not rule out the possibility of an
We find that the circumstances
individual individual lender getting into diffi-
which would create such a substan- institutions ... culties, especially if they rely to an
tial decline are a ‘triple whammy’ unusual degree on wholesale fund-
combining the spread of discounting who pursue a ing, have particularly low quality
which features in all our scenarios; a assets, or enter new and unfamiliar
housing market boom and bust; and strategy of areas of business.
a substantial erosion of the retail
franchise due to increased compe- a g g re s s i v e Conclusions
tition in retail deposit markets. To Despite intense competition for
reflect this erosion, we assume that wholesale business, mortgage margins are in
retail spreads fall a further 20 basis fact only slightly narrower than in
points per year after 1998, until by funded the 1980s. Although there has been
the year 2003 they are 100 basis considerable contraction in the
points below the level of our base-
expansion spread between mortgage lending
line. rates and wholesale funding rates,
In this case, the total risk the spread between mortgage
weighted capital ratio of our typical lending rates and average retail
lender falls to around 3%. While the deposit rates remains higher than in
lender would still be solvent, such the 1980s. The impact of cashbacks
an outcome would severely shake and interest rate discounts has not
the confidence of depositors and the been enough to alter the fact that,
markets. for most lenders who have access to
We cannot assign a probability a large pool of retail funds, mort-
to such an extreme combination of gage lending remains a safe and
events. Moreover, if capitalisation profitable business.
45