CAIN Web Service
ESRC Data Archive Bulletin:
An Overview of the Building Society Mortgage Survey
Text: Donal McKillop and Liam Barton ... Page Compiled: Fionnuala McKenna
Material is added to this site on a regular basis - information on this page may change
An Overview of the Building
Societies Mortgage Survey
Donal McKillop and Liam Barton
The principal source of detailed
official statistics on the private sector housing market is that
of the 5% Sample Survey of Building Society Mortgages (Archive
study number 33191). This survey is carried out by the Department
of the Environment in conjunction with the Building Societies
dates from the end of 1965. At different the questionnaire upon
which the survey is based has
been subject to revision. One such example of a mainline revision
was in 1982 when the analysis of the previous
tenure of borrowers was extended to identify sitting tenants.
An example of a more recent major revision was in
April 1992. The survey was broadened to obtain information not
only on first advances but also remortgages and further advances.
At the same stage the survey coverage was extended to encompass
not only building societies plus Abbey National plc but also the
returns of other mortgage loan institutions such as banks, household
mortgage companies and insurance companies. In short,
the survey has been extended to include members of the Council
of Mortgage Lenders. As yet results for this broadened coverage
have not been published.
A core feature of the survey in the past and
one which will remain for the foreseeable future is the 5% sample
survey of building society mortgage completions. For the purposes
of the survey, building societies are divided into
four strata. The stratum to which they are allocated is determined
by their volume of total assets. The first stratum consists of
large societies. This grouping is requested to complete questionnaires
on a sample of 5% of their new mortgages advances. In the second
stratum every second society is chosen and requested to complete
questionnaires on 10% of their mortgages. In the third stratum
every fourth society is chosen with the questionnaire
coverage, in this instance, 25%. The fourth
stratum which is made up of the smallest societies is excluded
from the analysis. Only one of the two indigenous Northern Ireland
building societies features in the survey. The Londonderry Provident
Building Society with an asset base in 1991 of 5 million falls
into the fourth stratum and is not covered. In contrast the Belfast
based Progressive Building Society, with an asset base in 1991
of 256.3 million, is allocated to the second stratum and is one
of those chosen to complete monthly questionnaires on 10% of its
mortgage advances. In actuality, however, the Progressive returns
on average 10 questionnaires per month which adds up to a 5% coverage
of its mortgage advances.
Although the survey coverage is supposed to
be 5% the actual outturn is much closer to 3.5% with this representing
a coverage of approximately 8,000 mortgages per quarter. One reason
that the sample proportion is consistently less than 5% is a consequence
of the exclusion of small societies from the analysis and under
returns by some societies, as in the case of the Progressive.
This, however, accounts for only a small part of the shortfall.
Another reason put forward is that advances included in the analysis
of the sample survey are restricted to those granted for the purchase
of dwellings for occupation, whereas the estimated totals include
all advances for the purchase of dwellings (Evans, 1975). In addition,
it is argued that there appears to be a shortfall in the combined
requisite number of questionnaire returns after the occurrence
of transfers of engagements and/or mergers between societies.
This latter point may be viewed as particularly pertinent given
the rapid decline in building society numbers over the last number
of years. For example, in 1991 alone there were
8 transfers of engagements and by the end of 1991 there were only
91 registered members of the Building Societies Association in
operation. This compares with 255 ten years ago.
The specific object of the survey is to collect
all the particulars of the "mortgage transaction". In
consequence the information is collated under three broad headings:
the mortgage advance; the dwelling; and the main applicant.
A total of some 116 regular computer tabulations are now produced
from the survey data for each quarter and calendar year, although
recently, prior to revisions, a total of some 300 tables were
produced. It is argued that the list of regular tabulations is
continually under review. Consequently as demand arises new tabulations
can be provided with the opportunity to produce them retrospectively.
The majority of both the key national and
regional results from the survey are published by the Department
of the Environment and the Council of Mortgage Lenders.
In the case of the Department of the Environment
the principal quarterly results from the survey are published
in the Department of Environment's Housing and Construction
Statistics. These figures are, in addition, released
by press notice usually within two months after the end of the
quarter to which they refer. Other results from the survey are
published annually in Housing and Construction Statistics
in the form of supplementary tables.
Broadly similar tabular information is also
provided by the Council of Mortgage Lenders in Housing Finance.
This is a quarterly journal on the mortgage market and provides
features on the housing and mortgage market plus, in the spring
issue, 25 tables "covering all aspects of housing and mortgage
market activity". The latest edition, Housing Finance,
No.14, May 1992 provides an analysis of building society mortgage
lending in 1991.
Some of the figures published in the above
sources are also available elsewhere, for example in Government
publications such as Social Trends. Furthermore, the data
obtained from the survey is also made available to the ESRC Data
archive at Essex.
So far the concentration in this overview
has been on building societies. Building societies now, however,
account for only 75% of the funds available for mortgage finance.
To accommodate this fact the above noted publications have introduced
tabular material on mortgage advances from other loan institutions.
In Housing and Construction Statistics, for example,
material has been incorporated on mortgage advances and dwelling
prices for both banks and insurance companies. The information
provided is obtained respectively from the Bank of England and
the British Insurance Association sample survey. For the future,
however, the revised extended survey which covers mortgage lenders
will eliminate the problem of the survey being constrained simply
to building societies.
In terms of the regional emphasis of the mortgage
survey the majority of variables are analysed by region and include
information specifically on Northern Ireland. Indeed across Housing
and Construction Statistics and Housing Finance there
is a wide range of published information already available for
Northern Ireland. Information is provided both in single form
and in terms of variables analysed in conjunction with others.
Examples of available single form information include the regional
distribution of: loans; housing stock; house completions; mortgaged
dwellings; age of dwelling; type of dwelling; income of borrowers;
previous tenure of borrower; age of first time buyer; sex of borrower;
average dwelling price; deposit paid and average advance. For
the most part the cross tabular information presented tends to
concentrate upon regional loan distribution, dwelling
type and borrower status vis a vis the other variables listed
One problem which invariably occurs, particularly
when national data is broken down into its constituent regional
parts, is whether the same size is large enough to yield reliable
results. Pannell (1992) states that, while for the most part the
sample size is large enough to yield meaningful results, caution
at times needs to be employed in interpreting the figures with
this particularly the case in respect of some of the figures for
Northern Ireland. Approximate coefficients of variation are reported
in Housing and Construction Statistics for average dwelling
prices, advances and incomes. It is clear from these values that
the likely sampling error both year-on-year and quarter-on-quarter
is much higher in Northern Ireland than in other regions of the
United kingdom. On average for Northern Ireland the maximum likely
sample error in year-on-year (quarter-on-quarter) percentage charges
in the various series is 5.4 (12.4). Computing the average series
error for each of other United Kingdom regions and then taking
the average regional score yields comparable values of 2.9 (6.5).
This reinforces the point that caution must be exercised in interpreting
figures for Northern Ireland.
Finally it should be noted that the revised
and extended mortgage survey as it currently stands omits some
of the main players in the mortgage market in Northern Ireland.
The most notable omissions are the Northern Bank and the Ulster
Bank. The explanation for their omission rests with the fact that
they are not members of the Council for Mortgage Lenders. The
Department of the Environment is, however, not constrained to
sample members of the Council of Mortgage Lenders only and consequently
it should be a relatively easy matter to accommodate these institutions
in the sample survey. Irrespective of the ease or otherwise of
including these banks in the sample, it is nevertheless
important that they should be, given their role in the mortgage
market in Northern Ireland and the fact that neither the London
and Scottish clearing banks nor the home loan companies are represented
in Northern Ireland.
In summary it can be said that a number of
problems arise vis-`a-vis the Northern Ireland
dimension of the Mortgage Survey. Most notable amongst these are
the sample size for certain variables, the under-return by the
only Northern Ireland-based society included in the sample and
the limited sample coverage of the revised survey. It should nevertheless
also be evident that the survey offers a broad spectrum of regional
specific data on the various facets of the mortgage transaction.
Evans, A.W., (1975), CSO
Studies in Official Statistics, No.26.
Pannell, B., (1992), "Survey of Mortgage Lending 1991",
Housing Finance, No.14.
Return to Contents
. Tables not available in these publications
can be obtained directly from the department of the Environment
although a charge is levied for non-academic users.