An assessment of bank lending to UK SMES in the wake of the crisis.
Armstrong, Angus ; Davis, E. Philip ; Liadze, Iana 等
The availability of bank finance to small and medium sized
enterprises (SMEs) is important to allow SMEs to start up and finance
investment for growth. To assess changes in such availability over
2001-12, we used data from a series of surveys that provide detailed
information on the characteristics of a sample of UK SMEs, their owners
and experiences of obtaining finance. Using econometric models, which
included controls for $ME characteristics and risk factors, indicators
of changes in the provision of bank lending over the time period
abstracting from borrower risk could be obtained. The results suggest
ongoing restrictions on the availability of SME bank finance up to
2012--which appear to have persisted into 2013. Further research using
macro data shows an impact of economic uncertainty on such finance. If
unresolved, these patterns could imply adverse effects on economic
performance in the short and long term.
Keywords: small and medium enterprises; bank lending;, credit
supply; banking crisis
JEL Classifications: E44; G01
Introduction
A key policy issue for the UK is whether tightening of credit terms
by banks to Small and Medium Enterprises (SMEs) (1) has been sustained
ever since the financial crisis, beyond that which can be justified by
changes in the perceived riskiness of borrowers. If there is evidence of
such a supply constraint, this is of relevance to policy makers as the
banking sector would be imposing an externality on the rest of the
economy. In this article we present our assessment of these questions
based on an econometric analysis of UK survey data over 2001-12. (2) Our
assessment follows two lines of enquiry: (1) characteristics of SMEs
likely to face supply of credit constraints; (2) distinguishing cyclical
and structural changes in lending conditions, mainly using indicators of
uncertainty at a macroeconomic level. The surveys do not allow clear
identification of the specific impact of the supply of bank lending to
SMEs (see Holton et al., this issue). Throughout we treat overdrafts
(mainly used for working capital) as separate from term loans (more
likely to be used for investment finance).
The article comprises a short overview of key aspects of SME
finance followed by an outline of our research methodology, charts to
illustrate lending conditions and then our main econometric results. Our
key results are as follows: (i) controlling for risk, the rejection
rates for loans and overdrafts have risen further since 2008-9; (ii)
controlling for risk, there is a sustained high level of overdraft and
loan margins since 2008-9; (iii) particularly for term loans, the
rejection rate has increased significantly since 2008-9 for low and
average risk firms and not significantly for high risk firms; (iv)
collateral requirements have not increased although arrangement fees are
somewhat higher than in 2008-9; (v) renewals finance appears to have
been more strongly affected by credit rationing since 2011 than
applications for new finance; (vi) uncertainty proxies affect the volume
of SME lending more than they do large firm lending, as does the risk
adjusted capital adequacy of banks.
I. Credit rationing and intermediation for SMEs
The supply of bank credit to SMEs has distinct characteristics
compared to credit to larger businesses (see also Beck, this issue).
First, lending to SMEs is generally riskier as they are often young
businesses, they often have less collateral available for security and
they are less likely to have pricing power in their product markets.
Second, SMEs are often more opaque than larger publicly quoted firms
because they have lower reporting requirements, have less need for
formal reporting structures and are subject to less outside monitoring
by equity investors. This creates some important information issues.
Third, the collateral or assets used to secure loans are likely to be
less liquid as they are more firm specific and they may entail
incomplete contracts (no clear distinction of the asset title). These
difficulties mean that the cost of bankruptcy and loss on asset disposal
may be greater for smaller businesses than for large.
The role of banks in the economy is best seen in the context of
institutional mechanisms to overcome information difficulties. This
implies a comparative advantage for banks over securities markets for
financing certain types of information-intensive borrowers such as SMEs.
This comparative advantage entails imperfect substitutability between
bank and market finance for SMEs and implies that availability of
finance from banks per se may affect real decisions for SMEs, such as
investment.
Asymmetries of information are substantial for SMEs--the lender
faces a problem of screening and monitoring borrowers. As these are
costly to the lender, the price of credit will tend to be higher, i.e.
there will be price rationing of credit for SMEs. The profit maximising
lender may even seek to impose quantitative restrictions on the amount
of debt the borrower can obtain, so-called 'equilibrium quantity
rationing of credit', because higher interest rates may give a
stimulus to adverse selection and risk taking (Stiglitz and Weiss,
1981). (3) Whereas price rationing is indicated by loan margins,
quantity rationing is likely to be indicated by loan rejection rates.
Higher costs of credit, quantitative credit rationing or lack of
collateral will have adverse effects on overall economic performance,
since SMEs' investment will tend to be limited to what is available
from holdings of liquidity and flows of internal finance (Campello et
al., 2010, show evidence for the US). SMEs with strong balance sheets
will invest more readily than those dependent on external finance.
Fraser (2012b) for the UK shows that lack of working capital boosted the
likelihood of SME failure and induced lower sales growth in 2008-9.
Risk aversion of lending officers may affect lending separately
from borrower characteristics when uncertainty is high. As suggested by
Baum et al. (2002), uncertainty may have a major effect on SME lending
since uncertainty will increase perceptions of default risk.
Accordingly, banks may withdraw from higher risk lending such as lending
to SMEs as well as real estate lending when there is heightened
uncertainty, and revert to a more conservative distribution between
loans and securities. The authors found this pattern empirically in the
US for household sector loans and real estate loans but not for
aggregate corporate lending.
In this context, it has been widely suggested in the UK that there
has been a regime shift from risk loving to risk aversion on the part of
bank lending officers in respect of corporate lending generally owing to
the 'shock' of the financial crisis and that this pattern may
not be reversed even when market and economic indicators of uncertainty
decline. This pattern may of course be underpinned by other factors such
as tighter ongoing and expected capital adequacy requirements on SME
lending, e.g. Basel III as well as losses affecting bank capital during
the crisis and recession and withdrawal of small and foreign banks from
the market in the wake of the crisis, thus reducing overall lending. (4)
Within SME loans, a priori considerations suggest that in periods of
uncertainty banks may prefer overdrafts to loans since the degree of
control is greater, (5) the term is shorter and penalties can be applied
more readily.
2. Data and research methodology
Our main data sources are surveys of SME finance, namely the UK
Survey of SME Finances (UKSMEF) for 2004, 2008 and 2009, and its
successor the SME Finance Monitor (SMEFM) for 2011Q1 and 2012Q2. To our
knowledge, these are the sole publicly-available sources of microdata on
SME lending in the UK. As recorded in Fraser (2012a), UKSMEF provides
detailed information on the characteristics of SMEs, their owners and
experiences of obtaining finance. The surveys are based on large,
representative samples of UK businesses with less than 250 employees.
The 2004, 2008 and 2009 surveys form a longitudinal panel. Of the 3,964
firms covered, 1,707 firms (43 per cent) were observed in two or more
surveys. In total there are 6,250 observations: 2,500 in the 2004 survey
and 3,750 in the 2008 and 2009 surveys. (6) UKSMEF was succeeded by the
SMEFM (BDRC, 2012). (7) The survey is undertaken each quarter with about
5,000 interviews of different SMEs (i.e. there is no panel structure to
the data). Quotas are set by business size, sector and region, to a
carefully constructed sample design which ensured that sufficient
interviews were conducted with SMEs of all sizes to allow for robust
analysis. The results have then been weighted to be representative of
SMEs with up to 250 employees and a turnover of less than 25 million
[pounds sterling].
Unfortunately the SMEFM omits some variables included in the UKSMEF
while adding others. In particular, UKSMEF includes both renewals and
new loans together while SMEFM allows them to be treated separately.
Accordingly we have included renewals and new facilities for loans and
overdrafts together in the 2001-12 data since these cannot be separately
identified in the UKSMEF. The rejections rate is defined as the rate at
which firms that actually applied for finance were refused credit either
by being refused completely or not receiving as much credit as they
requested (we omit from the denominator firms that did not apply for
credit).
In our econometric work, we estimate the determinants of rejection,
margins, arrangement fees and collateral for both overdrafts and term
loans for those firms seeking finance. This is done for the periods
2001-4, 2005-7, 2008-9, 2010-11 and 2011-12, where results are defined
relative to 2001-4, which we consider a relatively 'normal'
period, in contrast to excessively lax credit as in 2005-7 and the
tightening since 2008. Rejection and collateral are zero-one variables
and are estimated by probit, interest rate margins are estimated using
two stage least squares (8) and arrangement fees (as a proportion of
loan/overdraft size) are continuous positive variables with a
substantial proportion of zeroes and are estimated by tobit. Whereas
earlier work by Fraser (2012a) on UKSMEF from 2001-9 used panel
estimation techniques, we employ pooled estimation, given that SMEFM
data are not in panel form.
In all of these regressions we control for firm risk. This resolves
an important source of endogeneity where credit risk is ignored in other
studies of credit supply since risk has an effect on supply. We include
in particular Dun and Bradstreet risk ratings (as a direct measure of
lending risk), missed loan repayments and unauthorised overdrafts
(associated with delinquency) and sales, age of business, legal status,
owner's educational qualification and gender (business and owner
characteristics used in credit scoring) as controls. Unfortunately, in
contrast to Fraser (2012a), we cannot include some other key risk
variables due to their omission from SMEFM such as firm assets (relevant
for collateral), number of finance providers and length of relationship
(relevant for relationship lending) return on assets and the debt-assets
ratio (financial ratios used in credit scoring) as well as VAT
registration and ethnicity. (9) In this econometric section we also
assess the differences between risk classes in unexplained changes in
credit conditions. (10)
The final part of the analysis is to establish whether changes in
demand and supply are likely to be affected by uncertainty. Uncertainty
is used as a key non-cyclical variable (in terms of variation in
expected credit losses) likely to affect both supply and demand for
credit, while other variables capture cyclical effects per se. This work
involves both macro and micro datasets. We undertake macro modelling
work using as the dependent variable the British Bankers Association
(BBA) lending to small business (turnover up to 1m [pounds sterling])
series from 1990-2012. This enables uncertainty to be tested alongside
standard determinants of bank lending as in Barrell et al. (2009), Davis
and Liadze (2012) for the series SME lending, SME term lending, SME
overdraft lending, unincorporated business lending, (11) and as a
control, total lending to Private Non Financial Corporations (PNCFs).
Furthermore, we incorporate uncertainty into the micro analysis by
estimating various regressors, such as conditional (12) volatility
measures (13) derived from monthly data on changes in share prices, GDP
growth and inflation as also in Byrne and Davis (2005). The use of
volatility generated regressors does not suffer the endogeneity problems
of other generated regressors.
In Armstrong et al. (2013) we also test whether these uncertainty
measures can help to explain differences in rejection rates, collateral,
margins and fees in the different periods (bearing in mind such macro
factors are common to all firms in the sample) in addition to the
overall economic conditions as shown by firm characteristics. This
allows a distinction between the level of activity and uncertainty
associated with this activity and provides some insight into the nature
of credit supply and demand using as a basis the regressions for the
2001-12 period noted in section 2 above.
3. Trends in the data
The following charts show key variables by selected sub-periods and
include the first and fifth waves of SMEFM for years 2010-11 and 2011-12
respectively. All charts include renewals as well as new lending. The
proportion of firms applying for debt finance across the sample has
declined since 2008-9 (figure 1), both for overdrafts and term loans.
Note that application is not necessarily an accurate measure of demand
since some firms may be deterred from applying despite positive
financing needs (the difference being 'discouragement' as
defined in the SMEFM). Furthermore the composition of firms applying for
and receiving credit is changing over time, as analysed in the
econometrics in section 4.
Credit ratings (14) within the sample have worsened sharply in the
period 2010-12 compared to the earlier periods, with a much higher
proportion of firms falling into the above average risk category (figure
2). Minimal, low and average risk classes have all fallen relative to
the 2009 survey, with the fall in low risk being part of a trend
throughout the 2000s. A somewhat higher per cent of firms have been
non-rated.
In figure 3 the rejection rate is defined as the proportion of
firms which applied for credit and were either refused outright or
received less credit than they requested, as a proportion of firms
applying. The data suggest a dramatic rise in the rejection rate as a
proportion of applications in 2010-12, reaching the highest levels in
the June 2012 survey of any period since 2001. The shift is accentuated
by its combination with a falling level of applications (figure 1).
Debt margins for borrowing firms fell in 2010-11 from a peak in
2008-9 for term loans, before rising again to their second highest level
in 2011-12. For overdrafts the cost of credit was marginally higher in
2010-12 than in 2008-9 (figure 4). (Note that base rate has been at 0.5
per cent since 5 March 2009, so for the most recent periods the margin
is also a broad (l5) indication of the overall cost of finance.) Levels
in 2011-12 are much higher than at a broadly similar state of the cycle
in 2001-4; a similar, albeit more muted, pattern exists for total
corporate lending spreads over interbank rates and base rate. As regards
distinctions between types of financing, the margins for overdrafts tend
to be lower than those for loans, consistent with a lower level of
control over the borrower for loans, but there is a degree of
convergence in margins in the latest periods.
Overdraft fees peaked in 2007-8 but stood at a high level within
the sample period also in 2011-12, while percentage fees for term loans
have come down from highs in 2008-9 (figure 5).
Collateral protects the lender against default so may be expected
to rise when there is financial stress. That said, the percentage of
loans having collateral requirements is shown in figure 6 to be
structurally around 55 per cent for term loans, although a lower figure
is shown for 2010-11, possibly reflecting the devaluation of collateral
in the wake of the crisis. Overdraft collateral requirements are more
cyclical with a sharp rise in 2011-12 to the highest level observed
since 2008-9 and the second highest in the sample. (16)
The micro data referred to above are complemented by macro data
from the BBA (adjusted since June 2006 by BIS to avoid series breaks in
the data). Quarterly data before June 2006 are estimated by NIESR and
based on annual series. These data refer to SMEs with a turnover of up
to 1 million [pounds sterling]. Figure 7 shows there has indeed been a
boom and bust in SME term lending in recent years, albeit driven largely
by construction and real estate activities, while overdraft volume has
remained subdued. Looking back, overdraft lending has declined
consistently even in nominal terms since 1991, while term lending rose
to a peak in 2009. There may be effects from the ceiling of 1 million
[pounds sterling] on turnover, as the 'real' size of a firm
with such turnover is obviously quite a lot smaller in 2012 than it was
in 1990.
[FIGURE 7 OMITTED]
4. Econometric results
4. 1 Characteristics of SMEs likely to face supply of credit
constraints.
Our principal results are from the pooled micro regressions which
are shown in tables 1-5. We use the data up to 2009 from UKSMEF and then
waves 1 and 5 of the SMEFM, in each case estimated relative to 2001-4.
(17) To ensure unbiased results we employ robust or bootstrap standard
errors (Wooldridge, 2006). We commence with regressions for rejection
rates of firms applying for loans or overdrafts over 2001-12 in table 1.
The results for overdrafts over 2001-12 show that, controlling for
business characteristics related to the firm's risk profile, there
are significantly lower rejection rates for overdrafts in 2005-7 than in
2001-4, and much higher ones in 2008-9 (at the 10 per cent level) and
2011-12. Other significant determinants of rejection for overdrafts are
firm size (higher sales mean a lower rejection rate) and risk (average
and above average risk --as well as no risk rating--imply more frequent
rejection than low risk or minimal risk which is the default).
Delinquency, i.e. loan default in the past as well as unauthorised
overdraft borrowing, leads to significantly higher rejection rates,
while older firms on which there is better information are less likely
to be rejected. This may relate also to a longer relationship with the
bank. We also have a significant effect for hotels and restaurants,
which may be due to this sector being among the least profitable and
having worse than average risk among other industries (BDRC, 2012).
Results for term lending rejection are similar to overdrafts. Again
there is a strong positive effect on rejection for applications during
2008-9 and 2011-12, and also in 2010-11. The rejection rate for those
applying in 2010-12 is higher than in 2008-9 and in the case of 2011-12
significantly higher. This is suggestive of tighter credit conditions,
after controlling for firm risk, to which SMEs are vulnerable owing to
the degree of market power banks enjoy in this sector. As regards
control variables, risk, age and unauthorised overdraft borrowing again
come to the fore. Firm size is again negatively related to rejection,
but is significant only for firms with sales of over 1 million [pounds
sterling].
We ran the regression separately for the categories low, average
and high risk to see whether the credit rationing effects apply to all
risk-groups or only certain ones. The results are shown in table 2. For
space reasons, control variables are not reported in table 2 and
subsequent tables, but they were included in the regression. (18) There
is a significant increase in rejection for overdrafts in 2008-9 and
2011-12, controlling for other characteristics, for the average risk
group only. There is a significant effect for loans in 2008-9 and
2011-12 for both the low and the average risk category and in 2010-11
only for the low risk category. This is a potentially important finding
since it shows that it is not the highest risk category that has
undergone a significant change in the probability of rejection. Indeed
high risk firms have had a broadly unchanged rate of rejection, for
given firm characteristics, except for boosts to lending in the boom
period. It is the lower risk firms that have borne the brunt of higher
rejection rates in the period since the crisis, especially in the term
loan market. These data are consistent with a partial withdrawal by
banks from SME lending as an overall asset class. Reasons may include
uncertainty, risk aversion, the tightening of regulation; and banks
showing forbearance for larger 'zombie' firms leaving SMEs to
bear the brunt of deleveraging. Devaluation of (housing) collateral for
SME lending may also be playing a role.
Turning to margins (table 3) in 2001-12, our results are pooled
regressions run by two stage least squares with robust standard errors.
Employment is used to instrument loan size as discussed above. We have a
time pattern which is similar for loans and overdrafts, with
significantly lower margins than in 2001-4 prevailing in 2005-7 and
2007-8 controlling for firm characteristics, but much higher levels in
2008-9, 2010-11 and 2011-12. The time effects peak in 2010-11 for
overdrafts and in 2008-9 for loans, but the second highest level for
loans is in 2011-12. This suggests application for an overdraft is seen
as a risk factor for firms of any size.
As regards the control variables (detail is shown in Armstrong et
al., 2013), firm risk affects margins for both overdrafts and loans,
with above average risk leading to higher margins. Unauthorised
overdraft borrowing increases margins for subsequent overdrafts, as
would be expected. Higher sales (i.e. larger firms) lead to lower
margins for loans but not for overdrafts. There are some industry
effects, with higher margins for sectors seen as risky, such as
construction. (19)
Controlling for firm characteristics, the incidence of collateral
requirements for overdrafts (table 4) is significantly higher in 2008-9
than in 2001-4 but lower in 2005-7 and also 2010-11. The incidence in
2011-12 was similar to 2001-4, suggesting that collateral is not being
used in an exceptional manner to control credit demand or limit risk.
There are no significant time effects for term loans. Higher sales and
legal status as a limited company lead to collateral being required more
frequently as does higher risk. The former may relate to availability of
collateral in larger and incorporated firms. The risk effect is of
course more likely to be for direct protection of the lender, as is a
positive effect of unauthorised overdraft borrowing on collateral
requirements in overdrafts.
Arrangement fees for overdrafts are higher in 2007-12 than in
2001-4, controlling for firm characteristics (table 5). They are also
higher for firms with higher risk. Loan size is negatively related to
proportionate fees (implying 'economies of scale' in
provision) while incidence of collateral increases it (perhaps
reflecting costs of valuation and documentation). For term loans the fee
rate shows no significant unexplained change and rates are higher for
firms with unauthorised overdraft borrowing and higher than minimal
risk.
In Armstrong et al. (2013) we provide some further results, namely
comparison of the loan application variables for regressions with and
without renewals for 2010-12, for discouragement (2010-12), for
ethnicity (2001-9 and 2012 only) and for bank type (2001-9 only). These
are omitted here for reasons of length. Suffice to note some key
results: (i) conditions appear tighter for renewals than for all lending
in 2010-12, perhaps reflecting the granting of the original loan in the
easy money period 2005-7; (ii) there are no strong trends in
discouragement separate from the control variables; (iii) there is some
evidence that black applicants are less likely to receive loans than
others and (iv) newly nationalised banks in 2008-9 were less likely to
reject applications and charged lower margins than other major banks did
in that period, suggesting a role for governance in bank behaviour.
4.2 Robustness checks
We explored the robustness of the results by performing several
specification tests. For the main estimation for margins using 2001-12
data, we adopted as an instrumental variable for the endogenous variable
loan size the size of employment. In the first stage of the two stage
least squares procedure (not reported but available upon request), loan
size appears to be highly and positively correlated with the size of
employment. This shows its suitability as an instrument for loan size
for both overdraft and loan. Further, we tested the heteroskedasticity
in the model errors and find that the null hypothesis of homogeneity of
errors was to be rejected, implying the presence of heterogeneity and
the need for robust standard error procedure which is used in most of
our regressions to avoid bias (Wooldridge 2006). However, for some of
the estimations for 2001-12 STATA does not permit use of the robust
standard error procedure, therefore we used the bootstrap method
instead.
4.3 The role of uncertainty at a macroeconomic level
The final part of the analysis is to establish whether changes in
demand and supply of credit are likely to be affected by uncertainty.
Uncertainty is used as a key non-cyclical variable (in terms of
variation in expected credit losses) likely to affect both supply and
demand for credit, while other variables capture cyclical effects per
se. As noted, banks may withdraw from higher risk lending such as
lending to SMEs as well as real estate lending when there is heightened
uncertainty.
Macro estimation work looks at the volume of lending at a macro
level based on the data in figure 7 as well as the aggregate PNFC
lending data and the national accounts data for unincorporated business
lending. (20) We estimate lending as an error correction equation
depending on corporate profits and the cost of lending (long real rate
augmented by the corporate-government spread), as in Barrell et al.
(2009) and Davis and Liadze (2012). To these we add the uncertainty
proxy conditional variance of share price changes based on a GARCH (1,1)
(21) equation for quarterly changes in share prices (FTSE-100 index).
The unit root tests (22) showed all variables are trended I(1)
other than the conditional variance (stationary I(0)) and unincorporated
business loans (I(2) over a short sample). We accordingly enter the real
lending variables as differences, the variance as a level and also for
economic reasons we include the long rate as a level (interest rates
cannot be trended in the long run).
Interpreting these results, we see that in this specification the
negative uncertainty effect on lending is greater and more significant
for SME lending and for unincorporated businesses than for aggregate
corporate lending. This is suggestive of a greater sensitivity of small
firm lending to underlying uncertainty than is that of larger firms. One
explanation for this pattern may be that there is a greater degree of
monopoly power for SME lending on the part of banks, while large firms
have the option of accessing the bond market (Davis, 2001) which limits
price and quantity rationing of credit by banks to large firms. A
complementary explanation may be that periods of uncertainty have a
greater impact on the default risk of SMEs--and the quality of
information--than large firms. We may also be capturing demand side
effects whereby demand for loans declines during periods of uncertainty
since future profitability of investment projects is less certain.
Meanwhile, there is in each case a significant long-run relation between
corporate profits and lending stocks, and for PNCFs the long-term real
interest rate is significant also.
We tried alternative uncertainty proxies, namely conditional
volatility of inflation (using the Consumers Expenditure Deflator) and
growth (in GDP) in the macro regressions, again derived from GARCH (1,1)
estimation (table 7). The table is read across for the effect of each
alternative type of uncertainty for each type of lending, with the
overall specification (not reported in detail) being as in table 6.
Economic growth uncertainty has a similar effect to share price
volatility, with a significant impact on the types and total of SME
lending as well as on unincorporated business lending, and again no
effect on lending to all firms. On the other hand, there is a weaker
effect of inflation conditional volatility, with an impact largely on
unincorporated business lending.
Table 7 also reports results for the effect of the risk adjusted
capital adequacy of the banking system. Whereas for an individual bank
it may be anticipated that lower capital adequacy would lead to less
lending (i.e. a positive relationship) it is also the case that tighter
regulation as evidenced by higher capital adequacy can entail higher
spreads (Barrell et al., 2009) and tighter credit conditions owing to a
higher capital charge on lending. The latter effect seems to be
predominant in the sample overall, with a negative impact of capital
adequacy on lending. This is evidence of a further structural effect on
lending, although further research would be needed to fully verify it.
(23)
5. Conclusions
Our assessment offers evidence suggestive of ongoing tight credit
supply conditions in 2010-12, well after the height of the financial
crisis in 2008/9. Margins in particular are historically high, even
controlling for firm risk, suggesting banks have to some extent taken
advantage of the lower base rates to profit from SME lending (although
variations in other determinants of funding costs will also influence
the profits banks can obtain). For overdrafts, margins are higher, given
risk, than in 2008-9. Rejection rates allowing for risk have increased
even compared to 2008-9, suggesting quantity as well as price rationing
of credit especially for term loan finance. Banks have simply chosen to
reject many applications, which economic theory suggests may relate to
poorer information--although it may also link to risk aversion as well
as regulatory and market pressures to shrink balance sheets. The
rejection rate has increased particularly for low and average risk firms
and not significantly for high risk firms. Banks may have viewed lending
to the safer categories of SMEs as relatively more risky in the period
after the financial crisis than they did before, although the pattern is
also suggestive of a partial withdrawal from SME lending as an asset
class. The evidence suggests greater restriction in term loans than
overdrafts. This maybe because overdraft facilities are to protect
existing bank exposures (although we note arrangement fees for
overdrafts have risen). Comparing 2011-12 with a broadly similar phase
of the credit cycle in 2001-4, we see higher rejection rates, spreads
and fees underlining the unusual nature of the current situation.
Uncertainty as measured by the conditional volatility of share
prices and growth apparently has a greater influence on SME lending than
on the corporate sector as a whole. This implies a structural shift away
from SME lending as long as uncertainty persists. Capital adequacy has a
negative effect on SME lending absent for large firm lending, suggestive
of a greater actual or expected effect of regulation on SME lending.
There is tentative evidence of an uncertainty effect on firm-by-firm
data which implies higher variances of macroeconomic and financial
variables which entail tighter conditions for SME credit at a firm level
also. The uncertainty variable may be capturing inter alia the
structural shift to risk aversion that took place for bank lending
officers during the financial crisis; capital is again also significant.
Furthermore, it would be helpful to investigate alternative sources of
funds that may be acting as substitutes for bank credit like invoice
discounting, factoring, internet funding, friends and family and credit
cards. We would not expect these to be sufficient to offset the
tightening by banks and they may be much more costly also.
More research is needed to further develop the work on uncertainty
and to use further instruments to test for the endogeneity of the
findings in the survey-data study, which would seek to distinguish
supply and demand for credit explicitly. Unfortunately the nature of the
surveys and related data make these infeasible; this is an important
area of further development of the SME survey in that until the issue of
a supply constraint is fully demonstrated, the validity of analysis in
this area will be incomplete. That said, we concur with Fraser (2012a)
that the effects found for loan applications "seem to be genuine
supply-side effects since the econometric models included extensive
controls relating to the risk profile of the business/owner and their
financial relationships" (ibid, p66).
Overall, we suggest that the research is indicative of an
economically damaging shortage of finance for SMEs, reflecting
banks' attitudes to risk and pressures to delever, albeit also
indicative of banks' market power in the SME sector. Although
demand is also probably subdued, there remain high rejection rates and
margins on credit. More recent data suggest that these have persisted,
as for example the SME Finance Monitor report for 2012Q3, which shows
declining applications and much higher rejection rates (32 per cent for
term loans, 23 per cent for overdrafts) although Q4 shows some reduction
in rejection rates; Bank of England data up to February 2013 show a
continuing decline since June 2012 in aggregate gross as well as net SME
lending despite the 'Funding for Lending' scheme, although
spreads on new lending fell in 2013Q1. There has also been weak lending
to large firms; meanwhile BIS adjusted BBA data show a 2 per cent
decline in stock of term lending to SMEs and 11 per cent fall in the
stock of overdrafts between June and December 2012. If the situation is
not resolved, output, investment and employment will be lower than would
otherwise be the case, with adverse effects on economic performance in
the long term as well as the short term.
REFERENCES
Agur, I. (2013), 'Wholesale bank funding, capital requirements
and credit rationing', IMF Working Paper No WP/13/30.
Armstrong, A., Davis, E.P., Liadze, I. and Rienzo, C. (2013),
'Evaluating changes in bank lending to UK SMEs over 2001-12
-ongoing tight credit?', NIESR Discussion Paper No. 408 and UK
Department of Business, Innovation and Skills Paper BIS/I 3/857.
Barrell, R., Davis, E.P., Fic, T., Holland, D., Kirby, S. and
Liadze, I. (2009), 'Optimal regulation of bank capital and
liquidity: how to calibrate new international standards', FSA
Occasional Paper No. 38.
Baum, C.F., Caglayan, M. and Ozkan, N. (2002), 'The impact of
macroeconomic uncertainty on bank lending behaviour', University of
Liverpool Working Paper.
BDRC (2012), 'SME finance monitor QI 2012', BDRC
Continental.
Bester, H. (1985), 'Screenings vs. rationing in credit markets
with imperfect information', American Economic Review, 75, pp.
850-5.
Byrne, J. and Davis, E.P. (2005), 'Investment and uncertainty
in the G-7', Weltwirtschaftliches Archly, 141/1.
Campello, M., Graham, J. and Harvey, C. (2010), 'The real
effects of financial constraints, evidence from a financial
crisis', Journal of Financial Economics, 97, pp. 470-87.
Davis, E.P. (1991), 'Bank credit risk', Bank of England
Working paper No. I.
--(2001), 'Multiple avenues of intermediation, corporate
finance and financial stability', IMF Working Paper 01/115.
Davis, E.P. and Liadze, I. (2012), 'Modelling and simulating
the banking sectors of the US, Germany and the UK', NIESR
Discussion paper no. 396.
Department of Business, Innovation and Skills (2012), Business
Population Estimates for the UK and Regions 2012, London, BIS.
Fraser, S. (2012a), The Impact of the Financial Crisis on Bank
Lending to SMEs, Report prepared for the BIS/Breedon Review.
--(2012b), 'Are entrepreneurs financially constrained? A new
test based on finance gaps', CSME Working paper.
Holton, S., Lawless, M., McCann, F. (2012), 'Credit demand,
supply and conditions a tale of three crises', Central Bank of
Ireland working paper.
Stiglitz, J.E. and Weiss, A.M. (198 I), 'Credit rationing in
markets with incomplete information', American Economic Review, 71,
pp. 393-410.
--(1986), 'Credit rationing and collateral', in Edwards,
J. et al (eds), Recent Developments in Corporate Finance, Cambridge
University Press.
Wooldridge, J. (2006), Introductory Econometrics, A Modem Approach,
3rd edition, Thomson.
NOTES
(1) SMEs are typically defined as firms with less than 250
employees. According to Department of Business Innovation and Skills
(2012), SMEs in the UK at the start of 2012 represented 99,9% of the
total number of enterprises and account for 48.8% of the economic added
value and 59. 1% of employment in the UK private sector. BBA and Bank of
England data suggest that they account for around 10% of lending to UK
private non financial corporations.
(2) The surveys are the UK Survey of SME Finances (UKSMEF) for
2004, 2008 and 2009, and its successor the SME Finance Monitor (SMEFM)
which is quarterly from 201 I Q I to 2012Q2
(3) The key is that the interest rate offered to borrowers
influences the riskiness of loans in two main ways. First, borrowers
willing to pay high interest rates may, on average, be worse risks. They
may be willing to borrow at high rates because the probability that they
will repay is lower than average. This is again the problem of adverse
selection. Second, as the interest rate increases, borrowers who were
previously 'good risks' may undertake projects with lower
probabilities of success but higher returns when successful--the problem
of moral hazard, that the incentives of higher interest rates lead
borrowers to undertake riskier actions. See also Bester (1985) and
Stiglitz and Weiss (1986) on possible use of collateral to distinguish
borrower risk.
(4) One other possible cause of this shift is the reduction in
availability of wholesale as opposed to retail funding, see Agur (2013).
(5) Banks have the right to call in an overdraft very quickly.
(6) UKSMEF was conceived and developed by the Centre for Small and
Medium-Sized Enterprises (CSME), Warwick Business School; the first
survey was carried out by CSME in 2004 with funding from a large
consortium of private and public sector organisations led by the Bank of
England. A second survey was conducted by the University of Cambridge in
2007 and the third was again carried out by CSME in 2008 with funding
from the ESRC and Barclays Bank. The UKSMEF 2009, was conducted by the
Department for Business Innovation and Skills (BIS), IFF Research Ltd
(an independent market research agency) and Warwick Business School.
(7) SMEFM is financed by the major banks and undertaken by BDRC
Continental
(8) Number of employees or sales as a firm-size variable was used
as an instrument for loan size.
(9) Only available for the 2nd quarter of 2012.
(10) In our original research, the variable 'discouragement
from applying for finance' which is solely present in the 2010-12
surveys was assessed; we looked separately at renewals for 2010-12, we
also looked over the period up to 2009 at the differences between types
of banks in lending behaviour and assessing differences in lending to
ethnic groups in 2001-9 and 2012. A brief summary of key results is
given below, for detailed results see Armstrong et al. (2013).
(11) Source: Bank of England series code RPQB78D. The series is
largely lending to sole traders and hence is another proxy for SME
lending.
(12) Conditional volatility of a series measures the uncertainty
about future volatility taking into account what has already taken place
(i.e. it is 'conditional' on available information).
(13) The question of whether unconditional or conditional
volatility is most appropriate as a measure of uncertainty is an
important one. As argued in Byrne and Davis (2005), the key is the
distinction originally due to Knight between risk and uncertainty. Risk
can be defined as the danger that a certain contingency will occur, a
measure often related to future events susceptible to being reduced to
objective probabilities, while uncertainty is a term applied to
expectations of a future event to which probability analysis cannot be
applied, such as a change in policy regime or a financial crisis.
(14) External risk ratings by ratings agencies such as Dun and
Bradstreet and Experian use a variety of business information to predict
the likelihood of business failure in the 12 months following the
forecast. Such ratings are commonly used as background for decisions on
lending.
(15) Banks' funding costs are influenced by factors besides
base rate, including maturity of their borrowing, retail/wholesale
composition, deposit market competition and wholesale market perceptions
of bank risk.
(16) In Armstrong et al (2013) we depict the key variables with and
without renewals for the SME Finance Monitor over 2010-12. As one would
expect there is a higher rate of rejection if renewals are excluded.
However, there is much more of an upward trend in the rejection rate
including renewals than excluding them. So while credit may be
consistently tight for new loans it appears to be increasingly tight for
renewals. The rise in margins has been more pronounced with than without
renewals, although it is also apparent for new lending for term loans.
These patterns imply that renewals, that may be of loans previously made
under easy conditions prior to 2007, are facing increasingly tougher
conditions.
(17) Our results as presented are unweighted. The surveys also
provide weights to ensure that the samples are representative of all
SMEs in the UK. To account for that, we separately re-estimated the main
econometric results and found that using the weights does not
substantially change the results.
(18) Full details are available from the authors on request.
(19) Davis (1991) shows that property and construction already
stood out as risky for a large bank's dataset from 1970-90.
(20) As noted above, this comprises largely sole traders and is
hence another proxy for SME lending.
(21) GARCH results are available from the authors on request.
(22) See Armstrong et al. (2013).
(23) In Armstrong et al. (2013) we also provide tentative evidence
using our micro data of positive correlation between uncertainty and
changes in rejection, margins, collateral and fees that are not
explained by firm level risk variables. Capital adequacy is also
indicated to have an impact on credit tightening.
Angus Armstrong, E. Philip Davis, Iana Liadze and Cinzia Rienzo *
* NIESR. E-mail:
[email protected] and
[email protected].
This paper summarises a project prepared by NIESR for the Department of
Business, Innovation and Skills. We gratefully acknowledge assistance
from Stuart Fraser of Warwick Business School, advice from colleagues at
BIS and suggestions from participants in seminars at BIS and NIESR.
Table 1. Rejection rates 2001-12
Overdraft (p-value)
Bank debt applications
Applications in 2005-07 -0.410 ** (0.000)
Applications in 2007-08 0.124 (0.165)
Applications in 2008-09 0.162 (0.074)
Applications in 2010-11 -0.00376 (0.960)
Applications in 2011-12 0.238 ** (0.003)
Sales
Sales: 50,000 [pounds sterling]- -0.298 ** (0.005)
99,999 [pounds sterling]
Sales: 100,000 [pounds sterling]- -0.274 ** (0.003)
499,999 [pounds sterling]
500,000 [pounds sterling]-999,999 -0.242 * (0.022)
[pounds sterling]
1m [pounds sterling]-4,999,999 -0.434 ** (0.000)
[pounds sterling]
5m [pounds sterling] or more -0.223 (0.064)
Risk rating
Low 0.144 (0.157)
Average 0.312 ** (0.002)
Above average 0.629 ** (0.000)
Undetermined 0.617 ** (0.000)
Financial delinquency
Loan default 0.746 ** (0.000)
Unauthorised overdraft borrowing 0.687 ** (0.000)
Business Age
2-6 years -0.427 ** (0.000)
7-15 years -0.752 ** (0.000)
More than 15 years -0.889 ** (0.000)
Other business characteristics
Ltd Co 0.132 (0.087)
Partnership 0.0565 (0.546)
Limited Liability Partnership 0.148 (0.340)
Highest Qualification
Undergraduate 0.00332 (0.963)
Postgraduate 0.0139 (0.881)
Gender
Female -0.0640 (0.372)
Industry
Agriculture, hunting and forestry/ fish -0.0460 (0.742)
Construction 0.0925 (0.394)
Wholesale / retail 0.0540 (0.623)
Hotels and restaurants 0.477 ** (0.000)
Transport, storage and communication 0.151 (0.222)
Real estate, renting and business 0.0881 (0.408)
activity
Health and social work -0.0240 (0.865)
Other community, social and personal -0.0498 (0.695)
services
Region
East Midlands 0.0113 (0.928)
London 0.225 * (0.056)
North East -0.129 (0.303)
Northern Ireland -0.0401 (0.765)
North West -0.158 (0.211)
Scotland -0.0320 (0.794)
South East 0.0524 (0.660)
South West -0.0937 (0.431)
Wales 0.0746 (0.538)
West Midlands -0.00436 (0.971)
Yorkshire & Humberside -0.0998 (0.460)
Constant -0.829 ** (0.000)
Observations 4731
Loan (p-value)
Bank debt applications
Applications in 2005-07 -0.267 (0.055)
Applications in 2007-08 0.272 (0.061)
Applications in 2008-09 0.582 ** (0.001)
Applications in 2010-11 0.653 ** (0.000)
Applications in 2011-12 0.840 ** (0.000)
Sales
Sales: 50,000 [pounds sterling]- -0.313 (0.053)
99,999 [pounds sterling]
Sales: 100,000 [pounds sterling]- -0.147 (0.279)
499,999 [pounds sterling]
500,000 [pounds sterling]-999,999 -0.152 (0.325)
[pounds sterling]
1m [pounds sterling]-4,999,999 -0.348 * (0.022)
[pounds sterling]
5m [pounds sterling] or more -0.568 ** (0.001)
Risk rating
Low 0.130 (0.356)
Average 0.314 * (0.023)
Above average 0.356 * (0.016)
Undetermined 0.419 * (0.030)
Financial delinquency
Loan default 0.108 (0.589)
Unauthorised overdraft borrowing 0.276 ** (0.001)
Business Age
2-6 years -0.587 ** (0.001)
7-15 years -0.828 ** (0.000)
More than 15 years -1.000 ** (0.000)
Other business characteristics
Ltd Co 0.0534 (0.625)
Partnership -0.150 (0.287)
Limited Liability Partnership -0.0615 (0.815)
Highest Qualification
Undergraduate -0.0679 (0.527)
Postgraduate -0.186 (0.156)
Gender
Female -0.234 * (0.035)
Industry
Agriculture, hunting and forestry/ fish -0.117 (0.554)
Construction 0.202 (0.179)
Wholesale / retail 0.139 (0.364)
Hotels and restaurants 0.173 (0.271)
Transport, storage and communication 0.0672 (0.679)
Real estate, renting and business 0.107 (0.480)
activity
Health and social work -0.0545 (0.769)
Other community, social and personal 0.0415 (0.810)
services
Region
East Midlands 0.0518 (0.760)
London -0.0141 (0.936)
North East -0.421 * (0.023)
Northern Ireland -0.131 (0.450)
North West -0.313 (0.085)
Scotland -0.266 (0.143)
South East -0.0541 (0.737)
South West -0.153 (0.340)
Wales -0.0454 (0.782)
West Midlands -0.151 (0.342)
Yorkshire & Humberside -0.660 ** (0.002)
Constant -0.699 * (0.012)
Observations 2501
Notes: Dependent variable is the probability of rejection of an
application for finance. Effects are measured relative to: bank
debt applications in 2001-2004; sales: 49,999 [pounds sterling] or
less; risk rating: minimal; loan default: 0-4 number of times
unable to make repayments; unauthorised overdraft borrowing: 0-2
number of times has exceeded its overdraft limits; business age:
less than 2 years; industry: manufacturing; region: East. Other
business characteristics and highest qualification are 0/1 dummies.
*,** indicate significance at 5% and 1% correspondingly. Method:
pooled probit estimation with robust standard errors.
Table 2. Rejection rates 2001-12 for categories of risk
Low risk
Overdraft Loan
Applications in 2005-7 -0.642 * -0.252
(0.006) (0.361)
Applications in 2007-8 0.153 0.403
(0.348) (0.079)
Applications in 2008-9 -0.197 0.805 *
(0.313) (0.009)
Applications in 2010-11 0.0153 0.635 *
(0.920) (0.005)
Applications in 2011-12 0.244 1.009 **
(0.161) (0.000)
Average risk
Overdraft Loan
Applications in 2005-7 -0.260 -0.119
(0.099) (0.571)
Applications in 2007-8 0.205 0.456 *
(0.151) (0.043)
Applications in 2008-9 0.531 * 0.564
(0.001) (0.085)
Applications in 2010-11 0.0955 1.058 **
(0.499) (0.000)
Applications in 2011-12 0.367 * 1.039 **
(0.014) (0.000)
Above average risk
Overdraft Loan
Applications in 2005-7 -0.621 * -0.632
(0.004) (0.064)
Applications in 2007-8 0.00611 -1.006
(0.977) (0.063)
Applications in 2008-9 -0.246 -0.894
(0.261) (0.096)
Applications in 2010-11 -0.334 * 0.259
(0.032) (0.230)
Applications in 2011-12 -0.0541 0.396
(0.738) (0.079)
Note: Control variables shown in table I are included in the
regressions. p-values in parenthesis.
Table 3. Margins 2001-12
Overdraft (p-value)
Bank debt applications
Applications in 2005-7 -1.369 ** (0.000)
Applications in 2007-8 -1.136 ** (0.000)
Applications in 2008-9 1.388 ** (0.000)
Applications in 2010-11 1.874 ** (0.000)
Applications in 2011-12 1.630 ** (0.000)
Observations 2150
Loan (p-value)
Bank debt applications
Applications in 2005-7 -0.962 ** (0.005)
Applications in 2007-8 -0.669 (0.141)
Applications in 2008-9 3.246 ** (0.000)
Applications in 2010-11 2.221 ** (0.000)
Applications in 2011-12 2.307 ** (0.000)
Observations 649
Notes: Dependent variable is the margin over base rate. Effects
are measured relative to: bank debt applications in 2001-4; other
control variables are included as in table I . *,** indicate
significance at 5% and 1% correspondingly. Estimated by pooled
TSLS with bootstrap standard errors. Excludes SMEs with margins
greater than 30pp.
Table 4. Collateral requirements 2001-12
Overdraft (p-value)
Bank debt applications
Applications in 2005-7 -0.296 ** (0.000)
Applications in 2007-8 -0.0380 (0.595)
Applications in 2008-9 0.684 ** (0.000)
Applications in 2010-11 -0.187 ** (0.002)
Applications in 2011-12 0.0872 (0.212)
Observations 3943
Loan (p-value)
Bank debt applications
Applications in 2005-7 0.0988 (0.285)
Applications in 2007-8 -0.0763 (0.483)
Applications in 2008-9 0.104 (0.470)
Applications in 2010-11 -0.0734 (0.461)
Applications in 2011-12 0.000526 (0.996)
Observations 1731
Notes: Dependent variable is probability that firm will be required
to provide collateral to obtain finance. Effects are measured
relative to: bank debt applications in 2001-4; other control
variables are included as in table 1. *,** indicate significance at
5% and I % correspondingly. Estimated as pooled probit with
bootstrap standard errors.
Table 5. Arrangement fees as a share of size of overdraft/loan
for 2001-12
Overdraft (p-value)
Bank debt applications
Applications in 2005-7 0.00202 (0.576)
Applications in 2007-8 0.0103 ** (0.000)
Applications in 2008-9 0.00802 * (0.013)
Applications in 2010-11 0.0109 ** (0.000)
Applications in 2011-12 0.0146 ** (0.000)
Observations 2866
Loan (p-value)
Bank debt applications
Applications in 2005-7 0.00651 (0.341)
Applications in 2007-8 0.00709 (0.471)
Applications in 2008-9 0.0262 (0.122)
Applications in 2010-11 -0.00428 (0.420)
Applications in 2011-12 -0.00246 (0.694)
Observations 1235
Notes: Dependent variable is the arrangement fee as a proportion of
the size of the facility. Effects are measured relative to: bank
debt applications in 2001-4; other control variables included as in
table I . * ** indicate significance at 5% and I % correspondingly.
Estimated as pooled tobit with robust standard errors. Excludes
ratios greater than 9 for loans and greater than 6 for overdrafts.
Table 6. Aggregate corporate lending with conditional variance
of share price changes
Dependent variable:
Log difference of All firms SME term
lending to: (PNFCs) SME total loans
Constant 0.0765 ** -0.038 * -0.016
(3.9) (-2.2) (-0.5)
Log real loans(-1)-log -0.029 * -0.059 ** -0.032 *
real profitability (-1) (-2.5) (-5.2) (-1.6)
Long term real -0.0046 ** 0.0022 0.00005
lending rate (-1) (-2.8) (0.9) (0.02)
Log difference of 0.408 ** 0.124 0.055
real loans (-4) (4.6) (1.3) (0.5)
Conditional variance -1.11 -2.279 * -2.124 *
of share prices (-4)
R-bar2 (-1.1) (-2,4) (-2,1)
0.289 0.333 0.081
SE 0.022 0.022 0.023
Period 1991Q2- 1991Q2- 1991Q2-
2012Q2 2012Q2 2012Q2
Dependent variable:
Log difference of SME Unincorporated
lending to: overdraft business
Constant -0.065 * -0.0309
(-2.2) (1.6)
Log real loans(-1)-log -0.029 ** -0.058 **
real profitability (-1) (-3.6) (3.6)
Long term real 0.0016 -0.0014
lending rate (-1) (0.4) (0.9)
Log difference of 0.1552 0.596 **
real loans (-4) (1.4) (6.4)
Conditional variance -3.239 -2.208 **
of share prices (-4)
R-bar2 (-1,9) (3,3)-0.6
0.173
SE 0.039 0.015
Period 1991Q2- 1991Q2-
2012Q2 2012Q2
Sources: BBA, BIS, Bank of England.
Note: t-stat in parenthesis;*,** indicate 5% and 1%
significance levels correspondingly; estimated by OLS.
Table 7. Alternative uncertainty effects in macroeconomic
equations
Type of Share price Inflation
aggregate conditional conditional
lending volatility volatility
All firms -1.111 -0.029
(0.239) (0.022)
SME -3.239 * 0.021
overdraft (0.05) (0.336)
SMEs -2.28 * -0.012
(0.015) (0.34)
SME term -2.124 * -0.025
loans (0.037) (0.054)
Unincorp- -2.265 ** -0.051 **
orated (0.002) 0.000
business
Banking
Type of Growth sector RA
aggregate conditional capital
lending volatility adequacy
All firms -0.003 -0.004
(0.37) (0.113)
SME -0.015 ** -0.012 **
overdraft (0.001) (0.003)
SMEs -0.007 ** -0.007 **
(0.006) (0.002)
SME term -0.008 * -0.007 **
loans (0.031) (0.01)
Unincorp- -0.007 ** -0.007 **
orated (0.006) (0.002)
business
Note: Other variables, not shown, are as in table 6; p-values in
parenthesis.
Figure 1. Bank debt applications by sub-periods
Overdrafts Term loans
01-04 23% 14%
05-08 21% 11%
05-07 17% 11%
07-08 28% 14%
08-09 39% 11%
10-11 20% 10%
11-12 14% 7%
Note: Includes data on all SMEs An SME is defines to have a
demand for overdraft/loan: over 2001-8, if applied for overdraft/
loan; in 2009, if it used overdraft/term loan over the last 12 month;
over 2010-12, if it applied for new or renewed overdraft/loan.
Note: Table made form bar graph.
Figure 2. Credit rating (Dun and Bradstreet/Experian)
2004 2008 2009 2011 2012
Minimal 16% 11% 19% 17% 17%
risk
Low
risk 42% 31% 27% 21% 19%
Average
risk 32% 44% 28% 27% 27%
above
average
risk 9% 13% 12% 27% 28%
No risk
rating 1% 3% 13% 8% 9%
Note: Includes data on all SMEs.
Note: Table made form bar graph.
Figure 3. Bank debt debt rejection by sub-periods
Overdrafts Term loans
01-04 11% 5%
05-07 8% 6%
07-08 15% 9%
08-09 16% 14%
10-11 14% 18%
11-12 19% 23%
Note: Includes date on SMEs with bank debt
Note: Table made from bar graph.
Figure 4. Bank debt margins (a)
Overdrafts Term loans
01-04 2.0 2.3
05-07 0.6 1.3
07-08 0.7 1.8
08-09 3.6 5.3
10-11 3.7 4.0
11-12 3.7 4.2
Note: Includes data on SMEs with bank debt; excludes SMEs
with margins greater than 30pp. (a) Loan/overdraft margins are
calculated as a difference between interest rates charged on
loans/overdrafts and Bank of England base rate at the time of an
interview. Interest rate variables are a combination of a continuous
data and midpoint of ranges. 2011-12 includes margins supplied by
the survey as well.
Note: Table made form bar graph.
Figure 5. Arrangement fees as a per cent of size
Overdraft Term loans
01-04 1.1% 1.5%
05-07 1.7% 1.9%
07-08 2.9% 1.8%
08-09 1.6% 3.7%
10-11 1.7% 1.1%
11-12 2.0% 1.0%
Note: Includes data on SMEs with bank debt. Excludes outliers, i.e.
ration greater than 9 for loan greater than 6 for overdrafts.
Note: Table made form bar graph.
Figure 6. Bank debt involving collateral
Overdraft Term loans
01-04 44% 53%
05-07 36% 57%
07-08 46% 55%
08-09 78% 57%
10-11 42% 51%
11-12 55% 58%
Note: Includes data on SMEs with bank debt.
Note: Table made form bar graph.
