The financial crisis, bank lending and UK productivity: sectoral and firm-level evidence.
Riley, Rebecca ; Rosazza-Bondibene, Chiara ; Young, Garry 等
I. Introduction
There is a range of cross-country evidence suggesting that
long-term growth potential is damaged by major recessions associated
with financial crises (Reinhart and Rogoff, 2011). Oulton and
Sebastia-Barriel (2013) estimate that each year of a banking crisis
reduces output per hour worked by around 1 per cent for each year that
the crisis lasts.
Our main focus is on the slowdown in growth of aggregate labour
productivity. Labour productivity is estimated to have fallen sharply
during the recession of 2008-9, and to have recovered only sluggishly
after that. Figure 1 shows the unusually slow recovery of output per
worker following the most recent recession when compared with the three
other major UK recessions since the beginning of the 1970s.
Cross-country evidence suggests that the productivity slowdown has been
more pronounced in the United Kingdom than in other countries (Office
for National Statistics, 2014).
Understanding the continued weakness of productivity relative to
pre-crisis trends and the scope to make up past losses is of key
importance to macroeconomic policy. Now that the economy is beginning to
recover, a key question is whether any of the lost productivity growth
of recent years will be made up.
Any explanation for the recent weakness of productivity growth
needs to start from the observation that the macroeconomic position of
the UK economy, like most advanced economies during the Great Moderation
period, was broadly balanced with growth of around trend, inflation at
target and with output close to potential. This was in contrast to the
macroeconomic position leading up to all other postwar downturns when
there had been a prior imbalance between aggregate supply and demand and
the recession was caused by the policy tightening needed to bring
inflation under control. On this occasion, monetary policy was loosened
to offset the impact of a sharp contraction in global demand, together
with the consequences of tight credit conditions and greater uncertainty
on demand at home. Low interest rates, together with a lower exchange
rate, provided breathing space to incumbent businesses and allowed them
to absorb the demand shock. That in itself might have accounted for
weaker productivity in some businesses. But low interest rates would
typically encourage businesses to increase their capital intensity and
boost productivity. So it is unlikely that low interest rates alone
could account for the weakness in aggregate productivity.
[FIGURE 1 OMITTED]
Against that background it is possible to distinguish two broad
explanations for the weakness of UK labour productivity following the
financial crisis.
The first broad explanation emphasises the impairment of the
banking sector and the effect of tight credit conditions on the supply
side of the economy. (1) Figure 2 shows that bank lending to companies
fell more sharply in the aftermath of the most recent recession than it
did in the three other post-1970 recessions. According to this view, a
lack of credit availability stunted the development of
high-productivity, mainly young and small bank-dependent firms. The
absence of competitive pressure from such companies, together with low
interest rates and bank forbearance, then provided protection for older
established companies. Their continued survival led to congested markets
and reduced the profit opportunities available to more dynamic
businesses with the result that the normal reallocation of capital
towards stronger businesses did not happen and aggregate productivity
stagnated. According to this explanation, productivity would pick up and
wages grow without generating additional inflationary pressure once the
banking sector was repaired. Simply stimulating demand without repairing
the banking sector would be inflationary because the stagnation of
productivity reflected a weakness on the supply side of the economy.
[FIGURE 2 OMITTED]
The second broad explanation for the weakness of UK productivity
emphasises labour market flexibility and the willingness of workers to
accept nominal pay freezes in some instances and real wage reductions
more generally in order to keep their jobs in a weak demand environment.
(2) The relative weakness of wages would also have encouraged
generalised labour hoarding whereby businesses held on to employees to
avoid firing costs and kept them occupied looking for new business and
servicing existing customers. But this explanation goes beyond
traditional cyclical labour hoarding in that it may also account for why
businesses replaced workers who quit and maintained employment levels
even when output remained weak. According to this explanation,
productivity would pick up and wages grow without generating additional
inflationary pressure if demand could be stimulated.
In principle, these hypotheses can be distinguished by their
empirical predictions, especially by examination of disaggregated data.
If the productivity slowdown were caused by lack of access to credit it
would be associated with heterogeneous changes in productivity across
different types of companies and sectors according to their reliance on
the banking sector. According to this view, bank-dependent companies
would grow more slowly than otherwise or not get started at all, while
other companies might survive longer than they otherwise would. By
contrast, the productivity slowdown would be more widespread according
to the second broad explanation.
In practice, it is likely that both of these explanations have been
at work to some extent over the course of the recession and the
subsequent weak recovery. The aim of this paper is to assess the
evidence for and against the banking sector impairment explanation of
the weakness in labour productivity. To preview our findings, we find
clear evidence that the banking sector crisis affected credit supply to
businesses and caused bank lending to decline. This is a necessary
condition for the banking sector impairment explanation of the labour
productivity slowdown. But we do not find clear evidence of the
heterogeneity in performance between different businesses and industrial
structures that would have been expected if banking sector impairment
had been the key factor holding back productivity growth. This suggests
either that banking sector impairment is acting on businesses in ways
that are not well understood or (and) that other factors have also been
important.
Section 2 describes the evidence that the banking sector crisis
affected the financing conditions facing UK businesses--a necessary
condition for this explanation of the productivity slowdown. Section 3
assesses the evidence for whether the slowdown in productivity in
different sectors matches their dependence on banks. Section 4 assesses
the evidence at the level of individual businesses. Section 5 concludes.
2. Evidence of banking sector impairment on supply of finance to
companies
The global financial crisis that began in the summer of 2007 had a
profound effect on banks and the provision of finance to businesses in
the United Kingdom. The crisis revealed that the banking sector as a
whole was undercapitalised and led to the exit of some institutions, the
rescue of others and widespread injections of public and private
capital. One indicator of the intensity of the crisis is the cost of
insuring the unsecured debt of banks against the risk of default as
given by Credit Default Swap (CDS) premiums. Figure 3 shows the sharp
rise, prolonged elevation and recent fall in the CDS premiums of the
major UK banks.
Prior to the crisis, the CDS premiums of the major UK banks had
been close to zero, consistent with bank default being considered a very
low probability event by market participants. The prospect of default
meant that the cost that banks had to pay to fund themselves in
wholesale markets rose sharply for all banks, albeit by differing
amounts, with especially severe consequences for those banks that were
reliant on wholesale funding.
The higher cost of bank funding relative to interest rates on
completely safe assets such as government bonds could be expected to
have a number of consequences for the banks and their business
customers. In particular, it was likely to lead to a tightening of
credit conditions. Consistent with this, Gilchrist and Zakrajsek (2011),
in an investigation of credit conditions in the United States, found
that CDS premiums on bank debt are highly correlated with their
preferred indicator of credit conditions.
There are a number of channels by which unexpectedly higher bank
funding costs would be likely to impact on the business customers of the
banks. First, new loans would become more expensive as banks passed on
the higher cost of market funding. Second, while companies with existing
agreed loan facilities would be insulated from the effect of the higher
bank funding costs, they would nevertheless expect to pay higher spreads
on loans once those facilities expired. Third, to the extent that banks
were lending on pre-arranged loans at below their now higher cost of
funding, they would have an incentive to withdraw existing facilities if
that were contractually possible, and reprice them.
In addition to the effects of higher bank funding costs, it is
likely that bank credit supply became more restricted due to lower risk
appetite and reduced competitive pressure caused by the exit or change
in business model of some lenders. These factors may have led to
costlier terms and more stringent conditions on new bank loans than
would be justified by higher bank funding costs alone. They may also
have led to the withdrawal of banks from riskier forms of lending.
[FIGURE 3 OMITTED]
One manifestation of the sharp tightening of credit supply
conditions was the fall in the stock of loans to businesses shown
earlier in figure 2. While this was also likely to have reflected some
weakening of credit demand, Bell and Young (2010) found that tight
credit supply was likely to have been the dominant influence in the
three years immediately following the start of the crisis in summer
2007. HM Treasury (2013) also examined this issue in their more recent
assessment of the effect of credit conditions on the economy in the
context of the RBS bad bank review.
These studies of credit supply conditions draw on a number of
different sources: indicators of the price of bank credit; surveys of
lenders; surveys of businesses; and evidence from non-bank sources of
finance.
The price of bank credit
Measuring the impact of tighter credit supply on the cost of bank
finance to businesses is not straightforward. In particular, there are
no comprehensive data on quoted interest rates on new loans to
businesses split by credit quality. While there are good quality data on
the average lending rates actually paid by businesses, it is likely that
these rates understate the cost of new credit to the average business
when banks are focusing their lending mainly on collateralised lending
to good quality customers. Moreover, there is no clear distinction in
the available data between interest rates paid on new and existing
loans; for some lenders, new business includes companies drawing down
existing facilities with pre-arranged costs, or loans that have been
repriced in line with changes in reference rates. Reflecting both of
these factors, and given that existing facilities are likely to have had
lower spreads than those on genuinely new credit since the start of the
crisis, measured effective rates are likely to have underestimated the
rates at which companies have been able to arrange genuinely new loans
from banks in practice.
Figure 4 shows measures of the spread on different forms of
business debt finance. While the measured effective new lending spread
may be a good indicator of the cost of borrowing from banks for good
quality businesses drawing on existing facilities, the cost of genuinely
new borrowing is likely to have been better measured by the other
spreads according to the risk of the business. This would indicate that
the cost of new borrowing relative to Bank Rate peaked in around early
2009 and has since declined, albeit with a further rise in 2011 and 2012
in response to the effects of the Euro Area crisis on UK banks.
[FIGURE 4 OMITTED]
Surveys of lenders
Evidence from the lenders suggests that price and non-price terms
on loans rose during the financial crisis. Lenders responding to the
Bank of England's Credit Conditions Survey reported that spreads
increased markedly across all types of lending, particularly during 2008
and early 2009. The net percentage balances of lenders reporting
increased fees and commissions on loans to companies also rose.
According to the Credit Conditions Survey, the availability of lending
to businesses contracted markedly during 2007 and 2008, and has
gradually improved since then. Figure 5 plots the cumulated net
percentage balance from the Credit Conditions Survey alongside the Bank
of England's indicator of corporate credit spreads (Inflation
Report, February 2014, p. 14). These separate indicators broadly move
together and suggest that bank credit supply conditions were at their
most adverse for businesses in early 2009, and have since improved.
[FIGURE 5 OMITTED]
An independent review led by Sir Andrew Large into the lending
practices of RBS, the lender with the largest share of the SME market,
found widespread failure when it reported in November 2013. It concluded
that 'for some time RBS has not succeeded in supporting the SME
sector in a way that meets either its own financial targets, or the
expectations of customers and external stakeholders'.
Surveys of businesses
Surveys of businesses point to divergent trends in credit
conditions for large businesses, especially those with access to the
debt capital markets, and smaller businesses that are reliant on banks
for external finance. Respondents to the Deloitte Chief Financial
Officer surveys, mainly large businesses with capital market access,
indicated that bank credit became increasingly unattractive during
2007-8, but have become progressively more attractive since then. This
evidence suggests that the credit crisis had effectively ended by around
2011 for those types of companies. But this does not appear to be the
case for smaller businesses. Figure 6 shows responses to the Deloitte
CFO survey on the cost and availability of finance and compares these
with the responses of smaller businesses when the same questions were
added to the Federation of Small Business (FSB) quarterly survey.
Smaller businesses continue to report that credit is costly and not
easily available.
[FIGURE 6 OMITTED]
Even within the population of smaller businesses, there is
substantial variation in the extent to which businesses want to access
external finance and the conditions they face. The SME Finance Monitor,
the most comprehensive source of information on the financial conditions
facing SMEs, classifies a substantial proportion of SMEs, as 'happy
non-seekers', businesses not looking to increase their borrowing
facilities. In the September 2013 wave of the survey 78 per cent of SMEs
were 'happy non-seekers' of finance, based on their behaviour
in the twelve months prior to interview. These SMEs had neither sought
new or renewed loan or overdraft facilities in the twelve months prior
to interview, nor felt that anything had stopped them from applying.
Around 40 per cent of SMEs were classified as permanent non-seekers.
Seven per cent of SMEs were classified as 'would-be seekers'
of finance from banks, but were reluctant to apply for fear of being
rejected or because the process was felt to be burdensome.
But the SME Finance Monitor also points to the difficulty that
businesses who would like external finance have faced in arranging new
finance. Figure 7 shows that applications for loans and overdrafts from
businesses with an existing banking relationship were more likely to be
successful. In 2012, over half of applications from first-time
applicants were rejected. This suggests that new businesses would find
it difficult to raise finance in this environment. In addition existing
businesses would have difficulty in switching to a different lender.
[FIGURE 7 OMITTED]
In a detailed investigation of the SME Finance Monitor, Armstrong
et al. (2013), found that rejection rates for both overdrafts and term
loans were significantly higher in the period from 2008-9 onwards, even
controlling for risk factors, which is indicative of constraints to the
supply of credit.
It is possible though that some incumbent companies have faced
easier credit conditions than is warranted by their financial condition.
According to the 'zombie' hypothesis of Caballero et al.
(2008), banks may have shown forbearance towards businesses to avoid
crystallising losses that would appear on their own balance sheets. But
a recent Bank of England analysis (Arrowsmith et al., 2013) found that
outside the commercial real estate (CRE) sector only around 6 per cent
of companies, amounting to 14 per cent of bank exposure to non-CRE SMEs,
were benefitting from bank forbearance in 2013.
Non-bank sources of finance
One of the features of aggregate company finances since the
financial crisis has been the shift away from bank financing to other
sources. The decline in bank lending to companies since early 2008 has
been associated with less negative flows of other types of finance
alongside a rise in financial saving. While net external finance raised
has been weak, the change in the contribution from bank loans as a
source of finance since the crisis has been remarkable (figure 8). Some
large companies have been able to access the debt capital markets at
cheaper rates than the banks, providing a clear incentive to
disintermediation. In a recent study using company-level data, Farrant
et al. (2013) found that much of the finance raised was used to
restructure the balance sheets of the few, very large, companies able to
access the debt capital markets.
[FIGURE 8 OMITTED]
Summary of evidence on finance for business
Evidence on the impact of the banking crisis on credit conditions
facing businesses is summarised in table 1. It emphasises that there is
substantial heterogeneity in the extent to which different businesses
are likely to have been affected by the banking sector crisis. The
banking crisis is likely to have had the most adverse effect on new
businesses with little existing collateral, businesses seeking to expand
quickly and businesses with stretched balance sheets. But some, mainly
mature businesses are likely to have experienced better credit
conditions.
3. The effect of bank credit conditions on the performance of UK
companies: sectoral analysis
One means of identifying the impact of the banking sector crisis on
productivity is through an empirical analysis of productivity trends in
different industrial sectors. If the banking crisis has been important
in affecting productivity, then it might be expected that productivity
would have deteriorated by more in the most bank dependent sectors. The
aim of this section is to assess the evidence for whether the slowdown
in productivity in different sectors matches their dependence on banks
prior to the crisis and their vulnerability to a tightening of credit
conditions.
Bank dependence
In order to assess this issue, we first bring together evidence on
the bank-dependence of different industrial sectors in the United
Kingdom. We focus on the main private sector industries, but we exclude
those where productivity is likely to be affected by sector-specific
issues, such as agriculture, extraction and finance. Table 2 provides a
judgement-based ranking of one-digit industrial sectors according to
their bank dependence, with the most bank-dependent sectors at the top
of the table. The ranking is based on a number of different measures of
each sector's reliance on bank finance.
The ranking aims to classify industries according to the extent to
which businesses are vulnerable to a sudden shift in banking sector
credit conditions. In principle, a good measure of this would be the
proportion of industry output that is supplied by businesses who rely on
banks to provide the marginal source of finance. But, in practice, such
a measure cannot easily be calculated from existing data and so has to
be approximated by other measures.
The most common measure in the academic literature of the external
finance dependence of an industry is based on the proportion of capital
expenditures financed by external funds in an average firm, as
originally suggested by Rajan and Zingales (1998). This is intended as
an inherent, structural, time-invariant measure of the dependence of an
industry on external finance, reflecting technological factors such as
the initial scale of projects and their gestation periods. While this
measure is widely used in cross-country studies, it has certain
deficiencies (see, for example, von Furstenberg and von Kalkreuth,
2006). First, focusing on gross borrowing flows over a particular period
may misrepresent the exposure of an industry to banking shocks if that
borrowing is repaid quickly. For this reason it is better to focus on
debt stocks. Second, focusing on financing flows as a share of
investment will overstate the importance of external finance in
industries that are not very capital intensive. For this reason it is
better to focus on a measure of the stock of loans outstanding as a
share of output. Such a measure shows the bank liabilities of an
industry relative to the output that generates the means to service
those loans.
The first column of figures in table 2 shows this measure of
bank-finance dependence--the stock of gross loans outstanding relative
to the value of the output of the industry--for each industry prior to
the crisis. According to this measure, the accommodation and food
sector, consisting of hotels, pubs and restaurants, was the most
bank-dependent industrial sector at the start of the crisis; the
information and communication services sector was the least
bank-dependent. This measure fell sharply in the years following the
crisis in all industries, except information and communication services,
consistent with a widespread credit crunch.
There are some potential disadvantages with this measure that other
measures shown in the table might rectify. One disadvantage is that it
takes no account of the bank deposits that businesses in the same
industry might also hold. So an alternative measure uses in the
numerator gross loans net of deposits placed by businesses in the same
sector. This is shown in the third column of figures in table 2. This
shows that there are certain industries with high levels of debt before
the crisis, such as arts and entertainment, that had even higher levels
of bank deposits. Such industries should be better placed to withstand a
credit crunch since, in aggregate, businesses in that industry have
sufficient cash balances to draw on if debt becomes more costly or
difficult to access. Against this, it is also likely that the debt and
deposits are distributed unevenly, and possible that the businesses with
high levels of net debt are more capable of faster growth than the
businesses with net deposits that may be immune to the effect of tighter
credit conditions.
Another potential disadvantage with the gross debt to output
measure of bank dependence is that businesses with high debt levels
secured against property might not need credit to carry out normal
operations to the same extent as businesses that need to roll over bank
debt on a regular basis to fund working capital needs. An alternative
measure is the cumulative growth in the stock of loans in the five years
leading up to the financial crisis. This is shown in the sixth column of
figures in table 2. This measure shows certain industries expanded their
bank borrowing particularly rapidly in the run-up to the financial
crisis and so might be more vulnerable to a sudden stop in credit
growth. These industries included some of the business-facing service
sectors--especially administration and support services and professional
and scientific services--that had low levels of gross debt and negative
debt net of deposits. The construction industry also had strong growth
in gross debt prior to the crisis.
A stock equivalent of the Rajan-Zingales measure is the ratio of
bank debt outstanding in an industry relative to the capital stock,
shown in the fifth column of table 2. This is relatively high in the
accommodation and food industry, consistent with other measures of bank
dependence. But it is highest in professional and scientific services,
which otherwise does not stand out as being bank dependent. But this is
probably a misleading indicator in this case as while debt is high in
relation to capital, capital itself is relatively unimportant in
production in that industry.
A further measure of bank-dependence is the proportion of total
assets due to small and medium-sized companies in each industry
identified as having bank debt. This measure is derived from
company-level data. This measure also points to vulnerability in the
construction, accommodation and food, and wholesale and retail service
sectors.
Overall, there is some ambiguity about the ranking of different
industries according to their bank dependence. Some industries score
highly on some measures but not others. What does appear clear though is
that construction is one of the more bank-dependent sectors, while
information and communication is not.
Simple framework
In order to frame the discussion of how bank dependence might
affect the productivity of an industry, we sketch out a simple
framework. Using a simple Cobb-Douglas production function, labour
productivity in a sector is given by:
Y/N = A[(1 + [lambda]).sup.1-[alpha]][(K/N).sup.[alpha]] (1)
where Y is output, N is employment, [lambda] is a measure of total
factor productivity (which changes over time at the rate of labour
augmenting technical progress), K is the capital stock, [alpha] is the
capital share and A is a constant. By this expression, productivity
growth in the long run derives from technological improvements that are
usually taken to be exogenous, perhaps due to organisational change or
by taking on or training better-skilled employees (reflected in
[lambda]), and endogenous increases in the capital-labour ratio
(capital-deepening) in response to changes in relative factor prices. In
the short run, productivity may also change due to changes in the rate
of utilisation of factors.
According to this framework, banking sector impairment would affect
productivity by making it more difficult or expensive to finance capital
spending and cause affected businesses to choose a less
capital-intensive mode of production. If all businesses were the same
and the amount of labour available were fixed then this would reduce the
overall level of output that the economy could supply. The decline in
aggregate output would be a reaction to a lower capital stock brought
about by more expensive bank credit.
In an economy made up of different sectors and businesses with
varying needs for bank credit, persistent banking sector impairment
would lead to a shift in the share of output away from the more
bank-dependent sectors and businesses. Its effect would also depend on
how wages and the cost of capital responded to lower output. For
example, if the labour market was very flexible and wages fell relative
to the cost of capital in the whole economy, then labour productivity
would fall even in sectors that were not bank dependent as labour was
substituted for capital in response to the change in relative factor
prices. In this case, the decline in aggregate output would be a
reaction to a lower capital stock and to it not being allocated
efficiently.
This can be formalised since, using (1), productivity growth--the
change in log productivity--between any two dates is given by:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)
If it could be assumed that the rate of growth of total factor
productivity was constant over time in each sector and the only shock to
the optimal capital-labour ratio was due to the banking crisis, then
this would imply that sectoral productivity growth could be written as:
[DELTA] ln [y.sub.iT] = [DELTA] In [y.sub.iT-1] -
[[theta].sub.i][B.sub.T] (3)
where [[theta].sub.i][B.sub.t] represents the bank dependence of
each industry and the size of the aggregate banking sector shock. In
other words, under these assumptions, productivity growth would be equal
to what it was in the previous period less an amount that would vary
across industries according to their bank dependence.
So, using this simple framework, and the evidence on bank
dependence, the effect of banking sector impairment would be to reduce
productivity relative to what it would otherwise have been in all
sectors to some extent, but the largest declines relative to the
counterfactual would be in construction, manufacturing and the
consumer-facing service sectors (accommodation and food, wholesale and
retail), whereas the smallest falls would be in the business-facing
services sectors such as information and communication services.
Productivity and bank dependence
Table 3 summarises productivity performance in the one-digit
industries ranked according to their bank dependence, with the most
bank-dependent sectors at the top of the table.
On average, labour productivity in these industries had grown by 14
per cent in the five years leading up to the crisis, an annual rate of
around 3 per cent. It then fell sharply during the recession, recovered
a little and then stagnated. The swing in productivity--the turnaround
in the five-year growth rate--amounted to -19 per cent, similar in
aggregate to the deviation of productivity from the continuation of a
pre-crisis linear trend. The deterioration in productivity performance
appears to have been fairly widespread. The smallest decline appears to
have been in construction, where productivity growth was negative in the
run-up to the crisis. The largest declines were in the manufacturing,
transport and storage, and information and communication industries,
where productivity growth was strongest in the run-up to the crisis.
There does not appear to be any clear relationship between either
the turnaround in productivity growth or the gap between productivity
and a linear pre-crisis trend and the bank dependence of a sector. That
is, there is no evidence to support the relationship highlighted in
equation (3). On the face of it, this is evidence against the banking
sector shock being the primary cause of the weakness in productivity.
This assessment is based on a very simple counterfactual --that in
the absence of the banking sector crisis productivity growth would have
continued at the same rate as in the previous period. But it is possible
that the assumptions used to derive equation (3) may not hold in
practice. In particular it is likely that output and capital investment
in many of these sectors would have changed by differential amounts in
the absence of a banking sector crisis, especially as global economic
conditions had deteriorated. Any such changes would have altered the
path of labour productivity in the various industrial sectors and so
affected our assessment of the impact of the crisis. In order to
quantify the possible importance of this, we use pre-crisis
relationships to adjust for different post-crisis trends in output and
capital growth in two ways.
First, productivity is naturally affected by cyclical influences.
Productivity typically falls relative to trend in a cyclical downturn as
employment is slow to adjust to the weakness of demand and output. This
would suggest that some of the slowdown in productivity growth is likely
to have been caused by the weakness of demand. Indeed, Ashworth,
Goodhart et al. (2013) claim that the productivity slowdown is a largely
cyclical response to persistent weak demand.
To assess this, the productivity shortfall can be cyclically
adjusted by estimating what level of productivity would have been
expected in each industry given the normal relationship between
employment and output. The outcome is shown in the fifth column of
figures in table 3. As can be seen, this adjustment does reduce the size
of the estimated aggregate productivity shortfall considerably, although
it does not eliminate it. In fact, as is illustrated later for two
industries in figures 9 and 10, it explains most of the productivity
decline in the immediate aftermath of the recession. But it does not
explain the subsequent continued weakness; labour hoarding as demand
weakens can only be expected to last for a limited time. It is also more
significant in some industries than others, reflecting their different
rates of output growth since the crisis. For example, the largest gap
between output five years after the start of the crisis and its
pre-crisis level is in the transport and storage industry (seventh
column of figures in table 3). Once allowance is made for the sharp
decline in output experienced in that industry, productivity is
estimated to be 16 per cent higher than the pre-crisis relationship
would suggest. That is, given the fall in output, it is surprising that
productivity there did not fall by even more.
Second, as is emphasised by equation (1), labour productivity is
also affected by the extent of capital deepening in each industry.
Indeed this is the main channel by which the banking crisis is expected
to affect labour productivity. Consistent with this, the fast upward
pre-crisis trend in labour productivity in some industries was
associated with rapid investment and fast capital stock growth rather
than fast growth in total factor productivity (TFP). In deriving
equation (3) it was assumed that the pre-crisis capital deepening trends
would have continued if it were not for the banking crisis. But this may
not be the case. It is possible that investment and capital growth in
some fast growing industries was only temporarily strong and was about
to slow independently of the banking sector crisis. Moreover, the
increase in uncertainty in the post-crisis period was likely to have
caused businesses to postpone capital investment even if those
businesses were not reliant on the banking system to finance it. And it
is notable that in many industries --construction, wholesale and retail,
manufacturing, information and communication--labour productivity growth
simply stopped in the post-crisis period, irrespective of its pre-crisis
trend. This could have been associated with a sudden stop in capital
stock growth as investment fell sharply for reasons independent of the
banking crisis. So it is possible that the slowdown in productivity
growth in industries not directly affected by the banking
crisis--information and communication, for example--could be accounted
for by an independent reduction in investment.
A significant difficulty in assessing this is that capital stock
estimates have been especially unreliable in recent years and the ONS
has stopped updating them beyond 2010. In what follows, we use ONS
estimates of the net capital stock at constant prices until 2010. Beyond
that, we assume that the capital stock in each sector grows at the same
rate as in the previous two years. Pessoa and van Reenen (2013)
calculate TFP on the basis of different assumptions about capital stock
growth and find a relatively small fall in TFP after the crisis that
they attribute to under-utilisation of factors in the recession and
uncertainty-driven misallocation. But Oulton (2013) argue that the
estimate of capital per worker used by Pessoa and van Reenen is
incorrect as it is based on too high an estimate of the pre-crisis
capital stock. Clearly, the lack of reliable data on capital at the
sectoral and aggregate level makes it difficult to be confident in the
conclusions of this analysis.
Based on the available capital stock data, we estimate what level
of productivity would be expected in each industry given the normal
relationship between employment, output and capital stock growth. To do
this we embed equation (1) in an employment equation written in error
correction form which is estimated for each sector:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)
We then simulate what productivity would have been in each industry
if the pre-crisis relationship between employment, output and capital
had continued to hold. The outcome is shown in the sixth column of table
3. It shows that in some industries there is no longer a significant
abnormal weakness in productivity once account is taken of the normal
pace of adjustment of employment to movements in output and the capital
stock, implying that in these industries TFP is not unusually weak.
Figures 9 and 10 show the level of productivity relative to the
implied trend in a clearly bank-dependent industry (construction) and
one that is clearly not bank dependent (information and communication).
The figures suggest that much of the weakness of labour productivity in
those industries can be explained once the weakness of both output
growth and capital investment are taken into account. In other words,
labour productivity in those industries is not abnormally weak
conditional on the behaviour of output and measured capital
accumulation. That suggests that there was little change in total factor
productivity in those industries following the banking crisis.
[FIGURE 9 OMITTED]
[FIGURE 10 OMITTED]
One of the striking features of table 3 is that there does appear
to be a correlation between the size of the output shortfall relative to
its pre-crisis peak and the bank dependence of the industries. (3)
Output is generally below its previous peak in the more bank-dependent
sectors and above it in the largely business-related service sectors
that are less reliant on the banking sector. In other words, bank
dependence may explain why output growth has remained weak in some
sectors of the economy. And, given that all sectors are reliant to some
extent on the banking sector, it suggests that at least some of the
weakness in output in all industries might be due to the impairment of
the banking sector. Further, even in the largely bank-independent
information and communication sector, for example, the slowdown in
capital accumulation might be a reaction to the effects of the crisis
operating through other business sectors and impacting on demand and
confidence there.
4. The effect of bank credit conditions on the performance of UK
companies: company level analysis
The sectoral analysis provides some evidence consistent with the
supply capacity of the economy having deteriorated by more in the most
bank-dependent sectors of the economy. But the evidence is mixed and not
conclusive by itself. If the deterioration was caused by the banking
crisis then firm-level evidence should also point to consistent changes
in the composition of activity among businesses. This section attempts
to identify the impact of the banking sector crisis through an empirical
analysis of productivity trends using information on individual
businesses.
Framework
Increases in the productivity of private sector businesses derive
from two sources: first, improvements made within existing businesses,
perhaps due to the introduction of new capital, better technology,
organisational change or by taking on or training better-skilled
employees; and, second, improvements made in the types of businesses
that exist, where the process of market selection encourages the exit of
poor performers, the entry of highly productive newcomers and a shift in
market share to more productive incumbents. In a seminal study of
productivity growth in UK manufacturing from 1980-1992, Disney et al.
(2003) found that external restructuring accounted for around half of
labour productivity growth.
The importance of external restructuring in explaining a
significant component of productivity growth in normal times suggests
mechanisms by which credit market impairment might impact on this
process and cause overall productivity growth to weaken. This arises
because the process of implementing technological improvement requires
capital and better-performing businesses will need more capital in order
to expand.
This is likely to be more of an issue when better performing
businesses are younger and financially immature. Such businesses may
struggle to get access to finance without paying a substantial premium
relative to financially mature businesses with a track record of
retaining profits within the business and capital that may be used as
collateral for loans.
Khan and Thomas (2013) set out a formal, general equilibrium model
where collateral constraints limit the amount that young firms can
borrow even in normal times. Such constraints prevent these young firms
from reaching their desired capital stock quickly. Instead they
accumulate capital gradually and build up debt until they generate
enough profits to enable them to meet further investment spending out of
their own resources and begin to pay down debt. The presence of
collateral constraints in this model means that capital is inefficiently
allocated in that young companies have too little capital, are not as
productive as they could be and take up too small a share of the
economy.
Khan and Thomas show that a tightening of collateral constraints in
this model can reproduce some of the salient features of the recent
recession. In particular capital investment, employment and output all
fall and measured labour productivity declines even though, by
construction, there is no change to the technological capability of
individual companies (TFP). Moreover lending to companies falls and
young firms are more hindered in their investment activities relative to
the pre-shock economy. Because of this they take longer to outgrow
financial frictions and begin producing at a scale consistent with their
capability. This brings about a change in the distribution of capital
within the economy. The financially constrained companies remain smaller
for longer, while unconstrained, already large companies grow larger as
they expand to fill the gap left by constrained small companies. This
creates greater dispersion in the marginal product of capital which
highlights the gains that could be made from reallocation of capital
towards the constrained. In the economy as a whole, the average marginal
product of capital rises and productivity falls as the effect of lower
capital investment by young firms dominates the effect of greater
capital investment by mature ones.
Caballero et al. (2008) provide an alternative explanation for why
an impaired banking sector might contribute to a slowdown in aggregate
productivity. This also highlights the role of capital reallocation in
the growth process. Caballero et al. examine the impact of the Japanese
banking crisis on the macroeconomic stagnation that began in the early
the 1990s. They find that it had a surprising effect on job creation and
destruction. In particular they find that, rather than rising, job
destruction fell in sectors where there was a disproportionate number of
'zombie' companies who had been kept alive by banks seeking to
avoid crystallising losses. This had adverse implications for
productivity.
Unlike Khan and Thomas, who treat firm entry and exit as exogenous
factors, Cabellero et al. focus on the impact on overall productivity of
choices about firm entry and exit. In their model, potential new
entrants have a technological advantage over incumbents, but only enter
if they are likely to make sufficient profits to offset their start-up
costs. Caballero et al. use this model to analyse the impact of bank
forbearance where banks choose to protect poorly performing
incumbents--so-called 'zombies'--from the effects of a
recession. The effect is that exit of less productive firms is reduced,
while entry of potentially more productive firms is deterred. The
overall effect is that reallocation of capital to more productive uses
is hindered and aggregate productivity is held back.
The insights from this discussion of these channels from an
impaired banking sector onto companies are summarised in table 4. This
highlights that an impaired banking sector is likely to impact on
different types of businesses in possibly opposing ways. It stresses the
importance of looking at changes in the behaviour of different types of
business in identifying the impact of an impaired banking system on
productivity. In aggregate these changes should imply a reduction in
measured productivity in large part because the allocation of production
across firms becomes less efficient.
Data description
To investigate this further we use data on individual businesses
from the Annual Respondents Database (ARD), a micro-dataset based on a
register of businesses. Data on businesses in all major sectors of the
economy are available from 1997 up to 2011; data are available for
manufacturing back to 1974. Further details are available in Riley et
al. (2014).
The micro-dataset shows that there is significant turnover in the
business population. Looking across any two periods, businesses can be
broken down into those that exist throughout the period (survivors),
those that exist at the start of the period but not the end (exitors),
and those that exist at the end of the period but not the start
(entrants). Looking across different four-year periods (i.e. 2003-6 and
2007-10), between 65-70 per cent are survivors while around 30 per cent
are entrants or exitors. Survivors tend to be larger businesses and
account for around 85 per cent of total employment, whereas entrants and
exitors account for 12 and 15 per cent respectively.
Decomposition of overall productivity
If lack of access to credit and forbearance towards struggling
unproductive businesses were responsible for the weakness in
productivity after the crisis then this would be expected to show up in
the ways set out in table 4. In particular there would be a fall in the
extent to which productivity growth is explained primarily by
reallocation towards stronger businesses. To assess this we use the data
on individual businesses to break down overall productivity growth into
four basic components, following the methodology of Griliches and Regev
(1995), according to the following identity:
Overall productivity = average productivity growth
growth within surviving businesses
+ reallocation towards more
productive surviving businesses
+ reallocation towards new
businesses
+ reallocation from exiting
Businesses
[FIGURE 11 OMITTED]
The results of the decomposition of productivity growth on an
annual basis from 2002-11 are shown in figure 11. In the pre-crisis
period (2002-7), productivity growth in this sample of businesses
averaged 3.2 per cent per annum. Of this, the 'within'
component contributed 1.5 percentage points, while external
restructuring contributed 1.7 percentage points. This is similar to the
Disney et al. (2003) findings that on average around half of
manufacturing productivity growth is accounted for by growth within
companies and half by restructuring towards more efficient businesses.
New entrants typically detract from productivity growth when they first
begin their operations, while exits boost productivity growth reflecting
the lower productivity of those going out of business relative to
survivors. Year-to-year movements in productivity growth are driven
largely by movements in the within-business component, both before and
after the banking sector crisis.
According to the framework discussed earlier, banking sector
impairment would be expected to be revealed primarily by a fall in the
contribution of the reallocation components as bank-dependent dynamic
businesses are prevented from growing and taking market share away from
their less efficient rivals. But there is little evidence that this was
the driver of productivity weakness following the banking sector crisis.
As in the aggregate statistics, productivity weakened in 2008 and fell
sharply in 2009. While it recovered some lost ground in 2010 and 2011,
it still remained around 1 per cent below its 2007 level four years
after the crisis. There does appear to have been a small fall in the
reallocation components of productivity growth following the crisis,
consistent with some modest direct effect of the banking crisis onto
productivity growth, but the main driver of the weakness of productivity
is the within-business component. This fell especially strongly in 2009,
consistent with the cyclical behaviour of productivity identified in the
earlier sectoral analysis. These findings are broadly in line with those
reported by Barnett et al. (2014), although they find a slightly larger
negative contribution from reallocation after the crisis.
Riley et al. (2014) also find that there is some evidence that the
reallocation component of productivity growth fell by more in the
bank-dependent industries following the crisis, also consistent with
some impact of the banking crisis on productivity growth.
While the large negative contribution of the within component may
reflect a typical cyclical reaction of labour productivity to the
weakness of output growth and subdued capital investment, the
contribution of this 'within' effect would be expected to be
smaller in a banking crisis than in an equivalent-sized recession not
associated with a banking crisis. To assess this, table 5 compares the
decomposition of manufacturing productivity growth in the recent
recession with that of the early 1990s, which was driven largely by a
tightening of monetary conditions rather than a banking crisis. (4)
There are some similarities in the downturns in manufacturing
output and productivity growth in the recessions that began in 1990 and
2008. Manufacturing output fell very sharply in both cases and was
accompanied by a slowdown in productivity growth. In the sample of
businesses used for the decomposition, manufacturing productivity fell
by 11 per cent in 2009 and then rose by 13 per cent in 2010, back to
around its 2008 level. In the earlier recession, productivity rose on
average over the years that output was falling.
But the evidence is not consistent with the banking sector crisis
playing a key role in explaining the slowdown in productivity in the
most recent recession. To see this, compare the decomposition of
productivity over the period immediately following the recession. The
final column of table 5 shows that over the period from 1990 to 1992,
output in the manufacturing sector fell cumulatively by 10.9 per cent, a
little larger than the cumulative fall of 8 per cent from 2008 to 2010.
Despite the large fall in output, productivity rose cumulatively by 5.9
per cent in the early 1990s period, in contrast to an increase of 0.8
per cent over 2008-10, implying a much sharper reduction in employment
in the earlier period. The difference in productivity performance across
the two periods is almost completely due to the within component
contributing positively to productivity growth, by 2.3 percentage
points, in the former period, in contrast to a negative contribution of
3 percentage points in the latter period. Indeed, external restructuring
is calculated to have made broadly the same contribution to growth in
the most recent recession as in the early 1990s, contrary to what would
be expected if the recent productivity slowdown was driven by banking
sector impairment.
This evidence casts doubt on the banking crisis being a major cause
of the different performance of productivity in the most recent period.
This is not to say it did not have some effect. As we have seen, there
is evidence that the reallocation component of productivity growth fell
by more in the bank dependent industries following the crisis. But the
major driver of the weakness of productivity appears to a common
factor--possibly a lack of confidence driven by uncertainty--that has
not just affected businesses and sectors that are reliant on bank credit
to grow.
5. Conclusion
This paper has explored the evidence on the extent to which the
weakness in labour productivity since 2008 can be explained by the
banking sector crisis.
There are good reasons for expecting there to have been some link.
The slowdown in capital accumulation and weak growth since 2008 broadly
match the predicted response to a credit shock of sophisticated models
such as Khan and Thomas (2013). And cross-country analysis covering long
spans of data such as Reinhart and Rogoff (2011) and Oulton and
Sebastia-Barriel (2013) point to an empirical regularity that growth
slows in the wake of banking crises.
The necessary condition for there to have been some effect appears
to have been satisfied in that there is clear evidence that the banking
crisis impacted on the supply of credit, especially for small companies.
However, the supply of credit for large companies, who account for
roughly 60 per cent of investment in the UK, was back to normal by 2011.
That would have suggested the possibility of a large-company led revival
in productivity growth in 2012 and 2013 if banking sector impairment had
been the key factor holding back productivity growth.
More generally, we have found little clear evidence of the
heterogeneity in performance between different businesses and industrial
sectors that would have been expected if banking sector impairment were
the main impediment to productivity growth. While sectoral analysis
suggests that output has been weakest in the most bank-dependent
industries, consistent with some impact of tight credit conditions on
supply capacity, the firm-level analysis suggests little change in
reallocation of productivity growth across businesses. Moreover a
comparison of the most recent recession and that of the early 1990s
suggests that the difference in productivity is largely due to
differences in productivity within businesses rather than external
restructuring as might have been expected in a productivity slowdown
generated by banking sector impairment. Our on-going research using
company accounts data is looking at whether the slowdown in productivity
within surviving businesses is more pronounced for businesses with
borrowing relationships with the most distressed banks.
REFERENCES
Armstrong, A., Davis, P., Liadze, I. and Rienzo, C. (2013),
'An assessment of bank lending to UK SMES in the wake of the
crisis', National Institute Economic Review, 225, F4-F14.
Arrowsmith, M., Franklin, J., Gregory, D., Griffiths, M., Wohlmann,
E. and Young, G. (2013), 'SME forbearance and its implications for
monetary and financial stability', Bank of England Quarterly
Bulletin, 53(4), pp. 296-303.
Ashworth, J., Goodhart, C., Seeker, G., Stannard, I. and Baker, M.
(2013), 'Productivity slowdown largely cyclical: good news for the
medium-term outlook', Morgan Stanley research, 15 July 2013.
Barnett, A., Chiu, A., Franklin, J. and Sebastia-Barriel, M.
(2014), 'The productivity puzzle: a firm-level investigation into
employment behaviour and resource allocation over the crisis', Bank
of England Working Paper No. 495, April.
Bell, V. and Young, G. (2010), 'Understanding the weakness of
bank lending', Bank of England Quarterly Bulletin, 50(4), pp.
311-20.
Broadbent, B. (2012), 'Productivity and the allocation of
resources', speech at Durham Business School, September.
Caballero, R., Hoshi, T. and Kashyap, A. (2008), 'Zombie
lending and depressed restructuring in Japan', American Economic
Review, 98(5), pp. 1943-77.
Disney, R., Haskel, J., and Heden, Y. (2003), 'Restructuring
and productivity growth in UK manufacturing', Economic Journal,
113, pp. 666-94.
Farrant, K., Inkinen, M., Rutkowska, M., and Theodoridis, C.
(2013), 'What can company data tell us about financing and
investment decisions?', Bank of England Quarterly Bulletin, 53(4),
pp. 361-70.
Furstenberg, G. von and Kalckreuth, U. von (2006), 'Dependence
on external finance: an inherent industry characteristic?, Open
Economies Review, Springer, 17(4), December, pp. 541-59.
Gilchrist, S. and Zakrajsek, E. (2011), 'Credit spreads and
business cycle fluctuations', NBER Working Paper No 17021, May.
Griliches, Z. and Regev, H. (1995), 'Firm productivity in
Israeli industry 1979-1988', Journal of Econometrics, 65(1),
January, pp. 175-203.
HM Treasury (2013), 'RBS and the case for a bad bank: the
government's review', HM Treasury, November 2013.
Khan, A. and Thomas, J.K. (2013), 'Credit shocks and aggregate
fluctuations in an economy with production heterogeneity', Journal
of Political Economy, December.
Martin, B. and Rowthorn, R. (2012), 'Is the British economy
supply constrained II? A renewed critique of productivity
pessimism', Journal of Political Economy, 121 (6), December, pp.
1055-107.
Office for National Statistics (2014), 'International
comparisons of productivity--final estimates, 2012',
www.ons.gov.uk/ons/ dcp171778_353315.pdf.
Oulton, N. (2013), 'Medium and long-term prospects for UK
growth in the aftermath of the financial crisis', Centre for
Economic Performance, Occasional Paper No. 37, October.
Oulton, N. and Sebastia-Barriel, M. (2013), 'Long and
short-term effects of the financial crisis on labour productivity,
capital and output', Bank of England Working Paper No. 470,
January.
Pessoa, J. and van Reenen, J. (2013), 'The UK productivity and
jobs puzzle: does the answer lie in labour market flexibility?',
Centre for Economic Performance Special Paper No. 31, June 2013.
Rajan, R. and Zingales, L. (1998), 'Financial dependence and
growth', American Economic Review, 88(3), pp. 559-86.
Reinhart, C. and Rogoff, K. (201 I), This Time is Different Eight
Centuries of Financial Folly, Princeton University Press.
Riley, R., Rosazza-Bondibene, C. and Young, G. (2014),
'Productivity dynamics in the Great Stagnation: evidence from UK
businesses', paper presented at Royal Economic Society conference,
April.
Tong, H. and Wei, S. (201 I), 'Did unconventional
interventions unfreeze the credit market? Some international
evidence', available at
http://siteresources.worldbank.org/INTMACRO/
Resources/UnconventionallnterventionsT ongWei061812.pdf.
Young, G. (1996), The influence of financial intermediaries on the
behaviour of the UK economy', NIESR Occasional Paper No. 50,
December.
NOTES
(1) See, for example, Broadbent (2012) for an explanation of this
view.
(2) See, for example, Martin and Rowthorn (2012) for an explanation
of the labour hoarding view.
(3) The correlation between the change in output from its
pre-crisis level and the pre-crisis level of loans net of deposits as a
share of output is -0.5.
(4) Young (1996) calibrates the impact of tight credit conditions
on the early 1990s recession.
Rebecca Riley *, Chiara Rosazza-Bondibene * and Garry Young **
* National Institute of Economic and Social Research and Centre for
Macroeconomics. E-mail: r.riley@niesr.ac.uk. ** Bank of England, NIESR
and Centre for Macroeconomics. E-mail: Garry.Young@bankofengland.co.uk.
The financial support of the Economic and Social Research Council is
gratefully acknowledged (Grant: ES/K00378X/1). We are also grateful to
Nick Oulton. Martin Weale and participants at a seminar at Glasgow
University for comments on an earlier version of this paper. The views
expressed in this paper are those of the authors, and do not necessarily
reflect those of the Bank of England or the Monetary/Financial Policy
Committee. This work contains statistical data which is Crown Copyright;
it has been made available by the Office for National Statistics (ONS)
through the Secure Data Service (SDS) and has been used by permission.
Neither the ONS nor SDS bear any responsibility for the analysis or
interpretation of the data reported here. This work uses research
datasets which may not exactly reproduce National Statistics aggregates.
Table 1. Summary of evidence on business financial conditions
Type of business Evidence
Potential new entrants * High rejection rates for loan
applications by first-time applicants
Financially immature * Higher spreads on new loans and
incumbents overdrafts
* Lower rates on existing facilities as
lower Bank Rate is passed through
* Reduced credit availability for some
and pressure to repay existing
facilities
* Higher rejection rates for loan
applications than before the crisis
* Perceived need to adjust balance
sheets to tighter credit supply
conditions before existing
facilities are renewed
Financially mature * No financial pressure
incumbents * Lower interest rates on existing
facilities
* Use of debt capital markets to repay
bank debt
Poorly performing * Little evidence of bank forbearance,
incumbents although some businesses were
supported by their lenders
Table 2. Industries ranked by bank dependence
Gross loans to Gross loans net of
output deposits to output
2008Q1 2013Q3 2008Q1 2013Q3
Accommodation
and food 0.84 0.60 0.71 0.45
Construction (a) 0.72 0.49 0.44 0.24
Wholesale and
retail 0.41 0.27 0.18 0.03
Manufacturing 0.44 0.24 0.17 -0.08
Transport and
storage 0.32 0.24 0.17 0.02
Arts and
entertainment 0.71 0.29 -0.12 -0.25
Administration and
support 0.41 0.26 -0.15 -0.19
Professional and
scientific 0.38 0.16 -0.27 -0.43
Information and
communication 0.11 0.13 0.01 -0.16
Main sectors 0.39 0.24 0.09 -0.09
Gross loans Pre-crisis Share of assets
to capital lending due to SMES
growth with bank
finance
Pre-crisis 2005-7
Accommodation
and food 0.45 63 0.39
Construction (a) 0.1 1 91 0.49
Wholesale and
retail 0.29 63 0.40
Manufacturing 0.25 24 0.22
Transport and
storage 0.14 -8 0.28
Arts and
entertainment 0.25 67 0.35
Administration and
support 0.15 101 0.35
Professional and
scientific 0.96 106 0.09
Information and
communication 0.11 -8 0.27
Main sectors 0.18 50 0.28
Source: Bank of England and ONS.
Notes: All currency loans and deposits from Bank of England industrial
analysis (tables C1.1 and C1.2 in Bankstats). Output is value of
sectoral output in previous four quarters. Pre-crisis lending growth
is calculated from 2003QI to 2008QI. Final column based on Companies
Accounts data. SMEs defined as companies with total assets of less
than 11.4 million [pounds sterling], in line with Companies House
thresholds.
Table 3. Labour productivity in industries ranked by bank dependence
Productivity growth in
five years before and after 2008Q1
Pre- Post- Swing
crisis crisis
Accommodation and food 12 -7 -19
Construction -3 -2 +1
Wholesale and retail 13 1 -12
Manufacturing 26 0 -26
Transport and storage 7 -22 -29
Arts and entertainment 10 -5 -15
Administration and support 16 4 -12
Professional and scientific 26 -11 -37
Information and 29 1 -28
communication
Main sectors 14 -6 -19
Estimated productivity shortfall
at 2013Q3 (minus is a shortfall)
Relative Relative to Relative to
to linear output- output and
trend adjusted capital-
trend adjusted
trend
Accommodation and food -11 -5 +2
Construction -7 +1 -4
Wholesale and retail -12 -4 -14
Manufacturing -23 -21 -5
Transport and storage -23 +16 -8
Arts and entertainment -12 -7 +13
Administration and support 3 +6 +7
Professional and scientific -18 +1 -11
Information and -24 -24 -6
communication
Main sectors -15 -4 -5
Output
short
fall
2013 Q3
Accommodation and food -4
Construction -11
Wholesale and retail -2
Manufacturing -9
Transport and storage -17
Arts and entertainment -5
Administration and support +13
Professional and scientific +3
Information and +8
communication
Main sectors -3
Source: ONS and own calculations.
Table 4. Impact of an impaired banking system on businesses
Type of business Predicted effect
Potential new * Entry of weaker potential entrants
entrants deterred by low profitability in
'zombie-infested' industries
* Productivity of entrants lower than
would otherwise have been due to credit
restrictions
Financially immature * Bank borrowing unable to expand as
incumbents quickly as desired
* Capital investment held back
* Business unable to expand at desired
pace
* Return on capital likely to be higher
than average
* Labour productivity lower than would
otherwise have been
Financially mature * Expand to fill gap left by constrained
incumbents businesses
* Can finance investment from own
resources
* Less competitive pressure from emerging
businesses
* Survival of 'zombies' may dampen
profitability and investment intentions
Poorly performing * Likely to have lower productivity than
incumbents average
* More likely to survive if supported by
bank
Table 5. Decomposition of manufacturing productivity growth: 1988-92
and 2006-10
1988 1989 1990 1991 1992 1990-92
Entrants -0.5 -0.1 -0.1 0.2 -0.4 -0.3
Exits 1.2 0.8 0.7 0.5 1.3 2.5
Between 0.1 1.0 0.3 0.7 0.4 1.4
Within 6.0 -1.4 -0.5 -1.0 3.8 2.3
Total 6.8 0.3 0.5 0.4 5.0 5.9
Memo: Output 7.2 -0.2 -2.5 -6.5 -1.9 -10.9
2006 2007 2008 2009 2010 2008-20
Entrants -0.4 -0.1 -1.1 -0.4 -0.4 -1.9
Exits 1.8 1.9 1.1 1.2 2.0 4.3
Between 0.8 0.9 0.8 0.0 0.6 1.4
Within 1.4 6.2 -2.1 -12.1 11.2 -3.0
Total 3.7 8.9 -1.4 -11.2 13.4 0.8
Memo: Output -0.8 3.4 -0.1 -14.7 6.8 -8.0
Source: Annual Respondents Database, ONS, and authors' calculations.
Notes: Griliches and Regev decomposition of labour productivity.
Manufacturing. Britain. Firms are classified as live if they are
active and have 10 or more persons employed. Output figures calculated
on the same sample as labour productivity.
