1 FINANCIAL CONSTRAINTS AND EMPLOYMENT

155a
FINANCIAL CONSTRAINTS AND EMPLOYMENT GROWTH: EVIDENCE PRE AND
POST FINANCIAL CRISIS IN SPAINISH SMEs
Javier Plitt-Stevens
Universidad Politécnica de Cartagena
Antonio Duréndez Gómez-Guillamón
Universidad Politécnica de Cartagena
Área temática: a) Información Financiera y Normalización Contable.
Palabras clave: Job Creation, Financial Constraint, Employment Growth, Crisis
1
FINANCIAL CONSTRAINTS AND EMPLOYMENT GROWTH: EVIDENCE PRE AND
POST FINANCIAL CRISIS IN SPAINISH SMEs
Abstract
The massive job destruction hit the Spanish society since the beginning of the financial
crisis in 2008. Although many aspects of the firms have been investigated as potential
drivers of this employment shock, we focused on lack of external finance as one of the
main factors. From a sample of 68731 Spanish manufacturing SMEs in the period
2004-2013, we analyzed how internal and external financial constraints affect
employment growth. We found during the crisis the dependence of external funding the
significance is much higher and specially affects the Micro enterprises.
Keywords: Job Creation, Financial Constraint, Employment, Firm Growth, Crisis
1. Introduction
The massive jobs destruction hit the Spanish society since the beginning of the
financial crisis in 2008. The unemployment rate reached 26.94% in the second quarter
of 2013 while the average in EU15 was 11.2% becoming the second worst in Europe
and the first worst among young people1. This employment crisis has been causedamong others- by the lack of credit offer in the bank system. In others words, the
financial crisis triggered the employment crisis.
Spain has been seriously affected by this credit constraint shock. ECB Surveys on
Access to Finance of SMEs in the Euro Area have shown the increase in bank lending
rates and the collateral requirements in Italy, Greece and Spain during the period after
the financial shock. SMEs was affected and more specifically the micro firms, which
represent in Spain 95.7% of total companies – 3.8% more than the average of the
EU27.
Under this context of financial and employment crisis, the researchers and policymakers are still looking for some answers about the best way to stop or decrease this
record in unemployment representing the biggest concern among Spanish people
since the beginning of the crisis2. This article attempts to help in this subject
1
2
EUROSTAT Web page: http://epp.eurostat.ec.europa.eu
Barómetro del Centro de Investigaciones Sociológicas from 2009 until nowdays.
2
investigating how financial constraint affect the employment growth analysing what kind
of firms are more sensitive to the lack of credit in the Spanish manufacturing firms.
We set three different empirical approaches. First, we have compared the results
before (2004-2008) and during the crisis (2009-2013)3 testing if in the previous years to
the crisis, the financial constraint affected less than during the crisis, the job creation as
we expected. Second, we used the data set of 68731 Spanish SMEs which contains
significant information about micro firms in order to test if they are more affected by the
financial constraint than the rest of SMEs. Finally, we compared the behaviour under
external and internal financial constraint of all the manufacturing subsectors controlling
by size and age.
Employment growth is one aspect of the company’s performance affected by financial
constraint. Lack of liquidity (Internal Financial Constraint) has a high positive
significance in the job creation at firm-level (Carpenter and Petersen 2002, Oliveira y
Fortunato 2006, Hutchinson and Xavier 2006). On the other hand, the External
Financial Constraint showed by the high lending rate of banks or high interest payment
of the firms affected the performance of employment growth too (Nickell and Nicolitsas
1999 for UK, Benito and Hernando 2002, Hernando and Martinez-Carrascal 2005 and
Benito and Hernando 2008 for Spain, Musso and Schiavo 2008 for French firms).
Following these two strand of literature, we follow the approach of Rahaman (2011)
incorporating both concept in the same model (Internal and External Financial
Constraint) and concluding that the presence of accessing to bank loans alleviate the
dependence of the firm on internal resources while when the restriction in external
sources are high, the firms uses internal resources to finance growth.
A wide range of studies analyse the employment growth conditioning by the financial
restrictions using age and size as a control variables (Oliveira y Fortunato, 2006 and
2008; Ayygari, 2011; Benmelech et al., 2011; Botazzi et al., 2014). However, only few
papers analyse the role of the industry sector on the growth. Acemoglu (2002) and
Duygan-Bump (2011) used the classification based on Rajan and Zingales (1998)
about the financial dependence of the different sectors in order to compare if one of
these are more affected by financial constraint than the others. Voulgaris et al. (2014)
incorporated the analysis of the differences in manufacturing subsectors for the Greek
3
Voulgaris et al. 2014 divided the panel in two (2004-2007 and 2008-2011) assuming that the first year
of the cisis was 2008 in Greece. However, in our paper we have considered according with the GDP
growth that the first year of the Sanish crisis is 2009 when the recession officially started.
3
firms. Our work follow this strand of literature contributing with a deeper analysis for the
Spanish companies.
While many articles analysed the consequences of the crisis in the firm performances,
only few of them compare the behaviour of the employment growth before and during
the recession (Campello et al. 2010 using a survey of US, European and Asian large
firms, Choorodow-Reich 2013 for US SMEs depending the bank channel, Volk and
Trefalt 2014 for Slovenian firms, Voulgaris et al. 2014 for the Greek manufacturing
sector Lee et al. 2015 analysed the impact of financial constraint on the innovative
firms, during and after the crisis for UK SMEs).
For the Spanish case, Garicano and Steinwender (2013) and Bentolila et al. (2013)
dividing the sample under different criteria (The owned classification and the health of
the lending bank respectively) comparing the crisis roll in financial constraint and job
creation. Following the same strand of recent literature, we contribute investigating the
profile of the firms (in terms of age, size and subsector) before and during the crisis by
the internal and external constraint in order to create more jobs. We focused especially
in Micro firms.
We found the influence of this financial constraint raised during the economic crisis
overall the manufacturing sector, age and size of the firms. Specifically affects low and
low-medium technology based sector. We find that the high dependant of financial
resources sectors are not significant affected more than the rest (Rajan and Zingales,
1998; Acemoglu, 2002). Going further, we found that Machinery, Ships, Plastic,
Metallic Wood and Recycling are subsectors most affected during the crisis while none
of the subsectors seem to be affected by financial constraint in the period before the
crisis.
It’s remarkable the change on the sign of the age in the panel before the crisis and
after that. When the financial pressure was not too strong the age has a negative effect
on the employment growth (in line with most of literature which analyse those two
variables), however when the financial pressure is really strong, the age change into a
positive sign – older firms contribute more in the job creation than young firms.
Finally, we analyzed micro firms separately from the rest of the SMEs and we found the
same behaviour but the effect of financial constraints (internal and external) are
stronger than test it the whole sample of SMEs in line with Volk and Trefalt (2014).
4
Based our analysis in the GMM-System Method (Arellano and Bover 1995) and
(Blundell and Bond 1998) we have some limitations which are important to remark.
First, the original data base contained all firms which had data more than two
consecutive years. In order to get the Arellano-Bond second-order test, we decided to
exclude all the firms with less than 4 consecutive years (Arellano and Bond 1991).
Second, In order to avoid the outlier’s problems, we exclude the observation in the one
percent tail of each regression variables (Bond et al, 2003 y Cummins et al. 2006).
Finally, our study is focused only in the manufacturing companies which has the
limitation that not represent the sharing of the rest of the sectors in SMEs (Bernanke,
1996)4.
This paper is organized as follow. Section II discusses the literature related. Section II
provides information of the model, empirical approaches, variables and the sample. In
Section IV analyses the result and section V concludes and suggest some policies to
take into account.
2. Literature Review
2.1 Financial constraint and employment
Financial constraint has been analyzed since more than half of century. Modigliani and
Miller (1958) concluded that the financial position does not influence in the market
value of the company under perfect market (labour and financial). But the first modern
approach about this subject was published by Stiglitz y Weiss (1981) incorporating the
concept of credit rationing under imperfect markets.
Investment (or lack of investment) was the first consequence of the financial constraint
analyzed by the literature. Fazzari, Hubbard and Petersen (1988) wrote one of the first
and most controversial (Carpenter and Guariglia 2008)5 articles using empirical data
concluding with a positive relationship between cash flow and asset investment in a
quoted firms across US.
Carpenter and Pertesen (2002) developed a model of a firm growth (in assets) which
analyses the effect of the internal financial constraint proxies by the cash flow,
obtaining the same conclusion of Fazzari, Hubbard and Petersen (1988) for the SMEs
in US. In the same line of research, Hutchinson and Xavier (2006) extended the model
4
In Spain, Micro firms represents 95.7% in all industries while it represents 84.6% in manufacturing
firms.
5
See Kaplan and Zingales 1997 and Cleary 1999 also.
5
to the Europeans countries where the micro firms play a fundamental roll in the
industrial make-up. They compared two different economies such as Slovenia and
Belgium. Oliveira and Fortunato (2006) applied the same model in the Portuguese
SMEs using employment variation in two consecutive years as a proxy of firm growth.
More recently, Guariglia (2011) used the Carpenter and Petersen (2002) model to
applying to the Chinese manufacturing industry in a big data panel.
Derived from cash flow sensitive, Nickel and Nicolitsas (1999) analysed the effect of
financial constraints on the employment, wages and productivity using “Borrowing
Ratio” - defined as the Total Interest Payments over Cash Flow - as a measure of
external financial constraint. They showed that “Financial Pressure” affected negatively
the employment in a sample of UK manufacturing firms. In Spain, Benito and Hernando
(2002), Hernando and Martinez-Carrascal (2005) and Benito and Hernando (2008)
used the same approach to conclude external financial constraint affected the
employment also.
Other measures of financial constraint used in the literature was leverage defined as
Total leverage over Total Assets (Cantor 1990; Sharpe 1994, Bernanke et al. 1996).
Following the same line (Almeida and Campello 2007) consider long-term debt
maturing in the short run as a measure of financing constraints and find that the decline
in investments is larger for firms that need to refinance a large proportion of their longterm debt at the onset of the crisis. (Vermoesen et al. 2012) also use the Long-term
debt as financial constraint proxy.
Some others used interest burden, collateral and even age and size (Headlock and
Pierce 2010). For example Chen (2007) compare seven different measures of financial
constraint in order to see the effect of all of this on the employment decision in China
manufacturing companies.
Financial ratios such as Borrowing ratio, leverage ratio, interest ratio are used for many
authors as a proxies of the financial constraint, sometimes don’t reflect the exact reality
of what happened in the different firms. For example it is not clear if one company
didn’t apply for a loan or, on the contrary, this loan was refused by the bank (Storey,
1994).
In order to solve this lack of information, some papers used directly on surveys which
typically ask the Managers, the CEO or CFO about the external finance (Simonly and
Winker, 1999; Bechetti y Trovato, 2002; Savignac, 2006 Angellini and Generale, 2008;
Campello et al, 2009, Caggese and Cuñat, 2011). Using these surveys as a proxy of
financial constraint, dividing the sample into a financial constrained and non6
constrained firms and then regressing them attempting to test whether the dependant
variable (Employment growth in our case) is affected more in the constrained
subsample or not.
Surveys present some limitation also. First, the nature of the survey which is conducted
to the owner or manager of the company. In this case, the interest party is the person
who can respond not very precisely looking for a better result in their own benefit
according with the agency theory problems. For example the firms which have obtained
the loans can give some details about their operation and, on the other hand, the firms
with a higher interest rate or which the loan have been refused don’t want to give any
detail (Savginac, 2006).
The controversy of the use of different variables as a proxies for the external financial
constraint is still in the air. Modern line of research use a multivariables indexes as a
measure of firm’s financial constraint. Some authors developed their own index based
different dimensions of the company and some others used directly some kind of
rankings already used from the governments or credit institutions.
Cleary 1999, Lamont et al. 2001 and Whited and Wu 2006 constructed an Index of
financial constraint based on “a standard intertemporal model augmented to account
for financial frictions”. However, the first paper that build an Index based on five
different dimensions was Kaplan and Zingales (1997) using cash flow over assets,
market over book ratio, total debt over total assets, dividends over fixed assets and
cash over fixed assets.
Some years later, Musso and Schiavo (2008) developed an Index to measure the
degree of financial constraint based in seven different dimensions for French
manufacturing firms. This Index take into account these variables related with financial
constraints: Size, profitability, liquidity, cash flow generating ability, solvency, trade
credit over total assets and repaying ability. Sufi (2009) uses the introduction of
syndicated bank loan ratings in 1995 to study financial and investment policies. Coluzzi
et al. (2012) based on a survey and combining these results with another ratios that
obtain an index call “financing obstacles” that it is used as a measure of financial
constraint. In the same sense, more recently, Botazzi et al. (2014) used for
Manufacturing Italian firms, the Credit Rated Index that CeBi (Centrale del Bilanci)
produces for every company in their own database.
All these Indexes or Scores present some advantages comparing with one dimensional
variables as proxy of external financial constraint. The first and obvious one is that
summarized a wide range of dimension instead only one. Second, they are continues
7
and updated every year comparing with some of them which are not time-varying.
Third, in some sense, they provide a degree of financial constraint of the companies
and based on that they can divided the sample according with these degrees. In our
work we use a multivariable Index as a main external financial constraint proxy based
on Musso and Schiavo (2008).
As an extension of the articles cited before, Cleary (2007) presented a model where
analysed the internal and external financial constraint at the same time. Following this
line of research, Guariglia (2008) investigated the assets investment sensitivity to cash
flow depending of the degree of internal and external financial constraints for quoted
and non-quoted UK firms. In this sense, this paper differs from Cleary (2007) because
of the use of UK firms database containing mostly non-quoted firms (99%), while the
US database used by Cleary contain mostly quoted firms.
In this article, he used the cash flow as a proxy of internal financial constraint which is
the most extended in the literature (Schiantarelli 1995, Hubbard et al.1998, Bond and
Van Reenen 2005). For the robust test, he added another measure based on coverage
ratio. On the other hand, the external financial constraint was proxy by the size of the
company (Schiantarelli 1995) measured in total assets. Age have been chosen in order
to check the robustness.
While Guariglia (2008) explores how both internal and external financial constraint
affect the asset investment, Rahaman (2011) applied a similar analysis to the
employment growth. He chose the difference of equity funds in two consecutive years
as measure of internal financial constraint and the ratio of short-term debt over total
liabilities as external financial constraint proxy. Focused in micro firms, Volk and Trefalt
(2014) analyzed the access to bank credit and their influence in firm growth for
Slovenian SMEs. They also contribute investigating the roll of collaterals in the same
analysis.
The difficult situation in the southern European countries have same results in
unemployment rate, being Spain and Greece the most affected ones. For the Greek
case, Voulgaris et al. (2014) used the Rahaman (2011) model in order to investigate
how the crisis affected the influence of financial constraint on employment growth.
In this study, we investigate how the availability of internal and external sources
affected the job creation or destruction in the Spanish SMEs. Based on the Rahaman
(2011) model and dividing the complete dataset in the pre-crisis period and during the
crisis as Voulgaris et al. (2014) and we conclude with different behaviours of the firms
before and during the crisis. We noticed this differences especially in micro firms
8
2.2 Crisis, financial constraints and the effect on the employment
One of the real effect of the massive crisis started with the Lehman Brothers
bankruptcy in September 2008 was the lack of the liquidity in the financial market
system around the world. Europe was clearly affected as shown the ECB Survey on
Access to Finance of SMEs in the Euro Area in the years after the crisis. The banks not
only rationed credit (Stiglitz and Weiss, 1981) reducing drastically the rate of loans
granted, but at the same time all these credit granted was with a higher rate than
before the crisis.
In this context of financial constraints, the strong rise of the unemployment rate was
one of the main consequences in Europe. This has particularly affected Spain
(Bentolila et al. 2013) and specially the construction sector - one of the major in job
creation in Spain – which simply suddenly stopped and the real estate bubble burst,
destroying millions of jobs in the following years, reaching more than 5000000 jobless.
Several studies have found evidence of how the crisis directly affects the availability to
create jobs under the financial constraint scenario. For large corporate firms, Campello
et al. (2010) analysed the behaviour of the corporate spending plan (e.g. employment
and asset investment) in the pre-crisis period (2007 Q3 to 2008 Q3) and after (2008
Q4).They compare constrained and non-constrained firms and they concluded that the
difference is bigger during the crisis period. On the same line, Almeida et al. (2012) use
long-term debt maturity structure as a credit constraint proxy in order to test its impact
on the real firm behaviour in a pre-crisis period compared with a period inside the crisis
for large manufacturing companies. Chodorow-Reich (2013) analyzed the effect of
bank lending friction on employment outcomes but only during the crisis (2008-2009) in
US public companies.
On the other hand, Vermoesen et al. (2012) investigated how external financial
constraint limited the asset investment during the economic crisis for Belgium SMEs
firms. The importance in job creation of the SMEs, not only in Europe, but around the
world was the topic analyzed by Ayyagari (2011) from 2006 until 2010 highlighting the
SMEs as the engine of the economy during the crisis. Volk and Trefalt (2014) focused
in micro firms behaviour investigating the roll of the financial constraint in the Slovenian
firms employment. Voulgaris et al. (2014) followed the model presented by Rahaman
9
(2011) in order to analyse the behaviour in job creation and destruction in the periods
pre and post crisis in Greece.
In Spain, Garicano and Steinwender (2013) divided the manufacturing companies in
Spanish owned and foreign owned in order to compare the effects of financial
constraint in both. They conclude that lack of access to financial affect more in Spanish
owned firms reducing the possibility of firm growth in investment and employment too.
The same line was followed by Bentolila et al. (2013) dividing the firms into two groups
depending on the heath of the bank that they are trade with. Analysing both periods,
before and after the crisis, he concluded that having loans with a healthy bank implies
better performance in employment, especially during the crisis years.
In our paper we follow Voulgaris et al. (2014) approach dividing the panel into two
periods – pre-crisis and during the crisis – and analysing then the effect of internal and
external financial constraint on employment growth.
3. Methodology:
3.1 The Model:
Our growth model is based on Goddard, Wilson and Blandon (2002) which is an
extension of the Carpenter and Petersen (2002) model. According with this model, the
variables at the right hand side are lagged one period.
This model was used by Fortunato and Oliveira (2006) in Portuguese firms and in the
same line by Hutchingson and Xavier (2006) in Belgium and Slovenian Firms. Coluzzi
et al. (2012) used it for five different countries in Europe.
The model developed by Carpenter and Petersen (2002) was originally based in the
US manufacturing industry, but then Hutchingson and Xavier used an extension of this
model adapted for the European industry context, taking into account the micro firms of
less than 10 employees which represent a significant part of the manufacturing sector6.
Following the work of Coluzzi et al. (2012) we obtain the final equation model:
(3)
6
EUROSTAT, dar las cifras de micro-empresas para paises como españa, Portugal, UK y belgica
10
Where
is the variation of the employment in percentage (as a firm growth), x is
the vector containing the lagged independent variables: Internal Financial Constraint
and External Financial Constraint as the main independent variables.
We include in x vector Age and one variable of financial constraint (internally and
externally)
We could rewrite the equation for the econometric data panel purposes like this:
(4)
In our analysis we use different measures of financial constraint following the existence
literature. For the model we use one measure at the time order to compare different
options (internally or externally constrained).
In the Coluzzi et al. (2012) article used at the same time Cash Flow, Leverage, Debt
Burden and a dummy to distinguish if the firm is young or not. For the purposes of
financial constraint they developed a financial obstacles measures starting from a
direct survey.
3.2 Variables
3.2.1 Growth: Different measures, different results.
Delmar (1997) shows five different measures of firm growth, however most of the
papers in the literature use three: Sales, Assets and Employment (Davidsson and
Wiklund 2006). For example, Carpenter and Petersen (2002), Goddard et al. (2009)
and Coluzzi et al. (2012) use the Total Assets; and on the other hand other researchers
prefer the annual difference in sales as a growth indicator (e.g.: Delmar 1997; Coad
and Rao 2008). Finally, the growth measured in employment has some advantages
than the others two (Coad and Hölzl 2012). But, in the end of the day the researchers
choose the variable which is more important for the investigation or simply the most
available one. In our study we focused in the employment growth as a main subject of
our investigation, therefore we used this indicator as our main dependant variable.
In terms of absolute or relative growth, Delmar (1997) and Storey (1994) explain the
pros and cons of each measure. While absolute growth has a positive association with
initial firm size, the relative measure has a negative association. At the same time, the
use of the difference in logs is less affected by the heteroscedasticy from the
econometric point of view (Coad and Hölzl 2012). In order to mitigate the affection of
the size in the growth measures, Birtch (1987) and Schreyer (2000) developed a
combined index of absolute and relative, commonly called Birtch Index. According with
11
this, we can conclude that the results of the investigation depends of the different way
in measuring growth (Almus, 2002; Shepperd and Wiklund, 2009; Coad and Holzl,
2012).
As a result of this previous analysis we will use the relative measure of employment
growth expressed as follow:
(5)
Then we can transform it in log differences:
(6)
In this case we use the relative measure of employment growth defined as the variation
of the total employees of the firm during one period divided by the number of total
employees at the beginning of this period.
3.2.2 Financial constraint variables.
None of the approaches has a real consensus about which the best measure of
financial constraint is. Single and multivariable indexes are extracted from the results
data sheets that not reflect effectively the financial constraint position of the firm. On
the other hand we have the surveys that include the personal point of view of the
manager of the firm. In this case based on the information asymmetries and in many
cases the agency Theory problems it couldn’t be possible having totally the real
answers.
In our case we have used the multivariable Financial Constraint Index as a main
indicator of external financial constraint (Musso ans Schiavo 2008) and the difference
in internal funds in two consecutive years as an internal financial constraint measure
(Rahaman 2011). In the case of the Global Index, we lagged it one period, because
when the lender decided to grant a credit or not, use the latest information available.
In both cases, tested it different approaches using different variables for rubust results.
First, in the case of external financial constraint we tested the borrowing ratio (Nickell
and Nicolitsas 1999) and the ratio called “access to a bank credit facility” developed by
Rahaman (2011) and defined as Short-term bank loans and overdraft over Total
12
liabilities. Really in our specific case we replaced that with the ratio of short-term debt
over Total liabilities – adding the Trade Credit which is not included in the measure
used by Rahaman (2011). Second, for the internal financial constraint measure, we
also used cash flow as the most used measure (e.g. Fazzari et al. 1988, Carpenter and
Petersen 2002, Oliveira and Fortunato 2006).
Testing these different measures of the same dimension we tried to investigate if the
behaviour differs from variable to another and at the same time we evaluate the
robustness of the results obtained.
3.2.3 Control variables.
Size and Age are the most common and most important variables in order to analyse
the behaviour on the firm growth (Gibrat’s Law) and specially evaluating the
consequences in the employment growth. We use the logarithmic form of each one. As
a Size measure we decanted by the total assets (Rahaman 2011) at the year t-1.
We controlled the model by the sector of the manufacturing firms. We used different
kind of dummy variables. First, following the Rajan and Zingales (1996) we divided the
sample in three different groups according with the grade of financial dependence
(Low, Medium and High Financial Dependant). Second, we used the technology grade
for dividing the total firms in High, Medium High, Medium Low and Low technology
(Moreno and Coad 2015). Finally we use NACE-2 digits as a main division following
(Voulgaris et al. 2014).
All the variables included in the model and their expected correlation sing with the
dependent variable ae shown in Table 1.
Variables
Description of Variables
Expected
sing
Dependent Variable
Change in Employment yearly (log form)
Independent Variables
(-)
AGE
Years since firm born in log form
(-)
SIZE
Total Assets in log form
(+)
IFUND
Log Difference in Internal Fund
(+)
CF
Cash Flow over Total Assets
(+)
13
FC_INDEX
Index of External Financial Constraint
(+)
Bank Credit Facilities
Short-term bank loans and overdraft over Total liabilities
(+)
Borrowing ratio
Interest Payment over Cash Flow
(-)
Dummy Variables
Crisis Dummy
1 if Year > or = 2009, 0 otherwise
Financial Dependence
1 if NFD: non-financial Dependent, 0 otherwise
Grade Dummy
1 if MFD: Medium financial Dependant, 0 otherwise
1 if HFD: High Financial Dependant, 0 otherwise
1 if Low Technology, 0 otherwise
Technology Grade
1 if Low-Medium Technology, 0 otherwise
1 if Medium-High Technology, 0 otherwise
1 if High Technology, 0 otherwise
NACE-2 Digit classification for Manufacturing Sector
Industry Subsectors
Table I - Variables and Sign expected
3.3 Data Sample
The data were obtained by SABI (Iberian Balance Sheets Analysis System) database,
developed by Bureau Van Dijk. This database contains accounting and financial
information for Spanish firms. We focused in the manufacturing companies from 2004
until 2013. In the first place we download 92349 firms including micro-enterprises with
more than 1 employee. The only limitation was that the company had to have data
more than two consecutive years. We include all corporate forms (S.A., S.L., S.C.,
C.B., etc) and self-employment.
We reduced the firms until 73393 dropping the companies with lack of employee data
in the middle of the firm live cycle. We exclude the firms with less than four consecutive
years of information in the main variables in order to get results in the second order
Arellano test (Arellano and Bond 1991).
While we keep all the firms which are entries after 2004, we drop the exits before 2013
because we consider their behaviour is completely different than the companies want
to survive (Aterido and Hallward-Driemeier 2007). In order to avoid the outlier’s
14
problems, we exclude the observation in the one percent tail of each regression
variables (Bond et al, 2003 y Cummins et al. 2006).
Finally, we obtained an unbalanced panel of Spanish manufacturing firms of 68731
SMEs with 692117 observations during the period between 2004 and 2013, five years
before the crisis and five years during the crisis.
For the most of our approaches, we divided the panel in two – before and during the
crisis – the first one form the year 2004 until 2008 included and the second, from 2009
until 2013. This division presents a difference with others studies in terms of the year of
the beginning of the crisis. While Campello et al. (2010) start to analyse the crisis in the
fourth quarter of 2008 and Voulgaris et al. (2014) use the year 2008 as the first year,
we include the year 2008 in the pre-crisis period because GPD growth was still positive
(1.115).
For the second approach, when we compare the behaviour of different subsector of
manufacturing firms, we also divided the sample following the 2-digits NACE sectors.
3.4 Estimations
The structure of the shot panel given in this study presents the problem of the
correlation of the lagged dependent variable with the error term (Roodman, 2006). In
order to avoid issues with the estimation model we use GMM-system estimation,
developed by Arellano and Bover (1995) and Blundell and Bond (1998), for the
dynamic panel data with a big number of companies and small number of years. We
included in the model the year dummies variable following Roodman (2006).
We assumed all variables are endogenous implemented instruments for AGE, SIZE,
CF and FCINDEX in differences and in levels.
We tested the model using the Bond second order test and for each panel (Arellano
and Bond 1991) for autocorrelation in residuals and the Sargan test of over-identifying
restrictions for the validity of the instruments
4. Results:
Results are divided into three parts. First, we divided the sample in two periods, before
(2004-2008) and during the crisis (2009-2013) and we compared the results between
15
micro, small and medium-size firms. Second, with the same two panels, we compare
the behaviour of the all subsectors, especially which ones create more jobs and which
destroy more. Finally, we tested the model using different internal and external financial
constraint proxies.
4.1 Pre and Post Crisis: Results for Micro, Small and Medium-size firms.
In table 3 we show the results comparing micro, small and medium firms, before and
during the crisis.
The remarkable finding is the change in the sing of the age, which is negative
correlated with employment growth in the period previous of the crisis and negative
correlated during the crisis period for all size of the firms. However, age is significant at
the 5% level for micro firms, significant at 10% level for small and not significant for
medium-size firms in the pre-crisis period. This results suggest that while in the precrisis period the young firms generate more jobs than older ones, when the cycle
change the old companies lead the job creation. This results is in line with Voulgaris et
al. (2014) for Greek manufacturing firms analysing similar periods.
Internal financial constraints proxies by the difference in internal funds are significant
and positive for all the firms during the crisis and it is only significant for small firms at
1% level and medium firm-size at 10% level in the period before. The results of this
finding is contrary of Volk andTrefalt (2014) where the importance of internal funds for
firm growth in higher for micro firms. Similar results obtained Aterido and HallwardDriemeier (2007) for external financial constraint shown more dependence of micro
firms than small and medium. On the other hand, Moyi (2013) obtained a nonsignificant relationship between financial constraint and investment
for MSEs,
suggesting that micro firms used the loans to the operations of the firm and no for the
investments or in our case for creating new jobs. Other explanation of this finding could
be that micro firms have less access to formal finance, therefore they satisfied their
financial needs using another non-formal sources.
4.2 Analysing Sectors.
We divided the manufacturing sector following two different criteria. First, we divided
the sample according with the technology used in different sectors. With this division
we trying to see the influence of this technology following the articles related with a
high growth firms (Moreno and Coad, 2015). And second, we divided the
manufacturing industry following the 2-digits NACE sectors.
16
Table 4 indicates the results comparing the grade of technology. We observe that for
the high technology sectors, the effect of the External Financial constraint on
employment growth is statistically non-significant while for the rest are positive and
significant effect. In this case high technology industries has less dependence of
external financial constraint that medium and low tech firms.
If we compare in the periods before and during the crisis (Table 5), the situation is
similar, but in the case of the pre-crisis period none of the sectors are statistically
significant, which in general suggest that in the periods when the economy growth the
whole manufacturing sector is non dependant of external credit in order create jobs,
however in the period of crisis there are a relation between the credit constraint and the
employment growth.
Comparing the 19 manufacturing subsectors. We observe that Machinery, Ships,
Plastics, Metallic, Recycling, Wood and Textile shows a positive and significant
relationship between External Financial constraint and employment growth. In table 7
we compare the most significant subsectors before and during the crisis obtaining
similar results, for the period before the crisis none of the sectors shows significant
financial dependence in order to create jobs, however after that all the same sectors
shows the same behaviour than in total panel. Only machinery, vehicles and Metallic
sector had a positive and significant correlation with internal credit constraint.
4.3 Robustness check
We will use different measures of internal and external financial, constraint in order to
check the robustness of the model. For internal financial constraint we will add the cash
flow. For external financial constraint we used an Index of FC developed by Musso and
Schiavo (2008), the Borrowing ratio presente by Nickell and Nicilitsas (1999) and the
Bank credit Facilities showed by Rahaman (2011).
5. Conclusions:
In our paper we focus on the fundamental question for researchers and policy-makers:
Who can create more jobs if it is possible to release funds to the credit market?
Sometimes the governments have the possibility of “injecting liquidity” to the bank
systems in order to re-activate the economy after a crisis, but they have to be sure
which is the best profile of the companies to receive this credits.
Our study tried to address in this direction, comparing the behaviour of different SMEs
(Micro, Small and Medium) and who of these can create more jobs using external
17
funds. In principle, in the case of manufacturing Spanish firms, small firms (10-249
employees) seems to use the external credit to create jobs more than the others.
In this case, the results suggest that older firms of any of the groups generate more
employment than the young ones during the crisis. This results are very similar that
Voulgaris et al. (2014) found for the Greek manufacturing firms.
We analyzed the manufacturing subsector concluding with high technology sector is
the less affected by financial constraint in order to generate jobs. Then we compared
all the manufacturing subsectors and we can observe a big differences between them
and the behaviour before and during the crisis.
In this field, we will continue with our research crossing the data of the more dependant
sectors (Table 6 and 7) and the data of which of these sectors create more jobs in the
period 2004-2013 and then 2004-2008 and 2009-2013. With this matrix we will obtain
the more worthy sectors in order to supply liquidity for creating jobs.
At the moment we are testing the model adding different control variables to this basic
model. We added one macroeconomic variable as GDP Growth in order to control the
model for the business cycle and more performance variable such us leverage and
labour productivity among others.
The study will continue with marginal effects analysis following the Rahaman (2011) in
order to quantify in the end (if possible) the employment lost by the financial credit
shock at the beginning of the crisis. And in the same line, if the policy-makers can
reduce the lack of credit, they can reduce the unemployment as well applying this
resources to the right kind of companies.
References:
Almeida, H., & Campello, M. (2007). Financial constraints, asset tangibility, and
corporate investment. Review of Financial Studies, (2003).
http://rfs.oxfordjournals.org/content/20/5/1429.short. Accessed 9 June 2013
Almeida, H., Campello, M., Laranjeira, B., & Weisbenner, S. (2012). Corporate Debt
Maturity and the Real Effects of the 2007 Credit Crisis. Critical Finance Review, 1,
3–58.
Aterido, R., & Hallward-Driemeier, M. (2007). Investment climate and employment
growth: The impact of access to finance, corruption and regulations across firms
18
(No. 3138). http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1032567.
Accessed 5 June 2013
Ayyagari, M. (2011). Small vs. young firms across the world: contribution to
employment, job creation, and growth. World Bank Policy …, (April).
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1807732. Accessed 24 June
2014
Bentolila, S., Jansen, M., & Jiménez, G. (2013). When Credit Dries Up: Job Losses in
the Great Recession. http://www.bportugal.pt/enUS/EstudosEconomicos/Conferencias/Documents/2013LabourMarket/paper3_e.p
df. Accessed 17 June 2013
Campello, M., Graham, J., & Harvey, C. (2010). The real effects of financial
constraints: Evidence from a financial crisis. Journal of Financial Economics.
http://www.sciencedirect.com/science/article/pii/S0304405X10000413. Accessed
16 March 2013
Carpenter, R. E., & Petersen, B. C. (2002). Is the Growth of Small Firms Constrained
by Internal Finance? Review of Economics and Statistics.
doi:10.1162/003465302317411541
Carpenter, R., & Guariglia, A. (2008). Cash flow, investment, and investment
opportunities: New tests using UK panel data. Journal of Banking & Finance,
44(June), 0–29.
http://www.sciencedirect.com/science/article/pii/S0378426607004037. Accessed
19 May 2014
Chodorow-reich, G. (2013). The Employment Efects of Credit Market Disruptions :
Firm-level Evidence from the 2008-09 Financial Crisis. Review of Finance - Oxford
Journals, 17(1), 229–259. doi:10.1093/rof/rfr038
Coad, A., & Hölzl, W. (2012). 24 Firm growth: empirical analysis. … on the Economics
and Theory of the Firm. http://books.google.com/books?hl=en&lr=&id=uZlToe_bEoC&oi=fnd&pg=PA324&dq=firm+growth:+Empirical+analysis&ots=ZkTqZ
FRGqh&sig=iXs08MmX1okuL3E0QWt9f042Mp4. Accessed 24 June 2014
Coad, A., & Rao, R. (2008). Innovation and firm growth in high-tech sectors: A quantile
regression approach. Research Policy, 37(4), 633–648.
http://www.sciencedirect.com/science/article/pii/S0048733308000152. Accessed
14 June 2014
Coluzzi, C., Ferrando, A., & Martinez-Carrascal, C. (2012). Financing obstacles and
growth: an analysis for euro area non-financial firms. The European Journal of
Finance. doi:10.1080/1351847X.2012.664154
Davidsson, P., & Wiklund, J. (2006). 3. Conceptual and empirical challenges in the
study of firm growth. … and the Growth of Firms.
http://books.google.com/books?hl=en&lr=&id=2kkIZhXo1xwC&oi=fnd&pg=PA39&
dq=Conceptual+and+empirical+challenges+in+the+study+of+firms&ots=W6ZNDU
rwAs&sig=2qtlbeLgbO-N0y83nTdKwQtUxzU. Accessed 24 June 2014
19
Fazzari, S., Hubbard, R., & Petersen, B. (1988). Financing constraints and corporate
investment. From Brookings Papers on Economic Activity, 1, 141–195.
http://www.nber.org/papers/w2387. Accessed 13 March 2013
Garicano, L., & Steinwender, C. (2013). CEP Discussion Paper No 1188 February
2013 Survive Another Day : Using Changes in the Composition of Investments to
Measure the Cost of Credit Constraints, 2013(1188).
Goddard, J., Wilson, J. O. S., & Blandon, P. (2009). Panel tests of Gibrat’s Law for
Japanese manufacturing., 20, 415–433. doi:10.1016/S0167-7187(00)00085-0
Guariglia, A. (2008). Internal financial constraints, external financial constraints, and
investment choice: evidence from a panel of UK firms. Journal of Banking &
Finance, 44(June), 0–31.
http://www.sciencedirect.com/science/article/pii/S0378426607003974. Accessed
19 May 2014
M.Volk; P.Trefalt. (2014). Access to Credit as a Growth Constraint. Journal of Banking
and Financial Economics, 1(1), 29–39. doi:10.7172/2353-6845.jbfe.2014.1.2
Moyi, E. (2013). Credit and employment growth among small enterprises in Kenya.
International JOurnal of Business and Economics Research, 2(3), 69–76.
doi:10.11648/j.ijber.20130203.14
Nickell, S., & Nicolitsas, D. (1999). How does financial pressure affect firms? European
Economic Review, 43(8), 1435–1456. doi:10.1016/S0014-2921(98)00049-X
Oliveira, B., & Fortunato, A. (2006). Firm Growth and Liquidity Constraints: A Dynamic
Analysis. Small Business Economics, 27(2-3), 139–156. doi:10.1007/s11187-0060006-y
Rahaman, M. (2011). Access to financing and firm growth. Journal of Banking &
Finance, 35(3), 709–723.
http://www.sciencedirect.com/science/article/pii/S0378426610003377. Accessed 5
November 2014
Rajan, R., & Zingales, L. (1996). Financial dependence and growth (No. 5758).
http://www.nber.org/papers/w5758. Accessed 16 March 2013
Sharpe, S. (1994). Financial market imperfections, firm leverage, and the cyclicality of
employment. The American Economic Review.
http://www.jstor.org/stable/10.2307/2118044. Accessed 6 February 2014
Storey, D. J. (1994). New firm growth and bank financing. Small Business Economics,
6(2), 139–150. doi:10.1007/BF01065186
Vermoesen, V., Deloof, M., & Laveren, E. (2012). Long-term debt maturity and
financing constraints of SMEs during the Global Financial Crisis. Small Business
Economics, 41(2), 433–448. doi:10.1007/s11187-012-9435-y
Voulgaris, F., Agiomirgianakis, G., & Papadogonas, T. (2014). Job creation and job
destruction in economic crisis at firm level: the case of Greek manufacturing
20
sectors. International Economics and Economic Policy, 12(1), 21–39.
doi:10.1007/s10368-014-0287-6
21
22
Variable
Delta_Emp_~g
L1.
L2.
Pre_Micro
-.17788354***
-.04657115**
Post_Micro
-.18282022***
-.05333273***
Age_Log
L1.
-.18679307**
.05640513***
Size_Act_Log
L1.
.06895191**
.1898064***
I_Fund
Index_FC
L1.
Pre_Small
-.09534733**
.0013911
-.12206751*
.07758517**
Post_Small
Pre_Med
-.14897855***
-.04303363***
-.10516583
.02809756
.15692493***
-.16942376
.19844898***
.0956849
.28509106***
.47814054*
.53992519***
.04296282
.22786694***
.37336284***
.00041349*
.00093608***
.00016521
.00038072***
-.00106231
-2.6524975***
-.81561471
-3.3086182***
-.97643732
_cons
-.46258151
N
r2
r2_a
22964
67489
31588
Post_Med
.6713391***
81809
6158
-.12134216***
-.00724532
.08788365*
.00043092*
-4.9150555***
15278
legend: * p<0.05; ** p<0.01; *** p<0.001
Table 3 – Comparative Micro, Small and Medium-Size Firms before and during the crisis
23
Variable
Total_Tec
High
Med_High
Med_Low
Low
-.14705032***
-.04745541***
-.14535949***
-.04973375***
-.15873985***
-.0537591***
-.1448667***
-.16673384***
-.1504907***
Delta_Emp_~g
L1.
L2.
-.14879911***
-.04696819***
-.18454186
.07615901
Age_Log
L1.
-.15252467***
-.26740484***
Size_Act_Log
L1.
.18930656***
.09611138*
.20549168***
.17848348***
.17228862***
Index_FC
L1.
.00092244***
.00128758*
.00093629***
.0010396***
.0007699***
_cons
-2.2723062***
-2.5261645***
-2.1200214***
-2.0069173***
N
r2
r2_a
260269
-.79172058
1476
46563
84376
116756
legend: * p<0.05; ** p<0.01; *** p<0.001
Table 4: Manufacturing sector divided by Grade of Technology
24
Variable
Delta_Emp_~g
L1.
L2.
Age_Log
L1.
High_Pre
High_Post
M_H_Pre
M_H_Post
M_L_Pre
M_L_Post
Low_Pre
Low_Post
-.24933617*
.23709512
-.12927971
.0673649
-.11975568**
-.00137414
-.14617678***
-.04704119***
-.07631199**
-.01781075
-.15656035***
-.05397532***
-.10926696***
-.02376774
-.16182824***
-.05101232***
-.61024847**
-.19944895***
-.22945298***
-.07248397***
-.40859588***
-.04089872***
-.28781555***
-.08232148***
Size_Act_Log
L1.
.06481582
.10043469
.05558752
.25942802***
-.01648231
.25033668***
.08431999***
.19149399***
Index_FC
L1.
.00056733
.00101308
-.00008632
.00095306***
-.00015737
.00105908***
.00028206
.00078318***
_cons
.63590658
-1.0250309
-.15682167
-3.4589827***
-3.4304084***
-.42725821
-2.4498579***
N
r2
r2_a
377
1099
12021
34542
1.2585477***
22031
62345
30250
86506
.05; ** p<0.01; **
Table 5: Manufacturing sector divided by Grade of Technology before and during the crisis
25
Variable
Total
Aircraft
Delta_Emp_~g
L1.
L2.
-.18462622***
-.05167424***
-.39201234**
-.01073739
Age_Log
L1.
.05532339***
.11270595
Size_Act_Log
L1.
.16508566***
-.16447488*
I_Fund
.31321883***
Index_FC
L1.
.00090067***
.00443743*
.00048622
.0010438
_cons
-2.3285626***
1.7384633*
.22979446
.05694075
157
198
N
r2
r2_a
90453
Variable
Railroad
Delta_Emp_~g
L1.
L2.
Age_Log
L1.
-.33813579***
-.15528187**
Pharma
Computing
-.16024088
-.03368189
-.4213609***
-.22046277**
-.20050122***
-.06851687*
-.18815388**
-.39643206***
-.19982549**
.01785435
.020322
-.03957884
45
Machinery
-.19192174***
-.08923823***
-.2789735***
-.1414211*
Size_Act_Log
L1.
-.08450263
.18344992***
.02960247
I_Fund
.07152951
.19327716***
-.00051036
Index_FC
L1.
-.000399
.00138609***
N
r2
r2_a
969
-.15042572*
-.12418488
-2.4872359***
-.33360107
Delta_Emp_~g
L1.
L2.
Age_Log
L1.
.00032942
-1.7204786*
-.24735118***
-.11176342*
.05511337
-.15950001***
-.05607136*
-.13718308**
.19436864**
.09762962*
.28778732***
.09344727
.00056701
.00091613*
-2.7113131**
1706
-1.0001338
Petroleum
Mineral
Metalic
-.21747702***
-.04167558
.19186462
-.29796573***
-.20396677***
-.07284195***
-.19464811***
-.06538507***
-.02483832
-.21108067
-.1026593
.02656808
.0452837
.06481984
.13072291***
-.01777679
.09945481
.29472053***
.00069274*
.00088456***
.17176703***
.05951003
.0014067***
.00255385
-2.2842846***
-.35500632
-.70926541
28
4737
3234
Reclycling
Wood
-.16892918***
-.04387323**
-.0441316
Textile
-.20062154***
-.07054536***
-.1610609***
-.04605286*
-.06318858
-.06129983
Size_Act_Log
L1.
.11717659***
.13571379***
.23752579***
I_Fund
.14879108***
.18599698***
.00836119
Index_FC
L1.
.0011018***
.00074474***
.00099945***
_cons
-1.4786766***
-1.6440091***
-2.9513248***
N
r2
r2_a
Chemical
2803
655
Variable
-.00017165
Vehicles
961
.00294395***
10439
.1678464**
Plastics
.01992289
1.5974386*
.06346636
Ships
-.22469705**
_cons
Electrical
7808
14986
7169
Table 6: All Manufacturing subsectors
26
-1.8189532***
20459
Variable
Pre_Vehicles
Post_Vehicles
Pre_Chemical
Post_Chemical
Pre_Machinery
Post_Machin~y
Delta_Emp_~g
L1.
L2.
-.50712189
-.34624866
-.23306089***
-.08412392
-.253356**
-.14258453**
-.14846691***
-.03911533
-.20072542***
-.0912749
-.18793978***
-.08126559***
Age_Log
L1.
.32969453
.00796291
-.23199421
-.12631039*
-.04608172
Size_Act_Log
L1.
.35112809
.13920904
.07151497
.10859413*
I_Fund
.25777354
.23026476**
.06421469
Index_FC
L1.
-.00124816
.00055458
.00059078
.00036181
.00148988***
.0025171
_cons
-5.1585346
-1.8981966
-.35526565
-1.1819662
-1.0549585
-3.2252342***
4.127826
-.99134009
-.1680036
N
r2
r2_a
245
716
728
2075
2528
142
513
830
Variable
Pre_Metalic
Delta_Emp_~g
L1.
L2.
-.13818495**
-.0175046
Age_Log
L1.
-.21100081**
-.02860471
.00095744*
Pre_Ships
Post_Ships
Pre_Plastics
Post_Plastics
-.22462961***
-.04216275
-.26207021
-.12924708
-.27006533***
-.14158566**
-.2091872*
-.02871337
.04798048
-.08608594
-.07404551
-.33795252
.08854358
.23629009***
-.31953465
.06960991
.07071401
.19970228
.13889453*
-.46043006
.0404836
-.20603697
7911
.00340889***
.00117728
.0416769
.17144587**
Pre_Mineral
-.10647346
-.07848822
-.2209629***
-.07243665**
-.23310236
-.0499414
-.01512423
.08304156
.06510371
.05659073
.07403119
.00162817***
Post_Mineral
-.00122224
-2.4673551**
.00077819*
.8125982
-1.0992215
1263
3474
2404
Post_Metalic
Pre_Recycling
Post_Recycl~g
Pre_Food
Post_Food
Pre_Wood
Post_Wood
Pre_Textile
-.2004607***
-.07637463***
-.18463257**
-.00858189
-.16380874***
-.04517107*
-.1762615**
-.01396942
-.2070072***
-.07411049***
-.17292586**
-.01451847
-.20348765***
-.07513601***
-.22340763***
-.09324343
-.17127122
-.00305406
-.15353329***
-.13589002
-.04479585
.0886374***
.05961873
.14889455***
.07156532
.26112058***
.09439043
-.1407288
-.0188661
.04003152
Size_Act_Log
L1.
.00856581
.17835375***
.04304775
I_Fund
.07979615
.21033502***
.17750312
.07210171
Index_FC
L1.
.00041303
.00094047***
.00093894
_cons
.33125898
-2.4603093***
-.21014338
N
r2
r2_a
5284
15175
2059
.1467152***
-.1654578
.14833885*
-.475357*
-.46886964
-.01887569
.19891291
.00104752***
.00062505
.00043121
-8.012e-06
.00093196***
.0008851
-1.9625539***
-1.4777054
-.43643921
-1.8786753***
.16170049
5749
3466
-.79331042**
10633
3878
Table 7: More statistically significant manufacturing subsectors
27
11108
1792
Post_Textile
-.15708693***
-.04440165*
-.00497169
.00101336**
-3.3895663***
5377