Financial constraints, innovation performance, and sectoral disaggregation

Abstract

How do the effects of financial constraints on innovation performance vary by sector and firm characteristics? This paper uses innovation survey data from eleven European countries to examine the heterogeneity of these effects. So far, there has been a lack of cross-country micro-level studies exploring the effects of financial constraints on innovation performance in Western Europe and only little research about the variability of such effects between the broad sectors of production and services. Our results suggest that the impact of direct measures of financial barriers differs in production and service sectors, and also by the firm's export orientation. In particular, financial constraints appear to have more pronounced negative effects in the production sector than in the service sector. Among different types of firms, the response to financial constraints seems to be stronger for non-exporters.

Full text

Financial Constraints, Innovation Performance and
Sectoral Disaggregation
Georgios Efthyvoulou
Priit Vahter
ISSN 1749-8368
SERPS no. 2012030
Originally Published: December 2012
Updated: December 2013

Financial Constraints, Innovation Performance
and Sectoral Disaggregation∗
Georgios Efthyvoulou†
University of Sheffield
United Kingdom
Priit Vahter‡
University of Tartu
Estonia
This version: 28 December 2013
Abstract
How do the effects of financial constraints on innovation performance vary by sector
and firm characteristics? This paper uses innovation survey data from eleven European
countries to examine the heterogeneity of these effects. So far, there has been a lack of
cross-country micro-level studies exploring the effects of financial constraints on innova-
tion performance in Western Europe and only little research about the variability of such
effects between the broad sectors of production and services. Our results suggest that the
impact of direct measures of financial barriers differs in production and service sectors,
and also by the firm’s export orientation. In particular, financial constraints appear to
have more pronounced negative effects in the production sector than in the service sector.
Among different types of firms, the response to financial constraints seems to be stronger
for non-exporters.
JEL classification: L1; L2; O1; O3
Keywords: financial constraints; innovation; firm heterogeneity
∗This paper was developed as part of the SERVICEGAP project, which is funded by the European Commis-
sion, Research Directorate General as part of the 7th Framework Programme, Theme 8: Socio-Economic Sci-
ences and Humanities, Grant Agreement no: 244 552. Priit Vahter acknowledges also financial support from the
Ministry of Education and Research of the Republic of Estonia financed research project no: SF0180037s08. The
authors wish to thank Andy Dickerson, Ian Gregory-Smith, Arne Risa Hole, Antonio Navas, Mary O’Mahony
and an anonymous referee for helpful comments and suggestions. The authors are also indebted to the par-
ticipants of the 11th Annual Conference of the European Economics and Finance Society, the 11th Annual
Conference of Comparative Analysis of Enterprise Data & COST, the 2013 Conference of the Estonian Eco-
nomic Association, the workshop on “Innovation: From Europe to China” at the University of Kiel, the
INDICSER meeting in Budapest, the SERVICEGAP meeting in Mannheim, and seminars at the University of
Birmingham and the University of Sheffield. Finally, the authors are grateful to Eurostat for granting access
to the CIS non anonymised micro-data. The usual disclaimer applies.
†Corresponding author. Address: Department of Economics, University of Sheffield, 9 Mappin Street,
Sheffield, S1 4DT, United Kingdom; Tel: +44 (0) 114 222 3412; Email: g.efthyvoulou@sheffield.ac.uk
‡Address: Faculty of Economics and Business Administration, University of Tartu, Narva Rd. 4 - A115,
Tartu 51009, Estonia; Tel.: +372 55 660 540; Email: priit.vahter@ut.ee
1

Financial Constraints, Innovation Performance and Sectoral Disaggregation 2
1 Introduction
There is ample evidence in the economics literature that achieving sustained long-term pro-
ductivity and economic growth is intrinsically linked to research and development (R&D)
and innovation investment (Coe & Helpman, 1995; Cainelli et al., 2006; Coe et al., 2009).
Due to informational asymmetries with external investors and uncertain and lagged returns,
this type of investment is considered to be particulary sensitive to financial constraints (Him-
melberg & Petersen, 1994; Hall, 2002). The available empirical evidence, however, is not as
conclusive as one might expect. Some studies provide evidence that financial frictions have
a strong negative effect upon innovation (Mulkay et al., 2000; Aghion et al., 2008; Ouyang,
2011), while some others fail to reach the same conclusion (Harhoff, 1998; Bond et al., 2005).
A common feature of many earlier studies on this topic is that they try to identify which
firms are more or less financially constrained by investigating the sensitivity of their R&D
investment to internally generated cash flows. Kaplan & Zingales (1997), and more recently
Campello et al. (2010), stress that traditional cash flow based indicators may fail to reveal
financially constrained firms (for example, because of positive correlation of both cash flow
and investment with expected future profitability) and argue in favor of direct survey-based
measures of financial constraints. Another concern associated with the study of obstacles to
innovation is the presence of bias arising from the endogeneity of the financial constraints
variable and survey sampling issues. Recent papers that use direct indicators based on firms’
own assessments and address such econometric problems (see Savignac, 2008; Hottenrott &
Peters, 2012; Gorodnichenko & Schnitzer, 2013) point to significant negative effects of financial
barriers on the propensity of firms to engage in innovation.1
Although the aforementioned literature has provided important insights, the tests im-
plemented consider mainly one country or a group of countries with similar characteristics
and involve a relatively small number of firms.2The shortage of systematic studies based on
cross-country data renders it difficult to conclude that the reported effects are a universal phe-
nomenon. In addition, many existing studies on this topic focus on exploring the causal effect
of financial constraints on R&D investment. For example, Brown et al. (2012) find strong
evidence that financing constraints drive R&D below the socially optimal levels. While R&D
has strengths as a measure of innovation, it is an input (not the output) in the innovation
process, and as suggested by Griffith et al. (2006), it does not take account of the produc-
tivity or the effectiveness of effort. Furthermore, public R&D and public financial support at
different stages of the innovation process may sometimes even replace the firm’s own R&D
investment (David et al., 2000; Clausen, 2009). Therefore, considering the effects on direct
measures of innovation output can complement the findings of these studies and contribute to
a better assessment of the overall impact of financing difficulties on innovation performance.
Finally, the existing literature tells us little about the cross-sectoral variability of such effects.
Since the nature of the innovation process and the intensity of usage of various innovation
inputs may differ across industries, an accurate analysis of the impact of financial barriers
on innovation performance should also take into account the considerable differences between
the aggregate production and service sectors. The presence of distinct sectoral differences in
the response to financial constraints takes on particular importance in the European context,
where the share of manufacturing in employment and value added has decreased over the past
decades and economies have shifted towards services (Brandes, 2008). Moreover, an enquiry
1See Czarnitzki & Hottenrott (2010) for an overview of the empirical literature.
2A negative relation between financial constraints and innovation has been shown, for example, in French
manufacturing firms (Savignac, 2008), in German manufacturing firms (Hottenrott & Peters, 2012), in Por-
tuguese firms (Silva & Carreira, 2012), and in transition economies (Gorodnichenko & Schnitzer, 2013; M¨anna-
soo & Merik¨ull, 2014).

Financial Constraints, Innovation Performance and Sectoral Disaggregation 3
in this direction sheds light on the channels through which better access to external finance
can foster innovation activity and lead to higher productivity and economic growth.
The present article addresses these issues using European-comparable firm-level data from
the Community Innovation Surveys (CIS). In particular, it contributes to the literature in
two main aspects. First, we explore the relationship between direct measures of financial
constraints and innovation performance using data from a large number of firms in both
Western and Eastern European countries. Second, we examine whether these effects vary
between and within the broad sectors of production and services. To avoid the spurious
positive correlation due to firms not wishing to innovate (and thus without financial obstacles
to innovation), we restrict the sample to include only the potentially innovative firms3(to
be referred to as “innovative firms” from now on) and consider the effects on the propensity
to have innovation success (positive share in sales of innovative products) rather than the
propensity to engage in innovation activities. In addition, we tackle the endogeneity problem
by estimating the probability of being a successful innovator and the likelihood to face financial
constraints simultaneously using recursive-mixed-process estimators (Roodman, 2009).
By way of preview, the main findings can be listed as follows. First, the existence of
financial constraints (especially, due to limited availability of internal funds) is an important
hampering factor to innovation performance across European countries. Second, the role of
financial constraints appears to be stronger and statistically more robust among innovative
firms in the production sector than in the service sector. Within industries and especially
within the production sector, innovative firms that do not engage in exporting activities
appear to experience the greatest problems.
The paper proceeds as follows: Section 2 presents an overview of the related literature
and develops the main hypotheses to be tested; Section 3 describes the data used; Section 4
outlines the empirical model specification and the econometric techniques applied; Sections 5
reports the estimation results and investigates their robustness; Section 6 offers a discussion
of the study’s conclusions.
2 Literature Review and Hypotheses Development
2.1 Financial constraints and innovation
Different types of investment, including investment in R&D, can be financed from two sources:
external sources, such as bank loans and other forms of debt, and internal sources, such as
retained earnings or new equity. When firms decide about their optimal levels of investment,
they choose the capital structure that minimizes their long run cost of capital. Modigliani &
Miller (1958) state that in perfect capital markets with no taxes, bankruptcy costs or asym-
metric information, investment decisions are not dependent on capital structure. However,
such conditions do not generally hold and information asymmetries or principal-agent conflicts
(Stiglitz & Weiss, 1981; Jensen & Meckling, 1976) influence lending and investment decisions.
As a result, the cost of capital and the impact of financial considerations on the investment
decision may differ across different types of investment (Meyer & Kuh, 1957; Leland & Pyle,
1977; Myers & Majluf, 1984). Investment in R&D, compared to physical assets, is likely to be
more affected by financial factors (Himmelberg & Petersen, 1994; Hall, 2002; Hall & Lerner,
2010) because it requires large sunk costs (Alderson & Betker, 1996) and produces intangi-
3Following Savignac (2008), potentially innovative firms are the firms that wish to innovate. Specifically,
the corresponding sample includes: (i) the firms that report product (good or service) or process innovation,
(ii) the firms that report ongoing or abandoned innovation activities, or (iii) the firms that report obstacles
to innovation. The excluded firms are those that are not interested in innovation; that is, the firms without
innovative activities and without any obstacles to innovation.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 4
ble assets that can be difficult to use as collateral for external borrowing (Williamson, 1988;
Alderson & Betker, 1996). Furthermore, investment in innovation projects is characterized by
high degree of information asymmetries and agency costs that result in problems of adverse
selection (Leland & Pyle, 1977) and moral hazard (Jensen & Meckling, 1976). This drives
lenders to ask for a higher rate of return than in the case of less risky investments in physical
assets (Hall, 2002; Czarnitzki & Hottenrott, 2010).4As stressed by Myers & Majluf (1984),
although information asymmetries matter for external financing of all types of investment,
they are particularly significant in limiting financing of innovation investments due to the
complexity and specificity of the innovation process.
The discussion in the previous paragraph suggests that funding from external sources is
relatively more difficult and more expensive for innovation investment in general, and R&D
in particular, than for other types of investment. Consequently, firms have to rely mostly on
internal funds to finance their innovation projects (Hall, 2002; Hottenrott & Peters, 2012).
This also implies that if internally generated funds are not available (for instance, during
periods of negative liquidity shocks) firms find themselves particularly constrained in their
R&D investment decisions, and thus in producing successful innovative products. A clear
implication for our empirical analysis is that firms with limited internal funds are more likely
to be constrained in their innovation activities, and that the detrimental effect of internal
financial constraints on innovation output is likely to be stronger than the effect of external
financial constraints alone.5
The standard empirical strategy to identify financially constrained firms has been to ana-
lyze investments’ sensitivities to changes in available sources (usually cash flows) across groups
of firms with similar attributes (Fazzari et al., 1988; Hoshi et al., 1991). However, many pa-
pers question the use of investment-cash flow sensitivities, especially in studies that do not
control for the potential endogeneity of cash flow or ignore the possibility of external finance.6
Theoretical concerns about measurement errors raised by Alti (2003) and Moyen (2004) have
shifted the focus towards more comprehensive analyses and the mechanisms behind the ob-
servation that firms depend on their cash to invest more (see, for example, Brown et al.,
2009).7As an alternative, Kaplan & Zingales (1997) and Campello et al. (2010) support
the use of more direct approaches towards the identification of financially constrained firms
based on survey data. Survey-based responses provide specific information on firms’ financial
constraints without relying on particular assumptions about their investment behavior and
can be used to explore the linkages between financial barriers, firm characteristics and the
business environment. In our context, a firm is defined to be financially unconstrained if it is
able to implement its innovation projects at optimal scale, and constrained if it is unable to
do so due to shortage of funding (see also Hall, 2002; Czarnitzki & Hottenrott, 2011).
4The cost premium for funding R&D is most likely to be higher than that for ordinary capital investment
as the potential investors find it especially difficult to distinguish between good and bad R&D investments, as
compared to shorter-term and less risky projects (Leland & Pyle, 1977; Hall, 2002). This is, in its essence, a
standard lemons’ problem (Akerlof, 1970).
5External financial constraints can arise because of the firm’s increased need for external funds to finance
its investments (for example, in R&D) and because of the high cost of external funds; that is, the cost the firm
would incur conditional upon using external funds (Hennessy & Whited, 2007).
6There is also the argument that firms tend to smooth R&D spending over time, which leads to difficulties
in measuring the impact of changes in cash in one period on subsequent investments (Hottenrott & Peters,
2012).
7A closely related literature deals with the cost of external finance (see, for example, Hennessy & Whited,
2007).

Financial Constraints, Innovation Performance and Sectoral Disaggregation 5
2.2 Explaining variation across industries
The severity of financial constraints, as well as its effect on innovation performance, may
vary significantly across firms operating in production and service industries. This can be
attributed to the potentially different nature of the innovation process and the different inten-
sity of usage of various innovation inputs in the two broad sectors. First, there are important
differences in the role of external knowledge and innovation-related cooperation (Gallouj &
Weinstein, 1997; Gadrey & Gallouj, 1998). Several studies (see, for example, OECD, 2009)
argue that innovation in service firms (as compared to innovation in production firms) relies
more intensively on obtaining knowledge from outside sources, especially clients and suppli-
ers, and on having cooperative arrangements with external partners. In fact, many types of
services are co-produced with clients and have significant inputs from clients in the service
development (Tether, 2003, 2005; Gallouj & Savona, 2009). Second, service sector innovations
tend to have a more ad hoc nature and are often more of an incremental type as compared to
production sector innovations.8For instance, Leiponen (2008) provides evidence that R&D
in services is conducted mainly by informal teamwork on a more ad hoc basis and is less
institutionalized. Third, innovation in less knowledge-intensive services is less dependent on
large scale R&D projects (Gallouj & Weinstein, 1997; Tether, 2003), suggesting that higher
R&D spending in services is not as closely associated with higher innovation output as in
production. Fourth, innovation in services can sometimes be related to changes in work or-
ganization (organizational innovations), training of workforce and the application of ICT (see
Polder et al., 2009; Mothe & Nguyen Thi, 2010). Finally, innovation-specific expenditure on
materials and energy tends to be lower in services (with the exception of transportation ser-
vices); even though labour costs make up a large share of the expenditure, for instance, due
to the importance of interaction with clients (Tether, 2003, 2005).
In summary, the combination and intensity of use of the key innovation inputs can differ
substantially between the two sectors. The lesser need for R&D (which is highly sensitive
to financial constraints) and the stronger role of other cooperation-based inputs (which are
less dependent on the availability of funds) in service industries implies that the impact of
financial barriers on innovation success may be less pronounced in services than in production.
There is also suggestion in the literature that firms in service industries are more capable of
self-financing their innovation activities and use fewer bank loans for this type of investment
than firms in production industries (Dahlstrand & Cetindamar, 2000). This may be driven
by the fact that service firms require, on average, a lower initial investment (lower sunk costs)
and have to attain a lower minimum efficient scale than production firms (Silva & Carreira,
2010). Despite these sectoral heterogeneities, most of the existing empirical studies on the
effects of financial constraints on innovation performance concentrate on manufacturing or
production industries. One notable exception is a recent study by Gorodnichenko & Schnitzer
(2013) which shows, among others, that the causal effect of financial constraints is similar
for firms in services and manufacturing. It must be stressed, however, that this study uses
the propensity to engage in innovation activities as the innovation output and employs World
Bank’s BEEPS data for about 10,000 firms in emerging market economies where the service
sector has been underdeveloped. Clearly, further analysis and empirical evaluation along these
lines (that also considers cross-country data from developed countries, employs alternative
innovation measures and puts more emphasis on sectoral-level differences) is needed. The
present article seeks to do this.
8Because services can be often consumer-specific, it has to be noted that it is not always easy to differentiate
between service innovations and service variations (Tether, 2005).

Financial Constraints, Innovation Performance and Sectoral Disaggregation 6
2.3 Explaining variation across firm groups
The importance of financial factors in constraining innovation activity may also vary accord-
ing to firm-level characteristics. An obvious distinction is between large and small companies.
Recent attempts to analyze theoretically the impact of borrowing constraints on firm dynam-
ics include Cooley & Quadrini (2001), Albuquerque & Hopenhayn (2004) and Clementi &
Hopenhayn (2006). In Albuquerque & Hopenhayn (2004) and Clementi & Hopenhayn (2006)
lenders introduce credit constraints because of limited debt contract enforceability and be-
cause of asymmetric information on the use of funds or the return on investment, respectively.
Such market imperfections force firms to enter at a suboptimal small size. However, as they
pay off their debt and increase their equity value, they establish creditworthiness and build up
internal resources that enable them to expand to their optimal size. Consequently, according
to these models, large firms have weaker borrowing constraints: if they see an investment
opportunity in an innovation project, they will be in a better position to borrow and finance
it than small firms. In a similar vein, Hennessy & Whited (2007) formulate a dynamic struc-
tural model and find that large firms face lower bankruptcy and equity flotation costs than
small firms, which makes it less costly or less difficult for them to obtain external financing
for innovation investment. Finally, as argued by Gorodnichenko & Schnitzer (2013), larger
firms are less financially constrained in their innovation activities because they have higher
availability of internal resources and can therefore benefit from economies of scale in R&D
and marketing. Other empirical studies that look into this issue include Ughetto (2008) and
Hottenrott & Peters (2012) who show that external financial constraints are more binding
for R&D and innovation of small firms, and Scellato (2007) who finds that financing barriers
affect more strongly the patenting activities of small enterprises.
Another important distinction is between firms that engage in exporting activities and
those that do not. Exporters tend to be larger and more well-known, and thus, they may
have higher levels of collateral and enjoy better relations with external investors or lenders.
More importantly, exporters typically exhibit better financial health than non-exporters. The
main reasons for that are summarized in Greenaway et al. (2007) and Bridges & Guariglia
(2008). Exporting firms have access to both domestic and international financial markets and
are less subject to those financial constraints induced by tight monetary policy and recessions
at home. This allows them to diversify sources of financing and the associated risks, and
improves the availability and stability of internal funds for investment, including investment
in innovation projects. The more stable cash flow and the better availability of internal funds
can also relax external financial constraints, due to greater assurances to lenders that the firm
will be able to service its obligations. Furthermore, exporters need to have relatively high
productivity levels to be able to cover the substantial sunk costs of exporting (see Melitz,
2003; Helpman et al., 2004). Thus, being an exporter provides a signal that the firm can
generate enough profits in foreign markets to recover such costs and can further relax external
financial constraints (for instance, when it comes to borrowing for R&D). Finally, export
status can improve firms’ access to credit by signaling borrowers’ resilience to domestic and
foreign competition, as suggested by Bernini (2012).9Following these arguments, we expect
that financial constraints will have stronger effects on the innovation performance of non-
exporters, as these firms have to rely more on internal funds to finance innovation investment.
The present article will attempt to test this hypothesis. A positive answer would also support
the view that exporting and innovation are complements: a firm that does not engage in
exporting (and thus cannot materialize productivity gains from such complementarities) is
less likely to have innovation success under the presence of financial constraints.
9We note that there is also an opposite view, claiming that exporters may face tougher financial constraints
as their international activities can mean more risks for lenders (Feenstra et al., 2011).

Financial Constraints, Innovation Performance and Sectoral Disaggregation 7
3 Data
We employ cross-country micro-level data from the fourth Community Innovation Survey
(CIS4) which covers the period 2002-2004. The countries considered in our study are: Bul-
garia, Czech Republic, Estonia, France, Italy, Norway, Portugal, Romania, Slovakia, Spain,
Sweden. The rationale for the choice of these eleven countries is twofold: first, they all have
information on the dependent and key explanatory variables used in our empirical model; sec-
ond, they report a sufficiently large number of innovative firms and provide data for firms in
both production and service sectors. Unfortunately, for three large countries (namely, France,
Italy and Spain) the information required to carry out the same empirical analysis using
data from the next CIS wave (CIS2006 with observation period 2004-2006) is not available.10
Hence, while we use CIS2006 data for robustness checks, our analysis relies primarily on CIS4
for which the sample size is very large and enables a more detailed firm-level and industry-
level comparison. Due to their confidential character, the CIS data were accessed through the
SAFE Center at the premises of Eurostat in Luxembourg. In this way, we avoid the possibility
of micro-aggregation bias associated with the use of the publicly available micro-aggregated
CIS data (Mohnen & R¨oller, 2005).
The CIS data set has a number of advantages relative to data sets employed in previous
studies. First, it provides direct self-reported measures of firms’ financial constraints and inno-
vation, and thus, we do not need to rely on indirect proxies. Second, it is the only data source
that contains cross-country information on innovation activities in Western European coun-
tries. Third, it is based on a common survey questionnaire and methodology and includes
data on a large number of firms and a broad range of industries, which makes the corre-
sponding data set suitable for cross-industry and cross-country comparison. Fourth, it entails
information on both internal and external financial constraints, which allows us to identify
the channels through which financing barriers may affect innovation. In order to construct
instrumental variables for our measures of financial constraints, we also employ industry-level
data from Amadeus: a comprehensive database containing comparable financial information
for millions of companies across Europe. Our CIS4-based sample, which results from merging
these two sources, contains about 38,000 innovative firms, out of which about 28,000 are from
Western European countries. Notice that firms that operate in the financial intermediation
sector, which are frequently subject to prudential supervision and government intervention,
are not included in our sample.
4 Empirical Strategy
One distinctive characteristic of the CIS questionnaire is that it begins by asking all firms for
some general information and whether they have innovation activities (completed, ongoing,
or abandoned) or face any obstacles to innovation. Then, only the firms that provide positive
responses to these questions (that is, the firms that wish to innovate) are requested to answer
a large number of additional questions, such as those on public financial support, informa-
tion sources and cooperation. In the last part of the questionnaire, all surveyed firms are
asked about financial and non-financial constraints to innovation. As pointed out by Savignac
(2008), questioning the firms that do not wish to innovate, and hence do not meet any finan-
cial constraints, about such constraints may lead to a positive correlation between the two
variables. To cope with this issue, we restrict the sample to include only the firms that wish
10The CIS2008 (with observation period 2006-2008) is also not considered here as the CIS2008 questionnaire
does not include questions on factors hampering innovation activities.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 8
to innovate11 and consider the impact of financial constraints on innovation success rather
than the propensity to innovate.12
An econometric problem associated with the study of obstacles to innovation is the endo-
geneity of the financial constraints variable. This endogeneity may arise because both financial
constraints and innovation patterns may be affected by common elements of unobservable het-
erogeneity; for example, by firm-specific risk factors, such as the uncertainty associated with
the output of an innovation project, or the lack of information about the time needed to bring
an innovation project onto the market (Savignac, 2008). To address this problem, we consider
a recursive two-equation model that builds on the works of Piga & Atzeni (2007) and Savignac
(2008) and takes the following form:
I∗
isc =X′
iscβ1+ϑrRisc +ϑfF Cisc +ε1(1)
F C ∗
isc =X′
iscβ2+Y′
iscγ+ε2(2)
(ε1
ε2)∼N{(0
0),[1ρ12
ρ12 1]}
where I∗and F C ∗are the unobserved latent variables underlying I0and F C ,I0is a dummy
variable that equals one if the firm reports positive turnover from newly introduced or sig-
nificantly modified goods or services13 (‘Innovation Success’); F C is a dummy variable that
equals one if the firm reports that the lack of finance from either internal or external sources is
highly important in hampering its innovation activities (‘Financial Constraints’); Xis a vector
of control variables; Ris the observed R&D intensity, measured by the logarithm of R&D ex-
penditure as percentage of sales plus 0.1 (‘R&D Intensity’); Yis a vector of additional control
variables (to act as instruments for F C ); ε1and ε2are idiosyncratic error terms, assumed
to be jointly normally distributed; ρ12 is the correlation coefficient between ε1and ε2, which
accounts for the possible existence of omitted or unobserved factors that affect simultaneously
the two equations; and, i,s,cindex firm, industry and country, respectively. Vector Xin Eq.
(1) includes control variables commonly used in innovation production functions (see Cr´epon
et al., 1998; Mohnen & R¨oller, 2005; Griffith et al., 2006; L¨o¨of & Heshmati, 2006; Mairesse
& Mohnen, 2010); that is, an indicator of whether the firm has cooperative arrangements
on innovation activities14 (‘Cooperation’), a categorical variable reflecting different sources of
information for innovation15 (‘External Search’), an indicator of whether the firm uses design
pattern, trademarks, or copyright to protect inventions or innovations (‘Formal Protection’),
indicators of whether the firm has exporting activities (‘Export’) and is part of an enterprise
group (‘Group’), and, finally, the size of the firm, measured by the logarithm of the number
of employees (‘Size’). To capture unobserved heterogeneity, we also include industry (αs) and
country (αc) fixed effects.
However, another econometric concern that arises with the above model is that the R&D
intensity variable in Eq. (1) may also be endogenous relative to the dependent variable, and
thus correlated with ε1and ε2. This suggests that estimating only Eqs. (1) and (2) jointly
11This is similar to the approach followed by Brown et al. (2012), who concentrate their analysis on R&D
reporting firms.
12Notice that the CIS data set provides limited information for the sample of firms not wishing to innovate,
and thus, estimating first the impact of financial constraints on the propensity to innovate (using a two stage
approach) is not possible.
13As a measure of the commercial success of innovation, this variable outperforms simple indicators coding
engagement in some kind of innovation activity (Mohnen & R¨oller, 2005; Mairesse & Mohnen, 2010).
14Specifically, this indicator captures cooperation with at least one of the following partners: other enter-
prises within the same enterprise group, other enterprises in the same sector, suppliers, clients or customers,
government, and universities. It excludes, however, cooperation with R&D institutes.
15These include knowledge from within the enterprise group, from clients, suppliers, competitors, consultants,
universities, research institutions, conferences, professional associations and scientific journals.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 9
may produce inconsistent results and lead to misleading inferences. To address this issue, we
treat the potential endogeneity of Ras an omitted variable problem and employ a control
function correction method (see Blundell & Powell, 2003). The idea behind this method is
to derive a proxy that conditions on the part of Rthat is correlated with ε1and ε2, and use
this proxy as a covariate in Eqs. (1) and (2). Formally, we consider a modified version of the
above model that accounts for the endogeneity of both F C and R, as follows:
I∗
isc =X′
iscβ1+ϑrRisc +ϑfF Cisc +αs+αc+ρ13 ε3+ ¯ε1(3)
F C ∗
isc =X′
iscβ2+Y′
iscγ1+ρ23 ε3+ ¯ε2(4)
Risc =X′
iscβ3+Y′
iscγ2+Z′
iscϑz+ε3(5)
ε1=ρ13ε3+ ¯ε1
ε2=ρ23ε3+ ¯ε2
(¯ε1
¯ε2)∼N{(0
0),[1ρ12
ρ12 1]}
where Zis a vector of additional control variables (to act as instruments for R); ε3is an
idiosyncratic error term, such that ε1,ε2and ε3, as well as the tails ¯ε1and ¯ε2, are jointly nor-
mally distributed; and, ρ13 and ρ23 are the correlation coefficients between the corresponding
idiosyncratic error terms.
We estimate the proposed model using a two step procedure. In the first step, we estimate
Eq. (5) by least squares and use the resulting estimates to construct the control function ˆε3.
With the control function in place, in the second step we jointly estimate Eqs. (3) and (4)
using conditional recursive mixed process (cmp) estimators (see Roodman, 2009). The cmp
approach takes into account both simultaneity and endogeneity risks and produces consistent
estimates for recursive systems in which the endogenous variables appear on the right-hand-
side as observed. Since Eqs. (3) and (4) constitute a recursive process (imposed by the
instrumentation strategy), consisting of one structural equation (innovation success equation)
and one reduced-form equation (financial constraints equation), the analysis in the second step
is essentially a limited information maximum likelihood (LIML) estimation. The advantage
with this approach, as opposed to two-stage least squares and related linear methods, is the
gain in efficiency as it takes into account the covariances of the errors and uses the information
about the limited nature of the reduced-form dependent variable (Anderson, 2005; Roodman,
2009). In the special case where both dependent variables are binary (as above), the model is
fundamentally a bivariate probit model with endogenous dummy regressor (see Wooldridge,
2002). Since the second step uses an estimate of ε3from the first step, as opposed to the true
ε3, the asymptotic sampling variance of the second-step estimates needs to take this extra
source of variation into account. To do that we undertake 200 replications of the procedure to
bootstrap the estimated standard errors.16 Karaca-Mandic & Train (2003) derive the specific
form of the standard formulas for two-step estimators that is applicable to the control function
approach, and note that the bootstrap and asymptotic formulas (Newey, 1984; Murphy &
Topel, 1985) provide very similar standard errors for such applications.
The consistency of the proposed methodology depends on the validity of instruments
included in vectors Yand Z, which in turn, relies on two conditions. First, the instruments
must be determinants of the likelihood to face financial constraints and the value of R&D
intensity, respectively. Second, they must not be correlated with the unobserved factors
that may affect the propensity to have innovation success. It is easy to show that the first
condition is satisfied: the estimated coefficients on the instruments must have the expected
16The results remain qualitatively the same when we undertake different number of bootstrap replications,
such as 100 or 300.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 10
sign and be statistically significant at conventional levels of significance. To make sure that the
second condition is fulfilled, we choose variables which affect the firm’s innovation performance
only through the dependent variable in Eqs. (4) and (5). Specifically, vector Yincludes
four industry-level proxies of financial performance (calculated using cross-country industry
averages); namely, ‘Public Support’, measured by the number of different types of sources
of public funding for innovation, ‘Collateral’, measured by the logarithm of tangible assets,
‘Financial Risk’, measured by the gearing ratio, and ‘Profitability’, measured by the operating
cash flow ratio.17 We expect that firms that operate in industries that are highly dependent
on funding from a variety of public sources are more likely to report financial constraints.
In addition, following Canepa & Stoneman (2003), we expect that more profitable industries
are less likely to experience financial constraints (due to larger amount of internal funds),
whereas more risky industries and those with fewer realisable assets are more likely to be
financially constrained (due to more difficult access to external funds). Such industry-level
variables appear to be good instruments, since they can influence the amount of internal
funds and the attractiveness of firms to external investors but cannot influence the firm’s
innovation performance directly. On the other hand, vector Zincludes two variables that
exert positive influence on innovation success through the R&D channel; namely, ‘Continuous
R&D’, measured by an indicator of whether the firm engages in continuous intramural (in-
house) R&D, as opposed to occasional or no intramural R&D at all; and, ‘Cooperation R&D’,
measured by an indicator of whether the firm has cooperative arrangements on innovation
activities with R&D institutes.18 To test the validity of the aforementioned instruments, we
perform the Sargan-Hansen test for over-identifying restrictions in linear LIML models. Even
though there is no theoretical evidence to suggest that the assumptions necessary to perform
this test are satisfied in the bivariate probit with endogenous dummy regressor, previous
empirical studies argue that this is actually the best available diagnostic (Evans & Schwab,
1995; Y¨or¨uk, 2009).
Notice that in order to ensure that the sample is representative of the relevant population
of firms in each country, all regressions are weighted by country sampling weights. These
weights correspond to the inverse of the probability of selection; that is, the total population
of firms divided by the total number of surveyed firms in each country. Detailed definitions
and data sources for all variable described in this section are given in Table A.2. The cross
correlation matrix for these variables is displayed in Table A.4.
5 Empirical Results
5.1 Descriptive statistics: production versus services
We start by considering the differences across sectors with respect to financial obstacles and
innovation inputs. Tables A.1 and A.3 display the frequency distribution of financially con-
strained firms for production and service industries, as well as descriptive statistics of the main
variables for the two broad sectors. According to the figures, innovative firms in production
industries are more likely to report financial constraints than establishments in service indus-
17‘Collateral’, ‘Financial Risk’ and ‘Profitability’ are constructed using the cross-country 3-year average of
the corresponding firm-level variables based on Amadeus data for the periods 2002-2004 (for CIS4) and 2004-
2006 (for CIS2006). To ensure that the industry-level measures are not sensitive to extreme values, all firm-level
variables are first winsorized at 1% and 99%.
18This is similar to the approach followed by Griffith et al. (2006), who include the variable ‘Cooperation’
(capturing all types of cooperative arrangements on innovation activities) in the R&D intensity equation but
not in the innovation outcome equation. However, tests for over-identifying restrictions - based on our sample
- reveal that only cooperation with R&D institutes is a valid instrument, as cooperation with other types of
cooperation partners can also affect innovation success directly.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 11
tries, mainly due to the presence of higher barriers to external financing. In addition, firms
in both sectors are more likely to report internal financial constraints than external financial
constraints. Another interesting feature is the existence of apparent differences between the
two sectors concerning the type and intensity of use of innovation inputs. In particular, the
proportion of firms reporting that they are engaged in continuous intramural R&D and the
level of R&D intensity are greater in production than in services. Production firms also have,
on average, higher levels of collateral (measured by the value of tangible assets), are more
likely to engage in cooperative arrangements with R&D institutes, and draw on formal mea-
sures to protect returns from innovation more frequently. On the other hand, service firms use
a larger variety of sources of information to develop or improve their products or services, and
are more likely to engage in innovation-related cooperation with external partners, such as
customers or suppliers. A further look into the standard deviations from the sectoral means
reveals the dispersion and heterogeneity of firms that operate in service industries, especially
with respect to their financial performance (as captured by the indicators of financial risk and
profitability). The large variance of the R&D intensity variable in the case of service firms
reflects the high values of R&D investment for knowledge-intensive industries combined with
the much lower values for less knowledge-intensive industries, such as hotels and catering.
5.2 Main findings
We continue our analysis by estimating Eq. (3) (with the control function ˆε3embedded into
the specification) for the full sample of innovative firms using a univariate probit model (see
column (1) of Table 1). As a first point, we can notice that the traditional determinants
of innovation performance included in our model have the expected positive sign and are
statistically significant at the 1% confidence level. Specifically, the results suggest that coop-
eration and formal protection increase the probability of having innovation success by 4% and
13%, respectively. These results are similar to those found in papers considering the Cr´epon
et al. (1998)’s three-stage model or other models on the innovation value chain in European
countries (see Roper et al., 2008; OECD, 2009). Consistent with the literature on “open inno-
vation” (Chesbrough, 2003; Dahlander & Gann, 2010; Love et al., 2013) and the causal effects
of exporting (Salomon & Shaver, 2005; Damijan et al., 2008; Vahter, 2011), we also find that
external knowledge sourcing and export orientation play an important role: adding a new
type of external knowledge linkage and having exporting activities are associated with 2%
and 5% higher probability to be in the group of successful innovators, respectively. Finally, as
shown in the vast majority of studies on innovation (see, for example, Beckeikh et al., 2006;
Griffith et al., 2006), innovation performance is positively related with firm size and R&D ef-
fort. Turning now to our variable of interest (‘Financial Constraints’), we can see that it has a
surprising positive and highly statistically significant impact on innovation performance. This
lends support to the endogeneity argument and the need for a two-equation model: ignoring
the endogeneity of the financial constraints variable may render the estimates of a univariate
probit equation biased and inconsistent.
Column (2) of Table 1 presents the results of a bivariate probit estimation where the
endogenous financial constraints variable is instrumented using the specification of Eq. (4)
(with the control function ˆε3embedded into both Eqs. (3) and (4)). The evidence obtained
validates the above statement: once the endogeneity bias is corrected, we find a negative
(but statistically insignificant) relationship between financial constraints and innovation per-
formance while all other estimates remain virtually unchanged. As pointed out in Section
2, the impact of financial constraints is expected to be more pronounced for firms that do
not engage in exporting activities. To test this prediction, we re-estimate the specification
in column (2) for this group of firms. The results, displayed in column (3), indicate stronger

Financial Constraints, Innovation Performance and Sectoral Disaggregation 12
effects for non-exporters: the coefficient on ‘Financial Constraints’ has the expected negative
sign and appears to be statistically significant at the 1% confidence level. Qualitatively, the
corresponding estimate suggests that the likelihood to have innovation success is 40% lower
for non-exporters who face financial constraints. It must be stressed that in all specifications
of Table 1 (as well as in those of the subsequent tables), the instruments included in vectors Y
and Zhave the desirable properties. Specifically, they appear to be strong determinants of the
likelihood to experience financial constraints and the level of R&D intensity,19 respectively,
and to be uncorrelated with the error term of the innovation success equation (the p-value
of the over-identifying restriction tests is above standard levels of statistical significance).20
Another finding worth mentioning is that the variables ‘Size’ and ‘Production’ (coding firms
in production industries21) enter the financial constraint equation highly statistically signif-
icantly and with negative and positive sign, respectively, suggesting that the probability to
be financially constrained is higher for small firms and for firms that operate in production
industries.22
<Insert Table 1 here >
Drawing upon the last result, we now turn to investigate possible cross-sector heterogene-
ity of the causal effect of financial constraints on innovation success. To do that, we partition
the full sample of innovative firms into production and service industries and re-estimate the
regression specifications of columns (1)-(3) in Table 1. Columns (1)-(3) in Table 2 present the
results for production industries, whereas columns (1)-(3) in Table 3 present the results for
service industries. Two regularities stand out. First, the causal effect of financial constraints
on innovation performance is stronger and statistically more robust in production than in ser-
vices (see column (2)). Second, within the two sectors and particularly within the production
sector, firms that do not have exporting activities are more sensitive to financial constraints
(see column (3)). Specifically, in production, the estimated reduction in the probability of
having innovation success due to the presence of financial constraints is large (22% for the
full sample and 45% for the sub-sample of non-exporters) and statistically significant in both
specifications. In contrast, in services, the corresponding effects are relatively small, or of the
opposite direction, and fail to reach statistical significance. The observed heterogeneity in the
impact of financial barriers between the two sectors is consistent with previous studies docu-
menting that the innovation output in service industries is less dependent on R&D (which is
particulary sensitive to financial constraints) and more dependent on other, less costly inputs,
such as collaboration with clients and knowledge from outside sources. As shown in columns
(2) and (3), while ‘R&D Intensity’ has stronger effects on innovation success in production
industries, the impact of ‘Cooperation’ and ‘External Search’ is relatively more pronounced in
service industries. In addition, our results indicate that ‘Collateral’ in services is not as highly
associated with financial constraints as in production (see column (2)), which may capture
the fact that service industries require a lower initial investment and tend to use fewer bank
loans compared to production industries (see Silva & Carreira, 2010).
19For brevity, Tables 1, 2 and 3 do not display the results for the R&D intensity equation. However, in all
regressions the estimated coefficients on the instruments ‘Continuous R&D’ and ‘Cooperation R&D’ appear to
be positive and highly statistically significant (results available upon request).
20When we replace the industry-level public support variable with its firm-level counterpart, the Hansen-
Sargan test rejects the hypothesis that the instruments are correctly specified, confirming the validity of our
chosen instrument structure.
21Production industries include: manufacturing; mining and quarrying; electricity, gas and water supply;
and, construction. Service industries include: wholesale and retail trade, repair of motor vehicles, personal and
household goods; hotels and restaurants; transport, storage and communication; and, real estate, renting and
business activities.
22Notice, however, that the variable ‘Production’ controls mainly for unobserved heterogeneity, since sectoral
differences can also be captured, to some extent, by the other regressors in the financial constraints equation.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 13
As mentioned in Section 2, information asymmetries and the intangible nature of assets
created by innovation projects increase the cost of external fund raising for such investments,
and hence, firms, first and foremost, use internal funds to finance innovation projects as com-
pared to external debt. This, in turn, implies that firms with limited internal funds are more
likely to be constrained in their innovation performance, as they may have to leave some of
their innovation projects on the shelf (Hottenrott & Peters, 2012). This conjecture is sup-
ported by our results: when we re-define the ‘Financial Constraints’ variable to capture lack of
finance from internal sources, we find a monotonous increase in the responsiveness to financial
constraints both in the full sample of innovative firms and the sub-sample of production in-
dustries. Specifically, the estimated coefficients on ‘Financial Constraints’ and the associated
marginal effects reported in columns (4)-(5) of Table 1 and 2 appear to be economically and
statistically more significant compared to those in columns (2)-(3). As expected, implement-
ing the same tests using lack of finance from external sources as proxy for financial constraints,
produces weaker causal effects: the estimated coefficient on ‘Financial Constraints’ is smaller
in absolute value and reaches statistical significance only in the equations of non-exporters
(see columns (6)-(7) of Tables 1 and 2). Notice that the chosen instruments behave in the
predicted way across these new specifications; that is, ‘Profitability’ appears to be stronger
determinant of internal financial constraints, whereas ‘Collateral’ and ‘Financial Risk’ appear
to be stronger determinants of external financial constraints.
<Insert Table 2 and Table 3 here >
Are the reported findings sensitive to alternative definitions of the outcome variable? To
answer this question, we use as threshold for innovation success the value that corresponds
to the sample’s 50th and 75th percentiles of “share of sales with new products” (5% and
20%, respectively), and re-run the regression package of Tables 1, 2 and 3. Rows (1)-(3)
of Table 4 summarize the results on the financial constraints variable23 when we consider
the median as threshold value for coding successfully innovative firms (Im), while rows (4)-
(6) when we consider the upper quartile as alternative threshold value (Iq). Overall, re-
coding the ‘Innovation Success’ variable generates estimates which are similar to our baseline
estimates and lead to the same conclusions. Once again, we find that innovation performance
in production industries exhibits high responsiveness to financial constraints, especially when
we allow financial constraints to depend on the availability of internal funds. Furthemore,
we find that the strength of the response is particularly high for firms that do not engage in
exporting activities, although the difference between exporters and non-exporters seems to
become smaller as we move towards higher threshold values of innovation success. This result
may reflect the fact that all production firms, regardless of export status, may face difficulties
in achieving very high levels of innovation performance (and thus being included in the top
quartile of most successful innovators) when they are financially constrained.
The cmp procedure (Roodman, 2009) works for a large class of simultaneous-equation
systems where the equations can have different types of dependent variables. Thus, in order
to further explore the sensitivity of our results to the definition used for the outcome variable,
we treat ‘Innovation Success’ as a left-censored continuous variable and re-estimate Eqs. (3)
and (4) jointly using a combination of a tobit model with censoring from below at zero (for the
innovation success equation) and a probit model (for the financial constraing equation). Even
though using the informational content of “share of sales with new products” may allow us to
identify causal effects on the intensity of innovation success, shortcomings in the distribution
23For brevity and comparability, Tables 4 and 5 display only the results on our variable of interest. The
estimated coefficients on the remaining control variables and instruments are very similar to those reported in
the baseline specifications and do not change the inferences drawn from earlier findings.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 14
and range limits of this variable point to its subjective nature and suggest that we should
perhaps not draw too strong conclusions based on its continuous variation (see also Mairesse
& Mohnen, 2010). In particular, this variable: (i) has values that tend to be rounded (for
example, 10%, 20%); (ii) has a highly skewed distribution with a large mass of firms reporting
zero or very low level of innovative sales; (iii) may be plagued by outliers, as some countries
have a surprisingly large number of firms reporting a high percentage of innovative sales (even
100%).24 Nevertheless, replacing the binary indicator with its continuous counterpart and
implementing the aforementioned estimation strategy produces results that lead to the same
conclusions (see rows (7)-(9) of Table 4).
<Insert Table 4 here >
5.3 Robustness tests
To assess the robustness of the above findings we perform several tests. First, we check
whether our results hold when we take into account the level of financial market development
of the sampled countries, which is considered to be an indicator of the accessibility of external
finance. To do that, we exclude the Eastern European countries from our sample (namely,
Bulgaria, Czech Republic, Estonia, Romania and Slovakia) and run the same regressions as
in Tables 1, 2 and 3. As shown in rows (1)-(3) of Tables 5, the estimates on the financial
constraints variable are not much influenced by this exercise, suggesting that the reported
findings are not driven by countries with relatively lower level of financial development. Sec-
ond, we test the sensitivity of our results to different time samples by employing pooled data
from both CIS4 and CIS2006; that is, by adding data from CIS2006 for eight out of the eleven
CIS4 sampled countries. Despite the obvious problems with this approach (such as, consid-
ering firms that were surveyed in both waves), the results obtained confirm our key findings:
stronger response to financial constraints for production industries (particularly when we fo-
cus on the lack of finance from internal sources) and more pronounced effects for firms with
no exporting activities (see rows (4)-(6) of Table 5). In a third robustness check, we examine
if excluding the variable ‘R&D Intensity’ from the set of controls in Eq. (1) severely biases
our estimates of innovation sensitivity to financial constraints. The results in rows (7)-(9) of
Table 5 indicate that the answer is no: estimating Eqs. (1) and (2) using a bivariate pro-
bit model and allowing the R&D intensity to be included in the idiosyncratic error term ε1
produces similar results and does not change the inferences drawn.
Finally, we conduct further tests of robustness, such as dropping the non-manufacturing
industries (utilities, mining and quarrying and construction) from the aggregate production
sector,25 excluding the variable ‘Public Support’ from the list of instruments in Y, and running
the robustness tests described in this section using the alternative definitions of the variable
‘Innovation Success’. Once again, estimates based on these tests are very similar to the
baseline estimates (results available upon request).
<Insert Table 5 here >
6 Conclusions
This paper contributes to the literature in two main aspects. First, we use data from about
38,000 innovative firms in both Western and Eastern European countries and provide evidence
24Notice that the normality assumption is rejected even we exclude the 0% and 100% shares of innovative
sales.
25Firms in manufacturing industries constitute 90% of the production sector.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 15
that the presence of financial constraints is strongly negatively related to innovation perfor-
mance. This effect seems to be driven by limited availability of internal funds (rather than
limited access to external funds), which is consistent with the idea that innovation projects
tend to be financed by retained profits or equity, and thus the lack of funds from such sources
is a more binding constraint. Second, we show that the responsiveness to financial constraints
differs between production and service sectors, and also by the firm’s export status. Specifi-
cally, we find that: (i) innovative firms in production industries are significantly and robustly
more sensitive to financial frictions than those in service industries; (ii) within sectors, finan-
cial frictions are particularly detrimental for innovative firms with no exporting activities. The
differential impact of financial constraints between the two broad sectors can be explained, to
some extent, by sectoral differences in the type, combination and intensity of use of innova-
tion inputs. Specifically, the success of commercializing innovative products in service firms
appears to rely more on inputs that are less dependent on the availability of funds (such as,
collaboration with clients and suppliers and external sources of information) and less on R&D
investment which is highly sensitive to financial constraints. On the other hand, the stronger
effects of financial barriers for non-exporters may reflect the relatively lower productivity and
financial performance of these firms, which weakens their ability to overcome the sunk costs
of innovation investments. In addition, the less stable cash flow resulting from not having
international activities can signal lower ability to service their external debt and strengthen
the impact of external financial constraints on their innovation output. Finally, the absence
of complementary gains from engaging in both innovation and exporting can lead to higher
sensitivity to financial frictions.
Our results emphasize the role of financial constraints as one of the principal driving forces
behind low innovation performance for a significant portion of firms. Therefore, policies aiming
at enhancing access to external finance26 can have a strong positive impact on innovation
intensity in firms with limited internal funds, which may lead to a more rapid development
of new goods and services, and higher economic growth. On the other hand, the finding
that innovation sensitivity to financial frictions varies across sectors contributes to a better
understanding of sectoral heterogeneities, and provides micro-foundations for interpretation
of different effects on productivity and economic growth. In particular, financial frictions
affecting more strongly innovation performance in production industries (compared to service
industries) can account for possible productivity gaps between the two sectors27 and be seen
as one of the factors that cause different responses to financial crises (see Efthyvoulou, 2012).
Hence, further investigation into the mechanisms of how the occurrence of financial crises
affects firm-level and sectoral-level innovation performance is an important task for future
research.
26Examples of such policies include efforts to improve accounting standards and craft regulations that permit
firms to list on equity markets at an earlier stage (see Brown et al., 2012).
27The contributions of services and other sectors to the aggregate productivity growth in the United States
and Europe are discussed, for example, in van Ark et al. (2008).

Financial Constraints, Innovation Performance and Sectoral Disaggregation 16
A Appendix
A.1 Community Innovation Surveys
IWe consider a firm to be potentially innovative during the surveyed period if it answered
positively to at least one of the following: (1) introduced new or significantly improved
products (good or services) with respect to its capabilities, such as improved software,
user friendliness, components or sub-systems; (2) introduced new or significantly im-
proved process, distribution, method, or support activity for its goods or services; (3)
had any ongoing or abandoned innovation activities; (4) faced obstacles to innovation.
IThe key question about financial constraints to innovation is the following: “During the
surveyed period, how important were the following factors for hampering your innovation
activities or projects or influencing a decision not to innovate?”. We focus on two
factors: lack of funds within the enterprise group (internal financial constraints); lack of
funds from outside the enterprise (external financial constraints). The answer choices
are: (a) factor of high importance; (b) factor of medium importance; (c) factor of low
importance; (d) factor not experienced. We consider a firm to be financially constrained
if it answered that the lack of finance (either from internal of external sources) was
highly important in hampering its innovation activities or projects (in terms of leading
to delay, abandonment or not starting innovation projects).
A.2 Tables
Table A.1: Statistics for CIS4 firms
All Industries Production Services
Number of innovative firms 38482 25373 13109
Report impactaon innovation activities 17588 (45.7%) 11941 (47.1%) 5647 (43.1%)
Due to lack of
either internal or external finance 6497 (16.9%) 4364 (17.2%) 2133 (16.3%)
internal finance 4805 (12.5%) 3173 (12.5%) 1632 (12.5%)
external finance 3899 (10.1%) 2620 (10.3%) 1279 (9.8%)
both internal and external finance 2207 (5.7%) 1429 (5.6%) 778 (5.9%)
internal (but not external) finance 2598 (6.8%) 1744 (6.9%) 854 (6.5%)
external (but not internal) finance 1692 (4.4%) 1191 (4.7%) 501 (3.8%)
aImpact refers to serious delay, abandonment or not starting innovation projects.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 17
Table A.2: Description of variables for innovative firms
Variable Name Definition Source
Innovation Success (I0) 0-1 dummy variable, =1 if the firm reports positive turnover
from newly introduced or significantly modified goods or ser-
vices (“share of sales with new products”). Alternative defi-
nitions use the sample’s 50th and 75th percentiles of “share of
sales with new products” as threshold for innovation success
(denoted by Imand Iq, respectively).
CIS
Financial Constraints 0-1 dummy variable, =1 if the firm reports that the lack
of finance from either internal or external sources is highly
important in hampering its innovation activities
CIS
Cooperation 0-1 dummy variable, =1 if the firm has cooperative arrange-
ments on innovation activities with other enterprises or non-
commercial institutions (excludes cooperation with R&D in-
stitutes)
CIS
External Search number of highly important sources of knowledge or infor-
mation for innovation (ranges from 0 to 10)
CIS
Formal Protection 0-1 dummy variable, =1 if the firm uses design pattern,
trademarks, or copyright to protect inventions or innovations
CIS
Export 0-1 dummy variable, =1 if the firm sells goods or services in
other countries
CIS
Group 0-1 dummy variable, =1 if the firm is part of a firm group
(two or more legally-defined firms under common ownership)
CIS
Size number of employees in logs CIS
R&D Intensity [(R&D expenditure/sales)∗100 + 0.1] in logs CIS
Cooperation R&D 0-1 dummy variable, =1 if the firm has cooperative arrange-
ments on innovation activities with R&D institutes
CIS
Continuous R&D 0-1 dummy variable, =1 if the firm reports continuous en-
gagement in intramural (in-house) R&D
CIS
Public Support number of sources of public financial support for innovation
(ranges from 0 to 3: local, national, EU); industry-level av-
erage
CIS
Collateral = tangible assets in logs; industry-level average Amadeus
Financial Risk = ((non current liabilities+loans)/shareholders funds)∗100;
industry-level average
Amadeus
Profitability = (cash flow/operating revenue)∗100; industry-level average Amadeus
Industry Dummies set of industry dummies according to the firm’s main busi-
ness activities (NACE 2-digit level)
CIS
Table A.3: Summary statistics
All Industries Production Services
Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.
Innovation Success (I0) 0.65 0.47 0.67 0.47 0.61 0.48
Innovation Success (cont.) 21.29 29.96 21.08 29.16 21.70 31.44
Financial Constraints 0.16 0.37 0.17 0.37 0.16 0.36
Cooperation 0.31 0.46 0.30 0.45 0.33 0.47
External Search 1.47 1.51 1.45 1.49 1.50 1.55
Formal Protection 0.33 0.47 0.34 0.47 0.31 0.46
Export 0.56 0.49 0.63 0.48 0.40 0.49
Group 0.41 0.49 0.40 0.48 0.43 0.49
Size 4.18 1.39 4.26 1.36 4.03 1.42
R&D Intensity 0.23 1.85 0.30 1.72 0.10 2.07
Cooperation R&D 0.13 0.33 0.13 0.34 0.12 0.33
Continuous R&D 0.35 0.47 0.37 0.48 0.31 0.46
Public Support 0.13 0.07 0.14 0.06 0.11 0.08
Collateral 6.67 0.84 6.97 0.74 6.09 0.72
Financial Risk 104.78 41.32 107.75 10.68 99.03 41.94
Profitability 6.27 3.57 6.07 2.85 6.64 4.63

Financial Constraints, Innovation Performance and Sectoral Disaggregation 18
Table A.4: Cross correlation matrix for regression variables
Innovation Success Innovation Success Financial Cooperation External Formal Export Group
(I0) (cont.) Constraints Search Protection
Innovation Success (I0) 1.00
Innovation Success (cont.) 0.51 1.00
Financial Constraints 0.08 0.05 1.00
Cooperation 0.16 0.07 0.08 1.00
External Search 0.14 0.12 0.07 0.15 1.00
Formal Protection 0.17 0.05 0.05 0.18 0.14 1.00
Export 0.14 0.02 0.03 0.15 0.08 0.21 1.00
Group 0.03 -0.07 -0.04 0.23 0.00 0.18 0.20 1.00
Size 0.05 -0.06 -0.03 0.21 0.07 0.22 0.25 0.44
R&D Intensity 0.12 0.13 0.09 0.14 0.13 0.08 0.09 -0.06
Cooperation R&D 0.12 0.05 0.06 0.52 0.13 0.14 0.11 0.16
Continuous R&D 0.20 0.09 0.07 0.25 0.14 0.26 0.26 0.18
Public Support 0.18 0.10 0.07 0.13 0.06 0.12 0.20 0.07
Collateral 0.00 -0.07 -0.05 0.03 -0.05 0.00 0.16 0.07
Financial Risk -0.17 -0.09 -0.03 -0.18 -0.11 -0.10 -0.10 -0.10
Profitability -0.07 -0.03 -0.02 0.03 0.05 -0.01 -0.05 0.02
Size R&D Cooperation Continuous Public Collateral Financial Profitability
Intensity R&D R&D Support Risk
Size 1.00
R&D Intensity -0.10 1.00
Cooperation R&D 0.17 0.10 1.00
Continuous R&D 0.16 0.16 0.21 1.00
Public Support 0.05 0.26 0.10 0.29 1.00
Collateral 0.10 0.01 0.06 0.07 0.20 1.00
Financial Risk -0.12 0.03 -0.08 -0.06 -0.25 0.27 1.00
Profitability 0.03 0.02 0.02 -0.01 -0.02 0.08 0.03 1.00

Financial Constraints, Innovation Performance and Sectoral Disaggregation 19
Table 1: Bivariate probit model: all industries
Probit Bivariate probit
Internal or External FC Internal FC External FC
All Firms All Firms Non-Exporters All Firms Non-Exporters All Firms Non-Exporters
Coef. dy/dx Coef. dy /dx Coef. dy/dx Coef. dy/dx Coef. dy/dx Coef. dy/dx Coef. dy/dx
(1) (2) (3) (4) (5) (6) (7)
Equation for innovation successa
Financial Constraints 0.11*** 0.04 -0.43 -0.16 -1.09*** -0.40 -0.09 -0.03 -1.16*** -0.42 -0.23 -0.09 -0.71** -0.28
(0.00) (0.19) (0.00) (0.82) (0.00) (0.45) (0.02)
Cooperation 0.12*** 0.04 0.15*** 0.05 0.20*** 0.08 0.13*** 0.05 0.19*** 0.07 0.13*** 0.05 0.17*** 0.07
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
External Search 0.04*** 0.02 0.05*** 0.02 0.05*** 0.02 0.05*** 0.02 0.04*** 0.02 0.05*** 0.02 0.04*** 0.02
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Formal Protection 0.37*** 0.13 0.38*** 0.13 0.34*** 0.13 0.37*** 0.13 0.33*** 0.13 0.37*** 0.13 0.33*** 0.13
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Exports 0.15*** 0.05 0.16*** 0.06 0.16*** 0.06 0.16*** 0.06
(0.00) (0.00) (0.00) (0.00)
Group 0.14*** 0.05 0.13*** 0.05 0.15*** 0.06 0.14*** 0.05 0.17*** 0.06 0.13*** 0.05 0.18*** 0.07
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Size 0.12*** 0.04 0.11*** 0.04 0.03** 0.01 0.12*** 0.04 0.03* 0.01 0.12*** 0.04 0.05*** 0.02
(0.00) (0.00) (0.04) (0.00) (0.07) (0.00) (0.00)
R&D Intensity 0.81*** 0.29 0.80*** 0.29 0.58*** 0.23 0.81*** 0.29 0.58*** 0.23 0.81*** 0.29 0.61*** 0.24
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
ˆε3-0.75*** -0.27 -0.73*** -0.26 -0.52*** -0.20 -0.75*** -0.27 -0.53*** -0.21 -0.74*** -0.27 -0.57*** -0.22
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Equation for financial constraintsb
Public Support 0.11*** 0.12*** 0.08*** 0.10*** 0.14*** 0.15***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Collateral -0.10*** -0.09*** -0.05*** -0.03 -0.13*** -0.11***
(0.00) (0.00) (0.00) (0.10) (0.00) (0.00)
Financial Risk 0.08*** 0.14*** 0.03** 0.08*** 0.16*** 0.19***
(0.00) (0.00) (0.01) (0.00) (0.00) (0.00)
Profitability -0.04*** -0.06*** -0.06*** -0.07*** -0.03** -0.06***
(0.00) (0.00) (0.00) (0.00) (0.03) (0.00)
Size -0.09*** -0.10*** -0.11*** -0.12*** -0.06*** -0.06***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Production 0.06*** 0.07** -0.01 -0.01 0.06** 0.07*
(0.00) (0.02) (0.55) (0.61) (0.04) (0.05)
ˆε30.08*** 0.08*** 0.07*** 0.06*** 0.09*** 0.07***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Over-identification Testc0.79 0.43 0.52 0.08 0.95 0.45
Over-identification Testd0.22 0.22 0.85 0.22 0.85 0.22 0.85
Number of Firms 38060 38060 16621 38060 16621 38060 16621
Columns report estimated coefficients and associated marginal effects (evaluated at mean values). Bootstrapped p-values in parentheses. ***,**,* Statistically significant at the 1%, 5%
and 10% confidence level, respectively. aSpecifications include industry and country dummy variables. bSpecifications include all variables in vector X.cReports the Sargan-Hansen test
statistic [p-value], where H0: over-identifying restrictions in the financial constraints equation are valid. dReports the Sargan-Hansen test statistic [p-value], where H0: over-identifying
restrictions in the R&D intensity equation are valid. The Sargan-Hansen test is implemented in linear LIML models. All continuous regressors are taken in their standard normalised
form with zero mean and standard deviation, so that we can directly interpret the coefficients and marginal effects across the specifications.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 20
Table 2: Bivariate probit model: production industries
Probit Bivariate probit
Internal or External FC Internal FC External FC
All Firms All Firms Non-Exporters All Firms Non-Exporters All Firms Non-Exporters
Coef. dy/dx Coef. dy /dx Coef. dy/dx Coef. dy/dx Coef. dy/dx Coef. dy/dx Coef. dy/dx
(1) (2) (3) (4) (5) (6) (7)
Equation for innovation successa
Financial Constraints 0.12*** 0.04 -0.59* -0.22 -1.22*** -0.45 -0.68*** -0.26 -1.31*** -0.47 -0.22 -0.08 -1.15*** -0.42
(0.00) (0.08) (0.00) (0.00) (0.00) (0.64) (0.00)
Cooperation 0.12*** 0.04 0.15*** 0.05 0.16*** 0.06 0.14*** 0.05 0.14*** 0.05 0.13*** 0.04 0.14*** 0.06
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
External Search 0.04*** 0.02 0.04*** 0.02 0.02 0.01 0.04*** 0.01 0.02 0.01 0.04*** 0.01 0.02 0.01
(0.00) (0.00) (0.10) (0.00) (0.21) (0.00) (0.12)
Formal Protection 0.44*** 0.14 0.44*** 0.15 0.43*** 0.16 0.44*** 0.15 0.42*** 0.16 0.44*** 0.15 0.42*** 0.16
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Exports 0.17*** 0.06 0.18*** 0.06 0.18*** 0.06 0.18*** 0.06
(0.00) (0.00) (0.00) (0.00)
Group 0.10*** 0.03 0.08*** 0.03 0.19*** 0.07 0.09*** 0.03 0.21*** 0.08 0.09*** 0.03 0.19*** 0.07
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Size 0.16*** 0.05 0.14*** 0.05 0.07*** 0.03 0.13*** 0.05 0.06* 0.02 0.15*** 0.05 0.09*** 0.04
(0.00) (0.00) (0.00) (0.00) (0.05) (0.00) (0.00)
R&D Intensity 1.08*** 0.37 1.05*** 0.37 0.82*** 0.32 1.06*** 0.37 0.84*** 0.33 1.08*** 0.38 0.86*** 0.34
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
ˆε3-1.02*** -0.35 -0.98*** -0.34 -0.77*** -0.30 -0.99*** -0.35 -0.79*** -0.31 -1.01*** -0.35 -0.82*** -0.32
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Equation for financial constraintsb
Public Support 0.09*** 0.12*** 0.04** 0.06* 0.15*** 0.17***
(0.00) (0.00) (0.01) (0.06) (0.00) (0.00)
Collateral -0.10*** -0.07*** -0.03* 0.01 -0.15*** -0.11***
(0.00) (0.00) (0.08) (0.57) (0.00) (0.00)
Financial Risk 0.07*** 0.14*** 0.01 0.08*** 0.14*** 0.16***
(0.00) (0.00) (0.34) (0.00) (0.00) (0.00)
Profitability -0.08*** -0.09*** -0.14*** -0.13*** -0.02 -0.07*
(0.00) (0.00) (0.00) (0.00) (0.38) (0.05)
Size -0.07*** -0.08*** -0.12*** -0.17*** -0.05** -0.03
(0.00) (0.00) (0.00) (0.00) (0.01) (0.38)
ˆε30.06*** 0.07*** 0.05*** 0.07*** 0.07*** 0.06***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Over-identification Testc0.38 0.30 0.38 0.29 0.63 0.57
Over-identification Testd0.86 0.86 0.45 0.86 0.45 0.86 0.45
Number of Firms 25141 25141 9019 25141 9019 25141 9019
See notes for Table 1

Financial Constraints, Innovation Performance and Sectoral Disaggregation 21
Table 3: Bivariate probit model: service industries
Probit Bivariate probit
Internal or External FC Internal FC External FC
All Firms All Firms Non-Exporters All Firms Non-Exporters All Firms Non-Exporters
Coef. dy/dx Coef. dy /dx Coef. dy/dx Coef. dy/dx Coef. dy/dx Coef. dy/dx Coef. dy/dx
(1) (2) (3) (4) (5) (6) (7)
Equation for innovation successa
Financial Constraints 0.10*** 0.04 0.34 0.12 -0.37 -0.15 0.57 0.19 -0.18 -0.07 0.44 0.15 0.32 0.12
(0.00) (0.53) (0.54) (0.39) (0.83) (0.24) (0.45)
Cooperation 0.17*** 0.06 0.15*** 0.06 0.21*** 0.08 0.14*** 0.05 0.20*** 0.08 0.15*** 0.06 0.17*** 0.06
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
External Search 0.05*** 0.02 0.05*** 0.02 0.06*** 0.02 0.05*** 0.02 0.06*** 0.02 0.05*** 0.02 0.05*** 0.02
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Formal Protection 0.26*** 0.09 0.25*** 0.09 0.34*** 0.13 0.24*** 0.09 0.25*** 0.10 0.25*** 0.09 0.24*** 0.09
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Exports 0.12*** 0.05 0.12*** 0.04 0.11*** 0.04 0.12*** 0.04
(0.00) (0.00) (0.00) (0.00)
Group 0.16*** 0.06 0.16*** 0.06 0.14*** 0.05 0.17*** 0.06 0.14*** 0.06 0.16*** 0.06 0.15*** 0.06
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Size 0.04*** 0.02 0.05** 0.02 0.01 0.01 0.05*** 0.02 0.02 0.01 0.05*** 0.02 0.03 0.01
(0.00) (0.01) (0.57) (0.00) (0.47) (0.00) (0.23)
R&D Intensity 0.50*** 0.19 0.50*** 0.19 0.45*** 0.17 0.50*** 0.19 0.45*** 0.18 0.50*** 0.19 0.45*** 0.18
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
ˆε3-0.43*** -0.16 -0.43*** -0.16 -0.38*** -0.15 -0.43*** -0.16 -0.39*** -0.15 -0.43*** -0.16 -0.39*** -0.15
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Equation for financial constraintsb
Public Support 0.14*** 0.12*** 0.12*** 0.10*** 0.17*** 0.14***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Collateral -0.03 -0.09*** 0.01 -0.01 -0.07** -0.15***
(0.20) (0.00) (0.77) (0.62) (0.01) (0.00)
Financial Risk 0.09*** 0.14*** 0.04** 0.07*** 0.20*** 0.25***
(0.00) (0.00) (0.02) (0.00) (0.00) (0.00)
Profitability -0.04*** -0.06*** -0.05*** -0.06*** -0.04*** -0.05**
(0.00) (0.00) (0.00) (0.00) (0.00) (0.01)
Size -0.10*** -0.10*** -0.08*** -0.08*** -0.09*** -0.08***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
ˆε30.11*** 0.09*** 0.10*** 0.06*** 0.13*** 0.08***
(0.00) (0.00) (0.00) (0.02) (0.00) (0.00)
Over-identification Testc0.73 0.99 0.54 0.63 0.99 0.94
Over-identification Testd0.26 0.26 0.79 0.26 0.79 0.26 0.79
Number of Firms 12919 12919 7602 12919 7602 12919 7602
See notes for Table 1

Financial Constraints, Innovation Performance and Sectoral Disaggregation 22
Table 4: Alternative definitions of the innovation success variable
Threshold for ‘Innovation Success’: the sample’s 50th percentile of “share of sales with new products” (Im)
FC Industries Firms Coefficient P > |z|dy/dx No of firms
(1) Internal or External All All Firms -0.66 0.21 -0.26 38060
All Non-Exporters -0.91*** 0.00 -0.32 16621
Internal All All Firms -0.51 0.30 -0.20 38060
All Non-Exporters -1.03*** 0.00 -0.35 16621
External All All Firms -0.07 0.85 -0.03 38060
All Non-Exporters -0.39 0.28 -0.15 16621
(2) Internal or External Production All Firms -0.75*** 0.00 -0.29 25141
Production Non-Exporters -0.97*** 0.00 -0.33 9019
Internal Production All Firms -0.81*** 0.00 -0.31 25141
Production Non-Exporters -1.12*** 0.00 -0.37 9019
External Production All Firms -0.05 0.91 -0.02 25141
Production Non-Exporters -0.75* 0.07 -0.27 9019
(3) Internal or External Services All Firms 0.83 0.18 0.31 12919
Services Non-Exporters 0.23 0.75 0.09 7602
Internal Services All Firms 0.91 0.14 0.33 12919
Services Non-Exporters -0.80 0.38 -0.28 7602
External Services All Firms 0.63 0.12 0.24 12919
Services Non-Exporters 0.56 0.16 0.22 7602
Threshold for ‘Innovation Success’: the sample’s 75th percentile of “share of sales with new products” (Iq)
FC Industries Firms Coefficient P > |z|dy/dx No of firms
(4) Internal or External All All Firms -1.24* 0.08 -0.31 38060
All Non-Exporters -1.10*** 0.00 -0.26 16621
Internal All All Firms -1.09 0.17 -0.27 38060
All Non-Exporters -0.91* 0.05 -0.22 16621
External All All Firms -0.74 0.27 -0.20 38060
All Non-Exporters -0.47 0.36 -0.13 16621
(5) Internal or External Production All Firms -1.20** 0.03 -0.30 25141
Production Non-Exporters -1.30*** 0.00 -0.29 9019
Internal Production All Firms -1.22*** 0.00 -0.29 25141
Production Non-Exporters -1.20* 0.05 -0.25 9019
External Production All Firms -0.50 0.47 -0.15 25141
Production Non-Exporters -0.80 0.22 -0.19 9019
(6) Internal or External Services All Firms 0.43 0.53 0.15 12919
Services Non-Exporters -0.30 0.59 -0.09 7602
Internal Services All Firms 0.34 0.64 0.12 12919
Services Non-Exporters -0.62 0.25 -0.16 7602
External Services All Firms 0.51 0.27 0.19 12919
Services Non-Exporters 0.36 0.47 0.12 7602
Treat ‘Innovation Success’ as a left-censored continuous variable
FC Industries Firms Coefficient P > |z|dy/dx No of firms
(7) Internal or External All All Firms -0.14 0.64 -0.05 38060
All Non-Exporters -0.29** 0.02 -0.09 16621
Internal All All Firms -0.14 0.68 -0.05 38060
All Non-Exporters -0.29* 0.06 -0.09 16621
External All All Firms -0.03 0.91 -0.01 38060
All Non-Exporters -0.19 0.55 -0.06 16621
(8) Internal or External Production All Firms -0.11 0.27 -0.04 25141
Production Non-Exporters -0.27** 0.03 -0.08 9019
Internal Production All Firms -0.13*** 0.00 -0.05 25141
Production Non-Exporters -0.28*** 0.00 -0.08 9019
External Production All Firms -0.02 0.94 -0.01 25141
Production Non-Exporters -0.19 0.43 -0.06 9019
(9) Internal or External Services All Firms 0.64* 0.05 0.32 12919
Services Non-Exporters -0.31 0.53 -0.10 7602
Internal Services All Firms 0.66* 0.08 0.34 12919
Services Non-Exporters -0.32 0.44 -0.10 7602
External Services All Firms 0.59** 0.01 0.31 12919
Services Non-Exporters 0.65 0.14 0.32 7602
Columns report estimated coefficients, bootstrapped p-values and associated marginal effects (evaluated at mean
values). The marginal effects in rows (7)-(9) are for the expected value of the dependent variable conditional on
being uncensored. ***,**,* Statistically significant at the 1%, 5% and 10% confidence level, respectively.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 23
Table 5: Robustness tests
Exclude Eastern European countries
FC Industries Firms Coefficient P > |z|dy/dx No of firms
(1) Internal or External All All Firms -0.69 0.13 -0.27 28129
All Non-Exporters -1.35*** 0.00 -0.44 11607
Internal All All Firms 0.20 0.71 0.08 28129
All Non-Exporters -1.41*** 0.00 -0.44 11607
External All All Firms -0.34 0.25 -0.13 28129
All Non-Exporters -0.99*** 0.00 -0.34 11607
(2) Internal or External Production All Firms -0.79*** 0.00 -0.31 18122
Production Non-Exporters -1.38*** 0.00 -0.43 5969
Internal Production All Firms -0.68* 0.05 -0.27 18122
Production Non-Exporters -1.46*** 0.00 -0.43 5969
External Production All Firms -0.49 0.23 -0.19 18122
Production Non-Exporters -1.26*** 0.00 -0.40 5969
(3) Internal or External Services All Firms 0.43 0.55 0.16 10007
Services Non-Exporters -1.11 0.12 -0.39 5638
Internal Services All Firms 0.71 0.36 0.25 10007
Services Non-Exporters -1.30 0.21 -0.43 5638
External Services All Firms 0.45 0.30 0.16 10007
Services Non-Exporters 0.23 0.68 0.09 5638
Add data from CIS2006 for eight countries
FC Industries Firms Coefficient P > |z|dy/dx No of firms
(4) Internal or External All All Firms -0.57** 0.01 -0.22 51010
All Non-Exporters -1.08*** 0.00 -0.41 22665
Internal All All Firms -0.28 0.47 -0.10 51010
All Non-Exporters -1.12*** 0.00 -0.42 22665
External All All Firms -0.27 0.24 -0.10 51010
All Non-Exporters -0.64* 0.05 -0.25 22665
(5) Internal or External Production All Firms -0.59* 0.07 -0.22 33878
Production Non-Exporters -1.15*** 0.00 -0.43 12464
Internal Production All Firms -0.60** 0.02 -0.22 33878
Production Non-Exporters -1.24*** 0.00 -0.46 12464
External Production All Firms -0.20 0.63 -0.07 33878
Production Non-Exporters -0.95** 0.02 -0.37 12464
(6) Internal or External Services All Firms 0.11 0.85 0.04 17132
Services Non-Exporters -0.31 0.60 -0.12 10201
Internal Services All Firms 0.49 0.52 0.16 17132
Services Non-Exporters 0.02 0.98 0.01 10201
External Services All Firms 0.27 0.47 0.09 17132
Services Non-Exporters 0.35 0.41 0.13 10201
Exclude ‘R&D Intensity’ from the list of controls
FC Industries Firms Coefficient P > |z|dy/dx No of firms
(7) Internal or External All All Firms -0.51 0.11 -0.19 38060
All Non-Exporters -0.99*** 0.00 -0.37 16621
Internal All All Firms -0.60* 0.05 -0.23 38060
All Non-Exporters -1.16*** 0.00 -0.43 16621
External All All Firms -0.17 0.52 -0.06 38060
All Non-Exporters -0.56 0.11 -0.22 16621
(8) Internal or External Production All Firms -0.54 0.16 -0.20 25141
Production Non-Exporters -1.09*** 0.00 -0.41 9019
Internal Production All Firms -0.74*** 0.00 -0.28 25141
Production Non-Exporters -1.23*** 0.00 -0.44 9019
External Production All Firms -0.15 0.73 -0.05 25141
Production Non-Exporters -0.98*** 0.00 -0.37 9019
(9) Internal or External Services All Firms 0.20 0.74 0.07 12919
Services Non-Exporters -0.46 0.35 -0.18 7602
Internal Services All Firms 0.33 0.65 0.12 12919
Services Non-Exporters -0.86 0.30 -0.33 7602
External Services All Firms 0.33 0.41 0.12 12919
Services Non-Exporters 0.24 0.56 0.09 7602
Columns report estimated coefficients, bootstrapped p-values and associated marginal effects (evaluated at mean
values). ***,**,* Statistically significant at the 1%, 5% and 10% confidence level, respectively.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 24
References
Aghion, P., Askenazy, P., Berman, N., Cette, G., & Eymard, L. 2008. Credit constraints and the cyclicality of R&D
investment: evidence from France. Banque de France Documents de Travail, No. 198.
Akerlof, G.A. 1970. The market for “lemons”: quality uncertainty and the market mechanism. The Quarterly Journal
of Economics,84, 488–500.
Albuquerque, R., & Hopenhayn, H.A. 2004. Optimal lending contracts and firm dynamics. Review of Economic Studies,
71, 285–315.
Alderson, M., & Betker, B. 1996. Liquidation costs and accounting data. Financial Management,25, 25–36.
Alti, A. 2003. How sensitive is investment to cash flow when financing is frictionless? Journal of Finance,58, 707–722.
Anderson, T.W. 2005. Origins of the limited information maximum likelihood and two-stage least squares estimators.
Journal of Econometrics,127, 1–16.
Beckeikh, N., Landry, R., & Amara, N. 2006. Lessons from innovation empirical studies in the manufacturing sector: A
systematic review of the literature from 1993-2003. Technovation,26, 644–664.
Bernini, M. 2012. Competitive pressure, export status and financial constraints in transition economies. University of
Trento, mimeo.
Blundell, R.W., & Powell, J.L. 2003. Endogeneity in semiparametric and nonparametric regression models. In: Dewa-
tripont, M., Hansen, L.P. & Turnovsky. S.J. (eds.), Advances in economics and econometrics: theory and applications.
Cambridge University Press, Cambridge, UK.
Bond, S., Harhoff, D., & Van Reenen, J. 2005. Investment, R&D and financial constraints in Britain and in Germany.
Annals of Economics and Statistics,79/80, 433–460.
Brandes, F. 2008. The future of manufacturing in Europe: a survey of the literature and a modelling approach. European
Foresight Monitoring Network Brief, No. 137.
Bridges, S., & Guariglia, A. 2008. Financial constraints, global engagment, and firm survival in the United Kingdsom:
evidence from micro data. Scottish Journal of Political Economy,55, 444–464.
Brown, J.R., Fazzani, S.M., & Petersen, B.C. 2009. Financing innovation and growth: cash flow, external equity, and
the 1990s R&D boom. Journal of Finance,64, 151–185.
Brown, J.R., Martinsson, G., & Petersen, B.C. 2012. Do financial constraints matter for R&D? European Economic
Review,56, 1512–1529.
Cainelli, G., Evangelista, R., & Savona, M. 2006. Innovation and economic performance in services: a firm-level analysis.
Cambridge Journal of Economics,30, 435–458.
Campello, M., Graham, J.R., & Harvey, C.R. 2010. The real effects of financial constraints: evidence from a financial
crisis. Journal of Financial Economics,97, 470–487.
Canepa, A., & Stoneman, P. 2003. Financial constraints on innovation: a European cross-country study. EIFC Tech-
nology and Finance Working Papers, No. 11.
Chesbrough, H.W. 2003. Open innovation: the new imperative for creating and profiting from technology. Harvard
Business School Publishing, Cambridge, MA.
Clausen, T. 2009. Do subsidies have positive impacts on R&D and innovation activities at the firm level? Structural
Change and Economic Dynamics,20, 239–253.
Clementi, G.L., & Hopenhayn, H.A. 2006. A theory of financing constraints and firm dynamics. The Quarterly Journal
of Economics,121, 229–265.
Coe, D., & Helpman, E. 1995. International R&D spillovers. European Economic Review,39, 859–887.
Coe, D.T., Helpman, E., & Hoffmaister, A.W. 2009. International R&D spillovers and institutions. European Economic
Review,53, 723–741.
Cooley, T.F., & Quadrini, V. 2001. Financial markets and firm dynamics. American Economic Review,91, 1286–1310.
Cr´epon, B., Duguet, E., & Mairesse, J. 1998. Research, innovation, and productivity: an econometric analysis at the
firm level. Economics of Innovation and New Technology,7, 115–158.
Czarnitzki, D., & Hottenrott, H. 2010. Financing constraints for industrial innovation: what do we know? Review of
Business and Economics,55, 346–362.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 25
Czarnitzki, D., & Hottenrott, H. 2011. R&D investment and financing constraints of small and medium-sized firms.
Small Business Economics,36, 65–83.
Dahlander, L., & Gann, D.M. 2010. How open is innovation? Research Policy,39, 699–709.
Dahlstrand, A.L., & Cetindamar, D. 2000. The dynamics of innovation financing in Sweden. Venture Capital,2, 203–221.
Damijan, J.P., Kostevc, C., & Polanec, S. 2008. From innovation to exporting or vice versa? The World Economy,33,
374–398.
David, P.A., Hall, B.H., & Tool, A.A. 2000. Is public R&D a complement or substitute for private R&D? A review of
the econometric evidence. Research Policy,29, 497–529.
Efthyvoulou, G. 2012. The impact of financial stress on sectoral productivity. Economics Letters,116, 240–243.
Evans, W.N., & Schwab, R.M. 1995. Finishing high school and starting college: do Catholic schools make a difference.
The Quarterly Journal of Economics,110, 941–974.
Fazzari, S.M., Hubbard, R.G., & Petersen, B.C. 1988. Financing constraints and corporate investment. Brookings Papers
on Economic Activity,19, 141–206.
Feenstra, R.C., Li, Z., & Yu, M. 2011. Exports and credit constraints under incomplete information: theory and evidence
from China. NBER Working Papers, No. 16940.
Gadrey, J., & Gallouj, F. 1998. The provider-customer interface in business and professional services. The Service
Industries Journal,18, 1–15.
Gallouj, F., & Savona, M. 2009. Innovation in services: a review of the debate and a research agenda. Journal of
Evolutionary Economics,19, 149–172.
Gallouj, F., & Weinstein, O. 1997. Innovation in services. Research Policy,26, 537–556.
Gorodnichenko, Y., & Schnitzer, M. 2013. Financial constraints and innovation: why poor countries don’t catch up.
Journal of European Economic Association,11, 1115–1152.
Greenaway, D., Guariglia, A., & Kneller, R. 2007. Financial factors and exporting decisions. Journal of International
Economics,73, 377–395.
Griffith, R., Huergo, E., Mairesse, J., & Peters, B. 2006. Innovation and productivity across four European countries.
Oxford Review of Economic Policy,22, 483–498.
Hall, B. 2002. The financing of research and development. Oxford Review of Economic Policy,18, 25–51.
Hall, B.H., & Lerner, J. 2010. The financing of R&D and innovation. In: Hall, B.H. & Rosenberg, N. (eds.), Handbook
of the economics of innovation. Elsevier, North Holland.
Harhoff, D. 1998. Are there financing constraints for R&D and investment in German manufacturing firms. Annals of
Economics and Statistics,49/50, 421–456.
Helpman, E., Melitz, M., & Yeaple, S. 2004. Export versus FDI with heterogeneous firms. American Economic Review,
94, 300–316.
Hennessy, C.A., & Whited, T.M. 2007. How costly is external financing? Evidence from a structural estimation. Journal
of Finance,62, 1705–1745.
Himmelberg, C., & Petersen, B. 1994. R&D and internal finance: a panel study of small firms in high-tech industries.
The Review of Economics and Statistics,76, 38–51.
Hoshi, T., Kashyap, A., & Scharfstein, D. 1991. Corporate structure, liquidity, and investment: evidence from Japanese
industrial groups. The Quarterly Journal of Economics,106, 33–60.
Hottenrott, H., & Peters, B. 2012. Innovative capability and financing constraints for innovation: more money, more
innovation? The Review of Economics and Statistics,94, 1126–1142.
Jensen, M.C., & Meckling, W. 1976. Theory of the firm: managerial behavior, agency costs, and ownership structure.
Journal of Financial Economics,3, 305–360.
Kaplan, S.N., & Zingales, L. 1997. Do investment-cash flow sensitivities provide useful measures of financing constraints.
The Quarterly Journal of Economics,112, 169–215.
Karaca-Mandic, P., & Train, K. 2003. Standard error correction in two-step estimation with nested samples. Econometrics
Journal,6, 401–407.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 26
Leiponen, A. 2008. Is service innovation different? The Research Institute of the Finnish Economy Discussion Papers,
No. 1151.
Leland, H., & Pyle, D. 1977. Information asymmetries, financial structure, and financial intermediation. Jounral of
Finance,32, 371–387.
L¨o¨of, H., & Heshmati, A. 2006. On the relationship between innovation and performance: a sensitivity analysis. Eco-
nomics of Innovation and New Technology,15, 317–344.
Love, J.H., Roper, S., & Vahter, P. 2013. Learning from openness: the dynamics of breadth in external innovation
linkages. Strategic Management Journal, DOI:10.1002/smj.2170.
Mairesse, J., & Mohnen, P. 2010. Using innovations surveys for econometric analysis. National Bureau of Economic
Research Working Papers, No. 15857.
M¨annasoo, K., & Merik¨ull, J. 2014. R&D, credit constraints and demand fluctuations: comparative micro evidence from
10 new EU Members. Eastern European Economics, forthcoming.
Melitz, M.J. 2003. The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica,
71, 1695–1725.
Meyer, J. R., & Kuh, E. 1957. The investment decision: an empirical study. Harvard University Press, Cambridge.
Modigliani, F., & Miller, M. 1958. The cost of capital, corporation finance and the theory of investment. American
Economic Review,48, 261–297.
Mohnen, P., & R¨oller, L-H. 2005. Complementarities in innovation policy. European Economic Review,49, 1431–1450.
Mothe, C., & Nguyen Thi, T.U. 2010. The impact of non-technological innovation on technological innovation: do
services differ from manufacturing? An empirical analysis of Luxembourg firms. CEPS Instead Working Papers, No.
2010-01.
Moyen, N. 2004. Investment-cash flow sensitivities: constrained versus unconstrained firms. Journal of Finance,69,
2061–2092.
Mulkay, B., Hall, B.H., & Mairesse, J. 2000. Investment and R&D in France and in the United States. In: Deutsche
Bundesbank (ed.), Investing Today for the World of Tomorrow. Springer, Verlag.
Murphy, K.M., & Topel, R.H. 1985. Estimation and inference in two-step econometric models. Journal of Business &
Economic Statistics,3, 370–379.
Myers, S., & Majluf, N. 1984. Corporate financing and investment decisions when firms have information that investors
do not. Journal of Financial Economics,13, 187–221.
Newey, W.K. 1984. A method of moment interpretation of sequential estimators. Economics Letters,14, 201–206.
OECD. 2009. Innovation in firms: a microeconomic perspective. OECD, Paris.
Ouyang, M. 2011. On the cyclicality of R&D. The Review of Economics and Statistics,93, 542–553.
Piga, C., & Atzeni, G. 2007. R&D investment, credit rationing and sample selection. Bul letin of Economic Research,
59, 149–178.
Polder, M., van Leeuwen, G., Mohnen, P, & Raymond, W. 2009. Productivity effects of innovation modes. Statistics
Netherlands Discussion Papers, No. 09033.
Roodman, D. 2009. Estimating ful ly observed recursive mixed-process models with cmp. Center for Global Development
Working Papers, No. 168.
Roper, S., Du, J., & Love, J.H. 2008. Modelling the innovation value chain. Research Policy,37, 961–977.
Salomon, R.M., & Shaver, J.M. 2005. Learning by exporting: new insights from examining firm innovation. Journal of
Economics and Management Strategy,14, 431–460.
Savignac, F. 2008. Impact of financial constraints on innovation: what can be learned from a direct measure? Economics
of Innovation and New Technology,17, 553–569.
Scellato, G. 2007. Patents, firm size and financial constraints: an empirical analysis for a panel of Italian manufacturing
firms. Cambridge Journal of Economics,31, 55–76.
Silva, F., & Carreira, C. 2010. Financial constraints: are there differences between manufacturing and services? Estudos
Do GEMF, No. 16.

Financial Constraints, Innovation Performance and Sectoral Disaggregation 27
Silva, F., & Carreira, C. 2012. Do financial constraints threat the innovation process? Evidence from Portuguese firms.
Economics of Innovation and New Technology,21, 701–736.
Stiglitz, J., & Weiss, A. 1981. Credit rationing in markets with imperfect information. American Economic Review,71,
393–410.
Tether, B.S. 2003. The sources and aims of innovation in services: variety between and within sectors. Economics of
Innovation and New Technology,12, 481–505.
Tether, B.S. 2005. Do services innovate (differently)? Insights from the European Innobarometer Survey. Industry and
Innovation,12, 153–184.
Ughetto, E. 2008. Does internal finance matter for R&D? New evidence from a panel of Italian firms. Cambridge Journal
of Economics,32, 907–925.
Vahter, P. 2011. Learning by exporting: evidence based on data of knowledge flows from innovation surveys in Estonia.
William Davidson Institute Working Papers, No. 1011.
van Ark, B., O’Mahony, M., & Timmer, M.P. 2008. The productivity gap between Europe and the United States: trends
and causes. Journal of Economic Perspectives,22, 25–44.
Williamson, O. 1988. Corporate finance and corporate governance. Journal of Finance,43, 567–591.
Wooldridge, J.M. 2002. Econometric analysis of cross section and panel data. MIT Press, Cambridge, MA.
Y¨or¨uk, B.K. 2009. How responsive are charitable donors to requests to give? Journal of Public Economics,93,
1111–1117.