See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/228293372

The Effect of Public Subsidies on Firms

Investment-Cash Flow Sensitivity: Transient or
Persistent?
Article in Research Policy April 2010
DOI: 10.2139/ssrn.1596073

CITATIONS

READS

164

3 authors:
Massimo G. Colombo

Annalisa Croce

Politecnico di Milano

Politecnico di Milano

171 PUBLICATIONS 4,137 CITATIONS

40 PUBLICATIONS 254 CITATIONS

SEE PROFILE

SEE PROFILE

Massimiliano Guerini
Politecnico di Milano
19 PUBLICATIONS 88 CITATIONS
SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

Available from: Massimiliano Guerini

Retrieved on: 04 October 2016

The effect of public subsidies on firms investment-cash flow

sensitivity: transient or persistent?
Massimo G. Colombo*, Annalisa Croce** and Massimiliano Guerini***
Department of Management, Economics and Industrial Engineering,
Politecnico di Milano

Abstract
This work analyses the effect of public subsidies on firms investments in a longitudinal sample of 294 Italian
unlisted owner-managed new-technology-based firms (NTBFs), observed over a 10-year period from 1994 to
2003. We use a modified version of the Euler equation and GMM-system estimation techniques that duly take
into account the endogeneity of public finance. We find that small NTBFs rely significantly on internal capital
to finance their investments, while larger NTBFs do not. Receipt of public subsidies by small NTBFs results in
an increased investment rate, and reduced investment-cash flow sensitivity, in the immediately following year.
We interpret these results as an indication of the relaxation of financial constraints. Moreover, while the increase
in investment rate does not persist in the long run, the dependence of investments on cash flow remains
negligible after receipt of the first public subsidy. Nonetheless, small NTBFs are less likely to obtain public
support than their larger peers. This latter evidence raises some doubts about the global efficacy of
governmental interventions.

JEL codes: G31, H25, D92

Key words: Public finance; high-tech start-ups; financial constraints.
* Politecnico di Milano, Department of Management, Economics and Industrial Engineering, Via Lambruschini, 4/c, 20157, Milan, Italy.
ph: +39-02-2399-3248; fax: +39-02-2399-2710. E-mail address: massimo.colombo@polimi.it.
**Politecnico di Milano, Department of Management, Economics and Industrial Engineering, Via Lambruschini, 4/c, 20157, Milan, Italy.
ph: +39-02-2399-2743; fax: +39-02-2399-2710. E-mail address: annalisa.croce@polimi.it. Corresponding author.
***Politecnico di Milano, Department of Management, Economics and Industrial Engineering, Via Lambruschini, 4/c, 20157, Milan, Italy.
ph: +39-02-2399-2810; fax: +39-02-2399-2710. E-mail address: massimiliano.guerini@polimi.it.

We are grateful to Dirk Czarnitzki, Finn Hansson and Silvio Vismara for comments and suggestions on this work. Responsibility for any
errors lies solely with the authors.

Electronic copy available at: http://ssrn.com/abstract=1596073

1.

Introduction
Scholars agree that young high-tech firms (NTBFs, new technology-based firms) play a

crucial role in modern economies (Audretsch, 1995). It is also argued that capital market
imperfections arising from information asymmetries (i.e., hidden information and hidden action)
make it difficult for these firms, especially smaller ones, to obtain external financing (Carpenter and
Petersen, 2002a; Hall, 2002). In turn, financial constraints negatively influence firms investments and
performance, with obvious negative implications for social welfare. The available empirical evidence
that will be surveyed in Section 2 largely supports this view.
Accordingly, NTBFs have attracted considerable attention from policy makers at local,
national and supranational levels (see e.g., BEPA, 2008), with the presumption that public subsidies
can help these firms to overcome the above-mentioned financial constraints. However, whether public
subsidies are beneficial to NTBFs is questionable (Holtz-Eakin 2000). Public support may simply
result in the replacement of market failures with governmental failures. For instance, politicians may
use public programs to reward constituents, rather than to correct market failures (Cohen and Noll,
1991; see also Becker, 1983). Public interventions may also prevent the emergence of active venture
capital markets by crowding-out private funds (Leleux and Surlemont, 2003; Cumming and
MacIntosh, 2007). This problem is more likely to occur if government officers, who are responsible
for the allocation of public funds, suffer from a cherry picking syndrome: to avoid accusations of
misuse of taxpayers money, they subsidise only first-class projects that beneficiary firms would have
pursued anyway, even in absence of any subsidy (see Wallsten, 2000 and Lach, 2002 for evidence of
these crowding-out effects among U.S. small firms and large Israeli firms, respectively).
In this paper, we empirically investigate whether public support schemes in fact relax the
financial constraints of NTBFs, a research question that has received limited attention in the extant
literature (for exceptions see Hyytinen and Toivanen, 2005; Czarnitzki, 2006; Czarnitzki et al., 2009).
More precisely, following the approach originally proposed by Fazzari et al. (1988), we aim to detect
the treatment effect of public subsidies on the investment rate and investment-cash flow sensitivity
of NTBFs. We expect the removal of financial constraints after receipt of a public subsidy to result in

Electronic copy available at: http://ssrn.com/abstract=1596073

an increase of investment rate and a reduction of investment-cash flow sensitivity. Moreover, we

assess whether these allegedly beneficial effects are transient or persist over time. In examining this
latter issue, we will provide fresh insights into the mechanisms through which public support schemes
manage to relax financial constraints.
For this purpose, we analyse the investments in fixed and intangible assets within a
longitudinal hand-collected dataset consisting of 294 Italian owner-managed privately held NTBFs,
observed between 1994 and 2003. The sample is extracted from the RITA 2004 (Research on
Entrepreneurship in Advanced Technologies) database, developed at Politecnico di Milano. This
database is the most comprehensive source of data presently available on Italian NTBFs. In particular,
it contains information on all public subsidies received by sample firms from national governmental
institutions during the observation period. To detect financial constraints, we estimate an Euler
equation (Bond and Meghir, 1994) that allows us to control for unobserved investment opportunities
and adjustment costs. To take into account the potentially endogenous nature of public finance, we
resort to a system GMM (GMM-SYS) estimator for dynamic panel data models. Moreover, we
enlarge the usual set of internal instruments through the addition of variables that are a source of
exogenous variation for receipt of public subsidies, both across firms and over time, to allow better
control for selection based on unobservables. As far as we know, this is the first study that utilises this
methodology to assess whether public subsidies alleviate the financial constraints of NTBFs.
Our results highlight that small Italian NTBFs largely rely on internal cash flows to finance
their investments. Conversely, larger NTBFs do not seem to be financially constrained. Receipt of
public subsidies by financially constrained small NTBFs results in an increase of the investment rate
and in a reduction of the investment-cash flow sensitivity in the immediately following year. These
effects are both statistically significant and of large economic magnitude. Moreover, we find that the
dependence of investments on cash flows remains negligible from receipt of the first public subsidy
onwards, while the increase in investment rate does not significantly persist in the long term.
Nonetheless, despite these beneficial effects, small NTBFs are far less likely to obtain public support
than their larger counterparts. This latter evidence raises some doubts regarding the overall efficacy of
Italian governmental interventions in this domain.
3

The paper is structured as follows. In the next section, we briefly survey the literature on
firms financial constraints, and we consider the expected effect of public finance on the investment
rate and investment-cash flow sensitivity of NTBFs. In Section 3, we present the sample of firms; we
also briefly illustrate industrial policy measures in Italy that support NTBFs. In Section 4, we describe
the econometric methodology. In Section 5, we present the results of the econometric analysis, and
Section 6 concludes.

2.

Literature review and theoretical background

2.1.

The empirical literature on firms financial constraints

If capital markets were perfect, every profitable investment would be financed in equilibrium and the
source of financing would be irrelevant (Modigliani and Miller, 1958). As a corollary, the availability
of internal liquidity (i.e., current cash flows) would not affect the investment pattern of firms
(Jorgenson, 1963; Hall and Jorgenson, 1967). Conversely, if investors are less informed than
entrepreneurs, firms adhere to a pecking order when financing their investments (Myers and Majluf,
1984). First, they rely on internal sources of funds; then, when internal capital is exhausted, they turn
to the external capital source with the lowest cost, which usually is debt (at least for firms with low
leverage). Fazzari et al. (1988) argue that while the marginal opportunity cost of internal capital is
constant, the debt supply curve is upward-sloping, and greater capital market imperfections result in a
steeper slope. Under these circumstances, it would be expected the investments of firms that are more
financially constrained (i.e., they face a steeper debt supply curve) would be more sensitive to cash
flows. The authors also show that investment-cash flow sensitivity is higher for firms with low
dividend payouts, which allegedly have more binding financial constraints. Several studies replicated
the above analysis by grouping firms according to different proxies of information costs (see
Hubbard, 1998, for a comprehensive survey). Moreover, they considered different types of
investments, including R&D expenses (Hao and Jaffe, 1993; Himmelberg and Petersen, 1994;
Bougheas et al., 2003; see also the studies surveyed by Hall, 2002).1
1

Some other studies assessing the effects of financial constraints on R&D and innovation have relied on direct survey-based measures of the
existence of financial constraints. See e.g., Savignac, (2008) and Tiwari et al., (2009). These studies show that financial constraints indeed
hamper innovation.

Nevertheless, the approach proposed by Fazzari et al. (1988) presents some weaknesses. First,
Kaplan and Zingales (1997, 2000) theoretically demonstrated that the profit-maximising investment
choices of firms do not imply a monotonic relationship between financial constraints and the
sensitivity of investments to cash flows. Accordingly, for the 49 low-dividend firms included in the
Fazzari et al. (1988) study, they used detailed information from the annual reports and financial
statements to rank the extent to which these firms were financially constrained, finding that the
investments of the least financially constrained firms are the most sensitive to the availability of cash
flow. In the same vein, results that contradict Fazzari et al.s (1988) hypothesis were found by
Kadapakkam et al. (1998) and Cleary (1999; 2006), among others. Moreover, Almeida and Campello
(2007) noted that because firms can pledge their assets as collateral, both investments and borrowing
are endogenous: for financially constrained firms, the investment-cash flow sensitivity increases
concomitantly with the degree of asset tangibility, as these firms use cash flows to purchase
collaterisable assets, while it is unaffected by asset tangibility if firms are unconstrained.
Second, positive investment-cash flow sensitivity may simply derive from lack of proper
control for unobserved investment opportunities (see Hubbard, 1998). In empirical works these
opportunities are generally captured using average Tobins Q as a proxy for marginal q. However, the
relationship between the investment rate and Tobins Q may be non-linear if adjustment costs are not
well described by a quadratic function. Moreover, Tobins Q ceases to be a sufficient statistic for the
influence of investment opportunities when the assumptions of perfect competition and constant
returns to scale are not met. Finally, Tobins Q can suffer from measurement errors (Erickson and
Whited, 2000). Gomes (2001), Alti (2003) and Moyen (2004) showed that measurement errors and
identification problems can lead to significant investment-cash flow sensitivity even in the absence of
financing frictions. An alternative approach, which partially solves the problems mentioned above, is
to directly estimate the Euler equation (Abel, 1980). In this way, the assumption of linear
homogeneity of the net revenue function can be relaxed and the use of share price data can be avoided
(Bond and Van Reenen, 2007).
Third, Jensen (1986) pointed out that opportunistic behaviour by managers who misuse a firms
free cash flows could cause overinvestment and lead to a positive relationship between investment
5

rate and level of cash flows in the absence of any financial constraint. Even though overinvestment
and underinvestment problems stem from different theoretical considerations, they generate similar
empirical effects and are thus difficult to disentangle. Vogt (1994) reported evidence that both effects
are at work and that overinvestment and underinvestment dominate for larger and smaller firms,
respectively. Pawlina and Renneboog (2005) also found that the agency costs of ownership influence
the investment-cash flow sensitivity.

2.2.

Are NTBFs financially constrained?

In this work, we focus attention on NTBFs. The entrepreneurial finance literature argues that these
firms are most likely to be financially constrained. First, because of the technology-intensive nature of
their activity and the lack of a track record, they face hidden information and hidden action problems
(Carpenter and Petersen, 2002a; Hall, 2002). On the one hand, it is difficult for external investors to
gauge ex-ante the quality of the investment projects of these firms. Moreover, as imitation by
competitors of the innovative ideas of these firms can be very detrimental to their destiny, they are
reluctant to disclose relevant information to potential investors (Anton and Yao, 2002). On the other
hand, it is also difficult for outsiders to monitor ex-post the decisions made by entrepreneurs and to
discourage opportunism. Second, most of the assets of these firms are firm-specific or intangible.
Hence, they cannot be pledged as collateral (Berger and Udell, 1990). The arguments above especially
apply to smaller NTBFs. As these firms are typically less well-known than their larger peers, are often
unable to provide audited financial statements and have lower collateral relative to their liabilities,
they are more vulnerable to information asymmetries (Gertler and Gilchrist, 1994; Avery et al., 1998).
The few available empirical studies support the contention that NTBFs, especially smaller
ones, are financially constrained. Hao and Jaffe (1993) contrast the R&D investments of U.S. small
and large firms; they show that the former firms are liquidity-constrained, while the latter are not (see
Harhoff, 1998 for similar results on German firms). Himmelberg and Petersen (1994) analyse a panel
data set composed of 179 U.S. small non-VC-backed high-tech firms and find an economically large
and statistically significant relationship between R&D investments and cash flows. Using a panel of
more than 1,600 U.S. small manufacturing firms, Carpenter and Petersen (2002b) document that the

growth of firms total assets closely depends on the amount of the available cash flow; a cash flow
increase of $1 results in a slightly larger increase of firms total assets. Bertoni et al. (2010) examine
investment-cash flow sensitivity in a sample of Italian NTBFs, and find that the investment activity of
these firms reacts to internal cash flow shocks, but this sensitivity vanishes once firms obtain venture
capital (VC) finance from independent investors. Colombo and Grilli (2007) examine debt financing
in a similar sample. Their results conform to the pecking order hypothesis: NTBFs resort to bank
debt only when the estimated amount of finance needed by firms operations exceeds the amount
available internally.
It is important to emphasise that for NTBFs, the sensitivity of investments to cash flows can
be regarded as a more reliable indicator of binding financial constraints when compared to other
firms. On the one hand, the investment opportunities faced by NTBFs likely depend on the quality of
their business ideas, the innovative content of the technologies they are developing, and the
entrepreneurial talent of their owner-managers. These investment opportunities are unlikely to be
positively correlated with current cash flows. Therefore, it is implausible that high investment-cash
flow sensitivity results from the superior investment opportunities of the (few) NTFBs that have large
positive cash flows. On the other hand, most NTBFs are privately held. As ownership and control
generally are not separated, agency problems tend to be negligible. Therefore, for these firms free
cash flow abuses on the part of owner-managers are unlikely, and Jensens (1986) free cash flow
argument is not pertinent. Based on these considerations, in what follows we will interpret positive
investment-cash flow sensitivity as a sign of binding financial constraints that negatively affect the
investment activity of NTBFs.

2.3.

The role of public support in alleviating NTBFs financial constraints

Several previous studies analysed the effects of public support programs on firms R&D expenditures
and various indicators of firm performance, i.e., growth and productivity (see e.g., Lerner, 1999;
Wallsten, 2000; Lach, 2002; Gonzalez et al. 2005. See David et al. 2000, Klette et al. 2000 for surveys
of early studies, and Hussinger, 2008 for a review of more recent works). Conversely, studies that

have specifically examined whether public subsidies relax firms financial constraints are quite rare.
Hyytinen and Toivanen (2005) show that small Finnish firms operating in industries that are largely
dependent on external finance invest more in R&D and are more growth-oriented if more
governmental funds are available locally. These results support the view that i) capital market
imperfections hold back innovation investments and growth in this type of firm and ii) public support
helps alleviate the negative effects of these capital market imperfections. Czarnitzki (2006) analyses
the effects of firms internal financial resources, credit rating (interpreted as an inverse proxy of the
difficulty of resorting to external finance), and public funds on R&D expenses in a large sample
composed of small and medium firms located in West and East Germany. In the West Germany
sample, internal financial resources and public subsidies are found to positively influence R&D, while
bad credit rating has an opposite effect of large economic magnitude. Conversely, in East Germany
where public funds are more widely available than in West Germany, the sensitivity of firms R&D
expenses to internal financial resources is considerably reduced, credit rating plays no role, and public
funds are the driving force of R&D activity. Notably, receipt of public support leads to a 60% increase
in the probability of a firm being involved in R&D; the corresponding figure for West Germany is
24%. These findings again suggest that financial constraints indeed limit firms R&D activity, and
that public funds alleviate these constraints. Finally, in a recent paper, Czarnitzki et al. (2009) explore
the effect of public subsidies on the sensitivity of firms R&D expenses to internal financial resources,
distinguish between basic and strategic research. After showing that firms rely more on internal
funds for research investments than for experimental development activities, they find that subsidies
foster research activities and reduce sensitivity of investments to internal funds. Conversely, they did
not find any significant impact of public financing on investments in development activities.
Let us assume that small NTBFs are financially constrained. If public subsidies relax these
financial constraints, a greater amount of investments should be observed after receipt of the
subsidies, accompanied by a reduction of the sensitivity of investments to cash flows. There are three
non-competing explanations for these effects. First, the marginal cost of public funds is lower than for
other external sources of finance. Hence, for financially constrained firms, receipt of a public subsidy
renders profitable investment projects that would not be profitable in its absence. In addition to this
8

direct effect, public funds may have two additional indirect positive effects on firms investments.
First, a financially constrained firm receiving a public subsidy may use it (at least partially) to
purchase fixed assets that can then be used as collateral. This alleviates information asymmetry
problems, making it easier for the firm to obtain bank debt. Second, if public subsidies are provided
through a selective2 support scheme administered by a reputable governmental body and there is
fierce competition among applicants, they will have a certification effect, signalling to uninformed
external parties the good quality of the recipient firm. In turn, this signal will alleviate the hidden
information problems that make it difficult for these firms to obtain external finance.
In accordance with this latter argument, Lerner (1999) shows that U.S. small high-technology
firms receiving awards from the Small Business Innovation Research (SBIR) program exhibit greater
growth over a decade than those included in a matched control sample,3 but only if they are located in
a geographic area with high venture capital activity. This positive effect seems to be confined to the
first award received; the effects of subsequent awards were minimal. In addition, SBIR awardees are
found to be more likely to subsequently obtain venture capital, while venture capital investments did
not predict SBIR awards. Colombo et al. (2009a, 2009b) document that public subsidies obtained
through selective support schemes have beneficial effects on employment growth and total factor
productivity of Italian NTBFs, but only if they are obtained in the early years of the NTBFs lives,
when adverse selection problems are thought to be more severe. If selective schemes channel public
funds to older NTBFs or if support is provided through automatic schemes, the effects are negligible.
From the above reasoning, hypotheses H1 and H2 follow.
H1: Receipt of a public subsidy results in an increase in the amount of investments of small
NTBFs.
H2: Receipt of a public subsidy results in a decrease of the investment-cash flow sensitivity of
small NTBFs.

2
An automatic scheme gives financial assistance to all applicants fulfilling all the requirements specified in the law. In contrast, a
selective scheme provides financial support to selected applicants; applicants compete for financial subsidies and their projects are judged
by committees formed by experts who are appointed by the national authority.
3
Nonetheless, the finding that SBIR grants foster growth was not replicated by Wallsten (2000). After controlling for endogeneity of public
support in a multi-equations framework, it was found that SBIR grants did not positively influence firms employment growth, while
crowding out firms private R&D expenses.

If public finance relaxes the financial constraints from which small NTBFs suffer, the question
arises as to whether this effect is transient or persistent. Persistence over time of the allegedly positive
effects of public finance on the investments of small NTBFs would be an additional argument in
favour of public support to this type of firm. Although it is interesting to investigate the dynamics of
the effects of public funds on firms investment activity, this issue has been neglected by the extant
literature.
There are two reasons why public subsidies may have an enduring positive effect on the
investment activity of small NTBFs. First, let us assume that public funds have a short-term positive
effect on investments. If subsidised firms leverage this positive effect to become larger, they may be
able to reach a threshold size above which financial constraints are no longer binding. However,
whether public subsidies lead to increased growth of recipient firms is questionable (see e.g., Irwin
and Klenow, 1996; Lerner, 1999; Wallsten, 2000; Colombo et al., 2009a). More interestingly, we
mentioned above that public subsidies may allow beneficiary firms to acquire collaterisable assets. In
addition, award of a selective subsidy may signal a firms good quality to uninformed third parties. To
the extent that these indirect effects lead to the removal of the sources of financial constraints (i.e.,
hidden information and hidden action problems), we expect the associated benefits to persist over
time. We then derive hypotheses H3 and H4.
H3: Receipt of a public subsidy leads to a persistent increase in the investments of small
NTBFs.
H4: Receipt of a public subsidy leads to a persistent decrease of the investment-cash flow
sensitivity of small NTBFs.

3.

The data

3.1.

The sample

This work examines a sample of 294 Italian NTBFs observed from 1994 (or since their founding) up
to 2003. Sample firms were established in 1980 or later, were owner-managed at founding time and
remained so up to 1/1/2004, and operate in the following high-tech sectors in manufacturing and
services: computers, electronic components, telecommunications equipment, optical, medical and
10

electronic instruments, biotechnology, pharmaceuticals, advanced materials, aerospace, robotics,

process automation equipment, software, and internet and telecommunications services.

The

unbalanced panel dataset used in the empirical analysis includes 1,499 firm-year observations; there
are on average five observations per firm. We are interested in assessing the treatment effect of
public finance on firms investments. Therefore, we excluded from the dataset the observations
relating to venture capital-backed firms and firms that went through an IPO (i.e., all observations
relate to privately held, non-venture capital-backed firms).
The sample was extracted from the RITA 2004 directory, developed at Politecnico di Milano.
This directory has been used in several prior studies. In the absence of reliable official statistics, it is
the most complete source of data on NTBFs presently available in Italy and includes 1,974 firms (see
the Appendix for a detailed description). It is important to emphasise that because of the procedure
that was used to create this directory, lifestyle firms and firms that are created purely for tax-saving
objectives are unlikely to be included. These firms are very unlikely to apply for public funds; their
exclusion is an important strength of our dataset as it renders the estimates of the counterfactual (i.e.,
the investment choices that subsidised NTBFs would have made in absence of public support) more
precise. The sample analysed in the present study includes all RITA directory firms i) that participated
in a survey administered in the first semester of 2004 and ii) for which accounting data are available
for the entire observation period (sources: AIDA and CERVED database).
Table 1 reports the distribution of sample firms and RITA directory firms across industries,
geographic areas and foundation dates. Two 2 tests show that there are no statistically significant
differences between the distributions of the 294 sample firms across industries and geographical areas
and the corresponding distributions of the 1,974 RITA directory firms (2(4)= 3.81 and 2(3)= 5.26,
respectively). Conversely, sample firms are somewhat older than the population from which the
sample was drawn (2(3)= 39.01), with the foundation dates being more (less) concentrated in 19921997 (1998-2003). This probably because i) in Italy, all limited liability companies are obliged by law
to publish yearly accounting data, while for other firms, publication is not mandatory and ii) limited
liability companies are relatively less common among very young NTBFs. Note, however, that to

11

assess the effect of public finance on investments, firms need to be observed over time. Therefore, the
observation of very young firms provides limited insights into the issues analysed in this paper.
[Table 1 about here]
The sample is quite large and exhibits considerable heterogeneity in terms of the characteristics
of firms. Moreover, our dataset provides a rich set of industry-, location-, firm- and founder-specific
information on sample firms that can be used to build appropriate instruments for receipt of public
finance. However, only firms that survived up to the survey date could be included in the sample.
Survivorship bias may clearly distort the estimates of the effect of public finance on investment
activity. For instance, in the absence of public support, low-quality firms are more likely to both
exhibit low investment levels and to cease operations than their better-quality counterparts. If the
former firms receive a public subsidy and then manage to stay in business, a downward bias in the
effect of public support on investments follows. The same holds true if receipt of public subsidy by
high-quality firms with high investment levels renders them more attractive as acquisition targets. In
fact, we are not able to properly control for survivorship bias, as it is common in studies that use
survey-based data (for exceptions see Delmar and Shane 2006, Eckhardt et al. 2006). In Section 4.3
we will describe a partial control that was implemented in our estimates.

3.2.

Public subsidies to NTBFs in Italy

During the observation period, there was no large-scale national public support scheme in Italy
expressly targeted to NTBFs (like the SBIR program in the U.S. or the Jeunes Entreprises Innovantes
scheme in France). The only partial exception is the Law 297/1999 that aimed at favouring the
creation of academic start-ups. Indeed, the industrial policy of the Italian government relied on
horizontal measures directed to all firms, or was aimed at supporting specific sectors (e.g., the Law
808/1985 in support of the aerospace sector, dominated by large enterprises like Finmeccanica), or
focused attention on young or small enterprises independently of their sector of operations, with most
beneficiaries being small low-tech firms that abound in Italy. As a result, in the period under
observation here, sample NTBFs obtained financial support from 28 national policy schemes.

12

The policy measures implemented by the Italian government4 to support private firms differ
according to: i) their objective, ii) the evaluation method for applications, and iii) the financial
instrument through which subsidies are obtained by beneficiary firms. Regarding the latter two
dimensions, some measures relied on automatic schemes, and others on selective ones. Moreover,
financial instruments included tax credits, non-repayable grants, and subsidised loans (i.e., repayable
loans priced at below-market interest rates). As regards policy objectives, support to firms located in
the depressed areas of the South, which suffer from an economic and social gap with respect to the
other parts of the country, has traditionally been the most prominent one. Data provided by the
Ministry of Industry for the period 2000-2003 clearly document the importance of this objective (see
MAP, 2005).5 In this period, policy measures inspired by this objective accounted for about 46% of
the total amount of subsidies to private firms. The main schemes in this category, like the Law
64/1986 (Intervento Straordinario nel Mezzogiorno, Extraordinary Intervention for the South of
Italy) and the Law 488/1992, provided firms located in this geographic area with grants and other
forms of public finance to support their investments. Support to R&D and innovation was the second
most important objective of Italian industrial policy, with a 22% share of the subsidies awarded to
private firms in the 2000-2003 period. Most of the support was channelled through three instruments:
the FAR fund (Fondo Rotativo per le Agevolazioni alla Ricerca, Fund for Research Facilitations)
and the FIT fund (Fondo Rotativo per lInnovazione Tecnologica, Fund for Technological
Innovation), which were introduced by the Law 46/1982, and the Law 808/1985 targeting the
aerospace industry. FAR supported pre-competitive R&D expenses, while FIT had broader scope and
financed any type of innovation activity, including purchase of innovative machinery. FAR and FIT
accounted for roughly half of the total amount of the R&D and innovation subsidies granted to Italian
private firms in the 2000-2003 period and were widely used by NTBFs.6 The third most important

4
In Italy, several governmental institutions were responsible for the administration of public subsidies. They include: the Ministry
of Economics and Finance, the Ministry of Industry, the Ministry of University and Research, the Ministry of Labour and Welfare, the
Ministry of Agricultural Food and Forest Policies, the Ministry of International Trade, and the Institute for Foreign Trade (ICE). Quite
noteworthy is that in Italy, unlike other European countries, there is no public agency in charge of innovation policy measures.
5
Measures in support of the South date back to the early 1950s, when the Italian Government created the Cassa del Mezzogiorno (Fund
for the South), a public agency devoted to financing industrial development and public infrastructures in these regions. It ceased operations
in 1992. Reliable data that distinguish policy schemes according to their objectives are not available for the period before 2000. However,
the qualitative evidence provided by official documents confirms the prominence of this objective.
6
Law 808/1985 accounted for another 25%, but it mainly benefited large established firms. The remaining 25% was dispersed among a
plethora of schemes.

13

objective (9% of subsidies in 2000-2003) was the support of investments in fixed assets. The
remaining subsidies were channelled through schemes aimed at diverse objectives, including the
promotion of entrepreneurship and of the internationalisation of firms activities.
In spite of the variety of the policy measures implemented by the Italian government, it can be
presumed that if NTBFs were financially constrained, the receipt of a subsidy should have helped
them to relax these constraints, thereby positively influencing their investments. In fact, most policy
measures utilised by Italian NTBFs support investments in tangible or intangible (i.e., R&D and
innovation expenditures) assets.7 When this was not the case, the positive effect on investments may
have been indirect. On the one hand, beneficiary NTBFs may have been able to finance investments
through internal financial resources that would have been used for other purposes in absence of the
subsidy. On the other hand, receipt of a selective subsidy may have made it easier to obtain external
finance from other sources due to the quality certification effect. For these reasons, in assessing the
treatment effect of public finance on the investment activity of Italian NTBFs, we consider all
national policy schemes from which NTBFs obtained support.
A total of 66 sample NTBFs received public support from the Italian government in the period
1994-2003. Most of these firms received only one subsidy. Only 10 firms obtained two or more
subsidies. In Table 2, we illustrate the distributions of subsidised NTBFs by industry, geographical
area and foundation date. In Column III, we indicate the share accounted for by subsidised firms out
of the total number of sample firms in the corresponding category. From these figures, it is apparent
that public subsidies were not ubiquitous among Italian NTBFs. Moreover, the treatment was not
random. Results of 2 tests reject the null hypothesis that there are no differences between the
distributions of subsidised and non-subsidised firms by geographical area and foundation date (2(3)=
6.62) and 2(3)= 20.23, respectively). In the light of the discussion above about the objectives of
Italian industrial policy, the higher rate of subsidisation for firms in the South is hardly surprising.
The low subsidisation rate among very young NTBFs is more intriguing, and may reveal a cherry
picking attitude on the part of Italian governmental bodies in charge of administering support

7
According to Italian fiscal law, firms may decide whether to treat R&D expenses as investments or to expense them when they are
incurred. This latter option is more favourable for firms with positive net income.

14

schemes. Alternatively, high administrative costs may have discouraged younger and more time- and
resource-constrained firms from applying for support. Conversely, there are no significant differences
in the distribution of subsidised and non-subsidised firms by industry (2(4)= 6.48).
[Table 2 about here]
Table 3 reports the distribution of received subsidies across large and small firms. A firm-year
observation falls in the large firms or small firms category if in that particular year the focal firm
has total assets above or below, respectively, the overall median (i.e., firms might move between the
two sub-samples as their size changes over time). Out of the 1,499 observations in the dataset, there
are 85 firm-year observations in which one or more public subsidies were granted to the focal NTBF.
This indicates that the Italian government has not been overly generous with NTBFs. The larger
firms category accounted for 59 of these observations, while only 26 are smaller firms. Hence, the
likelihood of obtaining public subsidies is twice as large for larger firms than for their smaller
counterparts (7.9% and 3.5%, respectively). This difference may again be traced either to cherry
picking behaviour by Italian governmental institutions or to the presence of high application costs
that deter smaller firms.
[Table 3 about here]

4.

The econometric methodology

4.1.

Specification of the econometric models

To analyse the investment-cash flow sensitivity, the literature proposes several econometric models
(Hubbard, 1998; Bond and Van Reenen, 2007). As mentioned in Section 2.2, current cash flows
measure the availability of internal capital but may also be related to firms investment opportunities.
In the latter case, one cannot interpret the correlation between investments and cash flows as
documentation of financial constraints. Although for NTBFs this problem is less severe than for more
mature firms, it is important to include in the model some control for firms unobserved investment
opportunities. For this purpose, we estimate an Euler equation according to the model of Bond and
Meghir (1994):

15

I i ,t
K i ,t 1

I i ,t 1
= i + t + 1
K
i ,t 2

I
+ 2 i ,t 1

i ,t 2

CFi ,t
S
D
+ 3
+ 4 i , t + 5 i , t + i .t ,

i ,t 1
i ,t 1
i ,t 1

(1)

where Ii,t is the level of investments of firm i in period t, measured by the increase in the book value of
tangible and intangible assets net of depreciation, Ki,t is the end-of-period-t book value of firm is total
assets, CFi,t is firm is cash flow in period t after taxes but before dividends,8 Si,t is firm is sales
during period t and Di,t is firm is end-of-period-t total debt (i.e., the sum of short- and long-term
debt). If there are capital market imperfections and the external capital supply curve of NTBFs is
upward-sloping, we expect 3 to be positive, indicating financial constraints.
To assess whether firms investment rate and investment-cash flow sensitivity are affected by
public finance, we depart from the standard specification (1) in that we insert the dummy variable
PUB_FINi,t-1, which indicates whether firms i received public subsidies in year t-1 (PUB_FINi,t-1=1)
or not (PUB_FINi,t-1=0). More precisely, PUB_FINi,t-1 switches from 0 to 1 in the year that follows the
one in which the focal firm obtains the subsidy, switches back to zero in year two after receipt of the
subsidy and remains at zero thereafter.
I i ,t
K i ,t 1

I i ,t 1
= i + t + 1
K
i ,t 2

I
+ 2 i ,t 1

i ,t 2

CF
S
D
+ 3 i ,t + 4 i ,t + 5 i ,t +

i ,t 1
i ,t 1
i ,t 1

(2)

CFi ,t
+ i.t ,
+ 6 PUB _ FIN i ,t 1 + 7 PUB _ FIN i ,t 1

i
,
t
1

The coefficient 6 captures the increase in the investment rate generated by a public subsidy in the
immediately following year. The coefficient 7 measures the effect of public finance on the
investmentcash flow sensitivity in that year. We estimate equation (2) for the total sample and,
separately, for small and large firms.9 Smaller NTBFs allegedly are more financially constrained than
their larger counterparts; hence, we expect the effect of public finance on both investment levels and
investment-cash flow sensitivity to be more pronounced for small firms than for larger ones.

8
Other authors have used ex-dividend cash flows (e.g. Manigart et al., 2003). We opted for cash flows before dividends because our sample
is composed of unlisted firms. Managers of listed firms are more constrained than those of privately held firms to avoid a reduction in the
amount of dividends paid to shareholders, as this reduction may be perceived as a negative signal by investors. Conversely, in privately held
firms, dividends have no signalling role and all cash flows can be reinvested if some profitable investment opportunity arises.
9
A firm i is considered small/large in year t if its size, measured by total assets in year t, is smaller/larger than the median size, measured by
pooling all sample observations. The results do not significantly change if we consider median size by year. For the sake of synthesis, these
results are not presented here, but are available upon request from the authors.

16

So far, we have implicitly assumed that the receipt of public funds has a lagged impulse
effect on firms investment rate and investment-cash flow sensitivity. Thus, the receipt of public funds
engenders an instantaneous shock that materialises in the following year but does not persist over
time. This means that from year 2 after receipt of public funds onward, the investment choices of a
subsidised firm do not differ from those of a firm that never obtained public finance. To assess
whether receipt of a public subsidy permanently changes the investment pattern of NTBFs, we extend
equation (2) as follows:
I i ,t 1
= i + t + 1
K
i ,t 2

CFi ,t
S
D
+ 2 i ,t 1 + 3
+ 4 i ,t + 5 i ,t +

K i ,t 1

K i ,t 2
K i ,t 1
K i ,t 1
i ,t 1
CFi ,t
+ 8 PUB _ FIN _ pers i ,t 1 +
+ 6 PUB _ FIN i ,t 1 + 7 PUB _ FIN i ,t 1

K
i
,
t

CFi ,t
+ i .t .
9 PUB _ FIN _ pers i ,t 1

K i ,t 1
I i ,t

(3)

In this specification, PUB _ FIN _ pers i,t-1 equals 1 from year 2 after receipt of the public
subsidy up to the end of the observation period. If public subsidies have a persistent effect on NTBFs
investment choices over and above the transitory effect captured by parameters 6 and 7 , this
enduring impact is captured by parameters 8 and 9 . Again, 6 and 8 measure the effects of
public finance on firms investment rate, and 7 and 9 measure the effects on investmentcash flow
sensitivity. Note that should the persistent effect of public subsidies on firms investments equal their
short-term effect, 8 = 6 and 9 = 7 would be obtained. In estimating equation (3), we excluded
firms that in the observation period received more than one subsidy starting in the year in which they
obtained the second subsidy. The aim here is to disentangle the persistent effect of public subsidies
from their immediate (short-term) effect. The inclusion of firms that are serial beneficiaries of
public support would make it impossible to separately detect these two different effects.
Because equation (3) has a dynamic structure, to correctly evaluate the persistent effect of
public finance on investment rate it is useful to estimate the difference in the equilibrium investment
rate (i.e., for t) between a firm that received public finance and an otherwise identical firm that

17

did not. We can measure this difference through the following non-linear combination of the
estimated coefficients [for the derivation of this indicator see the Appendix (A2)]:

(1 1) 2 42 c~ (1 1) 2 42 c
I
,
=
K
22

(4)

where:
CF
c = 3
K

D

+ 4 S + 5
K
K

+ const

+ S + D
4 K 5 K

CF

+ 8 + 9
K

and
CF
c~ = 3
K

+ const = c + + CF
8
9

[Table 4 about here]

Table 4 reports some descriptive statistics for the variables used in our models. All of these
variables are normalised against the beginning-of-period-t stock of fixed and intangible assets. As
firms in our sample are relatively young and small, this value is sometimes close to zero, producing
extremely skewed and leptokurtic distributions of the variables. The presence of these outliers could
severely bias our results. To avoid this problem, we winsorised all variables (e.g., Dixon, 1960) with a
2% cut-off for each tail. In other words, for each variable we calculated the values corresponding to
the 2nd and 98th percentiles of its distribution and assigned these values to all observations falling
beyond them. This approach is useful because it reduces the impact of outliers and allows the use of a
larger number of observations than would be possible if outliers were deleted. Furthermore, it has
already been used in the investment literature (e.g., Baker et al., 2003), notably to assess investment
cash flow sensitivity (e.g., Cleary, 1999, 2006; Bertoni et al., 2010). Other cut-offs for winsorising
were computed, i.e., 1% and 5%, but a 2% cut-off offers the best compromise between smoothing
extreme values and maintaining sufficient variance.10 Descriptive statistics for winsorised variables
are also reported in Table 4.

10
Estimates using these different cut-offs are very close to those described in the next sections. They are available from the authors upon
request.

18

4.2. Estimation methodologies

In this work, we are interested in assessing the treatment effect of public finance on firms
investment rates and investment-cash flow sensitivity. Therefore, we have to deal with the potentially
endogenous nature of the public finance variables (i.e., PUB_FINi,t-1 and PUB_FIN_persi,t-1). In fact, a
firm that faces interesting investment opportunities is more likely to look for external finance,
including public subsidies, thus leading to a correlation between the two variables caused by reverse
causation (i.e., investments causing public finance). Moreover, the performance of an NTBF
(including its investment rate) is related to unobservable characteristics such as a brilliant business
idea, the development of a unique technology, or a team of smart owner-managers. If these
unobservables also influence the ability of firms to obtain public finance (for example, because of the
higher propensity of these firms to apply for public subsidies or the higher probability to be selected
by public bodies that administer the support schemes), a spurious correlation between public finance
and investment rate follows because of unobserved heterogeneity. An opposite bias in the estimated
treatment effect of public support is also possible if public subsidies are channelled by governmental
authorities to firms with poor investment opportunities, or if NTBFs with better investment
opportunities self-select out of public support schemes because of high application costs.
Both pooled ordinary least squares (OLS) and fixed-effects within-groups (WG) estimates in
these circumstances can lead to serious biases (Bond et al., 2001). Pooled OLS estimates are likely to
be biased because of lack of proper control for unobserved heterogeneity and endogeneity bias.
Through first differencing, WG estimates correct for time-invariant unobserved factors. However,
they do not take into account positive or negative shocks that may affect both the endogenous variable
and the covariates. We thus utilise the two-step system generalised method of moments estimation
(GMM-SYS, Arellano and Bover, 1995; Blundell and Bond, 1998) with finite-sample correction
(Windmeijer, 2005), which allows us to take into account the endogeneity of the covariates. In
addition to lagged levels of the series as instruments for first differences equations, the GMM-SYS
estimator employs additional moment conditions using first differences as instruments for variables in
levels. We consider covariates in the original Euler equation and all public finance variables to be
endogenous; therefore, instruments start from t-2.
19

However, the GMM-SYS estimator also presents some weaknesses. First, the use of a large
number of instruments results in significant finite sample bias, and measurement errors may cause
potential distortions in estimates. To deal with these problems, we limit the instrument set with
moment conditions in the interval between t-2 and t-4 (see Bond, 2002). Second, the moment
conditions of the GMM-SYS approach require that the instruments are uncorrelated with the error
terms, meaning that deviations of dependent and independent endogenous variables from the long-run
mean are not correlated with any unobservable and observable individual fixed effect. Results of the
Hansen statistic reassure us about the validity of the moment conditions used in all the estimations.
Moreover, pseudo-first stage regressions (shown in Appendix A.3) ensure the goodness of the GMMSYS estimator with respect to the GMM-DIFF estimator. Third, and most important, the standard
GMM-SYS estimator relies on the assumption of sequential exogeneity (see Wooldridge, 2000) in
that it assumes that future shocks to the firms investment opportunities are independent of receipt of
public subsidies, while past shocks can affect it. Therefore, it can produce (upward or downward)
biased estimates of the treatment effect of public finance by mistakenly attributing a selectionderived effect to the influence of public funds on firms investments. Thus, if NTBFs that face better
future investment opportunities are also more (or less) likely to obtain public finance, an upward (or
downward) bias for the estimated coefficients of PUB_FINi,t-1 in equation (2) and PUB_FINi,t-1 and
PUB_FIN_persi,t-1in equation (3) will follow. Similarly, if future shocks that positively affect both
investments and cash flows are positively (or negatively) correlated with receipt of public subsidies,
an upward (or downward) bias in the estimated effect of public finance on the investment-cash flow
sensitivity will follow. To address this problem we follow an approach inspired by Srensen (2007),
adding to the set of instruments of the GMM-SYS estimator three additional variables: the first two
variables reflect the availability of public funds in the geographic area of the focal NTBF and in the
industry in which it operates (PUB_FIN_Areai and PUB_FIN_Sectori)11; the third variable is the ratio
between the amount of yearly subsidies transferred to the private sector by the Italian government and
GDP (SUB_GDP; source: Eurostat). Good instruments need to be related to public finance variables,

11

These indicators represent the share of RITA NTBFs that obtained public funds out of the total number of RITA NTBFs that are located in
the same geographical area and operate in the same industry as firm i.

20

but should be independent of the error terms of the Euler equation. Firms with great future investment
opportunities may be located in geographic areas and operate in industries with abundance or scarcity
of public finance; moreover, these investment opportunities are unlikely to be correlated with the
amount of subsidies transferred by the state to the private sector. Independently of focal firms future
investment opportunities, the likelihood of obtaining public finance is higher if the firm operates in a
sector and geographical area with an abundance of public subsidies; it is also higher in years when
more subsidies are available. However, the effect of public finance on investment rate and
investment-cash flow sensitivity is independent of these variables, and so they are a source of
exogenous variation (in a similar vein, see also, Bottazzi et al. 2008, Chemmanur et al. 2009, Ivanov
and Xie 2010).12 As an additional robustness check, we replaced these variables with the estimated
likelihood of firm i obtaining public finance in year t (see Benfratello and Sembenelli, 2006 for a
similar approach in a different context). This was estimated through a logit panel data model.
Regressors include firm-specific characteristics (i.e., firm size and age, as well as a dummy indicating
firms that were located in a technology incubator when founded); founder-specific characteristics
capturing human capital (i.e., years of university education in management and economics and in
technical and scientific fields, years of technical and commercial work experience, management
experience, and a dummy indicating firms created by one or more individuals with experience in
academic research); the size of the founding team; and the instruments previously mentioned
(PUB_FIN_Areai, PUB_FIN_Sectori and SUB_GDP).
Finally, a possible survivorship bias in data may intervene in the investigated relationship. We
implemented a test to check for such bias in the spirit of Wooldridge (1995) and Semykina and
Wooldridge (2006): we estimated the Euler equations (2) and (3) via GMM-SYS, inserting as a
further control variable an inverse Mills ratio type of control factor. To compute this control factor,
we initially focused attention on the RITA 2000 sample. This sample, composed of 401 firms, was
selected according to the same criteria and strategy used for the sample examined in the present study
(Colombo et al. 2004). We considered the exit rate of sample firms in the period 20002003 and
12

Note that in accordance with the studies mentioned above, we make the assumption that the clustering of public funds in specific industry
or geographic markets is exogenous that is it is not driven by the investment opportunities unobserved to third parties faced by the NTBFs
that are in those markets and are potential candidates for receiving public funds. Therefore, it is uncorrelated with the error term of the Euler
equation.

21

estimated a probit model of firm exit in 20002003, conditional on survival up to the end of 1999; the
dependent variable in this model is the hazard rate of exit of sample firms in 20002003. The
independent variables include founders human capital (i.e., founders economic and technical
education, commercial and technical work experience and previous managerial experience), receipt of
public subsidies before 2000, firm size and age in 1999, number of founders, whether the firm was
located in an incubator when founded, and other controls (i.e., industry of the focal firm and
infrastructural characteristics of the area where the firm is located). We used the estimated
coefficients for this sample selection model to compute the inverse Mills ratio of survival for each
firm-year observation included in the sample analysed in the present work.13 This time-varying ratio
was then inserted in the Euler equations (2) and (3) to test for possible survivorship bias.
To evaluate the relevance of all our econometric models, we also implemented the Arellano and
Bond test for first- and second-order serial autocorrelation of residuals [AR(1), AR(2)]. If it is not
serially correlated, the difference of residuals should be characterised by a negative first-order serial
correlation and the absence of a second-order serial correlation. Our results confirm this as shown in
Section 5.

4.3. Tests
The effect of public finance on investment cash-flow sensitivity can be gauged through a simple linear
test on the parameters of the models. If an instantaneous change CF in CFi,t produces a change in the
investment rate, the coefficient of CFi,t/Ki,t1should be positive and significant. We interpret this
positive coefficient as an indication of financial constraints. As explained in Section 2.2, we expect
small NTBFs to be financially constrained, and public finance to help them to remove these financial
constraints. Hence, after receiving public finance, internal cash flow in these firms should no longer
have any effect on the investment rate (i.e., the coefficient of CFi,t/Ki,t1should not be positive and
significant). The relaxation of financial constraints engendered by public finance may be transitory or
persistent. In the transitory condition, the coefficient of the cash flow variable will not be positive

13

Formally (see Greene 2003): SURVIVAL

= ( ' w it ) ( ' w it ) -1 ; where wi is the vector of independent variables of the probit model on
it

firm exit, and () and () are, respectively, the density and distribution functions of the standard normal.

22

(and significant) only in the year immediately following the receipt of the public subsidy. Conversely,
in the persistent situation, the coefficient of this variable will not be positive and significant also for
the remaining observation period. Following this line of reasoning, we performed the following Wald
tests of the null hypothesis that a change in cash flow does not affect the investment rate:
In equation (2): 3=0
3 + 7 = 0
In equation (3): 3 = 0

for firms that did not obtain any public subsidies,

for firms that obtained public subsidies.
for firms that did not obtain any public subsidies,

3 + 7 = 0

for firms that obtained public subsidies (short-term effect),

3 + 9 = 0

for firms that obtained public subsidies (persistent effect).

The effect of public finance on investment level can be evaluated in a similar way by
performing Wald tests of the following null hypotheses:
In equation (2): 6=0.
In equation (3): 6 = 0.
Using the pseudo-Wald test performed with the delta method, we can test the null hypothesis that
public finance has no effect on firms equilibrium (i.e., steady state) investment rate as follows (see
equation (4)): I = 0.
K

5.

Econometric results

We illustrate in Table 5 the estimates of equation (2). Model I uses, in addition to internal
instruments, the exogenous instruments PUB_FIN_Areai, PUB_FIN_Sectori and SUB_GDP. In
Model II, these latter instruments are replaced by the likelihood of firm i obtaining public finance in
year t. In Model III, we add the inverse Mills ratio type of control for survivorship bias. Finally, in
Model IV, the public finance variables are considered as exogenous; these latter estimates are
presented merely for comparison purposes. In all models, the null hypothesis of absence of a negative
first-order serial correlation is rejected, while the null hypothesis of absence of a second-order serial
correlation is not. The Hansen tests also indicate that the null hypothesis of equality to zero of the
specified orthogonality conditions is not rejected. Moreover, the coefficient of the survivorship bias

23

control in Model III is not significant. Thus, unobserved factors that are positively or negatively
associated with a firms investment rate seem not to be correlated with the likelihood of exit of sample
firms.
In panel (a) of Table 5, estimates refer to the entire sample. Results give some support to the
view that NTBFs are financially constrained, as the coefficients of the cash flow variable capturing
the sensitivity of investments to cash flows for non-subsidised fir+ms, are positive and weakly
significant. Public finance does not significantly affect the investment rate. However, the investmentcash flow sensitivity of subsidised firms is not significant. We then split the sample into two subsamples, composed of large and small firms (respectively, panels (b) and (c) of Table 5). As
previously mentioned, an observation falls in the large firms or small firms sub-sample if in that
particular year the firm has total assets above or below, respectively, the overall median. Estimate
results for large firms suggest that these firms are not financially constrained, as the coefficient of the
cash flow variable, though positive, is not significant. Moreover, public finance does not significantly
affect either the investment rate or the investment-cash flow sensitivity. Results for small NTBFs are
more interesting. The investment rate of small non-subsidised NTBFs is found to be sensitive to cash
flows, as documented by the positive and significant (at 1%) coefficient of the cash flow variable.
This result is in line with the argument that these firms find it difficult to obtain external finance at
fair terms and are forced to rely mainly on internal capital to finance their investments. When small
NTBFs receive public funds, they are able to remove the financial constraints that would otherwise
inhibit investments. Accordingly, their investment rate increases and investments are no longer
dependent on internal cash flows. In fact, the coefficient of PUB_FINi,t-1 (i.e. 6 ) is positive,
significant at conventional confidence levels and of large economic magnitude.14 The coefficient of
the cash flow variable becomes negative in the year immediately following the receipt of public
finance (i.e. 3 + 7 < 0 ) and is no longer significant, as documented by the value of the Wald test
reported at the bottom of the table, indicating the removal of financial constraints.
[Table 5 about here]
14
Using the estimated coefficients reported in panel (c) of Table 5 (Model I), and average values for the variables of interest, we
find that in the year immediately following the receipt of the subsidy, the investment rate of a subsidised firm increases by 19% with respect
to the investment rate that the same firm would obtain in the absence of the subsidy.

24

Let us now consider whether the effects of public finance on the investments of small NTBFs
are transitory or persistent. For this purpose, we report in the last two columns of Table 6 the results
of the GMM-SYS estimates15 of equations (3), with and without the control for survivorship bias. The
estimates are run on a sub-sample that includes small NTBFs that either did not receive any public
finance or received it only once. In the first two columns we report estimates of equation (2) relating
to the same sub-sample of firms for comparison purposes. These latter estimates are qualitatively
similar to those illustrated in panel (c) of Table 5, and confirm that non-subsidised small NTBFs
largely rely on internal cash flows to finance their investments. After receipt of a public subsidy, the
investment rate increases and the sensitivity of investments to cash flows vanishes. Interestingly, the
first public subsidy received by small NTBFs has a larger treatment effect on both the investment
rate and the investment-cash flow sensitivity than the effect highlighted by the estimates reported in
Table 5 (which do not distinguish the first subsidy from the subsequent ones).
[Table 6 about here]
Let us now turn attention to the estimates of equation (3). The estimated one-year ahead
effects of public finance on the investment rate and the investment-cash flow sensitivity are
remarkably similar to those obtained from the estimates of equation (2). We are interested here in
assessing whether these effects persist over time. Regarding investment-cash flow sensitivity, the
results of the Wald tests reported at the bottom of Table 6 indicate that for subsidised small NTBFs,
the null hypothesis of no persistent positive dependence of investments on cash flow shocks cannot be
rejected at conventional confidence levels (the coefficient is positive but not significant). Thus, our
results suggest that receipt of a public subsidy leads to a permanent relaxation of the financial
constraints that small NTBFs suffer in absence of public support.
Let us now examine whether public subsidies have a persistent positive treatment effect on
the investment rate. Accordingly, we rely on the indicator reported in equation (4). In Table 7 we
report the estimated value of the equilibrium investment rate (i.e., in the steady state) for nonsubsidised and subsidised small NTBFs; this value is calculated using the coefficients of the equation

15
For the sake of synthesis, we do not present other GMM-SYS estimates in Table 6. The results are very close to those presented in Table 6
and are available from the authors upon request.

25

reported in column III of Table 6. In particular, we calculated the steady state investment rate for the
two categories of firms according to different levels of cash flow (tenth percentile, first quartile,
median, average and third quartile). We also calculated the investment rate that subsidised firms
would exhibit should the short-term effect of public finance persist unaltered over time (i.e., 8 = 6
and 9 = 7 ). The difference in the investment rate (reported in the third row of Table 7) of
subsidised and non-subsidised firms, though non-negligible in magnitude (+0.1298 at the median
value of the cash flow rate; +0.3184 at the 10th percentile), is not significant. This result suggests that
the effect of public finance on the investment rate of subsidised NTBFs is not persistent over time.
Interestingly, if the estimated short-term effect of public finance remained unaltered over time, the
positive effect of receipt of a public subsidy on the steady-state investment rate would be larger
(+0.4789 at the median value of the cash flow rate, although again not statistically significant;
+1.1424 at the 10th percentile, significant at 5%).
Table 7 also provides further evidence for the persistent effect of public finance on relaxation
of firms financial constraints. For non-subsidised small NTBFs, the estimated steady state investment
rate monotonically increases with increases in the value of the cash flow rate. The difference in the
investment rate of high-cash flow firms (III quartile) and low-cash flow firms (I quartile) is quite large
(+0.3277) and significant (at 1%). Conversely, for subsidised firms the corresponding difference is
much smaller (+0.0507) and non-significant. We interpret these results to suggest that i) nonsubsidised small NTBFs face binding financial constraints and ii) receipt of a public subsidy leads to a
persistent removal of these financial constraints.

6.

Discussion and conclusions

The aim of this paper was to empirically analyse whether public finance helps small, financially

constrained NTBFs to relax the financial constraints that bind their investment activity. Moreover, we
were interested in assessing whether the effects of public finance are transient or persist over time. For
this purpose, we considered a unique hand-collected longitudinal dataset that includes 294 Italian
NTBFs observed from 1994 to 2003. Out of these firms, 66 have obtained one or more public
subsidies. The longitudinal dimension of the dataset and the availability of a rich set of variables
26

allowed us to estimate a modified version of the Euler equation using GMM-SYS estimation
techniques with an enriched set of instruments, so as to better take into account the endogenous nature
of public finance.
Our results can be synthesised as follows. First, our estimates indicate that small NTBFs
significantly rely on internal funds to finance their investments. In fact, these firms exhibit a positive
and significant investment-cash flow sensitivity. Conversely, no such relationship emerges for larger
NTBFs. These findings are in the spirit of several previous studies that claimed that small NTBFs are
the firms most likely to be financially constrained (Carpenter and Petersen, 2002a; 2002b; Hall, 2002;
Hyytinen and Toivanen, 2005; Czarnitzky, 2006; Colombo and Grilli, 2007; Bertoni et al. 2010).
Second, we showed that receipt of public subsidies results in an increase of the investment rate
and a reduction of the investment-cash flow sensitivity in the immediately following year. The
increase in investment rate is both statistically significant and of large economic magnitude.
Third, we analysed the dynamics of firms investments after receipt of a public subsidy. For
this purpose, we focused attention on small NTBFs that in the observation period received public
subsidies only once. We found that the increased investment rate detected for these firms in the short
term does not significantly persist in the long term. However, the sensitivity of investments to cash
flow turned out not only to vanish immediately after receipt of the public subsidy, but also to remain
negligible in the longer term. We interpret this result to suggest that after receipt of a public subsidy,
the financial constraints that bind the investment activity of small NTBFs are permanently removed.
Our work contributes to the literature arguing that public finance is extremely important for
small and young high-tech firms (see e.g., Lerner, 1999; Lach, 2002; Colombo et al. 2009a, 2009b).
In particular, previous studies documented that public finance can alleviate the negative effects of
capital market imperfections on innovation and growth of these firms (Hyytinen and Toivanen, 2005;
Czarnitzky, 2006; Czarnitzky et al., 2009). Our findings confirm this view, and also go one step
further in indicating that the positive effects of public finance on the removal of financial constraints
of small NTBFs are not transient, but instead persist over time after receipt of the public subsidies.
This is original evidence in support of the importance of public subsidies for this type of firm.

27

Our results also document that public support does not have similar beneficial effects on
larger NTBFs. These firms are not financially constrained and accordingly, public support does not
affect their investment strategies. This evidence reinforces the view that public policy measures have
to be targeted to small, generally young NTBFs, which really need public finance for innovation,
growth and investment. In Italy as in most European countries, this has rarely been the case in the past
(Colombo and Grilli, 2006). Therefore, it is fair to acknowledge that despite the potential beneficial
effects of public support for NTBFs, governance failure while distorting fund allocation can prevent
these benefits from materialising (on these issues see also Schneider and Veugelers, 2008).
Much remains to be done in this field. There are three directions for future research that seem
very promising to us. First, it should be determined whether the results we obtained for Italian NTBFs
are generalisable to other countries. In this perspective, the creation of a cross-country dataset similar
to the one used in this work would be a fundamental step forward in providing a micro-econometric
assessment of the effects of public policy measures. In a similar vein, we focused attention here on
subsidies provided by the Italian national government. However, both local and supranational (e.g.,
the EC) public institutions are increasingly important sources of public support to NTBFs. Whether
the results presented here depend on the source of public support is an interesting research issue.
Finally, our results suggest that public finance is effective in persistently alleviating the negative
effects of capital market imperfections faced by small NTBFs. However, they do not tell us the reason
behind this positive effect. On the one hand, award of a public subsidy may have a quality
certification effect, reducing the information asymmetries between small NTBFs and external
investors. However, this effect is confined to selective schemes administered by reputable
governmental bodies. On the other hand, public subsidies independently of their nature (i.e., either
selective or automatic) may allow beneficiary firms to increase the amount of tangible assets; in turn,
these assets can be pledged as collateral to raise additional capital. Because of the lack in our dataset
of a sufficient number of firm-year observations relating to subsidised small NTBFs, we were not able
to discriminate between selective and automatic schemes, and so we were not able to distinguish
between these two effects. This clearly offers promising opportunities for future research. In
particular, a careful analysis of the treatment effect of different policy schemes would improve our
28

understanding of the specific mechanisms that lead to removal of the financial constraints that bind
the investment activity of small NTBFs, thereby providing useful new insights for the design of
effective policy measures in support of this type of firm.

29

References
Abel, A. B. (1980). Empirical investment equations: an integrative framework. CarnegieRochester Conference on Public Policy , 12 (1), 39-91.

Almeida, H., and Campello, M. (2007). Financial constraints, asset tangibility, and corporate
investment. Review of Financial Studies , 20 (5), 142960.

Alti, A. (2003). How sensitive is investment to cash flow when financing is frictionless?
Journal of Finance , 58 (2), 707-22.

Anton, J. J., and Yao, D. A. (2002). The sale of ideas: strategic disclosure, property rights, and
contracting. Review of Economic Studies , 69 (3), 513-31.

Arellano, M., and Bover, O. (1995). Another look at the instrumental variables estimation of
error-components models. Journal of Econometrics , 68 (1), 29-51.

Audretsch, D. B. (1995). Innovation and industry evolution. MIT Press: Cambridge, MA.

Avery, R., Bostic, R., and Samolyk, K. (1998). The role of personal wealth in small business
finance. Journal of Banking and Finance , 22 (2), 101961.

Baker, M., and Stein, J. C. (2003). When does the market matter? Stock prices and the
investment of equity-dependent firms. The Quarterly Journal of Economics , 118 (3), 9691005.

Becker, G. S. (1983). A theory of competition among pressure groups for political influence.
The Quarterly Journal of Economics , 98 (3), 371-400.

30

Benfratello, L., and Sembenelli, A. (2006). Foreign ownership and productivity: is the direction
of causality so obvious? International Journal of Industrial Organization , 24 (4), 733-751.

BEPA. (2008). Innovation and growth in the EU: the role of SME policy. European
Commission.

Berger, A. N., and Udell, G. F. (1990). Collateral, loan quality, and bank risk. Journal of
Monetary Economics, 25 (1), 21-42.

Bertoni, F., Colombo, M. G., and Croce, A. (2010). The effect of venture capital financing on
the sensitivity to cash flow of firms investments. European Financial Management , forthcoming.

Blundell, R., and Bond, S. R. (1998). Initial conditions and moment restrictions in dynamic
panel data models. Journal of Econometrics , 87 (1), 115-143.

Bond, S. R. (2002). Dynamic panel data models: a guide to micro data methods and practice.
Portuguese Economic Journal , 1 (2), 141-62.

Bond, S. R., and Meghir, C. (1994). Dynamic investment models and the firm's financial
policy. Review of Economic Studies , 61 (2), 197-222.

Bond, S. R., and Van Reenen, J. (2007). Microeconometric models of investment and
employment. In J. J. Heckman, and E. E. Leamer, Handbook of econometrics (Vol. 6A). Elsevier:
Amsterdam.

Bond, S. R., Hoeffler, A., and Temple, J. (2001). GMM estimation of empirical growth models.
Working Paper (CEPR Discussion Paper No. 3048).
31

Bottazzi, L., Da Rin, M., and Hellmann, T. (2008). Who are the active investors? Evidence
from venture capital. Journal of Financial Economics , 89 (3), 488-512.

Bougheas, S., Grg, H., and Strobl, E. (2003). Is R & D financially constrained? Theory and
evidence from Irish manufacturing. Review of Industrial Organization , 22 (2), 159-174.

Carpenter, R. E., and Petersen, B. C. (2002a). Capital market imperfections, high-tech

investment, and new equity financing. Economic Journal , 112 (477), F54-F72.

Carpenter, R. E., and Petersen, B. C. (2002b). Is the growth of small firms constrained by
internal finance? Review of Economics and Statistics , 84 (2), 298309.

Chemmanur, T. J., Krishnan, K., and Nandy, D. (2009). How does venture capital financing
improve efficiency in private firms? A look beneath the surface. Boston College, Working paper.

Cleary, S. (1999). The relationship between firm investments and financial status. Journal of
Finance , 54 (2), 67392.

Cleary, S. (2006). International corporate investment and the relationships between financial
constraint measures. Journal of Banking and Finance , 30 (5), 155980.

Cohen, L. R., and Noll, R. G. (1991). The Technology Pork Barrel. The Brookings Institution
Press: Washington, DC.

Colombo, M. G., Grilli, L., and Delmastro, M. (2004). Entrepreneurs' human capital and the
start-up size of new technology-based firms. International Journal of Industrial Organization , 22 (89), 1183-1211.
32

Colombo, M. G., and Grilli, L. (2006). Supporting high-tech start-ups: Lessons from Italian
technology policy. The International Entrepreneurship and Management Journal , 2 (2), 189-209.

Colombo, M., and Grilli, L. (2007). Funding gaps? Access to bank loans by high-tech start-ups.
Small Business Economics , 29 (1-2), 25-46.

Colombo, M., Giannangeli, S., and Grilli, L. (2009a). A longitudinal analysis of public
financing and the growth of New Technology Based Firms: do firms age and applicants evaluation
methods matter? Working paper.

Colombo, M. G., Grilli, L., and Murtinu, S. (2009b). Direct public subsidies and the
productivity of high-tech startups: exploring the governmental build efficiency function. Working
paper.

Cumming, D. J., and MacIntosh, J. G. (2007). Mutual funds that invest in private equity? An
analysis of labour-sponsored investment funds. Cambridge Journal of Economics , 31 (3), 445-87.

Czarnitzki, D. (2006). Research and development in small and medium-sized enterprises: the
role of financial constraints and public funding. Scottish Journal of Political Economy , 53 (3), 33557.
Czarnitzki, D., Hottenrott, H., and Thorwarth, S. (2009). Industrial research versus
development investment: the implications of financial constraints. ZEW Discussion paper N. 09-049.

David, P. A., Hall, B. H., and Toole, A. A. (2000). Is public R&D a complement or substitute
for private R&D? A review of the econometric evidence. Research Policy , 29 (4-5), 497-529.

33

Delmar, F., and Shane, S. (2006). Does experience matter? The effect of founding team
experience on the survival and sales of newly founded ventures. Strategic Organization , 4 (3), 21547.

Dixon, W. J. (1960). Simplified estimation from censored normal samples. Annals of

Mathematical Statistics , 31 (2), 38591.

Eckhardt, J., Shane, S., and Delmar, F. (2006). Multi-stage selection and the financing of new
ventures. Management Science , 52 (2), 220-32.

Erickson, T., and Whited, T. M. (2000). Measurement error and the relationship between
investment and q. Journal of Political Economy , 108 (5), 102757.

Fazzari, S., Hubbard, R.G., and Petersen, B. (1988). Financing constraints and corporate
investment. Brookings Papers on Economic Activity , 1988 (1), 141-206.

Gertler, M., and Gilchrist, S. (1994). Monetary policy, business cycles, and the behavior of
small manufacturing firms. The Quarterly Journal of Economics , 109 (2), 309-40.

Gomes, J. (2001). Financing investment. American Economic Review , 91 (5), 126385.

Gonzlez, X., Jaumandreu, J., and Paz, C. (2005). Barriers to innovation and subsidy
effectiveness. The Rand Journal of Economics , 36 (4), 93050.

Greene, W. H. (2003). Econometric Analysis. Prentice Hall, 5th edition.

Hall, B. H. (2002). The financing of research and development. Oxford Review of Economic
Policy , 18 (1), 35-51.
34

Hall, R. E., and Jorgenson, D. W. (1967). Tax policy and investment behavior. American
Economic Review , 57 (3), 391414.

Hao, K., and Jaffe, A. (1993). Effect of liquidity on firm's R&D spending. Economics of
Innovation and New Technology, 2 (4), 27582.

Harhoff, D. (1998). Are there financing constraints for R&D and investment in German
manufacturing firms? Annales d'Economie et de Statistique , 49/50, 421-56.

Himmelberg, C. P., and Petersen, B. C. (1994). R&D and internal finance: a panel study of
small firms in high-tech industries. Review of Economics and Statistics , 76 (1), 3851.

Holtz-Eakin, D. (2000). Public policy toward entrepreneurship. Small Business Economics , 15

(4), 283-91.

Hubbard, R. (1998). Capital-market imperfections and investment. Journal of Economic

Literature , 6 (1), 193-225.

Hussinger, K.. (2008). R&D and subsidies at the firm level: an application of parametric and
semi-parametric two-step selection models. Journal of Applied Econometrics, 23(6), 729-47.

Hyytinen, A., and Toivanen, O. (2005). Do financial constraints hold back innovation and
growth? Research Policy , 34 (9), 1385-1403 .

Irwin, D. A., and Klenow, P. J. (1996). High-tech R&D subsidies Estimating the effects of
Sematech. Journal of International Economics , 40 (3-4), 323-344.

35

Ivanov, V. I., and Xie, F. (2010). Do Corporate Venture Capitalists add value to startup firms?
Evidence from IPOs and acquisitions of VC-Backed companies. Financial Management, forthcoming.

Jensen, M. (1986). Agency costs of free cash flow, corporate finance and takeovers. American
Economic Review , 7 (2), 323-29.

Jorgenson, D. W. (1963). Capital theory and investment behavior. American Economic Review ,
53 (2), 24759.

Kadapakkam, P., Kumar, P. C., and Riddick, L. A. (1998). The impact of cash flows and firm
size on investment: the international evidence. Journal of Banking and Finance , 22 (3), 293320.

Kaplan, S. N., and Zingales, L. (1997). Do investment-cash flow sensitivities provide useful
measures of financing constraints? The Quarterly Journal of Economics , 112 (1), 169215.

Kaplan, S. N., and Zingales, L. (2000). Investment-cash flow sensitivities are not valid
measures of financing constraints. The Quarterly Journal of Economics , 115 (2), 70712.

Klette, T. J., Men, J., and Griliches, Z. (2000). Do subsidies to commercial R&D reduce
market failures? Microeconometric evaluation studies. Research Policy , 29 (4-5), 47195.

Lach, S. (2002). Do RandD subsidies stimulate or displace private R&D? Evidence from Israel.
The Journal of Industrial Economics , 50 (4), 369-90.

Leleux, B., and Surlemont, B. (2003). Public versus private venture capital: seeding or
crowding out? A pan-European analysis. Journal of Business Venturing , 18 (1), 81-104.

36

Lerner, J. (1999). The government as venture capitalist: the long-run impact of the SBIR
program. Journal of Business , 72 (3), 285-318.

Manigart, S., Baeyens, K., and Verschueren, I. (2003). Financing and investment
interdependencies in unquoted Belgian companies: the role of venture capital. In P. F. Butzen, Firms
Investment and Finance Decision. Cheltenham: Edward Elgar.

MAP. (2005). Relazione sulle leggi e i provvedimenti di sostegno alle attivit.

Modigliani, F., Miller, M. (1958). The cost of capital, corporation finance, and the theory of
investment. American Economic Review , 48 (3), 261-97.

Moyen, N. (2004). Investment-cash flow sensitivities: constrained versus unconstrained firms.

Journal of Finance , 59 (5), 206192.

Myers, S., and Majluf, N. (1984). Corporate financing and investments decisions when firms
have information that investors do not have. Journal of Financial Economics , 13 (2), 187221.

Pawlina, G., and Renneboog, L. (2005). Is investment-cash flow sensitivity caused by agency
costs or asymmetric information? Evidence from the UK. European Financial Management , 11 (4),
483513.

Savignac, F. (2008). Impact of financial constraints on innovation: what can be learned from a
direct measure? Economics of Innovation and New Technology , 17 (6), 553-69.

Schneider, C., and Veugelers, R. (2008). On young innovative companies: why they matter and
how (not) to policy support them. Working Paper, Department of Economics, Copenhagen Business
School.
37

Semykina, A., and Wooldridge, J. (2006). Estimating panel data models in presence of
endogeneity and selection: theory and application. Working paper.

Srensen, M. (2007). How smart is smart money? A two-sided matching model of venture
capital. Journal of Finance , 62 (6), 2725-62.

Tiwari, A. K., Mohnen, P., Palm, F. C., and van der Loeff, S. S. (2009). Financial constraints
and R&D investment: evidence from CIS . Working paper.

Vogt, S. C. (1994). The cash flow/investment relationship: evidence from U.S. manufacturing
firms. Financial Management , 23 (2), 320.

Wallsten, S. J. (2000). The effects of government-industry R&D programs on private R&D:

The case of the Small Business Innovation Research Program. The RAND Journal of Economics, 31
(1), 82-100.

Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step
GMM estimators. Journal of Econometrics , 126 (1), 25-51.

Wooldridge, J. (2000). A framework for estimating dynamic, unobserved effects panel data
models with possible feedback to future explanatory variables. Economics Letters , 68 (3), 245-50.

Wooldridge, J. (1995). Selection corrections for panel data models under conditional mean
independence assumptions. Journal of Econometrics , 68 (1), 115-32.

38

Tables
Table 1. Distribution of sample firms across industries, geographical areas and
foundation date

Industry
ICT manufacturing
Automation equipment and robotics
Biotechnology, pharmaceutics and advanced
materials
Software
Internet and telecommunications services
Total
Geographical area
Northwest
Northeast
Centre
South
Total
Foundation date
1980-1985
1986-1991
1992-1997
1998-2003
Total

RITA directory

Sample

N.
427
212

%
21.6
10.7

N.
60
30

%
20.4
10.2

96

4.9

14

4.8

539
700
1,974

27.3
35.5
100.0

95
95
294

32.3
32.3
100.0

RITA directory
N.
853
447
366
308
1,974

%
43.2
22.6
18.6
15.6
100.0

Sample
N.
145
65
45
39
294

RITA directory
N.
345
350
622
657
1,974

%
17.5
17.7
31.5
33.3
100.0

%
49.3
22.1
15.3
13.3
100.0
Sample

N.
52
61
130
51
294

%
17.7
20.7
44.2
17.4
100.0

39

Table 2. Distribution of subsidised NTBFs across industries, geographical areas and

foundation dates

Industry
ICT manufacturing
Automation equipment and robotics
Biotechnology, pharmaceutics and advanced
materials
Software
Internet and telecommunications services
Total

Subsidised firms
N.
14
7

%
21.20
10.61

4.55

27
15
66

40.91
22.73
100.0
Sample

Geographical area
Northwest
Northeast
Centre
South
Total

N.
31
11
9
15
66

Foundation date
1980-1985
1986-1991
1992-1997
1998-2003
Total

%
46.97
16.66
13.64
22.73
100.0
Sample

N.
17
19
25
5
66

%
25.76
28.78
37.88
7.58
100.0

Ratea
%
23.3
23.3
21.4
28.4
15.8
22.4
Ratea
%
21.4
16.9
20.0
38.5
22.4
Ratea
%
32.7
31.1
19.2
9.8
22.4

Legend (a) Rate: percentage of subsidised firms out of the total number of sample firms in the
corresponding category

40

Table 3. Subsidies granted to NTBFs by firm size

No. of firm-year observations

Large firms
Small firms
Total

750
749
1,499

Firm-year observations in
which one or more subsidies
were granted
N.
%
59
7.9
26
3.5
85
5.7

Note: A firm i is considered small/large in year t if its size, measured by total assets is smaller/higher than
median size, measured pooling all the observations in the sample.

41

Table 4. Descriptive statistics on regression variables

Std. Dev.
Variable

N.

Mean

Median Std. Dev. Between

Within

Skewness Kurtosis

Not winsorised
Ii,t/Ki,t1

1499

1.292

0.444

7.362

4.767

6.233

21.297

506.106

CFi,t/Ki,t1

1499

1.017

0.479

3.664

2.484

2.964

12.235

229.673

Si,t/Ki,t1

1499

23.700

10.928

86.982

57.824

66.971

27.019

892.861

38.358

1480.16

(Di,t/Ki,t1)2 1499 2427.442 24.772 69792.49 32964.47 61809.35

Winsorised 2% each tail
Ii,t/Ki,t1

1499

0.932

0.444

1.537

0.896

1343

3.693

18.402

CFi,t/Ki,t1

1499

0.872

0.479

1.515

1.236

1.097

2.917

14.438

Si,t/Ki,t1

1499

19.821

10.928

25.818

22.381

14.989

2.878

12.548

(Di,t/Ki,t1)2 1499 288.575

24.772

1002.3

847.658

659.04

5.405

32.906

Note: Ii,t is the increase in a firms book value of tangible and intangible assets net of depreciation between periods t1 and t.
Ki,t1 is the beginning-of-period-t book value of tangible and intangible assets. CFi,t is firm is cash flows after taxes and before
dividends in period t. Si,tis firm is sales in period t. Di,tis firm is total current and long-term debt at the end of period t.

42

Table 5. Short-term effect of public finance on firms investment rate and investment-cash flow
sensitivity
Panel a. Total sample
Model I
Ii,t1/Ki,t2
(Ii,t1/Ki,t2)2
CFi,t/Ki,t1

PUB_FINi,t-1(CFi,t/Ki,t1)

Si,t/Ki,t1
(Di,t/Ki,t1)

Model IV

-0.0018

-0.0313

-0.0260

-0.0204

(0.0828)

(0.0867)

(0.0828)

0.0076

0.0107

0.0099

0.0094

(0.0124)

(0.0110)

(0.0117)

(0.0111)

0.2023

0.2087

0.2156

(0.1142)

(0.1084)

(0.1133)

(0.1153)

-0.1142

-0.3494

-0.1109

-0.1072

(0.2233)

(0.2668)

(0.2221)

(0.2204)

-0.0037

0.2309

-0.0429

-0.0071

(0.2760)

(0.3515)

(0.2638)

(0.2877)

0.1690
2

Model III

(0.0925)

0.2097

PUB_FINi,t-1

Model II

***

0.0190

**

0.0164

**

0.0202

(0.0077)

(0.0086)

(0.0072)

(0.0082)

-0.0000

-0.0000

-0.0000

-0.0001

(0.0002)

(0.0002)

(0.0002)

(0.0003)

Inverse Mill's Ratio

**

0.1400
(0.1668)

Constant

Investment-cash flow sensitivity of

subsidised firms (Wald test):
Short-term effect(i.e.1 year ahead)
No. of observations
Hansen

0.1961

0.2045

0.4320

0.1619

(0.1159)

(0.1286)

(0.3130)

(0.1401)

0.2060

0.4332

0.1658

0.2084

(0.2970)

(0.3734)

(0.2764)

(0.3101)

1499

1499

1499

1499

122.09

101.35

120.49

119.86

[142]

[119]

[141]

[141]

AR(1)

-4.0227 ***

-3.9367 ***

-3.9784

AR(2)

0.4950

0.4554

0.4451

***

-3.9116

***

0.4755

43

Table 5. Panel b. Large NTBFs

Model I
Ii,t1/Ki,t2
(Ii,t1/Ki,t2)

CFi,t/Ki,t1
PUB_FINi,t-1
PUB_FINi,t-1(CFi,t/Ki,t1)
Si,t/Ki,t1
(Di,t/Ki,t1)

Model III

Model IV

-0.0103

-0.0683

-0.0444

-0.0558

(0.1539)

(0.1537)

(0.1517)

(0.1575)

0.0116

0.0170

0.0151

0.0160

(0.0191)

(0.0187)

(0.0188)

(0.0193)

0.0871

0.0927

0.1077

0.1007

(0.1369)
-0.3354

(0.1324)
-0.5155 *

(0.1294)
-0.2894

(0.1355)
-0.3269

(0.2380)

(0.2668)

(0.2422)

(0.2473)

0.0460
(0.2278)

0.1403
(0.2764)

0.3404
(0.2272)

0.0397
(0.2299)

0.0151

0.0151

0.0155

(0.0096)

(0.0101)

(0.0101)

(0.0101)

-0.0000

0.0001

0.0001

0.0000

(0.0003)

(0.0003)

(0.0003)

(0.0003)

0.0180
2

Model II

Inverse Mill's Ratio

0.4924
(0.3884)

Constant

Investment-cash flow sensitivity of

subsidised firms (Wald test):
Short-term effect (i.e.1 year ahead)
No. of observations
Hansen

0.1833

0.2678

0.9637

0.2465

(0.2075)

(0.2135)

(0.6945)

(0.2163)

0.1331

0.2330

0.1417

0.1403

(0.2085)

(0.2705)

(0.2101)

(0.2101)

750

750

750

750

150.86

135.68

145.19

146.85

[142]

[119]

[141]

[141]

AR(1)

-3.5861 ***

-3.5126 ***

-3.5465 ***

-3.5210

AR(2)

-1.1304

-1.2489

-1.2137

-1.2251

***

44

Table 5. Panel c. Small NTBFs

Model I
Ii,t1/Ki,t2
(Ii,t1/Ki,t2)

CFi,t/Ki,t1

0.0492

0.0557

(0.0940)

(0.0963)

(0.0938)

(0.0899)

-0.0024

-0.0034

-0.0023

-0.0030

(0.0147)

(0.0150)

(0.0147)

(0.0143)

***

0.8403

(0.4434)
-0.8718

**

(0.4159)
Si,t/Ki,t1
(Di,t/Ki,t1)

0.0240
2

Model IV

0.0590

(0.1375)

PUB_FINi,t-1(CFi,t/Ki,t1)

Model III

0.0511

0.4805

PUB_FINi,t-1

Model II

0.4704

***

***

0.4733

***

(0.1431)

(0.1397)

0.8736

0.8420

(0.5670)

(0.4539)

(0.4532)

-0.8624 **

-0.8874 ***

(0.4050)

(0.4339)

-1.5966

**

(0.7980)
***

0.4840

0.0275

***

0.0244

(0.1413)
*

***

0.7644

0.0277

(0.0091)

(0.0094)

(0.0090)

(0.0094)

-0.0003

-0.0003

-0.0003

-0.0004

(0.0002)

(0.0002)

(0.0002)

(0.0002)

Inverse Mill's Ratio

***

0.0895
(0.2633)

constant

Investment-cash flow sensitivity of

subsidised firms (Wald test):
Short-term effect (i.e.1 year ahead)
No. of observations
Hansen

0.0535

0.0187

0.1637

0.0120

(0.1468)

(0.1540)

(0.3748)

(0.1455)

-0.3913

-1.1263

-0.3784

-0.4141

(0.3903)

(0.8010)

(0.3803)

(0.4110)

749

749

749

749

131.07

112.52

129.75

125.83

[140]

[119]

[139]

[139]

AR(1)

-2.6515

AR(2)

1.2954

***

-2.6373
1.2724

***

-2.6547 ***

-2.6367 ***

1.2884

1.2828

Note: standard errors in parentheses; degrees of freedom in square brackets. ***, ** and * indicate, respectively, significance levels of <1%,
<5% and <10%. All estimates include year dummies (coefficients are omitted from the table). Estimates are derived from two-step system
GMM with finite sample correction (Windmeijer, 2005). AR(1) and AR(2) are tests of the null hypothesis of, respectively, no first- and
second-order serial correlation. Hansen is a test of the validity of the overidentifying restrictions based on the efficient two-step GMM
estimator. The linear test for cash flow shifts is defined in Section 4; its p-value is based on the delta method. I measures firms
investments in fixed and intangible assets. K is the book value of tangible and intangible assets. CF is firms cash flows after taxes and
before dividends. S is firms sales. D is firms total current and long-term debt. PUB_FINit-1 is a time-varying dummy variable equal to 1 if
firm i received public finance in year t-1. The dependent variable is the ratio between period t investments in fixed and intangible assets and
the book value of total assets at the beginning of period t. Lagged dependent variables, CF/K, S/K, D/K, PUB_FINit-1 and relative interaction
terms are treated as endogeneous. Instruments start from t-2 and are limited to t-4. Additional exogenous instruments are the availability of
public finance in the geographical area and in the industry in which the firm operates (PUB_FINArea, PUB_ FINSector) and the ratio
between the amount of yearly subsidies transferred to the private sector by the Italian government and GDP (SUB_GDP). Small and large
firms are, respectively, below and above sample median for total assets. All ratios are winsorised at the 2% threshold.

45

Table 6. Short-term and persistent effect of public finance on the investment rate and
investment-cash flow sensitivity of small NTBFs subsidised only once
Short-term effect only
(eq. 2)
0.0949
0.0944

Ii,t1/Ki,t2
(Ii,t1/Ki,t2)

CFi,t/Ki,t1

(0.0981)

(0.9068)

(0.0987)

(0.0972)

-0.0066

-0.0066

-0.0070

-0.0069

(0.0148)

(0.0147)

(0.0148)

(0.0147)

0.4314

***

(0.1319)
PUB_FINi,t-1

0.9854
-1.2075

0.4323

***

(0.1333)
*

(0.5441)
PUB_FINi,t-1(CFi,t/Ki,t1)

Short-term and persistent

effect (eq. 3)
0.0995
0.0984

1.0047

(0.5431)
*

(0.6983)

-1.2152

(0.6950)

(Di,t/Ki,t1)

**

Investment-cash flow sensitivity of

subsidised firm (Wald test):
short-term effect (i.e.1 year ahead)

**

Hansen test

-1.2885

1.0786

(0.8031)

0.2922

0.3243
(0.3947)

-0.3632

-0.3608

(0.5974)

(0.6734)
**

0.0218

(0.0090)

(0.0088)

-0.0002

-0.0002

-0.0002

-0.0002

(0.0002)

(0.0002)

(0.0002)

(0.0002)

0.0204

0.0183

(0.2700)

(0.2720)

0.0417

0.0636

0.0492

0.0643

(0.1290)

(0.3833)

(0.1311)

(0.3850)

-0.7761

-0.7829

-0.8587

-0.8811

(0.7050)

(0.7014)

(0.7623)

(0.7966)

0.0666

0.0702

(0.5558)

(0.6595)

732

732

732

732

129.87

129.03

128.14

126.44

[136]

[135]

[138]

[137]

AR(1)

-2.5913

AR(2)

1.2767

***

-2.5968
1.2764

***

-2.5964
1.2776

**

-1.3121

(0.3995)

0.0218

***

(0.5399)

(0.0087)

Investment-cash flow sensitivity of

subsidised firms (Wald test):
persistent effect
No. of observations

**

(0.0090)

Inverse Mill's Ratio

Constant

0.0219

1.0549

0.4311
(0.1353)

(0.7786)

PUB_FIN_pers,t-1(CFi,t/Ki,t1)
0.0218

***

(0.5218)

PUB_FIN_persi,t-1

Si,t/Ki,t1

0.4298
(0.1346)

***

-2.6010

**

***

1.2766

Note: standard errors in parentheses; degrees of freedom in square brackets. ***, ** and * indicate, respectively, significance levels of <1%,
<5% and <10%. All estimates include year dummies (coefficients are omitted from the table). Estimates are derived from two-step system
GMM with finite sample correction (Windmeijer, 2005). AR(1) and AR(2) are tests of the null hypothesis of, respectively, no first- or
second-order serial correlation. Hansen is a test of the validity of the overidentifying restrictions based on the efficient two-step GMM
estimator. The linear test for cash flow shifts is defined in Section 5; its p-value is based on the delta method. I measures a firms
investments in fixed and intangible assets. K is the book value of tangible and intangible assets. CF is firms cash flows after taxes and
before dividends. S is firms sales. D is firms total current and long-term debt. PUB_FINit-1 is a time-varying dummy variable equal to 1 for
firms i that have received public finance at year t-1. PUB_FIN_persit-1 is time-varying dummy variable equal to 1 for firms i that have
received public finance at year t-2. The dependent variable is the ratio between period t investments in fixed and intangible assets and the
book value of total assets at the beginning of period t. Lagged dependent variables, CF/K, S/K, D/K, PUB_FINit-1, PUB_FIN_persit-1 and
relative interaction terms are treated as endogeneous. Instruments start from t-2 and limited to t-4. Additional exogenous instruments are the
availability of public finance in the geographical area and in the industry in which the firm operates (PUB_FINArea,PUB_ FINSector) and
the ratio between the amount of yearly subsidies transferred to the private sector by the Italian government and GDP (SUB_GDP). The
sample include small firms (below sample median for total assets) that received public finance once during their life. All ratios are
winsorised at the 2% threshold.

46

Table 7. Investment rate at equilibrium of subsidised and non-subsidised small NTBFs

Non-subsidised firms
Subsidised firms
Increase in comparison with
non-subsidised firms
Subsidised firms:
hypothetical persistent effect
equal to the short-term effect
Increase in comparison with
non-subsidised firms

Cash flow rate

10
I quartile
Median
percentile
(CF/K=0.2
(CF/K
(CF/K=0.0
364)
=0.4792)
007)
0.4447
0.5532
0.6680

Average III quartile Difference

(CF/K=0.7 (CF/K=
III vs I
791)
0.9306)
quartile
0.8095

0.8809

0.3277***
0.0507

0.7632

0.7780

0.7977

0.8196

0.8307

0.3184

0.2268

0.1298

0.0101

-0.0502

1.5872

1.3740

1.1469

0.8653

0.7227

1.1425**

0.8209*

0.4789

0.0558

-0.1582

Note: Equilibrium investment rate are calculated according to equation (A3) for non-subsidised firms and (A6) and (A10) for firms who
received public finance, respectively using estimated persistent effect and assuming the short-term effect as persistent. The difference
between investment rate of subsidised and non-subsidised firms is estimated by equations (A7) and (A11), respectively. ***: significant at
1%; **: significant at 5%; *: significant at 10%.

47

48