OUTPUT ADDITIONALITY OF PUBLIC SUPPORT FOR
INNOVATION:
EVIDENCE FOR IRISH MANUFACTURING PLANTS
Working Paper No. 103
June 2009
Nola Hewitt-Dundas and Stephen Roper
Warwick Business School’s Small and Medium Sized Enterprise Centre
Working Papers are produced in order to make available to a wider public,
research results obtained by its research staff. The Director of the
CSME, Professor Stephen Roper, is the Editor of the Series. Any
enquiries concerning the research undertaken within the Centre should
be addressed to:
The Director
CSME
Warwick Business School
University of Warwick
Coventry CV4 7AL
e-mail stephen.roper@wbs.ac.uk Tel. 024 76 522501
ISSN 0964-9328 – CSME WORKING PAPERS
Details of papers in this series may be requested from:
The Publications Secretary
CSME
Warwick Business School,
University of Warwick,
Coventry CV4 7AL
e-mail sharon.west@wbs.ac.uk Tel. 024 76 523692
1
Output Additionality of Public Support for Innovation:
Evidence for Irish Manufacturing Plants
Nola Hewitt-Dundasa and Stephen Roperb
a
b
Queen's University Management School, Queen‟s University Belfast,
Belfast
Centre for Small and Medium Sized Enterprises, Warwick Business School,
Warwick
Corresponding Author:
Nola Hewitt-Dundas,
Queen‟s University Management School,
25 University Square,
Belfast, UK
BT7 1NN
E-mail: nm.hewitt@qub.ac.uk
Tel: +44 (0) 2890 975155; Fax: +44 (0) 2890 975156
Acknowledgements
The authors wish to acknowledge the valuable comments received from delegates at
the 5th British/Irish and Israeli Regional Science workshop - Pushing forward the
Frontiers, and the Higher Education Authority, Ireland for providing financial support for this
research.
2
Abstract
Public support for private R&D and innovation is part of most national and regional
innovation support regimes. In this paper we estimate the effect of public innovation
support on innovation outputs in Ireland and Northern Ireland. Three dimensions of
output additionality are considered: extensive additionality, in which public support
encourages a larger proportion of the population of firms to innovate; improved
product additionality in which there is an increase in the average importance of
incremental innovation; and, new product additionality in which there is an increase
in the average importance of more radical innovation. Using an instrumental variables
approach, our results are generally positive, with public support for innovation having
positive, and generally significant, extensive, improved and new product additionality
effects. These results hold both for all plants and indigenously-owned plants, a
specific target of policy in both jurisdictions. The suggestion is that grant-aid to firms
can be effective in both encouraging firms to initiate new innovation and improve the
quality and sophistication of their innovation activity. Our results also emphasise the
importance for innovation of in-house R&D, supply-chain linkages, skill levels, and
capital investment all of which may be the focus of complementary policy initiatives.
Keywords: Innovation, policy, additionality, evaluation, Ireland
JEL Codes: O38, O31, H25, L60
3
Output Additionality of Public Support for Innovation: Evidence for Irish
Manufacturing Plants
1. Introduction
Recent reviews of innovation policy regimes in different countries, have emphasised
the wide variety of approaches being used to support firms‟ innovation activity. EU
(2003), for example, suggests that the budget allocation for innovation support in the
UK mirrors relatively closely that in the US, with support balanced roughly equally
between fiscal incentives for innovation, subsidy measures and „integrated packages
of support‟. Other countries adopt different approaches with Finland emphasising
direct support measures (subsidies and loans), and France placing more emphasis on
direct credit and loan support1. In each case, however, grant-based incentives to
support R&D and innovation remain important, particularly as a means of stimulating
innovative activity among smaller firms, those innovating for the first time or those in
more rural or peripheral areas (Oughton et al., 2002). This reflects general arguments
that these types of firms are likely to be the most resource constrained. Or, that such
firms are more likely than most to experience market failures in terms of, say, the
availability of finance for innovation (e.g. Martin and Scott, 2000)2.
1
In the UK, for example, 29 per cent of the innovation budget comprises subsidy or grant schemes
compared to 22 per cent in the Netherlands, 25 per cent in France, 47 per cent in Finland and 42 per
cent in the US. New Zealand is a marked exception with almost no innovation grants and innovation
support offered primarily through fiscal measures (EU, 2003, Table 5).
2
For a more general discussion of the market failure justification for innovation policy see, for
example, Metcalfe (1997).
4
Support for the value of public investment in this area comes from a number of
studies which suggest a degree of additionality from public support for private R&D
activity and positive effects on business performance (e.g. Griliches, 1995;
Mamuneas and Nadiri, 1996)3. This effect can operate through a number of different
organisational mechanisms, however (Roper et al., 2004). First, and most obviously,
public support for private R&D may reduce the cost to firms of building up their
knowledge stocks (e.g. Trajtenberg, 2000), enhancing business performance (Klette
and Johansen, 1998) and firms‟ ability to conduct future research projects (e.g.
Mansfield and Switzer, 1984; Luukkonen, 2000)4. Second, public support for R&D
activity may contribute to developments in firms‟ human resources and innovation
activity (e.g. Freel, 2005)5. Third, public support for R&D or innovation may improve
firms' ability to absorb R&D results or knowledge from elsewhere (Veugelers and
Cassiman, 1999; Cassiman and Veugelers, 2002)6. Fourth, reputational or 'halo'
effects may also stem from receipt of public R&D support7. Fifth, public funding of
3
This clearly depends on the extent of additionality in publicly financed, private R&D, on which see
Griliches (1995). Interestingly, for the US, Mamuneas and Nadiri (1996) identify some differences
between sectors in this respect, finding a substitute relationship in low technology industries and a
weak substitute relationship in high-tech sectors. Thus there is evidence of some crowding-out,
particularly in the low-tech sectors. See also Goolsbee (1998), David et al. (2000).
4
Luukkonen (2000) also indicates that participation in the collaborative EU Framework programmes
by Finnish firms laid the basis for future R&D by contributing to firms' involvement and influence in
standards negotiations, viz. participation 'provided background information for standardisation
negotiations … [and] … facilitated their contacts, since the experts of the companies could get better
acquainted with each other, which again helped their interactions. It was a question of an intangible
impact' (p. 716)
5
Sakakibara (1997) p. 462, for example, indicates that the managers of publicly supported
collaborative R&D projects in Japan rated researcher training as the most important benefit which their
companies derived from their project. Somewhat surprisingly, this 'intangible' benefit from
collaborative R&D was seen as more important than 'increase in the awareness of R&D in general',
'breakthrough in a critical technology', and 'accelerated development of the technology'.
6
More specific evidence of the complementarity of publicly supported R&D and firms' other
internally-funded R&D activity comes from Ballesteros and Rico (2001). They conduct an
econometric analysis of the Spanish 'Concerted Projects' support scheme and demonstrate that Spanish
firms which were more R&D intensive were also more likely to make use of government funding for
collaborative university-business R&D projects.
7
Powell (1998), for example, points out, in the context of R&D collaboration in technology intensive
industries, that 'a firm's portfolio of collaborations is both a resource and a signal to markets, as well as
to other potential partners, of the quality of the firm's activities and product'. (p.231).
5
R&D may also create the potential for R&D cost savings through collaborative R&D
and the sharing of research results8.
A key issue, regardless of which of these mechanisms predominates, is the extent to
which any increment in R&D or innovation activity is „additional‟, i.e. generates a
positive and significant average treatment effect (ATE). In terms of public support for
innovation, however, it is arguable whether any single ATE estimate can capture the
range of possible impacts. Public support for innovation may, for example, encourage
firms to both commence innovation activity and increase the scale of their innovation
investments. To capture this potential range of effects we estimate three notions of
the additionality of public support for innovation outputs: extensive additionality
represented by the ATE on the probability of undertaking innovation; improved
product additionality represented by the ATE of public support on firms‟ incremental
innovation; and, new product additionality represented by the ATE of public support
for innovation on firms‟ radical innovation. Our estimation approach – enabled by the
availability of suitable instruments – follows an instrumental variables approach.
The remainder of the paper is organised as follows. Section 2 outlines the economic
and policy setting for our analysis which focuses on the impact of public support for
innovation in Ireland and Northern Ireland over the 1994 to 2002 period. This is
important in the context of the current paper both because our key focus is on the
effectiveness of policy intervention but also because systemic approaches to
innovation stress the importance of the historical, economic and institutional context
8
Irwin and Klenow (1996), for example, examined the Sematech collaborative R&D facility, set up by
the US semiconductor industry in 1987 with substantial financial support from government. They
compared the R&D intensity of Sematech member and non-member companies, and conclude that
participation in the Sematech collaboration reduced members' R&D spending by 9 per cent.
6
within which innovation takes place (e.g. Edquist, 2004). Section 3 outlines our
conceptual approach reflecting plants‟ decision to proceed with an innovation project.
Section 4 describes our data sources and empirical methodology, and Section 5
summarises our empirical results. Conclusions and implications are the focus of
Section 6.
2. Economic and Policy Context
Over the period considered here (1994 to 2002), Ireland and Northern Ireland
experienced very different patterns of economic growth, suggesting marked changes
in the market incentives for investing in R&D and innovation9. In Ireland – the Celtic
Tiger – GDP grew by an average of 7.2 per cent per year from 1990 to 2000, while
real GNP grew by 6.3 per cent per year, largely due to continued inward investment
and re-investment in the high-tech sectors10. For Northern Ireland, the 1990s was
marked by more steady output growth of 3.2 per cent per year from 1990 to 1998, but
this still compared favourably to growth rates in the UK (2.1 per cent pa) and EU (1.8
per cent pa) during the same period. (e.g. Morahan, 2002) 11. From 2000 to 2002, the
economic situation looked very different, however, with manufacturing output
actually falling in Northern Ireland, while the rate of growth of output in Ireland
slowed considerably.
Our study period was also marked by significant changes in R&D and innovation
policy in both Northern Ireland and Ireland, although in both areas the innovation
policy regime remained strongly interventionist, justified in terms of broadly defined
9
For a detailed comparison see O‟Malley and Roper, 2004.
GDP is conventionally used in making international comparisons. However, in the case of the
Republic of Ireland GNP is generally regarded as more meaningful, since it excludes the substantial
profits of foreign multinational companies that are withdrawn from the country.
11
UK and EU data from OECD Historical Statistics 1970-1999.
10
7
economic development objectives12. In Ireland, for example, the landmark Culliton
Report of 1992, argued that “without state support and incentives the degree of
investment in technology will be less than is desirable from the point of view of
national economic development” (Industrial Policy Review Group, 1992, p. 55).
Essentially similar policy rationales were evident in Northern Ireland, with both
government reports (e.g. IRTU, 1992) and academic studies (e.g. Harris and Trainor,
1995) reflecting low levels of R&D and innovation and the perceived need for public
intervention.
EU Objective 1 status – which benefited both Ireland and Northern Ireland
throughout most of the period considered here - was also an important influence in
the support regime for innovation in both areas. In Ireland, for example, the
Operational Programme for Industrial Development, 1989-93 provided funding for
capability development, while the subsequent 1994-99 Operational Programme had a
specific sub-programme for research and development13. In Northern Ireland too, EU
funding was an important component of public support for R&D and innovation,
directly funding infrastructure projects as well as providing co-funding for a number
of regional innovation support programmes14. Developments in R&D and innovation
support later in the 1990s continued the focus on developing innovation capability in
12
Although see Lenihan, Hart and Roper (2005) on the ambiguity of objectives of much Irish
industrial policy and consequent difficulties in ex post evaluation.
13
Cogan and McDevitt (2000: 11) describe EU involvement in R&D in Ireland as having been of
„critical importance‟ to Irish S&T policy. They describe three benefits of the EU involvement. Firstly,
they cite the organisational and institutional learning it engendered. They state, rather philosophically,
that Ireland missed out on the industrial revolution and somehow expected to catch up with other
nations by using imported innovation and without building up a domestic innovation and R&D
capability. Secondly, the EU structural funds brought with it a disciplined evaluation of policy,
something which was missing from policy prior to this. Thirdly, rather than concentrating on research
that had little bearing on Irish industry, the Structural Funds were geared towards stimulating a selfsustaining capacity for innovation.
14
See, for example, Roper (1998) on Compete and Roper (2001) for a more detailed overview of the
innovation support regime in Northern Ireland.
8
indigenously-owned firms. In Ireland, the RTDI scheme was launched in 1997, for
example, with wide ranging objectives one of which was to introduce firms to R&D
and innovation for the first time15. A similar emphasis has also characterised policy
priorities in innovation support in Northern Ireland, with a focus on encouraging
firms to engage in R&D and innovation for the first time. The Compete programme,
for example, which provides support for near-market innovation „has always attracted
significant interest from companies engaging in R&D for the first time and 54 per
cent of … applications to the programme were first time users‟ (Invest NI, 2003, p.
4).
Changes in the innovation support regimes of Ireland and Northern Ireland, and their
increasing focus on developing innovation capacity in indigenously-owned firms,
were also reflected in changes in the overall levels of government investment in R&D
and innovation support. In Ireland, government support for R&D and innovation
activity grew steadily through the 1990s from €5 m pa to around € 30 m pa (Figure
1), while that in Northern Ireland remained relatively stable at an annual overall level
of around €25 m pa.
---Figure 1 here --This changing economic and policy context has important implications for our
assessment of the effectiveness of innovation support measures. First, spatial
differences in the economic and policy context in Ireland and Northern Ireland
suggest the potential importance of allowing for location in our modelling strategy.
Second, the priority given to supporting innovation in particular groups of firms, e.g.
those engaging in R&D or innovation for the first time, is likely to introduce a
15
Following some institutional changes in 1998, the RTI scheme was implemented through Enterprise
Ireland, the agency specifically charged with developing the capacity of indigenous Irish firms.
9
potential bias into OLS estimates of the ATE suggesting the value of an instrumental
variables approach (e.g. Madalla, 1993; Wooldridge, 2002). This is therefore
explicitly allowed for in our modelling approach. Third, the focus of innovation
support policy in Ireland and Northern Ireland on indigenously-owned firms suggests
this group may be of particular interest in assessing policy additionality.
3. Conceptual Framework
Our objective here is to develop a framework within which the firm-level
additionality of public support for innovation can be examined. This may involve the
decision of a plant either to abandon, go ahead with, or modify its innovation
investment decisions in the light of the availability of public support. Focussing
specifically on product innovation, three dimensions of additionality are considered:
extensive additionality, in which a larger proportion of the population of plants are
innovating; improved product additionality in which there is an increase in the
average importance of incremental innovation; and, new product additionality in
which there is an increase in the average importance of radical innovation. In each
case the availability of public support for innovation might positively influence the
innovation decision by either changing the balance of costs and anticipated revenue
from innovation or by reducing either the technical or commercial risk. Standard
optimising conditions (see, for example, Levin and Reiss, 1994) would then suggest
that public support for innovation should have positive extensive, improved and new
product additionality effects16.
To evaluate the empirical significance of this proposition for public support for
innovation in Ireland and Northern Ireland we make use of the now familiar notion of
16
This is consistent with the broad empirical evidence reviewed by Griliches (1995).
10
an innovation or knowledge production function (e.g. Geroski, 1990; Harris and
Trainor, 1995; Love and Roper, 2001). This relates innovation outputs of various
sorts to knowledge inputs with the argument here being that, where it generates
additionality, public support for innovation will increase innovation outputs given any
level of innovation costs and benefits. In other words, if I is an indicator of plants‟
innovation output, the innovation production function can be stated as:
I   x 
'z 
(1)
where x is a vector of plant level control variables, and z is a binary treatment variable
taking value 1 if a firm received public support for innovation and 0 otherwise. In this
model the size, sign and significance of the coefficient on the treatment term (i.e. )
will give an indication of the additionality of public support. More specifically we
consider here three variants of equation (1) each of which relates to one of the notions
of additionality defined earlier. To examine the notion of extensive additionality we
consider a version of (1) in which the dependent variable is a simple binary indicator
of whether or not firms are undertaking innovation. Here, the question is whether
public support for innovation has significantly increased the probability that a firm
will be engaging in either improved or new product innovation. To examine improved
product additionality and new product additionality we consider models in which the
dependent variables reflect the importance of improved products and new products in
plants‟ sales. In each case additionality requires public support to be having a positive
and significant effect on plants‟ sales of either improved or new products.
In addition to the treatment terms we include in each model a set of plant-level
controls comprising three groups of variables which have been shown to be important
in other studies of innovation activity. These relate to plants‟ knowledge sourcing
11
activity, their market position and their resource base. Measuring the intensity of
plants‟ knowledge sourcing through R&D is relatively straightforward with standard
indicators (used by Crépon et al., (1998), Lööf and Heshmati (2001, 2002), and Love
and Roper (2001)) measuring R&D employment relative to total employment.
Measuring the intensity of knowledge sourcing through plants‟ supply-chain and non
supply-chain innovation linkages is more experimental, and here we follow Love and
Roper (1999, 2001) who develop intensity scores for the extent of plants‟ external
contacts17. Market position indicators include a size variable (employment) and its
square, reflecting the frequent finding of a quadratic relationship between size and
innovation, and dummy variable indicators reflecting the nature of plants‟ production
process (i.e. small batches, large batches or continuous production). The strength of
each plant‟s internal resource base is proxied by a variety of measures, including
whether or not the plant is part of a multi-plant group, whether it is externally owned
and whether there is any R&D relevant to the plant carried out elsewhere within the
group. These measures of intra-group resources have proved important in previous
research on innovation (Love et al., 1996; Love and Roper, 1999, 2001). Indicators
of human capital and physical capital inputs are also included (percentage of
workforce with a degree and percentage with no qualifications; capital investment
relative to turnover).
4. Data and Methods
17
More specifically, we construct intensity scores for each plant‟s knowledge sourcing through supplychain and non supply-chain collaboration based on the number of types of organisation with which the
firm is undertaking collaborative innovation activity. For example, we identify five types of potential
supply-chain partners (customers, suppliers, competitors, other group companies and joint ventures):
plants undertaking innovation collaboration with three of these types of partner 'score' 60 per cent;
plants collaborating with all five types of partner score 100 per cent.
12
Our empirical analysis is based on data from the Irish Innovation Panel (IIP) which
provides information on knowledge use, government support and the innovation
performance of manufacturing plants throughout the Ireland and Northern Ireland
over the period since 1991. The IIP comprises linked surveys conducted using similar
postal survey methodologies, sampling frames provided by the economic
development agencies in Ireland and Northern Ireland, and questionnaires with
common questions. Each survey covers the innovation activities of manufacturing
plants with 10 or more employees over a three year period with an average response
rate of 34.5 per cent18. In the current analysis to allow greater variable choice, we use
data only from the second, third and fourth surveys. Reflecting the dimensions of
additionality discussed earlier three innovation output indicators are of interest here:
the probability that a plant introduced either new or improved products over the
previous three years; the importance of new and improved products in plants‟ sales;
and, the proportion of new products in plants‟ sales. The former is intended to reflect
the additional effect of public support on incremental innovation, and the latter more
radical product change. Over the sample period (1994 to 2004), 56.4 per cent of
plants introduced either new or improved products in Northern Ireland (Ireland, 62.4
per cent), while 12.5 per cent of sales were accounted for by new products and 25.7
per cent (Ireland, 29.0 per cent) by new and improved products (Annex 1). Over the
same period 23.1 per cent of plants in Northern Ireland received public support for
their product development activities compared to 25.6 per cent in Ireland (Annex 1).
In each case we are interested in estimating the average treatment effect, but
recognise that there are likely to be aspects of selectivity in the receipt of public
18
Details of each wave of the survey can be found in Roper et al. (1996), Roper and Hewitt-Dundas
(1998), Roper and Anderson (2000), Roper et al., 2004).
13
support. In other words, as the previous discussion suggested, public support for
innovation may be targeted specifically at smaller plants, or those which are
innovating for the first time. In addition, it is possible that elements of endogeneity
may exist in plants‟ receipt of public support. If development agencies adopt a
„backing winners‟ strategy for example, innovating plants may be more likely than
non-innovators to receive public support. In either situation direct estimation of the
treatment effect is likely to lead to biased estimates of the scale of the treatment effect
(see Maddala, 1993, pp. 257-290 for a general discussion). In the estimation we
therefore adopt an instrumental variables approach, instrumenting plants‟ receipt of
public support for product development by their receipt of support for process
development, R&D and capital investment (Annex 1). In each model instruments are
chosen to comply with the ignorability assumption with the exogeneity of instruments
being verified by a Wald test. Models were estimated using the IV probit and IV
Tobit procedures implemented in Stata v9.
5. Empirical Results
Our results divide naturally into those related to the additionality of public support for
product innovation in all plants, and those related specifically to policy impacts on
indigenously-owned plants. As suggested earlier, the latter group is of particular
interest as it was a specific focus of policy in both Ireland and Northern Ireland over
the whole of the study period.
Policy effects on all plants
IV probit estimates of the extensive additionality effect of public support for product
development are given in Table 1 for Northern Ireland and Table 2 for Ireland.
14
Positive and significant ATEs – extensive additionality – is observed in Northern
Ireland with a positive but insignificant ATE in Ireland. Other factors also prove
important in determining the probability of innovating, with R&D and supply-chain
linkages having positive and significant effects in both Ireland and Northern Ireland,
but little evidence of any important non-supply chain linkages. This reflects the
findings of other survey-based examinations of innovation which typically stress
internal R&D as an element of knowledge creation and absorptive capacity (e.g.
Griffith et al., 2003) and the importance of supply-chain knowledge linkages (HewittDundas et al., 2002). Skill levels (e.g. Freel, 2005), capital intensity, and the nature of
the plant‟s production activities also prove important in determining the probability of
innovating in both Northern Ireland and Ireland. Interestingly, however, we observe
no significant scale, ownership or plant vintage effects on the probability of
innovation in either jurisdiction.
---Table 1 here --IV Tobit estimates of improved product additionality and new product additionality
for Northern Ireland and Ireland are given in Tables 1 and 2 respectively. In each case
the results are strongly reminiscent of those for extensive additionality, with a
positive and significant ATEs in Northern Ireland and positive but insignificant ATEs
in Ireland. Again R&D and supply-chain linkages also prove important in
determining the average level of improved and new product innovation although we
observe no positive non-supply chain linkages. Perhaps the most disappointing
element of this latter result is the lack of any clear innovation effect from investments
in public knowledge sources (e.g. universities, R&D laboratories) in Ireland and
Northern Ireland (although see Wrynn, 1997). In terms of improved product
innovation we also observe only weak scale effects, although plant vintage, the nature
15
of plants production activities and skill levels remain important (Tables 1 and 2). In
terms of new product innovation, scale effects – which we equate with plant‟s
resources – prove more important, with consistent skills and investment effects.
---Table 2 here ---
Table 3, part A summarises the ATEs of public support for product development in
all plants. Effects are universally positive and more strongly significant in Northern
Ireland than in Ireland. The implication being that the extensive, improved product
and new product effects of policy in Northern Ireland were generally stronger than
those achieved in Ireland over this period. It is difficult from our econometric
analysis to be sure why this pattern of policy effects emerges but three possible
explanations are suggested by our descriptive statistics. First, it is clear that in general
levels of innovation activity in the population of plants were generally lower in
Northern Ireland than Ireland over the sample period (Annex 1). This means that a
larger proportion of plants in Northern Ireland were engaged in either no innovation
or incremental innovation than in Ireland, increasing the potential for additionality
among these firms. Other potential explanations relate more directly to the
consistency and level of public support for innovation in Ireland and Northern Ireland
(Figure 1). In general terms, levels of public support in Northern Ireland were more
consistent than those in Ireland, creating perhaps a more stable planning environment
in which plants were more confident about developing innovation projects. Possibly
more important, however, is that on a per employee basis public support for
16
innovation in Northern Ireland (€225 per employee per annum) was more than double
that in Ireland (€105 per employee per annum) over the 1994-2001 period19.
----------------- Table 3 here --------------------Additionality in Indigenously-owned plants
Indigenously-owned plants in Ireland and Northern Ireland tend to be smaller than
their externally-owned counterparts, more concentrated in more traditional sectors
and less innovation active than externally-owned plants (e.g. Roper et al., 2004).
Moreover, as noted earlier continuing doubts about the technological capabilities of
indigenously-owned Irish businesses have made such plants a key focus of policy
interest. We might therefore expect both a concentration of public support for
innovation among these firms as well as, potentially, more significant ATEs. In fact,
however, the proportions of indigenously-owned plants in both Ireland and Northern
Ireland in our sample who received public support for product development are lower
than those among externally-owned plants. In Ireland the proportion of indigenouslyowned plants receiving public support for product development was 23.2 per cent
(externally-owned 23.5 per cent). In Northern Ireland, a larger difference between
groups was evident with 19.7 per cent of indigenously-owned plants receiving public
support for product development compared to 27.5 per cent of externally-owned
plants.
Estimating similar models to those in Tables 1 and 2 for indigenously-owned firms
only we observe positive and significant ATEs for the probability of undertaking
19
Average annual grant expenditure in Northern Ireland over the 1994-2001 period was €22.6m
compared to an average of €28.31 in Ireland. Manufacturing employment in Ireland in 2001 was
268,000 and in Northern Ireland was 100,186. Sources: Figure 1, Department of Trade, Enterprise and
Investment, Belfast; Central Statistical Office, Dublin.
17
product innovation by indigenously-owned plants in both areas (Table 3)20. In other
words, for this group of plants public support for product development was having
positive extensive additionality. Other factors also prove important, however, with inhouse R&D and supply-chain linkages again positive and significant along with skill
levels and the production orientation of the plants. Less important were business size,
vintage and whether or not the plant was part of a multi-plant group. Essentially
similar results are observed with respect of improved product additionality with
positive and significant ATEs and a broadly similar range of other factors proving
important in determining the extent of plants‟ incremental innovation activity. In
terms of new product additionality, we also observe positive ATEs from public
support for product development in Ireland and positive and (marginally)
insignificant results in Northern Ireland (Table 3).
In short, therefore, public support for product development in indigenously-owned
plants in both Ireland and Northern Ireland was having generally positive and
significant extensive, improved and new product additionality effects. The
consistency of the other factors which support innovation in these firms – in-house
R&D, supply-chain linkages, skills also emphasise the potential value of other types
of policy initiatives in supporting innovation in this group of plants. Freel (2005), for
example, discusses the role of training and workforce skills as an enabler of
innovation, a relationship strongly evident in our data. Other measures, applied with
differential emphasis in Ireland and Northern Ireland have been support for capital
investment (see, for example, Fielding (2003) on Northern Ireland), and attempts to
20
Full details of these models for indigenously-owned firms are available from the authors on request.
18
develop supply chain linkages such as the Irish National Linkages Programme. Our
results suggest that both are also likely to be important as innovation enablers.
19
6. Conclusions
Our analysis, based on data for manufacturing plants in Ireland and Northern Ireland
suggests the positive output effects of public support for product development. For
both all plants, and the more specific target group of indigenously-owned plants, we
see positive and generally significant ATEs suggesting positive extensive, improved
product and new product additionality effects. This is consistent with more general
evidence of the positive additionality effects of public support for R&D cited earlier
(e.g. Griliches, 1995; Mamuneas and Nadiri, 1996), and is clearly a welcome policy
message given the ubiquity of public support for innovation activity. Our results also
provide further support for the importance of external knowledge sources for
innovation. A clear hierarchy emerges, however, with in-house R&D and supplychain collaboration both more important than non-supply chain collaboration. In
addition to these knowledge sourcing variables, we find strong evidence that plants‟
innovation activities depend on their organisational context, skill-base, and capital
investment.
A number of policy conclusions follow from our analysis. First, it is clear that public
support for product development can be effective in both encouraging plants to
undertake innovation for the first time (extensive additionality) and improving the
quality of plants‟ innovation (improved and new product additionality). Investment in
such public support measures may therefore be seen as worthwhile either in isolation
or as part of a package of innovation support measures designed to overcome
perceived barriers to innovation (e.g. Hewitt-Dundas, 2006). Second, our analysis
again emphasises the importance of boundary spanning links for innovation. Policy
measure to support collaboration – particularly along the supply-chain - are therefore
20
likely to have positive innovation benefits by encouraging knowledge diffusion.
Third, measures designed to strengthen plants‟ internal R&D capability are likely to
contribute directly to innovation but also to increase absorptive capacity, allowing
plants to more effectively take advantage of external knowledge sources (e.g. Roper
et al., 2000; Griffith et al., 2003). Fourth, skill levels also emerge as important in
shaping plants innovation success. Measures to promote skill development are
therefore likely to be doubly effective in increasing wealth creation both by
promoting innovation directly (e.g. Freel, 2005), and by increasing the effectiveness
with which innovation activity is commercially exploited and knowledge absorbed
(Roper et al., 2008).
21
References
Ballesteros, J A and Rico, A M (2001) 'Public Financing of Co-operative R&D
Projects in Spain: The Concerted Projects under the National R&D Plan‟, Research
Policy, 30, 625-641.
Cassiman, B and Veugelers, R (2002) „Complementarity in the Innovation Strategy:
Internal R&D, External Technology Acquisition, and Co-operation in R&D‟, EARIE
Annual Conference, September 2002, Madrid.
Cogan, D., McDevitt, J. (2000) „Science, Technology and Innovation Policy &
Science and Technology Policy Evaluation: The Irish Experience‟, Paper presented at
the Converge Workshop Madrid, 5 October 2000.
Crépon, B Duguet, E and Mairesse, J (1998) 'Research, Innovation and Productivity:
An Econometric Analysis at the Firm Level', Economics of Innovation and New
Technology, 7, 115-158.
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, 497-529.
Edquist, C (2004) „Systems of Innovation – A Critical Review of The State of the
Art‟ in Handbook of Innovation, Fagerberg, J Mowery, D and Nelson, R Oxford
University Press, 2004
EU (2003) „Raising EU R&D Intensity‟, Brussels.
Fielding, D (2003) „Investment, Employment and Political Conflict in Northern
Ireland‟, Oxford Economic Papers, 55, 512-535.
Freel MS (2005) „Patterns of innovation and skills in small firms‟, Technovation, 25,
2, 123-134
Geroski, P A (1990) „Innovation, technological opportunity, and market structure‟,
Oxford Economic Papers, 42, 586-602.
Goolsbee A (1998) „Does government R&D policy mainly benefit scientists and
engineers?‟ American Economic Review, 38, 423-428.
Greene, W H (1997) „Econometric Analysis’, Macmillan, New York.
Griffith, R, S Redding, and J Van Reenan. 2003. R&D and Absorptive Capacity:
Theory and Empirical Evidence. Scandinavian Journal of Economics 105 (1):99-118.
Griliches, Z (1995) „R&D and Productivity: Econometric Results and Measurement
Issues‟, in Stoneman, P (ed.) „Handbook of the Economics of Innovation and
Technological Change‟, Blackwell, Oxford.
22
Harris R I D and Trainor M (1995) „Innovations and R&D in Northern Ireland
manufacturing: a Schumpeterian approach‟, Regional Studies, 29, 593-604.
Hewitt-Dundas, N Andreosso-O‟Callaghan, B Crone, M Murray, J and Roper, S
(2002) 'Learning from the Best: Knowledge Transfers from Multi-national Plants in
Ireland: A North-South Comparison', NIERC/Euro-Asia Centre, University of
Limerick.
Hewitt-Dundas, N. 2006. Resource and capability constraints to innovation small and
large plants. Small Business Economics 26:257-277.
Industrial Policy Review Group (1992), A Time for Change: Industrial Policy for the
1990s (The Culliton Report), Dublin: Stationery Office.
Industrial Research and Technology Unit (1992) „Innovation 2000’, Department of
Enterprise Trade and Investment, Belfast.
Invest NI (2003) Industrial Research and Technology Unit, Annual Accounts 2001 –
2002, Invest NI, Belfast.
Irwin, DA and Klenow, P J (1996) 'High-Tech R&D Subsidies - Estimating the
Effects of Sematech', Journal of International Economics, 40, 323-344.
Klette, TJ, Johansen, F (1998) „Accumulation of R&D capital and dynamic firm
performance: a not- so-fixed effect model‟, Annales d'économie et de statistique, 4950, 389-419.
Lenihan, H Hart, M and Roper, S (2005) „Developing An Evaluative Framework For
Industrial Policy In Ireland: Fulfilling The Audit Trail Or An Aid To Policy
Development?‟, ESRI Quarterly Commentary, Summer 2005, 69-86.
Lööf, H and Heshmati, A (2001) 'On the Relationship Between Innovation and
Performance: A Sensitivity Analysis', SSE/EFI Working Paper No 446, Stockholm
School of Economics
Lööf, H and Heshmati, A (2002) 'Knowledge Capital and Performance
Heterogeneity: A Firm Level Innovation Study', International Journal of Production
Economics, 76, 61-85.
Love J H, Ashcroft B and Dunlop S (1996) "Corporate structure, ownership and the
likelihood of innovation" Applied Economics, Vol 28, 737-46.
Love, J. H. and Roper, S. (1999) 'The determinants of innovation: R&D, technology
transfer and networking effects', Review of Industrial Organization, vol. 15, 43-64.
Love, J. H. and Roper, S. (2001) 'Location and network effects on innovation success:
evidence for UK, German and Irish manufacturing plants', Research Policy, 30, 64361.
23
Luukkonen, T (2000) 'Additionality of EU Framework Programmes', Research
Policy, 29, 711-724.
Maddala, G (1993) „Limited Dependent and Qualitative Variables in Econometrics’,
Cambridge University Press, New York.
Mansfield, E and Switzer (1984) 'Effects of Federal Support on Company Financed
R&D: The Case of Energy', Management Science, 30, 5, 562-571.
Mamuneas, T P, and M I Nadiri. 1996. Public R&D policies and cost behaviour of the
US manufacturing industries. Journal of Public Economics 63:57-81.
Martin, S Scott, J T (2000) „The nature of innovation market failure and the design of
public support for private innovation‟, Research Policy, 29, 4-5, 437-447.
Metcalfe, S (1997) „Technology Systems and Technology Policy in an Evolutionary
Framework‟, in „Technology, Globalisation and Economic Performance’, Archibugi,
D and Michie, J (eds.), Cambridge University Press.
Morahan, T., 2002. "Progress in the NI economy – A UK regional comparison", DEL
Labour Market Bulletin, No. 16, pp 39-46.
O‟Malley, E and Roper, S (2003) „The Basis for Trade – A Cross-border Comparison
of Trade and Productivity‟, August, InterTradeIreland, Newry.
Oughton C, Landabaso M and Morgan K (2002) „The regional innovation paradox:
innovation policy and industrial policy‟, Journal of Technology Transfer, 27, 97-110.
Powell, W W (1998) „Learning from Collaboration: Knowledge and Networks in the
Biotechnology and Pharmaceutical Industries‟, California Management Review, 40,
3, 228-240.
Roper, S (1998) „Public Support for Near-Market R&D: The Northern Ireland
Experience‟, Regional Studies, 32, 3, 295-299.
Roper, S (2001) „Innovation, Networks and Plant Location; Evidence for Ireland‟,
Regional Studies, 35, 3, 215-228.
Roper, S and Anderson, J (2000) „Innovation and E-Commerce – A Cross-Border
Comparison of Irish Manufacturing Plants‟, NIERC Research Report 17.
Roper, S and Hart, M (2004) „Small Firm Growth and Public Policy in the UK: what
exactly are the connections?‟, Working Paper, Aston Business School.
Roper, S and Hewitt-Dundas, N (1998) „Innovation, Networks and the Diffusion of
Manufacturing Best Practice‟, NIERC Research Report 14, April, NIERC, Belfast.
Roper, S and Love, J H (2006) Innovation and regional absorptive capacity: the
labour market dimension, Annals of Regional Science, 40, 2, 437-447.
24
Roper, S Ashcroft, B Love, J H Dunlop, S Vogler-Ludwig, K and Hofmann, H (1996)
‘Product Innovation and Development in UK, German and Irish Manufacturing
Firms’, NIERC/Fraser of Allander Institute.
Roper, S Hewitt-Dundas, N and Savage, M (2004) „Innovation, Best Practice
Adoption and Innovation Networks - A Comparison of Northern Ireland and the
Republic of Ireland‟, InnovationLab Ireland Ltd, Belfast.
Roper, S, J H Love, J Du, and J H Love. 2008. Modelling the Innovation Value
Chain. Research Policy 37 (6-7):961-977.
Sakakibara, M (1997) 'Evaluating Government Sponsored R&D Consortia in Japan;
Who Benefits and Why?, Research Policy, 26, 447-473.
Sakellariou, C (2006) „Education policy reform, local average treatment effect and
returns to schooling from instrumental variables in the Philippines‟ Applied
Economics, 38, 4, p. 473-481.
Trajtenberg, M (2000) „R&D Policy in Israel: An Overview and Reassessment‟
NBER Working Paper No. 7930, October 2000
Veugelers, R. and Cassiman, B. (1999). Make and buy in innovation strategies:
evidence from Belgian manufacturing firms, Research Policy, vol. 28, 63-80
Wooldridge, J (2002) „Econometric analysis of cross section and panel data‟, MIT
press.
Wrynn, J (1997) „Foreign Direct Investment to a Peripheral Country – The Case of
Ireland‟ in Fynes, B and Ennis, S (eds) „Competing From the Periphery‟, Oaktree
Press, Dublin.
Youtie, J Shapira, P et al (2005) „Innovation in Manufacturing: Needs, Practices, and
Performance in Georgia, 2002-2005‟ GaMEP Evaluation Working Paper: E200502,
August 2005.
Summary of Figures and Tables
Figure 1: R&D and Innovation Grant Support in Ireland
and Northern Ireland (€m pa)
Table 1: Impact of Public Support on Product Innovation: Northern Ireland
Table 2: Impact of Public Support on Product Innovation: Ireland
Table 3: Summary of policy effects by type of additionality and area
Annex 1: Data Descriptives
25
Figure 1: R&D and Innovation Grant Support in Ireland
and Northern Ireland (€m pa)
45
40
35
30
Ireland
25
€m pa
Northern
Ireland
20
15
10
5
0
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
Notes and Sources: Northern Ireland, nominal sterling data from IRTU Annual Reports. See Roper (2001) for details; currency conversion to €
using market exchange rates. Ireland, nominal grants data provided by Forfas.
26
Table 1: Impact of Public Support on Product Innovation: Northern Ireland
Probability of innovation
(IV Probit)
Coeff
t-value
Improved product innovation
(IV Tobit)
Coeff
t-value
New product innovation
(IV Tobit)
Coeff
t-value
1.183
2.390
29.880
2.350
17.010
1.770
Plant level Controls
In-house R&D
Supply-chain links
Non-supply chain links
Plant size (employment)
Plant size (emp, squared)
Plant vintage
Production one-offs
Production small batches
Production large batches
Part of multi-plant group
Externally-owned
Workforce with degree
Workforce with no qual.
Capital Inv per employee
0.767
0.011
0.003
-0.001
0.000
0.000
-0.477
0.259
0.124
-0.145
0.203
0.016
0.000
-0.001
3.530
3.860
0.670
-0.710
0.580
0.050
-3.140
2.010
0.850
-0.710
0.940
1.900
-0.120
-0.220
23.203
0.294
-0.050
-0.003
0.000
-0.074
-16.698
6.113
3.570
-8.482
-5.806
0.421
-0.028
0.112
4.380
4.140
-0.530
-0.210
-0.160
-1.440
-3.430
1.670
0.900
-1.510
-1.000
2.250
-0.460
0.600
13.003
0.159
-0.055
-0.015
0.002
-0.046
-13.518
7.054
3.015
-7.635
-5.607
0.075
0.004
0.078
3.420
3.310
-0.900
-1.730
1.520
-1.100
-3.600
2.700
1.080
-1.840
-1.280
0.470
0.090
0.730
Constant
-0.536
-2.010
-6.777
-1.010
-4.347
-0.710
Public Support
n
Equation χ2(.)
Log Likelihood
Exogeneity χ2(1)
684
184.15
-2520.03
1.07
(ρ=0.300)
660
264.81
-9282.8
1.73
(ρ=0.188)
656
166.06
-8765.24
0.75
(ρ=0.38)
Notes: Survey responses are weighted to give representative results. Models are estimated on a pooled sample covering the 1994 to 2002 period.
Models also include industry dummies (not reported) following Youtie et al. (2005): food and textiles includes Nace 15-19; Materials based
industries includes Nace 20-23, 26, 36; Science based industries are Nace 24; Metals and engineering includes Nace 27-29 and electrical and
transport includes Nace 30-35
Source: Irish Innovation Panel
27
Table 2: Impact of Public Support on Product Innovation: Ireland
Probability of innovation
(IV Probit)
Coeff
t-value
Improved product innovation
(IV Tobit)
Coeff
t-value
New product innovation
(IV Tobit)
Coeff
t-value
Public Support
0.196
0.520
9.830
1.240
8.423
1.400
Plant level Controls
In-house R&D
Supply-chain links
Non-supply chain links
Plant size (employment)
Plant size (emp, squared)
Plant vintage
Production one-offs
Production small batches
Production large batches
Part of multi-plant group
Externally-owned
Workforce with degree
Workforce with no qual.
Capital Inv per employee
1.111
0.008
-0.002
0.000
0.000
0.001
-0.300
0.200
0.248
-0.075
0.050
0.009
-0.001
0.006
8.780
3.610
-0.460
0.140
0.470
0.440
-2.130
1.890
2.110
-0.500
0.310
1.630
-0.450
2.010
30.520
0.215
0.020
0.004
0.000
-0.075
-10.343
6.885
4.650
2.488
2.354
0.079
-0.071
0.013
8.030
3.730
0.260
0.330
0.110
-1.330
-2.320
2.220
1.430
0.600
0.570
0.550
-1.450
0.360
17.479
0.125
0.015
-0.003
0.001
-0.098
-9.311
4.860
2.615
-1.272
2.239
0.164
-0.061
0.019
6.200
2.980
0.240
-0.330
0.360
-2.940
-3.130
2.190
1.150
-0.430
0.750
1.560
-1.820
0.660
Constant
-0.431
-1.920
-9.768
-1.490
-7.366
-1.630
n
Equation χ2(.)
Log Likelihood
Exogeneity χ2(1)
1007
205.12
-6004.7
1.22
(ρ=0.269)
958
240.22
-24280
0.22
(ρ=0.63)
958
202.11
-22602.153
0.01
(ρ=0.911)
Notes: Survey responses are weighted to give representative results. Models are estimated on a pooled sample covering the 1994 to 2002 period.
Models also include industry dummies (not reported) following Youtie et al. (2005): food and textiles includes Nace 15-19; Materials based
industries includes Nace 20-23, 26, 36; Science based industries are Nace 24; Metals and engineering includes Nace 27-29 and electrical and
transport includes Nace 30-35
Source: Irish Innovation Panel
28
Table 3: Summary of policy effects by type of additionality and area
All Plants
Extensive Additionality
Incremental Additionality
Radical Additionality
Indigenously-owned plants
Extensive Additionality
Incremental Additionality
Radical Additionality
Northern
Ireland
Ireland
+
+
+
(+)
(+)
(+)
+
+
(+)
+
+
+
Notes: + denotes significant positive effect; (+) denotes positive but insignificant
effect.
29
Annex 1: Data Descriptives
Northern Ireland
N(N=1156)
Mean
Std. Dev.
Innovation Indicators
Sales of new product (% of total)
Sales of new and improved products (% of total)
Product Innovator (1/0)
In house R&D (1/0)
Public Support:
For product development (1/0)
For process development (1/0)
For R&D (1/0)
For capital investment (1/0)
Business Level Controls
Supply Chain Innovation Links (% mean)
Non- supply Chain Innovation Links (% mean)
Employment 50-99 (1/0)
Employment 100-249 (1/0)
Employment 250+ (1/0)
Plant age in years (mean)
Production Mainly One-Offs (1/0)
Production Mainly Small Batches (1/0)
Production Mainly Large Batches (1/0)
Multi-plant group (0/1)
Externally-owned (0/1)
Workforce with degree (%)
Workforce with no qualifications (%)
Capital Investment per employee (£000)
Industry Dummies
Food and textile industries (1/0)
Materials based industries (1/0)
Science-based Industries (1/0)
Metals and Engineering (1/0)
Electrical and Transport Sectors (1/0)
Ireland
(N=1571)
Mean
Std. Dev.
12.503
25.727
0.564
0.421
20.620
33.305
0.496
0.494
15.009
28.963
0.628
0.489
22.960
34.426
0.483
0.500
0.215
0.132
0.085
0.289
0.411
0.339
0.279
0.453
0.233
0.150
0.101
0.221
0.423
0.357
0.301
0.415
19.239
9.978
0.174
0.125
0.257
35.209
0.224
0.430
0.271
0.673
0.226
7.270
47.615
4.645
30.187
22.594
0.379
0.330
0.980
32.096
0.417
0.495
0.444
0.469
0.418
10.034
32.741
11.820
22.784
12.437
0.185
0.129
0.299
31.233
0.201
0.408
0.285
0.543
0.302
9.472
45.651
7.013
30.661
24.122
0.389
0.336
1.052
29.455
0.401
0.491
0.452
0.498
0.459
12.864
32.677
21.109
0.237
0.380
0.087
0.121
0.139
0.425
0.485
0.282
0.327
0.346
0.203
0.304
0.090
0.183
0.192
0.402
0.460
0.287
0.387
0.394
Notes: Survey responses are weighted to give representative results. Figures are
pooled values covering the 1994 to 2002 period.
Our use of industry dummies follows Youtie et al. (2005): food and textiles includes
Nace 15-19; Materials based industries includes Nace 20-23, 26, 36; Science based
industries are Nace 24; Metals and engineering includes Nace 27-29 and electrical
and transport includes Nace 30-35
Source: Irish Innovation Panel
30