Definitions

Innovation

An authoritative definition of innovation is provided in the Oslo manual (OECD 2005), a methodological manual by the OECD on the measurement of technological innovation. An innovation is ‘the implementation of a new or significantly improved product (good or service), or process, a new marketing method, or a new organisational method in business practices, workplace organisation or external relations.’ This definition is inspired by Schumpeter (1934). Note the subtle difference between invention and innovation. An invention is an idea made manifest, whereas an innovation is an idea applied in practice.
Although the various components of innovation are not mutually exclusive, using this rubric enables us to get a clearer picture of what we are referring to when we talk of ‘innovation’. Product innovation refers to the creation of new (or improved) goods or services that are launched on to the market. Whilst both goods and services are included in this aspect of innovation, much of the literature is dominated by innovations in physical goods. Process innovations, on the other hand, refer to changes in the way in which goods and services are produced. Market innovation refers to improved ways of sourcing supplies of raw inputs or intermediate goods and services, as well as opening up new market opportunities (which could relate to either creating new domestic or export markets). A final type of innovation is organisational innovation, which refers to changes in the architecture of production, and accounts for innovations in: management structure; corporate governance; financial systems; and changes in the way workers are paid.
Innovation is thus a broad phenomenon not confined to technological product and process innovations. In fact, Fagerberg (2009) argues that many of the most important innovations throughout history have been of the organisational kind such as, for instance, how the Japanese automotive industry reorganised the entire value chain in the period following the end of the Second World War.
A further important distinction in the meaning of innovation is whether the innovation is new-to-the-world or new-to-the-firm. Overwhelmingly, most innovations are new-to-the-firm and the dominant source of productivity is new-to-the-firm innovations. We expect that firms that engage in new-to-the-world innovations also engage in many innovations that imitate and copy other firms. However, the dominant view, based more on casual empiricism than hard data, is that the most profitable firms are the largest source of new-to-the-world innovations.
This report focuses on technological innovations. Market innovations, which are influenced by trade and investment policies; and organisation innovations, which are influenced by regulation and the public provision of education, health and transport services, inter alia, represent large and diverse policy areas. While we believe a review of these policies is beyond the scope of this report, the reader should bear in mind that reforms and innovations from these sources are as important for societal well-being as technological innovations.

9BMeasuring innovation

While the aim of government policies is always to optimise certain activities, innovation policies almost always aim to increase the level of innovation. Technological innovation activities are ‘all of the scientific, technological, organisational, financial and commercial steps, including investments in new knowledge, which actually, or are intended to, lead to the implementation of technologically new or improved products and processes’ (HOECD 2002H).
The innovation process is not limited to investment in R&D. According to HBrouwer and Kleinknecht (1997)H expenditures on R&D represent about one quarter of total innovation expenditure, and half of the latter consists of investment in capital expenditures. According to the OECD (2002) non-R&D innovation activities include ancillary services (i.e. scientific and technical education and training, libraries and museums, translation and editing of S&T literature, surveying and prospecting, data collection on socio-economic phenomena, testing, standardisation and quality control, client counselling and advisory services) and innovation activities which occur after the experimental development stage (i.e. patent filing and licensing, market research, manufacturing start-up, tooling up and redesign for the manufacturing process).
Measured R&D depends on the accounting and measurement policies of individual firms as there are few rules about how firms must organised their journals for innovation expenses. However, most firms are guided either by national accounting standards or national taxation rules – the two are not the same.
The Generally Agreed Accounting Principles (GAAP) define R&D according to whether it meets the definition of an ‘asset’ and can thus be said to contribute towards intangible assets. GAAP requires that asset expenditures: are separable (i.e., implying contractual or property rights); have the power to obtain future economic benefits; have the power to restrict the access of others to the benefits; have a 50 per cent or higher probability that future benefits will eventuate; and have been a cost from an external party. Research costs generally fail this test and are therefore normally expensed. This means that they are not separately accounted for and cannot be distinguished from production wages and administrative costs. Only downstream portions of the commercialisation process which have high probability of generating future income will be included as an R&D asset.
The ABS by contrast follows that Frascati manual definition, which includes risky, unsecured upstream research but excludes downstream activities. The former comprise basic research, applied research and experimental development.F F The ATO typically follows the ABS definition.F F The upshot is that the differences in standards make it unclear which guidelines firms are following in their own internal accounting, and what they include when they complete company reports and ABS surveys. In addition, changes over time to ATO rules, including the financial incentive to claim tax rebates, have meant that data based on this measure will not necessarily give a consistent series of data over time. Table 1 gives a potted summary of the different measures of R&D.
Table 1: Measures of R&D

As there is no universally agreed upon measure of innovation, researchers rely on proxy measures such as patents, R&D expenditure, research personnel, and technology balance of payments. Each of these measures captures a particular aspect of the process of technological change. Patent and trade mark data has been increasingly used in research because of its increased availability through the release of electronic administrative databases, but has been criticised because it’s scope if limited to patentable subject matter technologies, inter alia. Griliches (1990, p. 1661) however notes that ‘…in this desert of data, patent statistics loom up as a mirage of wonderful plenitude and objectivity’. Schmookler agrees with the general sentiment of Griliches ‘…we have a choice of using patent statistics cautiously and learning what we can from them, or not using them and learning nothing about what they alone can teach us’‖ (Schmookler 1966, p. 56). (cited in Gedik 2010, p. 102).
The ad hoc nature of much innovation data means that researchers can be limited in their choice of dataset. What is available either narrows the research question or forces researchers to ‘shoe-horn’ available data into their preferred economic model (which typically results in estimates with large standard errors). However, it also means we have no way of distinguishing between the absence of an economic relationship and measurement error. To overcome this problem, work continues on developing new and improved measures of innovation. The papers by Dodgson and Hinze (2000), Kleinknecht et al. (2002) and Jensen and Webster (2009a) advocated the use of multiple indicators of innovation in order to overcome the problems of single indicators. Appendix A provides a summary of existing innovation measures: their uses and shortcomings.

10BThe ‘ideal’ innovation policy

There can be two rationales for innovation policy: i) to solve a market failure; and ii) to build and sustain the national innovation system (NIS). In the market failure-based perspective, every policy measure must be justified both by the identification of some form of market failure, and by an argument that explains how the policy can bring the system closer to its optimal state (Soete et al. 2010). The national innovation system has been defined by Lundvall (1992) as ‘... the elements and relationships which interact in the production, diffusion and use of new, and economically useful, knowledge ... and are either located within or rooted inside the borders of a nation state.’ This second approach recognises that innovation is supported by non-market-based institutions such as socio-economic, political and cultural systems that work together and are an essential ingredient in innovation outcome. The extensiveness of innovation is affected by the societal education level and entrepreneurial base, as well as the regulatory environment, amongst other things. Policies justified on the grounds of the NIS perspective do not seek to address a specific market failure. Instead, they are primarily aimed at ensuring that all the elements of the system are in place and working smoothly with each other.F F Many of the NIS policies relate to programs under the control of the States.
Both the market failure and NIS theories of industrial economics have their underpinnings of the work of 19th century economist Alfred Marshall. In his two manifestos, The Principles of Economics and Trade and Industry, Marshall gave extensive descriptions and analyses of the role of knowledge, immaterial capital (which we now refer to as ‘intangible assets’ or ‘intangible capital’), learning-by-doing and networks, along with one of the first exposés of marginal analysis.
What can be called the market failure approach has evolved as an abstraction of Marshall’s ideas using the method of limiting tendencies. Essential elements of the economy are reduced to properties such as non-exclusivity, non-rivalry, pure competition and uncertain risk. These traits are viewed as abnormalities to be compared with the normal or ‘ideal’ (purely competitive) situation. An extreme version of this school adheres to assumptions of hyper-rationality – that is, the assumption that firms have perfect foresight and will literally always behave in a profit-maximising way. However, most industrial economists do not employ this extreme assumption, thus leaving more room for policy intervention.
The NIS approach also has its roots in Marshall but is based more on descriptive reality and uses the method of representative conditions. It views the economy as being in a constant state of transformation and thus either assumes the absence of equilibrium, unstable equilibrium or multiple equilibria. In this ‘evolutionary’ view of the economy, the industrial system is regarded as a whole rather than a set of properties that may operate separately from one another. These economists put most emphasis on a package of policies that are complementary and work towards moving a market (or technology area) from a low to high equilibrium point.
Although distinct in some ways, these branches of the economics discipline can overlap and are not necessarily independent of each other. In fact, we argue – based on the logic articulated by Marshall – that they are not distinctly different. Therefore, we treat the “systems failure” arguments as a subset of “market failure” – a subset which we refer to as “coordination failures”.F F Although the literature on coordination failures is far less advanced than the literature on other market failures, we will draw upon all of the relevant streams of the economics literature, where necessary, in order to draw out the key insights from the literature.
Our over-arching theoretical framework is that a policy intervention is deemed ‘ideal’ if:

• It eliminates large deadweight losses;
• Displacement effects are small (displacement = innovation that would otherwise take place);
• It does not create perverse moral hazard incentives (e.g. by encouraging rent-seeking or game-playing); and
• It does not adversely distort private costs and benefits elsewhere in the system.

Figure 1: Losses due to an Externality and Subsidy

Figure 1 presents a situation where an innovative activity is the output. We assume there is a positive externality on the consumption of that activity (therefore there is a difference between marginal social benefit [MSB] and marginal private benefit [MPB]). MC is the marginal cost each ‘piece’ of innovative activity. Under a market situation we have a deadweight loss as indicated by the triangle. If a public per unit subsidy is invoked to drive production to the optimal level (O*), then we eliminate the deadweight loss, but create a displacement effect (where production that will be undertaken under a free market situation receives a subsidy). In general, there will always be displacement effects from a program. Eliminating them completely would incur excessive costs in terms of complexity and administration (if indeed possible at all).
A good policy intervention will achieve net benefits for the community after taking account of all the impacts. Note that, in practice, efforts to address market failure are never perfect. They suffer from government failure in implementation (lack of knowledge to target the intervention, inability to provide incentive-neutral financing, political pressure by interest groups for beneficial treatment, etc.) and might have unintended side-effects, creating collateral costs that outweigh the benefits (Ketels 2009, p. 21). Important characteristics of successful innovation policies are their certainty and their stability. Without these, business decision makers are unable to assess risk and opportunity and make the trade-offs necessary for investment in new technologies (Marcus 1981).
An overly strong focus on market failures, as in neoclassical economic models, to justify government intervention may not be desirable. Opponents of the market-failure approach often argue that the innovation process is subject to so many market failures that neoclassical economic theory is an inappropriate tool for analysing the dynamics of innovation. The NIS approach is often put forward as an alternative paradigm to justify policy intervention (see e.g. Six Countries Programme 2009 and Dodgson et al. 2010). In this report, we adopt a pragmatic approach to innovation policy. We acknowledge that the viability of many policy instruments designed to address market failures is often substantially dependent on a well-functioning Australian system of innovation.F F However, in many cases, it is necessary to use the abstract marginal analysis approach to illustrate why non-intervention is sub-optimal.
It is also important to mention that, in addition to market failures, there are potential government failures. That is, even though the ideal policy interventions can be designed, the government may not be able to implement them due to problems associated with bounded rationality, principal-agent problems between the policy makers and the bureaucracy, and imperfect information. In our Report, we ignore these issues despite their obvious importance.
Figure 2: On Specific Targets in Innovation Policy

2BThe Innovation Process

Innovation is a dynamic process which has no beginning and no end: it is a continual process in which products and processes are in a constant state of flux. The innovation process follows a complex pathway which involves feedback loops generated by: learning by doing; trial and error; and discontinuities in production. That said, in order to make discussions tractable and clear, we often speak of a spectrum of activities from high-end or upstream basic science through to applied application, development and commercial activities.
A good understanding of the innovation process is needed to identify sub-optimalities. In this section we explain how an idea progresses from concept to the creation of a new product/process, all the way through to a market launch (or even export to a foreign market). In addition, we will examine the operation of the ‘market for technology’, where we will focus our attention on how technology is traded: how buyers and sellers meet, and the importance of long-term collaboration as a way of building trust between buyers and sellers. Finally, we examine our understanding of the determinants of successful collaborative arrangements – whether that be between firms (such as a joint venture) or between universities and firms.
In Figure 3 we present a brief snapshot of some recent Australian innovations.
Figure 3: Examples of Australian Innovations

11BWhy do innovations occur?

Innovation is arguably the way firms compete. The ability of firms to compete via price reduction is limited to what is financially sustainable and therefore price cutting can only be a short-term competitive strategy. If firms want to cut costs over the long term, or increase profit margins via the production of better products or by developing more markets, then they must innovate. Most profit maximising activity is essentially about creating a monopoly advantage for the firm, and innovation, which begins by being something new, is the genesis of this advantage.
This said, not all firms or all industries, innovate (or compete) with equal vigour and success. The overwhelming majority of innovations are those that are new-to-the-firm rather than new-to-the-world. In this section, we review theories about what drives firms to innovate. A good starting point is to look at studies which try to understand what makes some firms decide to attempt to undertake innovation activities and other firms to not do so. And what skills and capabilities shape this decision? In a recent article, Woerter (2008) summarised a number of hypotheses that have been proposed to address the issue:

• The Schumpeterian hypotheses (Schumpeter 1934, 1975), which focus on the size of the firm and the level of concentration in which the firm operates as key factors determining its innovative behaviour;F F
• The demand-pull hypothesis (Schmookler 1966), which proposes the significant role that market conditions play such as the size of the market and changes in prices;
• The technology-push hypothesis (Phillips 1966; Rosenberg 1976) which argues that conditions underlying knowledge production processes are essential;
• The financial-constraints hypothesis (Nelson 1959) and the related hypotheses which are based on issues surrounding the risks of R&D and the risk preferences of the institutions involved (Mansfield 1968); and
• The technology-related, supply-side factors hypothesis, which combines appropriability, tacitness of knowledge, technological opportunities and uncertainties (Dosi 1988).

Woerter (2008) then continued by citing recent empirical studies, based on the European Community Innovation Survey (CIS) data, aimed at providing formal tests of these hypotheses. Support for the ‘firm-size hypothesis’ is at best inconclusive, with a tendency for the effect to be negative. This is consistent with other non-CIS empirical studies which tried to sort out whether firms acquired market power because of successful innovation or whether market power enabled firms to make innovation profitable (i.e. Kamien and Schwartz 1982; Mansfield 1984; Levin and Reiss 1984; Acs and Audretsch 1987, 1988, 1991). Many studies that find market structure and/or firm size to be significant determinants of R&D intensity do not control for the underlying conditions of opportunity and appropriability (Phillips 1966; Sutton 1991; Scherer 1967; Cohen 1995; Bosworth and Rogers 2001).
In contrast, both the demand-pull and technology-push hypotheses have some support. There have been a series of economic studies that have tried to estimate the role of more deep-seated determinants such as the opportunities proffered by the scientific sector and how easily firms can appropriate their R&D profits (Levin and Reiss 1984; Pakes and Schankerman 1980). This avenue of research appears to have produced more consistent results than the earlier studies, in part because the theoretical directions of the effects are less ambiguous. However, it still leaves open the question of what governs scientific opportunity and natural appropriability. For example, it may be that size, and the underlying financial resources it implies, enhances the scope of an enterprise’s opportunity and appropriability sets.
Another smaller but concurrent stream of economic research considers why firms do not innovate rather than why they do. The key hypothesis concerns the financial hurdle for firms which desire to invest in highly uncertain and collateral-free projects such as R&D. As with the scientific opportunity and appropriability theories, there is a clear a priori prediction of the effects of retained earnings and gearing levels, and therefore empirical studies tend to find reasonably robust findings. There is a widespread view, and solid evidence, that raising debt to fund R&D is particularly costly and difficult because large R&D activities typically produce uncertain and distant collateral which, more often than not, are absent from balance sheets (Schumpeter 1943; Hall 2005; Scellato 2007; Canepa and Stoneman 2008; Carreira and Silva 2010). Evidence from Australia supports this: Palangkaraya et al. (2010) find that internal funding is the most common way patent applicants fund their research. However, Thomson (2010) used Australian data and did not find that working capital influenced the level of R&D.
There are few econometric Australian studies which explore the determinants of firm innovation; exceptions are Griffiths and Webster (2010) and Thomson (2010). The former found that most of a firm’s R&D activity is explained by internal factors such its managerial style and its competitive and appropriation strategies. The growth in product demand and internal sources of funds were significant, but smaller, in magnitude. The importance of the internal operation of the firm is also supported by Woerter (2008). He finds evidence in support of another hypothesis related to fifth point above. This hypothesis argues that if firm innovation is driven by its perceptions about the problems it faces, and if the firm’s perceptions depend on working routines which are influenced by the characteristics of the firm – such as the size of its employment and physical capital – then one could expect that industries with a greater variety in terms of firm characteristics would be relatively more innovative than industries with more homogeneous firms.

12BHow do innovations occur?

Some innovations occur ‘automatically’ within the firm in the sense that the managers see a problem or opportunity and then devise a process for producing the design, undertaking its development and then oversee the ‘manufacture’ and sale of the idea. A few firms have clear pathways for this process to occur within their organisation. Other innovations are not automatic and their successful conclusion depends on serendipity rather than planning. This is especially true when several distinct parties are involved in the value-added chain, or where prior experience by the players in the area is limited. Typically these are ideas that emanate from public sector organisations and SMEs. In these situations, the efficacy of brokers in the market for technology, or go-betweens, can be critical.
Apart from specific case studies, we are limited in our knowledge about how innovation occurs to large-scale surveys of different facets of the innovation process. We only have facets because the very heterogeneous nature of innovation means it is hard to generate stylised facts about common pathways (compared with say, the pathways for skill acquisition). Unfortunately, there is also a tendency for the literature to suffer from ‘high-tech myopia’ (the idea that economic growth and employment is mostly the result of research-intensive industries, as defined by their R&D intensity) and then to concentrate only on firms undertaking R&D or patenting. However, since we do not have comprehensive and widely reported measures of innovative activity, we cannot verify whether these so-called research intensive firms are the major contributors to productivity. How biased R&D or patent counts are as a measure of the broader term ‘innovation’ depends on the use to which the data is put (see Jensen and Webster 2009a). Often the estimated measurement bias is based on the researchers ‘expert opinion’ rather than empirical work. An exception is the work by Kirner et al. (2009), who used the 2006 German Manufacturing Survey data of 1663 firms and find that, while ‘low-technology’ manufacturing firms lag behind medium- and high-tech firms with respect to their product innovation, they may perform better at process innovation. This finding is perhaps explained by the tendency for low-tech innovations to involve processes which are not primarily based on formal research and technological development. Instead they tend to be practical and experience-based, usually involving implicit knowledge (Heidenreich 2009).
There is a view that variation between firms’ innovative intensity depends on the maturity of the industry. However, the importance of this source of variation may alter from one setting to another, as illustrated by the finding of McGahan and Silverman (2001). In their study they investigate the activity of US publicly-traded firms during the 1980s through to mid-1990s. They find that firms’ patenting activity does not decrease as the industry in which they operate matures in terms of the underlying technology life cycle. They conclude that industry maturity does not appear to lead to a switch from product to process innovation, nor does it imply lower firm innovative activities compared to those in emerging industries.
Our qualitative experience suggests that there are three main types of innovation process.

• A problem or opportunity is identified, solved and implement within the firm. Many incremental process innovations are of this type.
• A problem or opportunity is identified by a firm but parts of the innovation process are subcontracted to specialist research, design and manufacturing organisations who undertake research, development, translation, extension and pre-commercial activities on the originating firm’s behalf.
• A problem or opportunity is identified by a research organisation who then sells or licenses the findings to subsequent organisations with complementary specialist capabilities. The innovation process is owned by a series of parties who share in the risk of the whole process. Many of these innovations are radical and involve new scientific discoveries.

One well-documented facet of how innovation occurs is the source of knowledge which inventors build upon. In 2007, IPRIA surveyed all Australian inventors listed on patent applications submitted to the Australian Patent Office between 1986 and 2005. This survey collected information on commercialisation outcomes of 3,736 inventions.F F Essentially, the Inventor Survey provides us with a picture of innovation beyond that which can be provided by firm-level data such as R&D expenditure and patenting counts. It reveals that customer product users are the most important source of ideas and knowledge for Manufacturing and, to a slightly lesser extent, Resources and Services inventors (see Figure 4). Scientific literature is predominantly important for Resources and Services sector firms.
Figure 4: Importance of Knowledge Sources, Patent Applications Filed by Organisations Located in Australia by Sector, 1986–2005

Source: Figure 6.6. Palangkaraya et al. (2010).

Table 2 uses data from the Melbourne Institute Business Survey to compare the sources of knowledge used by companies in Australia with a similar US survey. The sources of knowledge are classified into open-learning (networks, customers and suppliers) and closed-learning (licensing of new technologies; hiring other organisations’ workers; patent disclosures; publications and technical meetings; and firm-sponsored R&D). This comparison shows that, on average, Australian firms rate open-learning styles more highly than closed-learning styles.F F Among the former, networks with other organisations was the most highly rated source of information, while among the latter, the most highly rated learning source was hiring skilled workers. All pair-wise means are statistically significant. Learning styles in our survey show little variation across firms either by size or by industry (although these results are not presented here). However, the responses from the US Levin et al. survey show little relation to the Australian list. In an analysis of the Australian data, Jensen and Webster (2009b) found that firms which favour closed-learning practices tend to rely more upon patents and secrecy and eschew lead-time and brands as ways to capture profits. Firms that favour open styles of learning operate in the opposite manner. The manner in which firms seek to stem the flow of knowledge from their firm affects how they behave and learn from other firms.
Table 2 : Rated Importancea of Firms’ Sources of Learning

Notes: a Scale is based on a Likert scale with anchors 1 (= very ineffective) and 7 (= very effective); b Informal networks with other organisations and Formal cooperation/networks with other organisations; c Lead suppliers and customers; d Patent disclosures and Publications or technical meetings; e In-house R&D and Reverse engineering.
Source: Jensen and Webster (2009b).
Figure 5 reveals that compared with other firms, Manufacturing firms, especially large ones, spend a smaller amount of time researching the ideas behind their inventions. Public sector organisations (which are not classified into sectors) spend the most amount of time researching the idea behind the patent. All in all, SMEs appear to spend more time on research than large firms, which may reflect greater overall caution by SMEs when it comes to patenting compared to large firms. Alternatively, it is possible that SMEs are slower at generating patented inventions compared with large companies.
Given the importance of networks and learning for innovation, one would expect to see high levels of collaboration among innovating firms. However, we do not find this. Figure 6 illustrates the propensity of the ABS BLD firms to collaborate with other businesses for innovation purposes.F F Collaboration helps firms tackle some of the important barriers to innovation, such as the inability to obtain access to finance due to the uncertainty of innovation and the inability to absorb external technology due to the partly tacit nature of the underlying knowledge. Collaboration can be especially important for smaller firms, which are more likely to face these difficulties. However, collaboration exposes these firms to contracting and transaction costs (Coase 1937), pernicious opportunistic behaviour (Williamson 1985) and expropriation of intellectual property (Arrow 1962).
Figure 5: Average Years Spent Researching the Invention, Patent Applications Filed by Organisations Located in Australia, 1986 to 2005

Source: Figure 6.8. Palangkaraya et al. (2010)
A recent study by Thomson and Webster (2011) found that Australian firms were more likely to collaborate over the development stage (of innovation) if they: were an SME or a large but highly geared firm; had relatively limited experience patenting in that technology; or considered the particular innovation to be technologically risky. Frenz and Letto-Gillies (2009) reviewed relevant international empirical studies from the empirical literature and found that almost all of them confirmed the significance of external collaboration with the users and external sources of technical expertise. The studies they reviewed also pointed to the importance of both formal and informal networks for innovation. Despite these potential benefits, Figure 6 demonstrates that collaboration is not very common among Australian innovating SMEs. On average across size categories and industry groups, only about four per cent of innovators report innovation-related collaboration. The figure also indicates that, in comparison to small innovating firms, medium innovators are more likely to collaborate in their innovative activities.

Figure 6 : Proportion of Innovating SMEs which Collaborate with Other Businesses for the Purpose of Innovation 2006–07, by Employment Size and Industry Group

Note: An innovating firm is a binary variable =1 if the firm has introduced any new or significantly improved goods/services or operational process in the respective year; =0, otherwise.
Source: Figure 4.6. Palangkaraya et al. (2010)
Data on what firms are actually doing are consistent with Tether (2002) who finds that most UK firms still appear to develop their new products, processes and services without forming (formal) co-operative arrangements with other firms or institutions. The extent of co-operative research arrangements for innovation may depend on the type of firm, and on what is meant by innovation. For example, firms which conduct R&D in order to introduce innovation which is ‘new-to-the-market’ rather than ‘new-to-the-firm’ are much more likely to engage in co-operative arrangements for innovation. Otherwise, most firms still appear to develop their new products, processes and services without forming (formal) co-operative arrangements with other firms or institutions.
An econometric analysis conducted by the Commonwealth Department of Industry, Tourism and Resources investigates how collaboration and other factors influence innovation novelty in Australian businesses. The analysis uses information from the Innovation Survey 2003, presented in ABS (2005). The database comprises only innovating firms, which are identified with a variable that is constructed by counting the number of different innovation-related activities, such as: the acquisition of machinery and equipment; training related to new goods or services; and substantial new design work. The analysis employs an ordered categorical probit model with the probability of introducing the highest degree of novelty (new-to-the-world innovation) as the dependent variable.
The model predicts that collaboration, while controlling for other firm characteristics such as size, foreign ownership and R&D intensity, is associated with a statistically significant increase in the chance of achieving new-to-the-world novelty. Collaboration was found to be more common and important to frontier and creative innovation than to relatively minor modification of goods, services and processes and purely adoptive innovation. The principle conclusion is that, in comparison to non-collaborating businesses, collaborating firms are more likely to introduce new-to-the-world innovation (however we cannot attribute causality here).
The persistence of innovation – the frequency and consistency of innovation produced by a particular firm – also gives us a clue to how innovation is occurring. In theory, there are a number of reasons why particular firms become persistent innovators. First is the cumulative causation or ‘success-breeds-success’ phenomenon (Nelson and Winter 1982). That is, innovative success yields profits that are reinvested in R&D. The second possible source of persistence is related to the idea that knowledge accumulation is intrinsically cumulative. If we look at Freeman’s list above, one may expect that successful innovators who satisfy the identified characteristics at one period of time may indeed satisfy the list at different periods.
Figure 7 defines large companies as ‘persistent innovators’ if they have at least one patent or design application in three (or more) consecutive years over the period from 1996 to 2007; ‘One-time innovators’ as those firms which had only one patent or design application during the period; and ‘sporadic innovators’ covers the remaining firms (i.e. firms with two patent or design applications over the period, as well as those firms with three or more applications which were not made in consecutive years).

14B Characteristics of (successful) innovators

Freeman (1991) summarises an interesting list of characteristics which make for successful innovators. According to this list, successful innovators are those who:

• Pay attention to the special needs and circumstances of users;
• Integrate the development, production and marketing activities;
• Link with external sources of scientific and technical information and advice, even when they typically have their own in-house R&D;
• Commit high quality R&D resources to the innovative project; and
• Bestow high status upon the ‘business innovator’ and, particularly in large organisations, have strong commitment from top management.

Terziovski et al. (2002) conduct a case study of a product development project (the ‘Bushranger’ Project) at Varian Australia Pty Ltd (a company with $140 million turnover which exports 95 per cent of its products). They find that Varian Australia focuses on optimising two critical success factors of product innovation, namely (a) meeting and exceeding customer needs and expectations by innovating new products and accelerating the cycle time from conceptualisation to market launch, and (b) establishing cross-functional, multi-disciplinary teams. Cebon (2008) summarises the findings from 11 Australian case studies and finds that successful innovators: paid attention to the market needs rather than the technology; managed risks, including IP risks, well; attuned corporate governance to innovation needs; had active support from, and at least considered understanding of their situation by, financiers; and were able to launch the product into the Australian market before going overseas.