Mapping, measuring and managing African national systems of innovation for policy and development: the case of the Ghana national system of innovation

Abstract

The systemic approach to innovation as key to economic development, in the context of the knowledge-based economy is increasingly evident and is of strategic value in terms of industrial and development policy (Bartels et al. 2012). The national system of innovation (NSI) of a country is therefore vital for enabling increased competitiveness through accelerated economic performance (Bartels and Voss 2005) and is crucial for developing countries hoping to catch up with advanced industrialised economies (Arocena and Sutz 2000). Innovation and NSI, previously seen linearly, are now viewed with a systemic network perspective. This perspective yields improved policy craft (Balzat 2002). In order to allocate limited resources effectively and efficiently through targeted policy, to increase industrial development and competitiveness, there is the need to map, measure and thereby manage the interactions of the core actors and barriers to innovation within the NSI (Bartels et al. 2009). The approach herein encapsulates the simultaneous application of a single data acquisition survey instrument (DASI) to the four core actors in the NSI (government, medium-and high-technology industry, knowledge-based institutions, arbitrageurs).

Keywords:

• national system of innovation
• evidence-based science
• technology and innovation policy
• economic and industrial development

Introduction

This paper addresses the challenges of policy analysis with respect to the mapping, measuring and managing and thus monitoring, the national system of innovation (NSI) for the formulation of coherent, evidence-based science, technology and innovation policy. The case in point is the methodology applied to the Ghanaian national system of innovation (GNSI) (Koria and Koseigi 2011). It presents a four-dimensional ICT-intense methodology (Koria et al. 2012) for mapping, measuring and hence managing NSI. This methodology takes its departure from Leydesdorff and Etzkowitz's (1996) triple helix model and elaborates the model as a triple helix model type 4 model (TH-4).

The approach first encapsulates the four core actors in the NSI: government (GOV); industry, specifically medium-and high-technology industry (MHTI); knowledge-based institutions (KBIs); and arbitrageurs (ARBs) (knowledge-brokers, venture capital, financial institutions). Secondly, it articulates the simultaneous application of a single data acquisition survey instrument (DASI) to respondents from the four core actors in the national system of innovation. Thirdly, it reflects the four phases of innovation policy of which a significant part is the involvement of knowledge-brokers (KBs), venture capital (VC) and financial institutions (FIs). The identification of the four core actors of the NSI advances the triple helix model (Etzkowitz and Leydesdorff 2000) as a TH-4 embedded within an environment of diffused information communications technology (ICT) (Koria et al. 2014).

National systems of innovation or national innovation systems (NSI/NIS) are crucially important to long-term economic competitiveness (Stern et al. 2000) and are of strategic policy concern for an increasing number of countries. This concern is accentuated by the global innovation divide (Sachs 2003) and the ever-widening ‘digital divide/digital inequality’ (White et al. 2011, DiMaggio and Hargittai 2001). The operative core of the NSI concept goes back a very long way and features in the economic work of Ricardo, Veblen, Schumpeter, Keynes, Solow and Romer. The original conceptualisation of the NSI can be attributed to Adam Smith in his 1776 analysis in terms of ‘knowledge creation in relation to directly productive activities but also specialised services of scientists’ (Lundvall et al. 2002, 5). The modern articulation of the concept is provided by List. According to Freeman (1995, 5):

Friedrich List's conception of ‘The National System of Political Economy’ (1841) … might … have been called ‘The National System of Innovation’. The main concern of List was with the problem of Germany overtaking England and, for underdeveloped countries (as Germany then was in relation to England), he advocated not only protection of infant industries but a broad range of policies designed to accelerate, or to make possible, industrialization and economic growth. Most of these policies were concerned with learning about new technology and applying it.

We map and measure NSI with one DASI applied simultaneously to all actors. Our TH-4 approach adds a distinctive feature which is missing from the traditional triple helix model, defined by the Institute for Triple Helix Innovation, as follows: ‘Academia, government and industry constitute the three helices that engage in triple helix innovation’ (Taxonomy of Triple Helix Innovation, White Paper, The Institute for Triple Helix Innovation). While this definition does not restrict the types of industry in triple helix innovation, we argue first that the role of abitrageurs is so central to innovation that they must be formally and distinctively included (Stern et al. 2000, Delgado et al. 2012, Gaba and Bhattacharya 2011). Secondly, we argue that, as per the diffusion of innovation paradigm (Rogers 2003), it is MHTI (ISIC Rev. 3), embodying innovators, early adopters and the early majority, that possesses the economies of scale and scope, as well as the economically significant capability and capacity to innovate (even though innovation can take place in low-tech industry). Furthermore, MHTI not only represents a disproportionately high percentage of the contribution to GDP but also, in developing countries, is more ICT-connected than low-tech industry. Thirdly, we aver that communications infrastructure in the form of diffused ICT is of crucial importance in intensifying the connectedness of NSI actors (Hilbert et al. 2010) and facilitating the flow of knowledge and resources between actors (Koria et al. 2012).

Literature review

Background to the recent evolution of literature

The intense debate on the roots of economic growth (Nelson and Phelps 1966), whether institutional and/ or technological, invokes a growing intellectual acceptance of the NSI as a vitally important research priority for crafting development policy (Ushakov 2012, Samara et al. 2012, Groenewegen and Van der Steen 2006). This debate has expanded the architecture of NSI literature to include the economics of innovation and technology, systems of innovation (at different ‘nested’ and networked vertical and horizontal levels of operation including local, sub-regional, national and supra-regional on the one hand and sector and technological on the other hand), industrial dynamics, organisational and policy dimensions of structural change, national innovation capacity (Bartels et al. 2012, Chang and Lin 2012) and transfer of technology. The international business facet, manifest as the transfer of technology and innovation through internalisation (Buckley and Carter 2004, Buckley and Hashai 2004, Dunning 2003, Buckley and Casson 2002) across the organisational boundaries of the firm via foreign direct investment (FDI) and foreign market servicing strategies, by multinational enterprises (MNEs), is equally of great consequence and a critical research concern. However, the dynamics of innovation inside MNEs within countries are beyond the scope of the paper.

A measure of the intensifying relevance of NSI is provided by expanding practitioner and research coverage. The 1999 DRUID conference on ‘National Innovations Systems, Industrial Dynamics and Innovation Policy’ suggested eight dimensions of NSI. The DRUID 2012 conference on ‘Innovation and Competitiveness – Dynamics of Organizations, Industries, Systems and Regions’ presented 67 parallel paper sessions ranging from 1 – Theoretical Perspectives on Regional Clusters to 67 – Emergence and Evolution of Technologies and Industries. The taxonomy of NSI has also been extended over organisational structures (‘soft’, ‘hard’), communication, co-operation, co-ordination, command and control, spatial dynamics, knowledge dynamics, knowledge stock, technology dynamics, functionality, industry-market dynamics and path dependency functions (Daneke 1998).

However, two provocative perspectives on NSI are cogent. First, according to Gatignon et al. (2001, 2), ‘Innovation and technical change are at the core of dynamic organizational capabilities … Yet after more than 30 years of research on innovation and organisational outcomes, fundamental concepts and units of analysis are often confused and/or ambiguous’. Secondly, according to Moldaschl (2010, 8) innovation theories make no sense because in ‘the last two decades … the semantics of innovation has quasi overflowed, lost any limits and thus inevitably become more meaningless’. It is not surprising, therefore, that, within this dynamic field, attention is focused more on conceptualisation and measurement issues at the different echelons: meta-, macro-, meso-, micro- and firm-level (Chaminade et al. 2012, Montalvo and Moghayer 2012, Freeman 2002, Carlsson et al. 2002, Sornn-Friese 2000, Patel and Pavitt 1994) rather than on empirical investigations of systems of innovation (Becheikh et al. 2006). Furthermore, given the centrality of institutions (i.e. the ‘rules of the game’), the contribution that distinguishes between institutions and organisations – following the separation of institutions and competencies (Patel and Pavitt 1994) – is of significant value (Edquist and Johnson 1997).

Despite the wealth of contributions, statistical assessments that use factor analysis to scrutinise NSI are fewer than expected (Bartels et al. 2012, Chang and Lin 2012). This is firstly due to the multiple challenges of measurement and secondly to the difficulty in selecting variables relevant to and capable of being responded to by all four core actors in the NSI. Notable exceptions model indicators of technological advancement in growth accounting and Patel and Pavitt (1994) use correlations to analyse sectoral specialisation. A seminal exception to the less than expected volume of quantitative evaluations is Nasierowski and Arcelus (1999), who employ factor analysis and structural equation modelling (SEM) to analyse NSI relationships. Also noteworthy is the SEM modelling of inter-firm competencies (Nooteboom et al. 1998) and the confirmatory factor analysis of Gatignon et al. (2001 2002). However, none of the above map and measure simultaneously all four core NSI actors. Therefore, more comparative analysis is called for; particularly, ‘much more NSI work on economies that are either underdeveloped or in the process of catching-up needs to be done’ (Sornn-Friese 2000, 10). Descriptive analyses, which hallmark innovation reports, can thus be complemented by rigorous quantitative analysis of the interdependencies among NSI actors, along with the factors that enable (or disable) the NSI, in terms of data, information, statistics and knowledge (DISK).

The main contributions have focused almost exclusively on NSI in industrial countries. While fruitful, the exclusive focus on advanced industrial country NSI, as suggested by Arocena and Sutz (2000), should carry an injunction against adoption of ‘Northern’ NSI characteristics without adaptation through local cultural and institutional lenses in the ‘South’ (Da Motta e Albuquerque 1999).

It is therefore our intention to expand the common NSI configuration (OECD 1999, Nasierowski and Arcelus 1999, Leydesdorff and Etzkowitz 1996, Lundvall 1992) in application to developing countries and present the factors or variables for policy recommendations. Figure 1 illustrates the interface between public (government) and private (investor) interventions in the dynamics of innovation.

Figure 1. Policy stages in the dynamics of innovation. (Adapted from Foxon et al. 2004).

The literature and its evolution

The first formal but unpublished use of the term ‘national systems of innovation’ was by Freeman in work in 1982 for an OECD expert group on Science, Technology and Competitiveness (Lundvall 2003, Carlsson 2006). Arguably, the operable key analytical framework for dealing with NSI was presented by Andersen and Lundvall (1988) as a system of: (1) backward linkages in flows of information; (2) learning by doing and searching; (3) distinctions between industrial subsystems at different stages in terms of life cycle; and (4) the open economy (Lundvall et al. 2002).

This conceptual frame, focusing on the development of technology and user-producer interactions, produced one definition of NSI as ‘ … the elements and relationships which interact in the production, diffusion and use of new and economically useful, knowledge are either located within or rooted inside the borders of a nation state’ (Lundvall 1992, 2). There are many definitions of NSI (UNIDO 2012, 16, Bartels et al. 2012, Achim and Popescu 2009). A more comprehensive definition, in policy, systemic and organisational capital terms, is provided by Bartels et al. (2012, 6) as ‘the envelope of conforming policies as well as private and public organisations, their distributed institutional relations and their coherent social and capital formations, that determine the vector of technological change, learning and application in the national economy.’

The two definitions above – separated by two decades and encompassing micro-level interactive production elements and relationships and macro-level conforming policies that determine technological change, learning and application – are contoured by interdisciplinary approaches within long-term comparative economic performance (von Tunzelmann 1997), national competitiveness (Porter 1990) and growth accounting (Greenwood and Jovanovic 1998, Jovanovic and Rob 1989, Arrow 1962, Solow 1960). These approaches have brought sharply into relief the indispensability of technological innovation and organisational capital in industrial dynamics and development (Squicciarini and Le Mouel 2012).

Economic performance is fundamentally determined by ‘deep determinants’ of income, which, according to Rodrik and Subramanian (2003, 32), are: geographygeology (or space) – the key to climatic advantages and natural resource endowments; integration – the role of international trade as a determinant of productivity change; and institutions – which determine the roles of property rights, the rule of law and the ‘rules of the game’ in society. These deep determinants in turn influence five drivers of development: multi-lateral organisation, macro-economics, factor inputs and suppliers, industry characteristics and competitive scope, which operate non-linearly (Porter 1990), across geo-economic space, as a complex adaptive eco-system of innovation, at supranational or meta-, national or macro-, cluster or meso-, industry or micro- and firm levels (Bramwell, Hepburn and Wolfe 2012). The crucial property enabling innovation in terms of capacity – and innovativeness in terms of capability – is the quality of reciprocating vertical and horizontal linkages in the innovation eco-system, connecting actors and the innovation assets infrastructure and facilitating the flow of DISK in the NSI. The density, distribution, directionality and strength of the linkages thus determine ‘the extent to which the potential for innovation induced by the common innovation infrastructure is translated into specific innovative outputs in a nation's clusters’ (Stern et al. 2000, 13).

Within these levels, the NSI conforms a geo-economic space of firms and strategies in a complex ecology wherein technological functions are essential (Ricart et al. 2004, 191, Wang and von Tunzelmann 2000). The ‘new’ empirics of growth (Durlauf and Quah 1998) emphasise these underpinnings, first identified in endogenous growth and technological change (Romer 1990). Contemporaneously, the institutionalised and spatially distributed network of market and non-market actors engaged in technology-related change-generating activities have attracted the rubric NSI (Dosi et al. 1988, Freeman 1987), with variations in the nomenclature depending on the unit or level of analysis.

Further elaborated since the 1990s within institutional literature (Hollingsworth and Boyer 1997), NSI is now viewed as vital to the effective co-ordination of inter-organisational relations in which knowledge is the currency of exchange in technological development (Edquist 1997, Lundvall 1995). This is due to the consistent affirmation that DISK and innovation, in co-determining patterns and trajectories of economic growth, are the key salient ingredients in the modern economy (Quatraro 2012, OECD 2011, Vania 2004). Nations must therefore access and develop new technologies if they do not wish to find themselves on the wrong side of the bifurcation in cross-country income distributions since the 1960s (Barro 1998, Durlauf and Quah 1998).

The central role of NSI in the dynamics of economic growth thus well acknowledged (Freeman 2002), invokes noteworthy portrayals in the taxonomy of structure and process in NSI. Among the earliest, Sábato and Botana (1968) indicated that relations between the government, the productive structure and the scientific and technological infrastructure are key to innovativeness. Pavitt's (1984) concept of innovative industrial patterns suggested the dimensions: supplier dominated; scale intensive; specialised supplier dominated; and science-based. In 1992 the OECD elaborated a taxonomy of industry and high- or low-tech dimensions (OECD 1992). Useful as they are, taxonomies based on size, sector and locational classifications have given way to those with systemic knowledge, human, social and organisational capital dimensions (van Ark and Hulten 2007). These latter taxonomies place greater emphasis on the emergent properties, complex adaptive system and non-linear properties of the NSI (Dasgupta 2002, Etzkowitz and Leydesdorff 2000, Lundvall 1998).

Archibugi and Iammarino (1999) develop a taxonomy of the globalisation of innovation with respective dimensions, level of formality and distance from innovation process and actions and forms. They indicate that global techno-scientific collaboration, an activity manifest at the interfaces of NSI actors, is archetypically performed by universities, public research centres and national and international firms. These are the actors in Leydesdorff's (2001, 1) ‘neo-evolutionary model of a triple helix of university-industry-government relations’.

The taxonomy of Braadland and Anders (2002), based on skills and the extent to which innovation is systemic, classifies the firm level component of NSI with the metric skilled worker or engineering density. Furthermore, Kaiser and Prange (2004) propose a territorial configuration to delineate NSI. In their taxonomy, regulatory policies and spatiality constitute the dimensions of NSI. Bjørnskov and Svendsen (2004) use the extent of decentralisation and availability of social capital to delineate the economic performance of Scandinavia relative to Eastern and Western Europe. In contrast, Asheim and Coenen (2004) and Munk and Vintergaard (2004) propose a taxonomy at meso-, or cluster-level in which the knowledge base, organisational nature, institutional characteristics and involvement in innovation are the four dimensions of classification.

On one hand, these taxonomies of NSI accentuate the complex externalities and dynamics of knowledge and innovation with their formal as well as informal organisational characteristics and direct and indirect involvement in processes of innovation (Munk and Vintergaard 2004, 4, Braadland and Anders 2002, 8). On the other hand, the taxonomies bring into relief ‘the complex dynamics … composed of subdynamics like market forces, political power, institutional control, social movements, technological trajectories and regimes. The operations (therefore) can be expected to be nested and interacting’ (Etzkowitz and Leydesdorff 2000, 113). The evolutionary model (Leydesdorff 2001, 2012), with functional and organisational dimensions (science-economy vs. public-private), reflects NSI actors and linkages that need accurate portrayal for policy analysis (OECD 1999). Herein, knowledge generation and diffusion by organisations and knowledge deployment moderated by institutions are viewed as the NSI capability that leads to innovativeness. NSI capacity is seen as the strength of links between actors, supply and market factor conditions, communication infrastructures, education/training and the macro- economic and regulatory contexts that lead to innovation. This bilateral perspective informs our method of mapping and measuring NSI (Koria et al. 2012).

As the NSI encapsulates and articulates the effectiveness and efficiency of technological change in the economy, it is important to recognise three basic constructs. First, translating systemic innovation into competitive performance is a function of close and stable relationships between economic actors (Richardson 1972). Secondly, these relationships are contextualised by non-market interactions (Lundvall 1985, Williamson 1985). Thirdly, the relationships and interactions are differentially moulded by specificities of national institutional ‘rules of game’ (Edquist 1997). To these three, a distinctive fourth construct may be added – differences in stocks and quality and rates of flows, of industrial knowledge as well as their management carry implications for the relationships between economic actors (Asheim and Coenen 2004).

It is therefore obvious that in circumscribing the concept of NSI, the social system constituted by institutional arrangements, resource capacities and proprietary functional capabilities is constructively important (Van de Ven 1993). With this perspective, Park and Park (2003, 403), pointing to the significant correlation between R&D and industrial structures, suggest their notion of NSI as ‘the structural and functional profiles of a nation that determine its innovative capability and economic performance.’ Echoing Van de Ven (1993), Wijnberg (1994, 316) uses a content rich ‘structure-conduct-performance model’ to identify the NSI in terms of the policy framework that affects the evolution of industrial development. Respectively, resource capacities or structure, institutional arrangements or conduct and proprietary functional capabilities or performance may be taken as constituting the NSI. For Wijnberg (1994, 318) the NSI is ‘a set of policies that influence the structural characteristics, the dominant type of innovation in particular industries, the risk-taking propensity of enterprises in these industries and the selection environment in which these industries develop’.

As sophisticated performance of the resource elements within NSI is associated with advanced economic and industrial development (Freeman 2002, Furman, et al. 2002) and vice versa (Hall 2004, Kremp and Mairesse 2004, Comin 2004), it is not surprising that innovation and NSI research are largely concentrated on and in, OECD countries (Mohnen et al. 2006, Hall and Mairesse 2006). A major contribution to the literature has been the ‘RISE’ project (1999). Farina and Preissl (1999) depict a taxonomy with private–public and knowledge generation–knowledge application as principal dimensions. They indicate that NSI resource capacities (R&D oriented) are relatively less important than proprietary functional capabilities. Crucially important for developing countries, they conclude that ‘the success of innovation is increasingly dependent on functions that are not R&D, like organisational re-adjustments, training of personnel, design of prototypes and procedural blue-prints, market research, availability of venture capital, integration of existing equipment and stocks of knowledge with new knowledge inputs’ (Farina and Preissl 1999, Chapter 6, 24).

The NSI literature on developing countries does not depart radically from the mainstream ideas identified above. Arocena and Sutz (2000, 58) point to the ‘four main aspects of the NIS concept’ as ‘an ex-post concept’, one that ‘carries a normative weight’ and is ‘relational’, as well as being path dependent and evolutionary and is therefore ‘a policy subject’. However, the interesting implication of the normative aspect, from the perspective of developing countries, is that ‘there are some general “good ways” and some ways that appear to be “better” than others’ and that it is important ‘to avoid copying or just following the latest policy fashion’ (Arocena and Sutz. 2000, 59). The concept of NSI in developing countries’ industrial development has received deserved attention in terms of case studies (Schiller 2006, Yeh and Chang 2003, Intarakumnerd et al. 2002) and policy (Edquist 2001).

Cassiolato and Lastres (2000) discuss NSI within the international production network of spatially distributed production context of policies to attract FDI and MNEs, which results in lessening emphasis on endogenous innovative activities. They conclude that, as the generation of innovation by MNEs is still dominated by the home country bias (Tsuge and Bartels 2003), simply exposing domestic firms to either FDI or MNEs sub-contracting/supply relationships (Bartels, et al. 2009) is necessary but insufficient. They highlight ‘the importance of continuous public and private policies and efforts aiming at promoting the capacity to acquire and use knowledge and to innovate’ (Cassiolato and Lastres 2000, 51). This is highly applicable to African countries.

Given the nuances in the conceptualisation of NSI, it is not surprising that empirical measurement is meagre in comparison with voluminous articulations of the concept (Becheikh et al. 2006). With this background and the definition of NSI by Bartels et al. (2012), Figure 2 –Triple Helix Type 4 – illustrates our model for mapping and measuring NSI.

Figure 2. Triple Helix Type 4 (an extension of the Triple Helix of Etzkowitz and Leydersdorff 1996).

Methodology

Collecting evidence through surveying

Basically, ‘evidence based policy making refers to the notion that policy intervention and direction are underpinned by an understanding of how things develop’ (Gera et al. 2006, 58) arrived at, in the case of NSI, by mapping and measurement. Essentially there are two forms of data collection, interviews and self-administered questionnaires (Lozar and Vehovar 2008). Data collection processes are complex and require careful consideration of a multitude of parameters as well as issues of statistical reliability and validity. The main issues in data collection methods are: cost, coverage and quality of response.

The accelerated proliferation of the ‘new techno-economic paradigm, centred on information and communication technologies (ICT), have [sic] accelerated and deepened both the codification of knowledge and the spread of information’ (Lastres et al. 2005, 1) and overall quality of data acquisition. The ability to access DISK is a key factor in developing competitiveness and engaging globally.1 However, the ‘digital divide’ is partly due to the minimal availability of ICT access in developing countries as compared to developed countries. Our methodology exploits free open source software (FOSS) which assists in bridging the digital divide. FOSS by its constitution – non-rivalry and non-excludability – acts as a public good. Mueller (2003) notes the positive correlation between the growth of a FOSS developer base and innovative capacities.

Mapping and measuring NSI using FOSS

This section addresses the FOSS application Lime Survey. Figure 3 illustrates the methodology.

Figure 3. Stages in the GNSI methodology. (Adapted from Koria & Köszegi 2011).

In survey design, questionnaire length is an important predictor of response rate (Tomaskovic-Devey 1994, Berdie 1973). Regarding the GNSI survey, variables were constructed from the NSI literature,2 and the DASI was created using a multi-step process: (1) Performing in 2007 a comprehensive NSI literature survey (including all innovation surveys since 2000). (2) From this, 300 NSI variables were extracted, which were reduced to 138 variables3 (Bartels et al. 2009). (3) Using the 138 variables, a survey instrument was designed to measure NSI actor perceptions (Clason and Dormody 1994, Garland 1991) using a five-point Likert scale. Strong empirical evidence supports the treatment of ordinal variables as conforming to interval scales (Labovitz 1967 1970 1971). (4) The DASI was then refined through peer review.4 The resultant version of the DASI is herein referred to as DASI-V1. (5) The DASI-V1 was then reverse translated into French and Spanish for the sake of accuracy and embedded into the web-based electronic survey FOSS Lime Survey. (6) The DASI-V1 was launched in seven emerging market economies (EMEs)5 on the basis of the Survey of Surveys of Innovation6 which had examined innovation surveys in EMEs. A second Survey of Surveys of Innovation in EMEs was conducted in 2012. The updated Survey of Surveys of Innovation shows that of the 128 surveys conducted since 1990 in EMEs,7 none can be strictly defined as a NSI survey, in terms of the same DASI being applied to all four core NSI actors.8 The crucially important characteristic differentiating our methodology from other methodologies, including the Frascati and Oslo manuals, is the fact that our survey applies contemporaneously the same DASI-V1 to the three core actors of the NSI as well as to a fourth actor, arbitrageurs, acknowledged to play a crucial role of intermediation between sources of knowledge and commercialisation of knowledge; and, furthermore, the DASI-V1 is applied using ICT. The GNSI survey obtained valid and reliable responses as shown in Table 1.

Table 1. GNSI Universe and Convenient Sample of Respondents

The methodology is based on the ‘triple helix model’ (Etzkowitz and Leydesdorff 1996), but includes the extra intermediary body, arbitrageurs (Koria & Köszegi 2011). Arbitrageurs, not represented in the traditional model, are of crucial importance as innovation requires internal and external knowledge and arbitrageurs provide links, resources and even technical knowledge so that firms can improve their performance, survival rate as well as accelerate and increase the effectiveness of their innovation processes (Zook 2003, Baygan and Freudenberg 2000, Hargadon 1998). Their resource allocation role is based on advantages in information asymmetries (Williamson 1969 1971 1973).

The next point of discussion is response rate. Low response rates are seen as problematic (Harzing 2007). Response rates differ significantly cross-sectionally across professions, occupational groups and countries. Evidence suggests that response rates by managerial staff are lower than those of non-managerial staff (Baruch 1999). Cycyota and Harrison (2006) identified an overall top manager response rate of 32%. Various strategies exist to increase response rates (Dillman 2000, Zwane et al. 2010) and steps were taken to maximise our response rate. For clarity and ease of response, the direction and strength of measurement scales were carefully considered within the DASI-V1. Therefore, the DASI-V1 for the GNSI deploys a five-point Likert scale with a midpoint, thus reducing the bias towards both extreme answers and towards false negatives (Matell and Jacoby 1972, Garland 1991).

From survey delivery options of mail, telephone, interactive voice response and internet,9 we chose the internet justified by: (1) covering larger sample sizes than conventional mail surveys (Berrens 2003), (2) shorter time (Cobanoglu 2001), (3) superior quality of retrieved data, in terms of non-responses and including discreetly conditionality (Olsen 2009) and (4) higher reliability of data due to elimination of data entry (Bartels and Lederer 2009, Muffo et al. 2003).

Analysis

In analysis the cross-tabulations the five-point Likert scales are dichotomised into the limits of the measurement scale of statistically significant variables as follows:

• Very Important - Irrelevant (VI-I)

• Very Strong – Very Weak (VS-VW)

• Very Positive – Very Negative (VP-VN)

• Very High Innovativeness – Very Low Innovativeness (VHI-VLI)

• Very Highly Successful – Not Successful (VHS-NS)

• Very High Constraint – Very Low Constraint (VHC-VLC).

Neutral was assigned to the negative side on the basis that a neutral perception by an expert respondent is not positive.

The variables analysed, reported and discussed are:

• Actor importance and strength of inter-, intra-actor linkages (cross-tabulation)

• Factor constraints on innovation (factor analysis).

Cross-tabulation permits the examination of statistically significant observations (i.e. the inter-linkages between GNSI actors and NSI variables) using the Chi-square test of significance reported at a confidence level of 95% or above to indicate significant systematic relationships.

Factor analysis reduces observed variables into factors within a pattern matrix (clusters of inter-correlated variables) with ‘mutual interdependence’ (Gaur 1997). The factors represent the underlying structure that influences the variation of variables in the data, sample and hence the population and universe of Respondents (Kim and Mueller 1978, 54, Kaiser 1974, Dziuban and Shirkey 1974, 359, Rummel 1970).

Factor analysis permits ‘the correlations or covariance between a set of observed variables [that] arise from the relationship of these variables to a small number of underlying, unobservable, latent variables, usually known as the common factors’ (Everitt 2002, 140) to be captured. We use exploratory factor analysis, as a priori constraints are not imposed on the data and this enables parsimonious reduction while maintaining the underlying pattern of the variation of variables (Hair et al. 1998). We use the Kaiser criterion (Kaiser 1960) for the number of factors extracted. Furthermore, for robust explanatory power, we require the extracted factors to explain at least 50% of the total cumulative variance. As we do not stipulate the factors to be uncorrelated, we use oblique rotation.

To maintain statistical significance, variables with less than 0.55 coefficient loading (equal to 30.25% of the variance accounted for by the factor) are suppressed. Cut-off criteria for factor loadings remain a matter of debate (Cudeck and O'Dell 1994, Bowles 2006, Hair et al. 1998). Shapiro et al. (2002) indicate several factor coefficient loading cut-offs.10 Heuristics suggest that loadings of>0.30 are salient and cut-off selection between 0.30 and 0.60 is representative in the literature (Bagozzi and Yi 1998, Swisher et al. 2004).

Results and discussion

The results, discussion and policy implications are presented, after which a complement of select policy recommendations is indicated.

Cross-tabulations

To portray actor inter-, intra-linkages, the convention used in our findings is as follows:

• Government inter-linkage with knowledge-based institutions; proactive inter-linkage, i.e. government to knowledge-based institutions is GOV–KBI

• passive inter-linkage (from government perspective) KBI–GOV

• Government intra-linkage GOV–GOV.

We find that GOV respondents have no statistically significant assessment of other actors’ inter-linkages. The MHTI view is striking for the absence of significant relationships KBI–GOV, as is that from KBIs, who do not have a significant assessment of MHTI–GOV (or vice versa) inter-linkages. Finally, ARBs do not have a significant assessment of the inter-linkages KBI–GOV, MHTI– KBI (or vice versa), MHTI–GOV (or vice versa). It should be noted that from an actor-centric view, ARBs indicate only two proactive linkages with respect to GOV and MHTI, notably [GOV]ARB–GOV, 83.4% and MHTI [BE] ARB–BE, 50.0% (see Figure 4 below). While there is no significant perception by MHTI of KBI–GOV relations, the majority of KBIs perceive VI-VS relationships between KBI–GOV and GOV–KBI with the exception of RI–ISTC linkages (only 26.4% KBIs assess this as VI-VS). Clearly, these two views of MHTI and KBIs are asymmetric.

Figure 4. Actor-Centric Assessment of Inter-Linkages (Very Important-Very Strong)

The policy implications of this asymmetry in actor importance and linkage strengths are: (1) insufficient information exchange between MHTI and KBIs with respect to KBI–GOV relations, which raises the policy question of whether there is sufficient dialogue that facilitates information and knowledge exchange; (2) ARBs are isolated from the GNSI and play no significant role in terms of intermediating knowledge transfers through modalities such as intellectual property rights (IPRs) and licensing regarding IPRs emanating either from KBIs, or flowing between KBIs and MHTI; and (3) the ARBs’ intra-linkages, perceived as VI-VS, have very few (if any) significant externalities.

With respect to MHTI perception of the tri-lateral relationship between GOV–ARB–KBI, in which there are more significant relations (ARB–GOV, GOV–ARB, ARB– KBI, KBI–ARB) than GOV–KBI or KBI–GOV. The minority of MHTI perceive VI-VS inter-linkages. There are ten significant linkages in GOV–ARB–KBI (unidirectional) compared to two GOV–KBI (bidirectional).

Again, this asymmetry suggests the following policy implications as absence of: (1) reciprocating relations of communications, co-operation, co-ordination and exchange functions formalised through well-functioning standing committees and conferences between GOV-ARB-KBI; and, (2) operative high-performance councils on Science, Engineering, Technology and Innovation, as well as on economic and social research and on the ‘knowledge brokering’ role of ARBs.

With respect to the actor-centric view regarding VI-VS inter-linkages (see Figure 4) there are no significant relations between ARBs and KBIs, only one ARB–GOV and one ARB–MHTI.

The key policy implication is that the isolation of ARBs infers, at best, a very limited intermediary role either in creating DISK, or as ‘pumps’ for flows of DISK, in the GNSI. At worst, ARBs have no functioning intermediation role. Specifically: (1) the absence of significant linkages ARB–KBI, KBI–ARB means that ARBs do not have access to DISK created by and held within, KBIs. Therefore, ARBs are prevented from acting as conduits to MHTI or investing directly in KBI-hosted spin-offs; (2) the [BE]ARB–BE linkage has less depth to it in the absence of ARBs’ access to DISK from KBIs; and (3) the [GOV]ARB–GOV linkage is devoid of the practicability of ARBs being able to persuade convincingly GOV towards policies that enhance the stocks and flows of knowledge in and throughout, the GNSI (i.e. from KBIs to MHTI directly, or indirectly, via ARBs, e.g. through advocacy and lobbying pressure).

Notwithstanding the overall weakness of inter-linkages among actors in the GNSI, from a triangular perspective the relationship GOV–ARB–KBI, the densest of this relationship is along the axis KBI–GOV. This reflects the traditional role of GOV in funding KBIs.

The policy implications of this public goods provision by GOV, in the context of the isolation of ARBs and their insubstantial intermediating role and overall VI-VW systemic inter-linkages, are: (1) very low returns from the expenditure in treasury, organisational effort and transaction costs,11,12,13 and (2) externalities – the fundamental reason for providing the public goods – are extremely limited, thus reducing the effectiveness and efficiency of the GNSI.

Importantly, the perception of KBI–GOV relations by KBIs is asymmetric to the perception of GOV–KBIs relations. KBIs perceive a bi-directional relationship, whereas GOV perceives no significant relationship (see Figure 4).

Policy implications from the asymmetry between KBIs and GOV regarding their inter-linkages have profound consequences. These policy implications are: (1) the GOV framework of incentives for KBIs (fiscal, monetary, regulatory, standards and performance) is mostly ineffective in that GOV demands little from KBIs in return for providing financial support to KBIs (and students) in Science, Technology, Engineering, Mathematics and Information Technology (STEMIT) programmes; (2) GOV-supported institutions supporting technical change (ISTC) inter-linkages with KBIs are largely ineffective (with respect to [ISTC]RI–ISTC; 61.4% of KBIs respondents indicate VI-VW); (3) the performance required from KBIs by GOV is limited at best and at worst does not encourage KBIs to engage proactively with other GNSI actors; (4) recalibration of STEMIT under-and post-graduate courses to the inter-disciplinary needs of MHTI; (5) reconfiguring the national service programme toward internships in MHTI for STEMIT students; and (6) conditioning financial support (research ‘top up’ grants, etc.) on joint research with MHTI.

Whereas the majority (51.1%–56.3%) of KBIs perceive KBI–GOV (bidirectional) as VI-VS, only a minority (26.4%) of KBIs perceive the crucial transformational RI–ISTC linkage14 as VI-VS (the majority of KBIs (61.4%) assess [ISTC]RI–ISTC as VI-VW) (see Figure 5).

Figure 5. Actor-centric assessment of inter-linkages (Very Important – Very Weak)

The policy implications of this VI-VW inter-linkage between RI–ISTC include: (1) truncated relations with demand and factor markets and with MHTI in the commercialisation of KBI's IPRs; (2) VI-VW KBI passive [HE]BE–HE inter-linkages; (3) the stocks of RI and KBIs IPRs find little or no receptive outlets either in ISTC or MHTI and hence there is little or no flow of intellectual property and knowledge within the GNSI; and, (4) limited GOV performance requirements from RI and ISTC.

MHTI see a significant bi-directional relationship between GOV and KBIs, but none with respect to KBI– GOV. Critically, while, GOV respondents do not mirror this view, KBIs do have a view of KBI–GOV inter-linkages (see Figure 4). This divergence between MHTI and GOV with respect to GOV–KBI and asymmetry between MHTI and KBIs with respect to KBI–GOV inter-linkages is indicative of discordance within the GNSI and, its pre-adolescent stage of evolution in terms of the TH-4 GOV–KBI–MHTI–ARB transactional and transformational linkages.

Policy implications arising from the policy analysis and of the actor-centric mapping in terms of the TH-4 relations include: (1) conspicuous gaps in GOV–MHTI and ARB–KBI (and vice versa) linkages; (2) noting that GOV Respondents have no significant assessment of inter-linkages among other actors in the GNSI, the policy levers available to GOV are, at best, articulated insufficiently well and at worst are far too remote for efficient policy craft and effective policy direction; (3) notably, ARBs mirror MHTI view of a significant bi-directional inter-linkage GOV–KBI in terms of [GOV]GOV–RI and [GOV]HE–GOV; [GOV]GOV–HE, [HE]GOV–HE and [GOV]HE–GOV, [HE]HE–GOV; and (4) the absence of actor-centric ARB–KBI (and vice versa) and GOV–MHTI ( and vice versa) inter-linkages points to limited ability on the part of GOV to enforce innovation policy with respect to KBI–ARB inter-linkages and MHTI.

In summary, the policy implications of the gaps identified in mapping and measuring the GNSI may be grouped into: (1) information asymmetries; (2) co-ordination failures; (3) lack of significant externalities; (4) glacial flows of DISK; (5) an ineffective framework of incentives; (6) lack of connectivity between actors; and, (6) unarticulated policy levers.

Figures 4 and 5 map and measure statistically significant actor-centric assessments of their inter-linkages with other actors (i.e. how one actor views its inter-linkages with another actor) in proactive, that is, for example from the perspective of GOV (GOV–KBI), or passive, that is, for example (KBI–GOV), along the dimension importance of actor and strength of actor-actor inter-linkages measured as VI-VS and VI-VW. The diagrams require viewing in tandem.

The policy implications of the asymmetrical dimension of actor importance and strength of actor-actor inter-linkages include: (1) a GNSI that is seriously deficient overall along the axes GOV–MHTI (and vice versa), GOV–ARB, GOV–KBI, ARB–KBI (and vice versa), MHTI–ARB, in bi-directional terms (proactive and passive inter-linkages); (2) this deficiency is compounded by the isolation of ARBs, absence of MHTI– ARB and KBI's passive MHTI–KBI, inter-linkages; (3) the inter-actor dialogue on innovation and innovation policy is therefore far from complete with respect to GOV–MHTI, GOV–ARB, GOV–KBI, MHTI–ARB (and vice versa) and ARB–KBI (and vice versa) inter-linkages; and (4) the lateral side of the TH-4 (GOV–MHTI) on which innovation policy, industrial policy and industrial innovation should be manifest is entirely missing.

The policy recommendations to address these asymmetries, defects and deficiencies include; (1) initiation of a formal consultative process on innovativeness and innovation in the national economy; (2) harmonising standards, managerial requirements and governance across KBIs; (3) eliminating constraints preventing public-sector institutions from engaging in STEMIT activities with the private sector; and (4) adopting common performance agreements (linked to funding) that have external relationship indicators across KBIs.15

Latent factors to barriers to innovation – ALL respondents

The variables which constrain innovation and innovativeness in Ghana were measured on a Likert scale16 and factor analysed. The results are presented in Table 2: Latent factors to barriers to innovation (ALL) below.

Table 2. Factors underlying barriers to innovation

In Table 2, total variance explained (TVE) indicates the amount of variance (variation) accounted for by the factor. It signals the extent of the influence of the factor. The Kaiser-Meyer-Olkin (KMO) is a measure of sampling adequacy and indicates the robustness and reliability of factors. The Bartlett's test of sphericity (BTS) designates the significant confidence level regarding the coherence of factors, reproducibility and generalisability of the results17 (Kaiser 1974, Dziuban and Shirkey 1974, 359, Kim and Mueller 1978, 54, Rummel 1970). Also the dominant heuristic for describing internal consistency and reliability using Cronbach's alpha, is indicated in Table 3 (George and Mallery 2003, Kline 1998, Cortina 1993).

Table 3. Internal consistency of factor

Four factors are responsible for 58.7% of the total variance of the variables constituting barriers to innovation in the GNSI. The reliabilities between 0.757 and 0.833 are deemed ‘acceptable’ and ‘good’. The coherence of the factors is indicated by the ‘meritorious’ KMO of 0.817 and the highly significant BTS. From a dimensional perspective, the four factors suggest that the barriers to innovation and innovativeness in the GNSI are: (1) organisational capital; (2) market transactional dysfunctions; (3) fiscal shortfalls; and (4) organisational risks.

In Table 2, Factor 1 <Skills–ICT Capability/Capacity> is the highest, most significant barrier to innovation in the GNSI, in which ‘Quality of Technically Trained Manpower’ is the most crucial variable. Factor 1 accounts for 63.5% (0.797²×100) of the variation in this variable.

Factors 2 <Unsophisticated Markets>, 3 <Deficient Fiscal Policy> and 4 <Reduced Organisational Risks> are also significant barriers to innovation; however, individually each explains less than a third of the TVE of Factor 1. These factors account for 79.6%, 59.3% and 64.5% of the variation in the variables ‘Lack of Demanding Customers’, ‘Lack of Finance’ and ‘Excessive Perceived Economic Risk’, respectively.

Factor 1 affirms our TH-4 model and underlines the critical importance of talent and the diffuseness of ICT for enhancing the stocks and flows of DISK and skills. These forms of intangible or organisational capital (Corrado et al. 2006, Kim 2007, Sullivan and Sullivan 2000) are crucial in cohering the NSI (Niosi 2002).

Factor 2, consistent with findings in the literature (Bartels et al. 2012, Delgado et al. 2012, Porter 1990), demonstrates the consensus on the importance of markets for driving innovation through demanding customers, innovative customers and competition.

The key policy implication is that without threshold levels in skills-ICT capability/capacity economy-wide innovativeness and innovation is extremely difficult to attain.18 Specifically, policy implications include: (1) in resource constrained circumstances, the crucial choice is for fiscal and monetary incentives, as well as regulation, standards and performance requirements, to be directed to improve the most significant Factor 1 <Skills-ICT Capability/Capacity>, through improving the ‘Quality of Technically Trained Manpower’ and the ‘Rate of Access to ICT’; and (2) in terms of the sequencing of policy implementation19; F1 <Skills-ICT Capability/Capacity> is relatively short term (1-3 years) given the capacity aspect of ICT; F2 <Unsophisticated Markets> and F3 <Deficient Fiscal Policy> are medium-term (3-5 years) given the legislative aspect of fiscal policy; F4 <Reduced Organisational Risks> is long term (5-10 years) given organisational culture and time taken to change institutional behavior.20

Conclusions

The results of our TH-4 approach to mapping and measuring the GNSI show evidently that firstly, in terms of actor importance and strength of inter-linkages, the GNSI is asymmetrically oriented with a bias to KBI–GOV and GOV–MHTI. Secondly, significant relationships GOV–KBI, GOV–MHTI and MHTI–GOV, ARB–KBI and KBI–ARB, MHTI–ARB, GOV–ARB are missing. Thirdly, the factors that underlie the barriers to innovation and innovativeness are identified as absence of skills, markets, finances and presence of risks.

In mapping and measuring empirically the GNSI, we have found significant gaps, asymmetries, low densities and weaknesses in the distribution of intra- and inter- linkages in the multifaceted relationships among and between the core actors. We have also identified significantly high barriers to innovation. These deficiencies point to a GNSI that is far from a well-integrated TH-4 model. The policy implications of these deficiencies have serious consequences for the rate at which the country may develop economically despite recent indications of confidence reflected in increased investment flows (UNCTAD 2013). The findings may be articulated in the following dimensions: (1) high information asymmetries – which signify that risks cannot be adequately priced with confidence that encourages innovativeness by entrepreneurs; (2) extensive co-ordination failures – which mean misapplication of precious resources as well as truncated public goods provision; (3) serious lack of significant externalities – that point to isolation of innovative activities; (4) glacial flows of DISK – which denotes that investments in human capital are largely unrealised; (5) an ineffective framework of incentives – which results in the inability of government to shape the innovation landscape to its purposes; (6) lack of connectivity between actors – which means that barriers to innovation are likely to be persistent; and (6) unarticulated policy levers – which suggest that the government strategic intentions are unlikely to be achieved.

NSI and the triple helix approach are increasingly the salient ingredient of economic development policy (Natario et al. 2012, Chang and Lin 2012). Our mapping and measurement of the NSI using the TH-4 framework and applying the DASI to all four actors simultaneously provides a comprehensive view of the GNSI. Our methodology is not only parsimonious in transaction costs but also results in an enhanced rate of survey responses. The use of FOSS to deploy the DASI-V1 is an innovative methodology that is advantageous in resource constrained developing economies, attempting to generate coherent evidence-based policies on science, technology and innovation.21

These results and recommendations enable policy makers to target specificities and to use the DASI-V1 to monitor policy implementation and effectiveness. More importantly they enable policy makers to decide effectively on the inevitable trade-offs involved in policy craft for economic development on the basis of robust evidence.

Notes

1 See the Communications Infrastructure measurement in the Competitiveness Framework Structure of the determinants of national competitiveness in Delgado et al. (2012, Appendix Table A1, 45).

2 Under the direction of Frank L. Bartels in the UNIDO Statistical Research and Regional Analysis Unit.

3 Thus achieving a high level of internal and construct validity.

4 In 2007 the questionnaire was sent to Prof. J. Howells at the Centre for Research on Innovation and Competition (CRIC), UK and Prof. S. Mani at the Centre for Development Studies, India, for peer review, additional suggestions and inputs.

5 As classified by Institute of International Finance (IIF), Egypt, Morocco, Chile, Peru, Malaysia, Thailand and the Ukraine – these countries were chosen as either no survey had been conducted or not for a long time.

6 The Survey of Surveys of Innovation was conducted by Ms. Simone Carneiro, UNIDO Consultant in 2007.

7 A breakdown of the 128 surveys conducted indicates: 60 Emerging Europe, 34 Latin America 19 Asia and 15 Africa/ Middle East.

8 Etzkowitz, H. (2003), Research groups as ‘quasi-firms’: the invention of the entrepreneurial university. Research Policy 32: 109–121, Leydesdorff, L. (2005). The triple helix model and the study of knowledge-based innovation systems. International Journal of Contemporary Sociology 42(1), Shinn, T. (2002), The triple helix and new production of knowledge: pre-packaged thinking on science and technology. Social Studies of Science 32(4): 599–614, Leydesdorff, L. and Meyer, M. (2006), Triple helix indicators of knowledge-based innovation systems: introduction to the special issue. Research Policy 35(10): 1441–1449.

9 Given the specific targeting of respondents in the four communities respectively, the survey participation invitation email, sent en-masse to all respondents, contains a link leading to the electronic questionnaire. To enhance data reliability and validity, the link was equipped with an authorisation token to restrict the access of people who had not received a token and ensure that each respondent was only able to answer the questionnaire once.

10 These range from≥0.400 to≥0.600.

11 % GDP spent on R&D – South Africa 0.9, Ghana 0.2, Kenya 0.4, Tanzania 0.4, Botswana 0.5. Sources: The World Bank (2012), World Development Indicators. Research and development expenditure % of GDP, 2005–2007, Washington, DC: The World Bank.

12 The expenditure on KBIs by government with respect to research and development is erratic while allocations to CSIR-STIPR are utilised 81% for staff costs and 9% for research (UNCTAD, 2011).

13 According to UNCTAD (2011) the low expenditure on R&D is hardly compensated for by either the private sector efforts which amount to about 2% of all funding for R&D or arbitrageurs. These figures represent the importance given to Science, Technology, Engineering, Mathematics and Information Technology (STEMIT) in national priorities; and the serious challenges facing the implementation of the technology and innovation policies indicated in the Ghana Industrial Policy.

14 That is, the transformation function of RI and ISTC in respect to innovation in contrast to the transaction, or exchange, function via market mechanisms and price signalling (Dunning 2003).

15 For a comprehensive view of recommendations see UNIDO (2012), Evidence-based policy making: the Ghana national system of innovation – measurement, analysis & policy recommendations, UNIDO, Vienna 2012, authored by Frank L. Bartels and Ritin Koria.

16 1 – Very High Constraint, 2 – High Constraint, 3 – Neutral, 4 – Low Constraint and 5 – Very Low Constraint.

17 Arbitrageurs did not produce a factor result as the number of respondents, while entirely adequate for cross-tabulation, did not meet the statistical requirements necessary to conduct a factor analysis.

18 With respect to the global competitiveness index (GCI), Ghana's ranking across a range of indicators relevant to innovation is: GCI 2010–2011 (out of 139) = 114; GCI 2009–2010 (out of 133) = 114; GCI 2008–2009 (out of 134) = 102. Innovation and sophistication factors = 100 (Business sophistication = 97; Innovation = 99); Basic requirements = 122 (Institutions = 67; Infrastructure = 106; Macroeconomic environment = 136; Health and primary education = 122); Efficiency enhancers = 96 (Higher education and training = 108; Goods market efficiency = 75; Labour market efficiency = 93; Financial market development = 60; Technological readiness = 117; Market size = 83). Source: World Economic Forum, 2010. The Global Competitiveness Report 2010–2011.

19 Notwithstanding the electoral cycle, or the time taken for legislative and regulatory processes to place policy on statute via parliamentary fiscal and monetary decisions (white paper, green paper, committee stage, bill and law). It is fully recognised firstly that such temporal characteristics are subject economically to the consequences (time delay, dislocation, discontinuities) of: (1) exogenous shocks; (2) market failures; and (3) government failures. Secondly, policy business plans and managerial actions are expected to be of a ‘rolling’ nature in order to attain, through incremental advances, as well as accelerated spurts, higher levels of innovativeness and innovation throughout the economy in the long-term.

20 While Ghana has improved its performance in the World Bank ‘Doing Business’ variables since 2004 much more needs to be done. Scrutiny of Ghana's performance in the ‘Doing Business’ surveys shows relative decline in the face of absolute improvements from 2004–2012 in: starting a business; dealing with construction permits; getting electricity; registering property; getting credit; paying taxes; trading across borders (but costs have risen); enforcing contracts; resolving insolvency. No improvement in protecting investors. However, despite these improvements, Ghana's rank position has slipped between 2011 and 2012 in all the above categories except: getting electricity; enforcing contracts and resolving insolvency. With reference to a key variable in innovation, starting a business, Ghana slipped 19 rank positions between 2011 and 2012. See IFC (International Finance Corporation) and World Bank, 2012. Doing Business 2012: Doing Business in a More Transparent World. Economy Profile: Ghana. Washington DC: The World Bank Group.

21 See final technical report, national innovation systems of BRICS countries, IDRC Center File 10422-1001, September 2007 to October 2010.