1 Introduction

Many regional policy instruments integrate elements operating on the principles of triple helix, especially: networking, industrial clusters, cluster initiatives, learning regions, innovation systems at the national and regional level and others. These systemic tools often incorporate other designated instruments. Thus, supporting their formation and their effective use should be able to produce a significant positive synergistic effect.

According to many studies related to innovation systems (for an overview of these studies see Tödtling and Trippl 2005), regions (defined as smaller than a national region and larger than a local unit) are considered to be the key to innovation systems working effectively for the following reasons (Cooke et al. 2000):

  • First: regions differ according to their industrial specialization and their innovation performance (Howells 1999; Breschi 2000; Paci and Usai 2000).

  • Second: knowledge spillover effects play the key role in the innovation process and are usually geographically bounded (Audretsch and Feldman 1996; Bottazzi and Peri 2003; Asheim and Coenen 2005; Stejskal and Hajek 2015).

  • Third: the growing importance of “tacit” knowledge has been indicated (Polanyi 1966; Howells 2002; Gertler 2003; Matatkova and Stejskal 2013) for a successful innovation process. The latter is often influenced by interventions due to political representation or by public administration institutions. However, interventions due to political representation are more often seen at the regional or local level.

The regions are the most suitable area (space) for innovation. Next, it is necessary to define the framework and instruments that enhance the innovation process (Cooke et al. 1997; Morgan 2007; Sternberg and Arndt 2001; Antonioli et al. 2014). The original paradigm for national innovation systems was thereby temporarilyFootnote 1 refuted and attention was transferred to the concept of the regional innovation system (RIS), which was introduced in the 1990s.

There are many scholars who analyzed the regional innovation systems and of course define it (for the overview see Cooke 2006). Majority of them is in line with Cooke’s definition (Cooke 2006):

RIS are useful for studying economic and innovative performance; they are also functional tools to enhance the innovation processes of firms. They do this by knitting together knowledge flows and the systems on which they rely, building trust and confidence in institutional reliability; and above all, they do it by generating institutional self-knowledge and a certain kind of collective dissatisfaction with the status quo. RIS comprise a set of institutions, both public and private, which produce pervasive and systemic effects that encourage firms in the region to adopt common norms, expectations, values, attitudes and practices, where a culture of innovation is nurtured and knowledge-transfer processes are enhanced (Matatkova and Stejskal 2011b).

Asheim and Coenen (2005; in Stejskal and Matatkova 2011b) divide the RIS this way:

  • territorially embedded regional innovation systems,

  • regionally networked innovation system,

  • regionalized national innovation system.

Territorially embedded regional innovation systems are similar to grassroots RIS by Cooke (2006), the best examples of this type are networks of small and medium enterprises (SMEs) in industrial districts. These systems provide bottom-up, network-based support through, for example, technology centers, innovation networks, or centers for real service providing market research etc. (Storper and Scott 1995 in Asheim and Coenen 2005).

Regionally networked innovation system means that firms and organizations are also embedded in a specific region and characterized by localized, interactive learning. This type is very similar to network RIS by Cooke. We can say that a networked innovation system is a result of policy intervention to increase innovation capacity and collaboration.

Regionalized national innovation system is different from the two systems above in two main points. First, parts of industry and the institutional infrastructure are more functionally integrated into national or international innovation systems. Second, the collaboration between organizations within this type of RIS conforms more closely to the linear model, as the co-operation primarily involves specific projects to develop more radical innovations-based on formal analytical-scientific knowledge. Cooke named this type of RIS system dirigiste RIS. The concrete example of this system could be technopoles or science parks. For more information see Asheim and Coenen (2005b).

Braczyk et al. (1998), Asheim and Coenen (2005), Cooke (2006) divide the RIS according to the size of the region’s incorporated companies, their financing methods or the territorial limits of the regional innovation system. It is also possible to divide regional innovation systems according to the degree of their infrastructure development within the region:

  • RIS with hard elements but without any soft infrastructure elements,

  • RIS with highly developed hard and highly undeveloped soft infrastructure,

  • RIS with highly developed hard and partially developed soft infrastructure,

  • RIS with highly developed hard and highly developed soft infrastructure,

  • RIS with a developed network for knowledge diffusion.

Many other authors tried to create own divisions of RISes. There are two scholars many times mentioned in references (Braczyk et al. 1998; Asheim and Coenen 2005b). The first division is according to Braczyk (in Cooke 2006). He says that there are three types of RIS emerged (Matatkova and Stejskal 2011a):

  • localist,

  • interactive,

  • globalized.

The localist type has few major public innovation or R&D resources, but may have smaller private ones. There will be high degree of associativeness among entrepreneurs and between them and local or regional policymakers.

The mix of public and private research institutes and laboratories in the interactive RIS is balanced, reflecting the presence of larger firms with regional headquarters and a regional government keen to promote the innovation base of the economy.

The innovation system in globalized RIS is dominated by global corporations, often supported by clustered supply chains of rather dependent small and medium-sized enterprises (SMEs). The research reach is largely internal and private in nature rather than public, although a more public innovation structure aimed at helping SMEs may have developed.

The second division is provided by Cooke (2004 in Cooke 2005) and it is based on the government dimension. There are three forms of RIS again:

  • grassroots,

  • network,

  • dirigiste.

Grassroots is where the innovation system is generated and organized locally, at town or district level. Financial support and research competences are diffused locally, with a very low amount of supra-local or national coordination. Local development agencies and local institutional actors play a predominant role.

A network RIS is more likely to occur when the institutional support encompasses local, regional, federal and supranational levels, and funding is often guided by agreements among banks, government agencies and firms. The research competence is likely to be mixed, with both pure and applied, blue-skies and near-market activities geared to the needs of large and small firms.

A dirigiste system is animated mainly from outside and above the region itself. Innovation often occurs as a product of central government policies. Funding is centrally determined, with decentralized units located in the region and with research competences often linked to the needs of larger, state-owned firms in or beyond the region.

2 Characteristics of RIS

There are many definitions of the RIS. Cooke (2002) describes the RIS as the wide infrastructure that helps in the innovation creation processes realized in interactions among many entities. Hudec (2007) states that RIS (from systematic point of view) is defined as the system that stimulates the innovation abilities of firms in a region and aims at the economic and social development and the level of the competitiveness.

Stejskal and Matatkova (2011b) offer that we should try to imagine RIS as a framework which includes, according to Cooke (2002), two sub-systems:

  • the knowledge application and exploitation sub-system,

  • the knowledge generation and diffusion sub-system.

The first is principally concerned with firms while the second is mainly concerned with public organizations like universities, research institutes, technology transfer agencies, and regional and local governance bodies responsible for innovation support practices and policies. In reality, there may be some overlaps since firms conduct knowledge creation activities, especially where they have formalized R&D laboratories, and universities and public or private research institutes conduct knowledge application activities.

Cooke et al. (2000), Cooke and Memedovic (2003) in Tödtling and Trippl (2005) add to above mentioned subsystems another one. The third dimension is the regional policy because policy actors at this level can play a powerful role in shaping regional innovation processes, provided that that there is sufficient regional autonomy to formulate and implement innovation policies. Tödtling and Trippl (2005) further add that in the ideal case, there are intensive interactive relationships within and between these subsystems facilitating and continuous flow or exchange of knowledge, resources and human capital. On the other hand, there are several types of RIS problems and failures such as deficits with respect to organizations and institutions and lack of relations within and between subsystems (Matatkova and Stejskal 2011b).

Therefore, the RISs encompass (as already showed above) the institutions from both the private and public sector. These institutions we can call “basic components” of every RIS. Due to these necessary parts of the network we can determine whether there is some RIS in selected regions. The RIS existence and the evaluation (level of development) was discussed by many economists, i. e. Cooke et al. (1997), Cooke (2001), Doloreux (2002), Andersson and Karlsson (2004), Doloreux and Parto (2005). On the basis of their work we can define the basic components of the RIS, which we can summarize into three fundamental groups: (a) the core of the RIS, (b) auxiliary and complementary organizations and (c) infrastructure, institutions and technical support.

According to the above mentioned, the regional innovation system is composed of three fundamental layers:

  1. (a)

    entrepreneurs,

  2. (b)

    supporting organizations,

  3. (c)

    environment and infrastructure.

In layer (a) companies, businesses and firms that are localized in the region are included. They should be focused on the creating of innovation, i.e. those who produce the market innovations, produce the patents, or spend public and private funds for research, development and subsequent development of innovations. In the layer (b) supporting organizations layer we include those organizations which helps and support the firms included in the first layer and provide complementary support services to them. The supporting organizations are primarily providers of knowledge, cooperating organizations for subcontracting, institutions for collaboration (they are the central part of industrial clusters and manage the cluster activities; Stejskal and Hajek 2012).

The layer (c) “environment and infrastructure” consists of three sub-layers (separate sub-system):

  1. (a)

    Institutions making up the innovation environment (or ecosystem)

    • Institutions forming the legal framework for business, preparing the strategic documents that support innovative business activities, innovation absorption, creativity, and development of innovation in firms;

    • Facilitators providing facilitation of the entities in RIS. These organizations are established to support the industrial clusters or business networks births,

    • Institutions and organizations that make up the convention, customs and usage in the ethics in business. They are often higher education providers (universities), often also entrepreneurial esprit chambers. These organizations support the social capital.

  2. (b)

    Incentives and initiatives

    • Public incentives to innovation creation and development or infrastructure suitable for innovations financially,

    • Private incentives that have decided to financially support the ideas of firms that do not have sufficient investments or capital. we can include venture capital or business angels in this group.

  3. (c)

    Hard and soft infrastructure

    • Fixed infrastructure (industrial zones, technological parks, scientific research parks, innovation and high-tech centers, etc.),

    • The infrastructure necessary for high-technology use (technological centers, testing and research centers or other scientific research centers and laboratories),

    • Knowledge infrastructure (high schools, universities, and other knowledge organizations that allow horizontal or vertical transfer of knowledge between knowledge producer and firms recipients).

In all the layers we can find private organizations (firms), followed by public institutions (mostly regional governments or their representatives—regional development agencies) and other supporting public (often private or NGO) agencies, which are necessary components of a favorable innovation environment. Collaborating ties among the entities in the RIS are often referred to as triple (sometimes quadruple) helix (Leydesdoff and Etzkowitz 1996).

Every RIS should have, for example industrial clusters, the specialization (be focused on productions of something special). All authors cited above regard the RIS as a general system that is fixed into the socioeconomic environment of the region and integrated in the system that involves entities from the various sectors. We cannot completely agree with the general view of the RIS. We believe that the RIS should focus on some range of industries and this focus should be reflected by regional (public) policy, which is one of the RIS’s subsystems. It will increase the efficiency of public policy and also the efficiency of financing because it cannot be assumed that the rule “all-does not fit-to all” will always be applicable.

The important components of each RIS are special activities resulting from geographical proximity, trust and willingness to cooperate. We cannot miss also the communication links between subjects of the RIS. These components determine the efficiency and quality of results arising from RIS existence in region.

3 Methods for RIS Analysis

There is no one shot method to be used universally for analysis and evaluation of the regional innovation system. Numerous authors have employed various methodologies when it comes to regional innovation system assessment. This piece of writing will take a critical review of some of the various methods that have been used to analyze the regional innovation system.

3.1 Participatory Evaluation

This method for assessing the regional innovation system is quite new and has not been widely accepted if we assess how credible it is (Diez and Esteban 2000). This method actively calls for allowing actors that are involved in the regional innovation system the chance to share their views and ideas when it comes to knowing how the regional dynamics of knowledge flow and innovation. The Participatory evaluation method is seen as an inner approach that does not rely on external factors or actors. This method is built on the premise that, regions are composed of numerous actors and stakeholders who are constantly interacting in the so when we want to get a clear understanding of how the system is working we need to involve all the active participants during the evaluation process. The active participation of the entities will ensure that outcomes achieved by the evaluation will be effective because it helps the regional actors in the process to perform the current evaluation and therefore come out with their results that can change the assessment into new ways of doing things.

The evaluation process is an important component of the learning process and this allows us to get a clear understanding based on the perspective of all the participants. It is precisely the very participants in the policy of economic development who contribute to understanding and learning about the processes of change underlying the program and to the development of a new awareness regarding the policy under evaluation (Diez 2001). Evaluation ceases to be an exercise of assessment where the predominant perspective comes from only one angle, that of the objectives of the policy designer as the only criteria for evaluation, and becomes an exercise stimulating the appearance of a learning process (Kuhlmann 1998).

We can summarize that the knowledge creation and transfer takes place inside and outside of the region (there is a so-called regional migration of knowledge). This knowledge “movement” helps to motivate the public organizations (regional governments, NGOs, agencies) to support these knowledge-based activities (described for example Finne et al. 1995; Diez 2001). This is the example of so-called participative development (if the funds are used and shared, we can called it participatory budgeting). The spill-over effects are learning during the co-operation and practice, and at the same time there is a significant cultivation of public policy that re-emphasizes the importance of knowledge as a production factor.

3.2 Interdisciplinary Methodology/Network Analysis

The interdisciplinary methodology has been described as the “appropriate tool” that can be used to evaluate network capital in the regional innovation system (Krätke 2002). Social network analysis is the mapping and measuring of relationships and flows between people, groups, organizations, computers or other information/knowledge processing entities (Krebs 2002). Social network or network analysis centers on the arrangement of relationships among actors and assess how resources are exchange among the various actors (Scott 1991; Wasserman and Faust 1994). The RIS is composed of numerous interactions among the various social entities and this result in the creation of network capital. So to evaluate the how this social network thrives, the interdisciplinary methodology can be used. Social Network Analysis has therefore proven to be useful because it enables the visualization of how people are connected, thereby enabling users of this methodology to find out how best people and institutions interact to share knowledge in the RIS. This methodology is built on the belief that social network are very important for the collaborating entities (Wassermann and Faust 1994) and society as a whole because of the end product that leads to transformation of the entities and society as a whole.

This analytical tool can be used to identify the vital properties of the RIS (Wassermann and Faust 1994; Jansen 1999). For a better and comprehensive understanding of networks and the participants involved, one needs to evaluate where the network is taking place (its location) and composition of actors that make up the network. These procedures provide us with a better understanding into the various roles and categories in a network—who constitute the connectors, where are the clusters and their makeup, who forms the center of the network, and who is on the periphery. This methodology can be relied upon in RIS when we endeavor to assess the rate at which knowledge and information flow across functional and institutional borders as in triple helix. It can also be useful when we want to find out who knows who (social relationships) and who might know what (expertise) in groups where individuals play key roles. One advantage of using this methodology in RIS is that, it provides it helps us to understand and simplifies the complex nature of interorganizational networks. It allows for comparative analysis by first of all mapping the already established network and its properties.

This methodology is able to generate data about network by using surveys. Since the network consists of industries and institutions, surveys will be able to determine the networked relationship by questioning the various actors involved. If the network structure is known, then an evaluation of its properties can follow to establish the extent of how they are interconnected and what role does the various actors play in the network can also be known. Haythornthwaite (1996) used the network analysis to study how information is exchanges in social networks and concluded that, the network analysis helped to create awareness of already established information exchange paths, and that information sources can act on information opportunities and alter information directions to improve the delivery of information services.

The overview of the case studies is presented in Table 1.

Table 1 Overview of interdisciplinary methodology/network analysis studies
Full size table

3.3 Cluster Analysis

Over the past two decades cluster analysis technique has been usage has increased (Everitt 1979; Gower 1967). Cluster Analysis also known as taxonomy analysis or segmentation analysis based on the techniques ability to produce classification (Everitt 1979). “Cluster analysis groups data objects based only on information found in the data that describes the objects and their relationships. The goal is that the objects within a group be similar (or related) to one another and different from (or unrelated to) the objects in other groups. The greater the similarity (or homogeneity) within a group and the greater the difference between groups the better or more distinct the clustering” (Nowak et al. 2008). According to (Romesburg 2004), cluster analysis refers to combinations of mathematical models that can be utilized to group objects that are similar into the same group. All objects have their attributes which might not be the same, but when has many objects, there is bound to be different attributes, so these can be arrange to for a cluster. Cluster analysis is the best and widely used research method when it is necessary to examine the similarity of the objects.

In the RIS, clusters analysis strongly focuses on the all the linkages and interactions that exist among various actors and people that results in the efficient creation of innovation, new products and services (Roelandt and Den Hertog 1999). The cluster in reference here is not assumed to be the same as happens in other forms of interaction they are very similar and linked in the value chain. Clusters can either be horizontal or vertical (cross-sectorial) network that consist of industries that are not the same but complementary firms that have a specific specialization that can result in the creation of innovation (Morgan 1997). The cluster analysis approach differs from other conventional research approaches because it takes into account collaborations and knowledge flow within the network (Rouvinen and Ylä-Antilla 1999). Comparatively, the conventional research approaches have focuses on networks that have homogenous firms producing same products, but the cluster have proven to be a reliable alternative because, it offers a different view in the RIS in the sense that, it places premium on the interaction-based theories of innovation which many authors now called “triple helix” (see Leydesdorff 2012; Vaivode 2015). This dynamic nature of the cluster analysis has made it a reliable alternative to the other traditional research approaches (Roelandt and Den Hertog 1999). Another reason that has made cluster analysis so important is its focus on vertical relationship and interdependence of actors who may not necessarily be similar firms or institutions (Roelandt and Den Hertog 1999).

Many studies have used cluster analysis methodology (Punj and Stewart 1983; Ketchen and Shook 1996; Feser and Luger 2003; Beuther and Sutherland 2007). The cluster analysis was used by Fesser and Bergman (2000) to study 23 national industry cluster template and the results proved that template clusters are useful to discover gaps and knowledge about extended product chains and therefore represents a useful first step in the detailed examinations of local cluster patterns. Arthur (1994) also used the cluster analysis to study the effects of Human resource system on manufacturing performance and turn over and concluded that “human resource system moderated the relationship between turnover and manufacturing performance”.

The overview of the case studies is presented in Table 2.

Table 2 Overview of cluster analyses
Full size table

3.4 Data Envelopment Analysis

Data envelopment analysis or DEA for short has increasingly become a famous management tool since the method first came into practice (Charnes et al. 1978). Many studies have been done in relation to DEA (see Banker et al. 1984; Dyson and Thanassaoulis 1988; Seiford and Thrall 1990; Anderson and Peterson 1993; Banker 1993). According to Boussofiane et al. (1991), “DEA is a linear programming based techniques used for measuring the relative performance of organizational units where the presence of multiple inputs and outputs makes comparison difficult.” The mathematical component of the DEA make it a useful tool that can be used to control and assess past activities and also useful for future planning. They have proved to be very vital for “ex post” evaluation of efficiency in management circles (Banker et al. 1984).

The DEA can also be employed to assess the performance of activities carried out by organization using output and input data (Lertworasirikul et al. 2003). In the knowledge based economies, universities produce knowledge using inputs in the form of labour (tutors), computers etc. to create output (knowledge). When one is given output and input data, it becomes easy to establish how the organization will perform using the DEA technique. They have become “powerful tools” that is used to measure efficiency and have since then been used to evaluate the efficiency of educational and research institutions in terms of their knowledge production functions (Lertworasirikul et al. 2003). The DEA is in the sense that it helps to characterize efficiency and inefficiency of decision making units (Zhu 2001).

To measure organization efficiency has been a source of worry for many years because there was no clear cut formula that provided the solution (Farrell 1957). As a mathematical model, it is not faced with deficiencies, (Andersen and Petersen 1993) have concluded that the DEA methodology has been very successful in determining the relative efficiency in decision making units but the method does not allow us to rank how efficient these units are. In addition Kao and Liu (2000) have also described the use of DEA to measure efficiency as very difficult because of its (DEA) use of complex economic and behavioral entities. This becomes more difficult when multiple outputs and inputs need to be aggregated in isolation to determine efficiency.

In a study to evaluate the comparative efficiency of ten Chinese third-party logistics providers 3PLs, Zhou et al. (2008) used the DEA approach and concluded that there was a decline in efficiency of Chinese 3PLs and this coincided with a steep decline in transportation activities as a result of the outbreak of the deadly SARS virus. The study also found out that technical expertise and sales opportunities directly correlate with operational efficiency of 3PLs at the same time, there was no direct correlation between the size of 3PLs and their performance. Abbott and Doucouliagos (2003) also used the DEA model to evaluate the efficiency of Australian universities. Their result proved that irrespective of the blend of input and outputs, Australian universities recorded high levels of efficiency relatively when compared one by one. In a study to measure the performance of 500 manufacturing firms in Turkey Düzakın and Düzakın (2007) used the DEA methodology and came out with the conclusion that during 2003 nine firms efficiently performed in Turkey, and out of these nine firms ranked among themselves. Furthermore, each of the firms in the analysis was ranked within each industry, and the results were that 65 firms were efficient among the industries.

The overview of the case studies is presented in Table 3.

Table 3 Overview of inputs and outputs in data envelopment analyses
Full size table

3.5 Case Studies

The case study methodology can also be used to evaluate the regional innovation system. The case study approach has been defined by many scholars (see below). Robson (2002) defines the case study as “a strategy for doing research which involves an empirical investigation of a particular contemporary phenomenon within its real life context using multiple sources of evidence”. The case study as an approach can be adopted for a study based on the research questions and the objectives the researcher wants to achieve. The case studies are pertinent when the research being undertaken addresses either a descriptive question or an explanatory question (Shavelson and Towne 2002). The case study therefore seeks to provide a rich description and detailed explanation of the reason behind a complex phenomenon, and why they have happened or remained as they are.

The case study is a more appropriate methodology for evaluating the RIS because it provides more detailed information comparatively to the other methods. This information gathered from individual cases can be compared to find out why the differences exist. It also allows researchers to collect data from multiple methods such as surveys, interviews, and observations among others that can be validated through triangulation. The required data for the case study are likely to come from diverse and not a singular source of evidence (Denscombe 2003; Yin 2003).

Case study research assumes that scholars need to study the conditions and factors what appear in similar case studies to understand them more closely. The major limitation of case study approach is that it does not allow for generalization since findings are unique to the particular case as against the other cases. It however provides in-depth information and enough bases for improvement in the case under study.

Huggins et al. (2011) used the case study in their study on small firm-University Knowledge Networks using evidence from the United Kingdom and the US. They used this methodology to study 16 Small and Medium Scale enterprises (SMEs) from the UK and US (8 SMEs in the UK, and 8 SMEs in the US). They used the firm level case study to compare these firms and generated data from semi-structured interviews with Chief Executive Officers of these companies. Their study found out that, the bulk of firms were <10 years old, but their global customer base indicated that they were innovative firms as they have started exporting their products contributing to the regional economies supporting the empirical evidence that innovative firms are very important in economic development (Siegel et al. 2003).

The overview of the case studies is presented in Table 4.

Table 4 Overview of case studies
Full size table

3.6 Regression Models

Regression analysis is a quantitative research technique used research or studies that involve modeling and examining several variables, where the relationship consists of a dependent variable and independent variables (Mosteller and Tukey 1977). The regression analysis is mainly used to get a detailed understand of the relationship that exist between a dependent variable and an independent variables (Ai and Norton 2003). Regression analysis allows researchers to identification and classification of relationships among multiple components (Schneider et al. 2010). This technique has become a key to economic statistics and it’s mainly used to achieve several objectives like predicting, forecasting, and finding the effect of one causal variable on another (Sykes 1993).

Regression analysis is preferred among statisticians because it allows users to make assumptions and it easily solves problems that are very complicated of because this method is very flexible (Oliver 2014). There are many types of regression techniques. The basic ones include linear regression, nonlinear regression, and the least squares method. According to Schneider et al. (2010), the linear regression is used to evaluate the linear relationship between a dependent variable and other independent variables.

3.7 Comparative Studies

Many authors believe that the RIS are specific entities that should be analyzed and evaluated individually. The findings should be compared with similar (and also foreign) regions. The researchers seek for the similarities (hits) or differences, and the analysis of the causes and consequences. The overview of the most important studies that dealt with RIS is given in the Table 5 below.

Table 5 Overview of comparative studies
Full size table

The overview of the case studies is presented in Table 5.

3.8 Qualitative Content Analysis

Qualitative content analysis has been defined as “a research method for the subjective interpretation of the content of text data through the systematic classification process of coding and identifying Themis or patterns” (Hsieh and Shannon 2005). Zhang et al. (2005) claim that “these three definitions illustrate that qualitative content analysis emphasizes an integrated view of speech/texts and their specific contexts. Qualitative content analysis goes beyond merely counting words or extracting objective content from texts to examine meanings, themes and patterns that may be manifest or latent in a particular text. It allows researchers to understand social reality in a subjective but scientific manner.” There are some international studies what used the qualitative content analysis.

The overview of the case studies is presented in Table 6.

Table 6 Overview of qualitative content analyses
Full size table

The practice shows that RIS analysis is not a simple process. Many studies have not been mentioned at all in this part of the publication, because they were too focused on specifics of individual regions and often cannot be generalized as the widely applicable methodology. Many of these studies tried to apply a combination of qualitative and quantitative approaches.

4 Application of the WSA Method for Regional Innovation Systems in Selected Regions of the Czech Republic

Methodological approach published in Nekolova, K., Rouag, A., & Stejskal, J. (2015). The Use of the Weighted Sum Method to Determine the Level of Development in Regional Innovation Systems – Using Czech Regions as Examples. Ekonomický časopis, 63(03), 239–258.

Regional innovation systems are suitable and often used tool of regional policy also in the Czech Republic. The importance of these systems is even more emphasized after joining the EU. The significant decentralization of the regional policy was realized after 2004 and the emergence of RISs is good example of this trend (the same trend was noted in Western countries in past). The regional innovation strategies were created in all Czech regions (NUTS 3), i.e. documents in which the strategy how to create and promote RISs are contained. However, the emergence of regional strategies was left in the hands of the regional governments. This caused that the quality of strategies in different regions is different. It determines that the application in the coming years is not always good and efficient. The suitable conditions for the RIS emergence are created in all Czech regions; in some regions created RIS latently (clear evidences of RIS existence are missing).

In 2016, the national Czech government decided to create a central regional innovation strategy (RIS3) and in all regions there the regional innovation strategies were initiated. These new versions of regional RIS3 strategies are based on the national RIS3 strategy. The regional characteristics and specifics are taken into account by close cooperation (the national coordinators of RIS3 strategy collaborated with regional representatives). The RIS3 has to be the key conditionality for approving the operational programs and boosting the investments to the research, development, innovation and ICT (financed from EU Structural funds in programming period 2014–2020). After past experiences, we afraid that the strategies will lead to investment, but without noticeable positive effect (the goals of RIS3). Therefore, we need to develop methods that help to analyze the quality of the RIS, to support and to assess the regional innovation system development and level.

4.1 WSA Method Characteristics

The weighted sum method (WSM) is based on the principle of utility maximization (Fiala et al. 1997). This method has been simplified by using only a linear utility function. Calculations are then manageable without the use of specialized software. First, we created a normalized criteria matrix R = (rij) whose elements are obtained from the criteria matrix Y = (yij) using the transformation rule, (1):

$$ {r}_{ij}=\frac{y_{ij}-{D}_j}{H_j-{D}_j},r\in 0;1,\forall i=1,\dots, pj=1,\dots, k $$
(1)

where r ij is the normalized value for the i-th alternative and j-th criterion, D j is the basal value, the lowest possible value an alternative acquires in the j-th criterion, H j is the ideal value, the best possible value an alternative acquires in the j-th criterion.

Obviously, r ij = 0 for the basal alternative, and r ij = 1 for the ideal alternative (Chyna et al. 2012). When using the additive form of multi-criteria utility functions, the utility of the option ai is then expressed by (2):

$$ u\left({a}_i\right)=\sum \limits_{j=1}^k{v}_j{r}_{ij},\forall i=1,\dots, p $$
(2)

where v j is the corresponding element from the weight vector, r ij is the normalized value gained from (1).

Obviously, the alternative with the highest utility value is considered as a compromise. In addition, the WSM makes it possible to arrange all the alternatives with respect to their utility values (Chyna et al. 2012).

The option that reaches the maximum utility value is selected as being the best, or the results can allow the variants to be classified according to their decreasing utility values.

As seen in Eq. (2), the vector of criteria weights must be determined for calculating utility. In the context of this analysis, we use the Fuller’s triangle method. The determination of weights is based on a pairwise comparison between criteria (Subrt et al. 2011). Because of the pairwise comparison, the number of comparisons is equal to:

$$ N=\left(\begin{array}{c}k\\ {}2\end{array}\right)=\frac{k\left(k-1\right)}{2} $$
(3)

Each comparison may be performed inside Fuller’s triangle. Criteria are numbered as serial numbers 1, 2,…, k. Users then work with the triangular diagram; the double lines formed by serial numbers are arranged in pairs so that each pair of criteria appears exactly once. The user indicates (by encirclement) which criterion is more important for comparing each pair. We mark the number of encirclements of i-th criterion as n i. The weight of the i-th criterion is then calculated as:

$$ {v}_i=\frac{n_i}{N};i=1,2,\dots, k $$
(4)

The main advantage of this method is the simplicity of the information required from users. If it is necessary to exclude zero weight, the number of encirclements may be increased by one with the condition that the denominator in Eq. (4) must also be increased accordingly.

4.2 The Definition of RIS Characteristics

Using study findings and detailed results coming out of references (e.g. Cooke et al. 1997; Andersson and Karlsson 2004; Doloreux and Parto 2005; Hudec 2007; Skokan 2010), Table 7 defines set characteristics for a “standard” form for the RIS.

Table 7 Regional innovation system characteristics
Full size table

If the set of characteristics cited above exists within one region, the authors agree that we can say that a regional innovation system exists in its basic form. At the same time, none of the authors mention the degree of development, precisely because the degree to which a characteristic has been achieved will vary from one RIS to another. Therefore, the degree to which they have been achieved increases the likelihood of positive effects being created when an RIS exists in a given region. For example, these effects can be observed via an increase in regional GDP or a decrease in the unemployment rate.

However, many of these effects bring positive measurable results over the long term, which precludes the causal analysis of economic indicator changes. Consequently, it is not relevant to analyze the effects of the RIS directly.

The RIS characteristics that have been defined (see Table 7) represent criteria which will be quantified and then used to constitute the members of the criteria matrix used when applying the WSM. The quantification of the criteria must be done on the basis of descriptive analysis and information obtained from expert assessments or controlled interviews with experts on regional issues.

Particular characteristics were grouped on the basis of results derived from research findings on RIS layers. The characteristics cited above also contain those of the triple helix (these concern enterprises, support organizations, knowledge and public organizations as well as the environment and investment infrastructure). Relationships and links are two of the most important characteristics and should not be overlooked.

For the purposes of this analysis, the characteristics mentioned above are divided into three groups (see Table 8). The first two groups describe characteristics that are necessary and supportive in the region (physical infrastructure including industrial zones, technological parks, scientific research parks, innovation centers, etc.) and institutions. The existence of these characteristics does not reflect whether the RIS is working or not. They only describe the physical substance of the RIS and can be used as a binary variable (whether present or not) or to quantify the number of institutions. The third group consists of characteristics that have a quantitative nature or contain characteristics whose quality significantly depends on the scope and quality of the individual RIS (typically, the number of patents). On the basis of their analysis, we can conclude that an existing RIS leads to cooperation, knowledge spillovers and a synergic effect and, thus, the creation of innovation. This type of RIS will have a positive impact as a result of the public interventions that have been created and supported.

Table 8 The weight assigned to each criterion based on the Fuller’s triangle calculation
Full size table

It is logical that each characteristic will not have the same meaning for RIS existence and operation. We need to assign a weight to each characteristic inside each group; this weight provides information about the significance of each characteristic. The Fuller’s triangle method was used to assign weights. Preference ranking was done by ten experts.

The expert evaluation of preferences makes it possible to determine the criteria weights and their appropriate grouping according to Eq. (4). The resulting weights are summarized in Table 8.

The sum of the weights assigned to groups I–III equals one, just as the sum of the weights within each group is also equal to one.

Next, the WSM was applied for determining the weight of each characteristic. The method’s application will be divided into three progressive steps corresponding to the division of criteria from the three groups cited above. All the steps of the analysis process will correspond to the WSM as explained above.

For the case study (realized in 2015) we chose six regionsFootnote 2 of the Czech Republic (NUTS 3 level):

  • Kralovehradecky (KHK),

  • Pardubicky (PK),

  • Jihomoravsky (JMK),

  • Moravskoslezsky (MSK),

  • Liberecky (LK),

  • Stredocesky (STC).

4.3 The Evaluation of Necessary RIS Quantitative Characteristics

Criteria included in the group of quantitative characteristics are listed in Tables 7 and 8. Descriptive analysis was provided by an expert appraisal from the creator of the Czech Republic’s RIS in April 2015. The results are summarized in Table 9.

Table 9 Necessary quantitative characteristics
Full size table

When establishing a criteria matrix, it is necessary to give a point value to each indicator. Scoring was used for the sequence of the regions according to the assessment of each criterion. The poorest result was recorded as zero and the best as three. After point evaluation maximizing all criteria, it is possible to establish an initial criteria matrix where rows and columns correspond to Table 9:

$$ \left[\begin{array}{ccccccccc}1& 2& 1& 1& 1& 3& 2& 0& 3\\ {}2& 1& 0& 0& 2& 3& 2& 0& 3\\ {}3& 3& 2& 3& 3& 3& 3& 3& 3\\ {}0& 2& 3& 2& 1& 3& 3& 3& 3\\ {}3& 2& 0& 3& 1& 3& 2& 2& 3\\ {}1& 3& 2& 3& 1& 3& 2& 1& 3\end{array}\right] $$

Criteria in this matrix are maximized; we can therefore determine the maximum value H and the minimum value D from each column j: H = (3; 3; 3; 3; 3; 3; 3; 3; 3); D = (0; 1; 0; 0; 1; 3; 2; 0; 3).

Using Eq. (1), the initial criteria matrix is transformed into a normalized criteria matrix. Elements of this matrix express the indicator value of each variant according to certain criteria:

$$ \left[\begin{array}{ccccccccc}0.33& 0.5& 0.33& 0.33& 0& 0& 0& 0& 0\\ {}0.67& 0& 0& 0& 0.5& 0& 0& 0& 0\\ {}1& 1& 0.67& 1& 1& 0& 1& 1& 0\\ {}0& 0.5& 1& 0.67& 0& 0& 1& 1& 0\\ {}1& 0.5& 0& 1& 0& 0& 0& 0.67& 0\\ {}0.33& 1& 0.67& 1& 0& 0& 0& 0.33& 0\end{array}\right] $$

The normalized criteria matrix makes it possible to calculate the indicator value cited in Table 9 in each region on the basis of Eq. (2). It is important for that calculation to determine the weighting vector v1; its compilation is based on values presented in Table 8: v 1 = (0.222; 0.167; 0.028; 0.042; 0.042; 0.181; 0.083; 0.152; 0.083). The following results are those for the RIS development level in the selected regions according to indicator value calculations. These results are presented in Table 12.

4.4 The Evaluation of RIS Supporting Quantitative Characteristics

This group of characteristics was also analyzed using an expert appraisal and focused on their level of development in the selected regions. The completed results are summarized in the Table 10.

Table 10 Supporting quantitative characteristics
Full size table

Once again, each criterion was evaluated using points and by following the same method used for the necessary quantitative characteristics. The results consist of a criteria matrix whose rows and columns correspond to Table 10:

$$ \left[\begin{array}{cccc}1& 3& 2& 3\\ {}0& 3& 1& 3\\ {}2& 3& 3& 3\\ {}3& 3& 2& 3\\ {}0& 3& 2& 3\\ {}2& 3& 2& 3\end{array}\right] $$

Because the criteria matrix is maximized, we can specify the maximum and the minimum values H and D for each column j: H = (3; 3; 3; 3); D = (0; 3; 1; 3).

The following is the normalized criteria matrix formed on the basis of the transformation formula, (1):

$$ \left[\begin{array}{cccc}0.33& 0& 0.5& 0\\ {}0& 0& 0& 0\\ {}0.67& 0& 1& 0\\ {}1& 0& 0.5& 0\\ {}0& 0& 0.5& 0\\ {}0.67& 0& 0.5& 0\end{array}\right] $$

The calculation of the effects’ values for regions resulting from Table 10 is computed according to Eq. (2) using the normalized criteria matrix. The value of each effect is then calculated according to weighting vector v 2. Values are compiled using Table 8: v 2 = (0.499; 0.167; 0.167; 0.167). The calculation of the effect values gives the results summarized in Table 12.

Quantitative characteristics are concerned only with innovation infrastructure. On their basis, we can decide whether organizations that contribute and diffuse knowledge in each region exist and to what extent they exist; they make it possible to evaluate each region’s innovation potential. Therefore, evaluating the use of this potential is made possible by the analysis of the third group of characteristics—the group of qualitative characteristics.

4.5 Evaluating the Effect of the Existing Qualitative Characteristics

The results of the experts’ appraisal for the cited criteria’s existence, their degree of evolution, all is summarized in Table 11.

Table 11 Qualitative characteristics
Full size table

The criteria were also point evaluated using the same methods. The result consists of a criteria matrix whose rows and columns correspond to Table 11:

$$ \left[\begin{array}{ccccc}1& 3& 3& 2& 2\\ {}0& 2& 3& 2& 1\\ {}3& 3& 3& 3& 3\\ {}2& 3& 3& 2& 3\\ {}1& 3& 3& 2& 2\\ {}1& 3& 3& 2& 2\end{array}\right] $$

Because the criteria matrix has been maximized, we can specify the maximum H and the minimum value D for each column j: H = (3; 3; 3; 3; 3); D = (0; 2; 3; 2; 1). Next follows the normalized criteria matrix formed on the basis of the transformation formula, (1):

$$ \left[\begin{array}{ccccc}0.33& 1& 0& 0& 0.5\\ {}0& 0& 0& 0& 0\\ {}1& 1& 0& 1& 1\\ {}0.67& 1& 0& 1& 1\\ {}0.33& 1& 0& 0& 0.5\\ {}0.33& 1& 0& 0& 0.5\end{array}\right] $$

The calculation of the effects’ values in the regions resulting from Table 11 is computed according to Eq. (2) using the normalized criteria matrix. The value of each effect is calculated according to weighting vector v 3, and values are compiled using Table 9: v 3 = (0.3; 0.133; 0.3; 0.067; 0.2). The calculation of the effects’ values gives the results summarized in Table 12.

Table 12 Effect values within each group
Full size table

4.6 The Assessment of RIS Level for the Selected Regions

The previous sections have also assessed the effects resulting from existing RIS characteristics. This step consists of the overall quantification of RIS effects. This part analyzes the key instruments that have been assigned to each group of the regional innovation system characteristics described in Table 8. The vector of their weight is v 4, and its value is the following: v 4 = (0.333; 0.167; 0.5).

The value of indicators within the selected regions obtained for each group of characteristics is summarized in Table 12.

The overall values of the effects resulting from the existing RIS in the selected regions are calculated using the weighted sum of each effect. The values are listed in Table 13.

Table 13 Overall indicator values for RIS development level
Full size table

4.7 Conclusions

The level of RIS development was determined by the level to which the defined characteristics had been developed. The level of RIS development was depicted by determining values using the WSM and by the descriptive analysis summarized in Table 13.

The use of the WSM is simple in terms of calculating and obtaining specific values. On the other hand, the use of this method has some drawbacks in that it does not show the effects resulting from each characteristic. It only gives the accumulated value for the effects of each indicator. Furthermore, using such a method requires the weighting vector to be expressed numerically. The results derived from the use of the WSM can be authenticated by the use of another multi-criteria evaluation of the alternative. This method consists of the analytic hierarchy process (AHP) for validating results and is appropriate because it works on the same principle as the WSM, and its results are easy to compare. The use of the AHP method provides more detailed values than the WSM. On the other hand, the application of the AHP makes it easier to evaluate the degree of RIS advancement.

There are some limitations for generalizability of the results. The disadvantage of this approach is the lack of any discussion or international comparison of results (the comparable results on a wide platform are lacking).The results should be verified by another method. The adjustment of weights and subjectivity of criteria evaluation are the weakness of this method. The removal of these weaknesses can be subject to further research in this area.