TERRITORIAL IMPACT ASSESSMENT: COHESION POLICY AND BALANCED TERRITORIAL DEVELOPMENT (CZECHIA)

The intent of this paper is to add to the current knowledge in the field of TIA modelling by presenting a case study of cohesion policy (CP) in Czechia, 2007-2013. The empirical results are mixed. While the territorial impact of CP interventions concerning the NSRF objective of a 'Competitive Czech Economy' is higher in main metropolitan regions, CP interventions concerning the NSRF objectives of an 'Open, Flexible and Cohesive Society' and of an 'Attractive Environment' have higher impacts in regions with more desire for CP interventions. Consequently, territorial impacts of the three NSRF objectives are contrary to one another, and the observed pattern of overall territorial impacts of CP interventions is patchy, almost mosaic-like. Additionally, the paper suggests some methodological ideas for TIA modelling drawing inspirations from the prominent TEQUILA model. In particular, the spatial distribution of SF is used to model the intensity of CP interventions in a territory. A methodology how to model the potential territorial impact and the desirability of CP interventions in a territory is also presented.


INTRODUCTION
Since its first mention in the Amsterdam Treaty, territorial cohesion has become increasingly significant in EU policy agenda (see, e.g., Davoudi, 2005;Cotella, Adams and Nunes, 2012). Servillo (2010) stressed the importance of linking territorial cohesion to economic development, while Davoudi (2005), Elissalde and Santamaria (2014), Colomb and Santinha (2014), and Faludi (2005) point out the relationship between territorial and social cohesion in the 'European Social Model', noting that nobody should be disadvantaged by their place of residence. However, despite its increasing significance, the concept of territorial cohesion is still somewhat vague because there is a variety of definitions in literature concerning this topic (see, e.g., Camagni, 2017;Servillo, 2010;Nosek, 2017;Golobič and Marat, 2011;Camagni, 2009).
Historically, the concept of territorial cohesion relates to the EU objective of preventing large territorial disparities (see, e.g., Camagni, 2009;Nosek, 2017;Servillo, 2010). With this line of reasoning, it is expected that policy instruments aim to support lagging regions in order to reduce territorial disparities and to achieve balanced territorial development and 'territorial justice' (see, e.g., Malý and Mulíček, 2016;Colomb and Santinha, 2014;Colomb and Santinha, 2014). As Colomb and Santinha (2014), Camagni (2009) note, special attention is given to some types of territories, e.g., rural areas, declining urban areas, remote territories and others. However, several additional aspects regarding the concept of territorial cohesion deserve special mention:  Firstly, the concept of territorial cohesion emphasises that public services -or services of general interest -should be provided within reasonable distances of all people. Hence, also people living in peripheral regions are expected to have access to these types of services (see, e.g., Colomb and Santinha, 2014;Cotella, Adams and Nunes, 2012;Servillo, 2010;Faludi, 2005;Colomb and Santinha, 2014).
 Secondly, all territories are considered to have developmental potential, which should be identified, strengthened and exploited. Hence, endogenous development and place-based development are associated with the concept of territorial cohesion (see, e.g., Bentley and Pugalis, 2014;Colomb and Santinha, 2014;Abrahams, 2014). It is worth noting that also territorial competitiveness closely relates to this aspect of territorial cohesion.
Different aspects of the concept of territorial cohesion are interlinked, and thus, there is a potential conflict between them. In this regard, the main discussion focusses on the tensions between territorial competitiveness aims and 'territorial justice' aspirations (see, e.g., Luukkonen and Moilanen, 2012;De Propris, 2007;Servillo, 2010). It was these tensions that broadened the concept of territorial cohesion to also include the emphasis on polycentric territorial development and territorial cooperation (see, e.g., Zaucha, Komornicki, Böhme, Świątek and Żuber, 2014;Davoudi, 2005;Medeiros, 2012). As Malý and Mulíček (2016) claim, polycentric territorial development recognizes the potential for metropolitan areas to generate economic competitiveness. Territorial cooperation and functional links improve 'territorial justice' through developing strong metropolitan areas in peripheral regions (see, e.g., Medeiros, 2012). Hence, polycentric development may be perceived as a bridging concept concerning cohesion and competitiveness (see, e.g., Malý and Mulíček, 2016;Veneri and Burgalassi, 2012).
 The associations between policy inputs and impacts, the intensity of policy interventions in a territory; and the desirability of policy interventions in a territory are all considered in calculating territorial impacts.
 As indicated by TIA models, other factors may also be taken into account such as territorial vulnerability, substitution effects, sustainability of impacts, and territorial closeness of effects.
The intent of this paper is to add to the current knowledge in the field of TIA modelling by Thirdly, the desirability of policy interventions in a territory is operationalized and measured using composite indicators that relate to the three NSRF objectives. Fourthly, results are discussed in light of balanced territorial development, which is one of the tenets of territorial cohesion, and these characteristics are used as a guiding framework. Hence, a TIA model based on robust empirical grounds is suggested. In this context, it is noteworthy that a number of authors such as Medeiros (2015), Golobič and Marat (2011) argue that TIA modelling has a lack of 'hard quantitative data' and an over-reliance on subjective-based judgements.
This paper is structured as follows: the second section provides the objectives and methods. The third section presents results, which are then discussed in the following section.
The last section provides a conclusion.

OBJECTIVES AND METHODS
The main objective of this paper is to assess territorial impacts of CP interventions using TIA modelling and to discuss results relating to balanced territorial development. The methodology is based on the theoretical framework presented in the introduction. The starting point is the equation inspired by the prominent TEQUILA model (see, e.g., Camagni, 2009): where TIM r shows territorial impacts of CP interventions on a region r; PIM r,o is potential territorial impact of CP interventions that relate to an objective o on a region r; and D r,o is the desirability of CP interventions that relate to an objective o in a region r. TIM r , therefore, aggregates territorial impacts for the three NSRF objectives. All calculations are based on 206 Czech regions between level LAU1 and LAU2.
The potential territorial impact PIM r,o is calculated as a product of two components: (1) the general impact of CP interventions on the three NSRF objectives; and (2) the intensity of CP interventions (refer to e.g., Camagni, 2009 for this approach). For this purpose, two matrices are used. The first matrix contains priority axes of thematic and regional operational programmes under the Convergence and Regional Competitiveness and Employment objectives (hereafter referred to as priority axes) in rows, and contains the three NSRF objectives in columns. The general impact of each priority axis on each NSRF objective is determined using the intervention logic described in the NSRF and also by using expert judgements. The impact is rated on a four-point scale (from 0 to 3) ranging from 'no impact -0' to a 'very strong impact -3' (see, e.g., Camagni, 2009;Medeiros, 2014 for the use of scales), and using a three-step procedure as follows.
Firstly, the impact of each priority axis on each NSRF objective is evaluated according to three criteria: (1) the first criterion relates to the question whether the impact of a priority axis on a NSRF objective is explicitly mentioned in the NSRF content; (2) the second criterion relates to the question whether priority axis indicators are of relevance to a NSRF objective; (3) the third criterion relates to the question whether the link between priority axis indicators and a NSRF objective may be regarded as a strong link, considering ex-ante expected outcome values. The number of 'yes' responses determines the impact of each priority axis on each NSRF objective on a four-point scale. Secondly, five experts independently explore the impacts of each priority axis on each NSRF objective as these were rated in the first step of the procedure. On this basis, suggestions for change are gathered and these are discussed in the third step of the procedure and eventually made.  The number of patent applications and utility models per 100.000 inhabitants ( Concerning the objective of a 'Competitive Czech Economy' ( fig. 1), an uneven spatial distribution of TIM r values is demonstrated. However, spatial hierarchy appears to play a role in this distribution, indicating that the main metropolitan areas have higher TIM r values.
Prague is a notable exception to this rule due to its ineligibility to receive funds under the generous Convergence objective. However, Prague's low value here is at least partially compensated by high values for regions in close proximity to Prague. Additionally, there is a tendency in the eastern regions to have higher TIM r values than those in the western regions.   Altogether, a complex picture arises because the CP interventions relating to the three NSRF objectives work contrary to one another. This can also be seen in the lack of statistical significance in all the pair-wise comparisons for the three NSRF objectives together (see tab. 2; the last column).

DISCUSSION
The empirical results presented in the preceding section can be embedded in a broader theoretical context. Firstly, a number of studies have emphasized the influence of spatial factors on regional inequalities in post-communist countries (see, e.g., Ezcurra, Pascual and Rapún, 2007;Maier and Franke, 2015;Czyz and Hauke, 2011;Krzysztofik, Tkocz, Spórna , & Kantor-Pietraga, 2016;Martinát et al., 2016;Ženka, Novotný, Slach and Květoň, 2015;Marková and Švihlíková, 2016;Skokanová, Havlíček, Klusáček, & Martinát, 2017;Navratil et al., 2018). Three factors are usually expected to be significant in this respect: (a) spatial hierarchy and the advantages of location in the main metropolitan regions; (b) the easternwestern gradient and the advantages of location close to the borders of western countries; and (c) the inherited spatial specialization and the structural disadvantages of particularly old industrial regions. The importance of these factors was also demonstrated in the TIA models constructed, i.e., the importance of spatial hierarchy for the objective of a 'Competitive Czech Economy', the importance of inherited spatial specialization for the objective of an 'Open, Flexible and Cohesive Society', and the importance of the eastern-western gradient for all the NSRF objectives. Generally, the influence of the three spatial factors needs to be considered in planning territorial impacts for CP interventions.
Secondly, the empirical findings are relevant for the debate about the relationship between two spatial objectives -territorial competitiveness and territorial balanced development (see, e.g., Vanolo, 2010;Colomb and Santinha, 2014). The constructed TIA models suggest that CP interventions work in either direction, depending on their thematic orientation (see, e.g., Klímová and Žítek, 2015;Hájek and Górska-Szymczak, 2017;Kaufmann and Wagner, 2005; Severová, Chromý, Sekerka and Soukup, 2012 for relatively low impact innovation-oriented interventions in lagging regions). Therefore, the combined effects of CP interventions can undermine their overall contribution to balanced territorial development (see also Novosák, Hájek, Horváth and Nekolová, 2017 for this conclusion). The weight given to particular types of CP interventions and their links to the two spatial objectives are crucial for evaluating which of the objectives prevail.
Thirdly, the TIA models constructed extend the methodology of TIA modelling in some directions. This is the primary way of treating the potential territorial impact of CP interventions (PIM r ), which are operationalized using 'hard data' relating to the spatial distribution of CP interventions. Moreover, NSRF is taken as the main source for the gauging general impacts of CP interventions on the three NSRF objectives, and also the weights of objectives are set in a way that differs from previous studies and it relies more on 'hard data'.
Generally, the constructed TIA models are less subjective in nature, thereby we can at least partially remove one of the drawbacks of TIA methodologies (see, e.g., Golobič and Marat, 2011;Medeiros, 2015 for the problem of subjectivity in TIA modelling).

CONCLUSION
The intent of this paper is to add to the current knowledge in the field of TIA modelling by presenting a case study of cohesion policy in Czechia (2007Czechia ( -2013. The findings point out the need to address the complex nature of territorial impact of CP interventions. The territorial impact of CP interventions relating to the NSRF objective of a 'Competitive Czech Economy' is greater in the main metropolitan regions, but CP interventions relating to the NSRF objectives of an 'Open, Flexible and Cohesive Society' and of an 'Attractive Environment' have greater territorial impacts in regions with higher desirability for interventions.
Consequently, there is a mosaic pattern of overall territorial impacts of CP interventions, with different conclusions regarding their contributions to balanced territorial development, and the hypothesis that CP interventions contribute to balanced territorial development cannot be conclusively accepted.
There are several political implications that can be drawn from this research. Firstly, the overall assessment of territorial impacts of CP interventions masks the complexities that arise from their thematic decomposition. Therefore, it is desirable to deal precisely with the thematic dimension of both CP interventions and TIA modelling. Secondly, it is important to define the relationship between the two spatial objectives of CP interventions: (a) territorial competitiveness; and (b) balanced territorial development. A particular question exists about whether the desirability of CP interventions relating to the competitiveness objective is greater in the main metropolitan regions or whether it is greater in lagging regions. Thirdly, implementing the ideas introduced in this paper can bring fruitful results in 'less subjective' TIA modelling.
In our opinion, the results of this study can be helpful in order to provide methodological guidance for practitioners. In particular, the methodology is useful for both ex-ante policy analyses and ex-post policy analyses of territorial impacts. While the former analyses provide information about the most suitable course of action, the latter analyses indicate whether the actual choice was the most suitable. However, there are some limitations of using the methodology and two of them are worth mentioning. Firstly, the matrix of general impacts of CP interventions on the NSFR objectives can be improved by calibrating the impacts against achieved outcome indicator values. Secondly, an outflow of SF to other regions (e.g., through public procurements) ought to be considered in order to enhance our understanding of the phenomenon.